diff --git a/README.md b/README.md index 1721cefa5..c4dfc8be1 100644 --- a/README.md +++ b/README.md @@ -37,7 +37,7 @@ ComfyUI is the AI creation engine for visual professionals who demand control over every model, every parameter, and every output. Its powerful and modular node graph interface empowers creatives to generate images, videos, 3D models, audio, and more... - ComfyUI natively supports the latest open-source state of the art models. -- API nodes provide access to the best closed source models such as Nano Banana, Seedance, Hunyuan3D, etc. +- [Partner nodes](https://docs.comfy.org/tutorials/partner-nodes/overview#partner-nodes) provide access to the best closed source models such as Nano Banana, Seedance, Hunyuan3D, etc. - It is available on Windows, Linux, and macOS, locally with our [desktop application](https://www.comfy.org/download), our [portable install](#installing) or on our [cloud](https://www.comfy.org/cloud). - The most sophisticated workflows can be exposed through a simple UI thanks to App Mode. - It integrates seamlessly into production pipelines with our API endpoints. diff --git a/app/assets/api/routes.py b/app/assets/api/routes.py index e25b8a57f..d43485861 100644 --- a/app/assets/api/routes.py +++ b/app/assets/api/routes.py @@ -18,7 +18,7 @@ from app.assets.api.schemas_in import ( AssetValidationError, UploadError, ) -from app.assets.helpers import validate_blake3_hash +from app.assets.helpers import normalize_tags, validate_blake3_hash from app.assets.api.upload import ( delete_temp_file_if_exists, parse_multipart_upload, @@ -117,6 +117,87 @@ def _build_validation_error_response(code: str, ve: ValidationError) -> web.Resp return _build_error_response(400, code, "Validation failed.", {"errors": errors}) +class InvalidTagFilterError(Exception): + """Invalid combination of tag-filter query parameters.""" + + def __init__(self, message: str, details: dict): + super().__init__(message) + self.details = details + + +# Caps the per-tag EXISTS fan-out; deliberately covers the legacy spellings too. +MAX_TAG_FILTER_TAGS = 100 + + +def _resolve_tag_filters( + q: schemas_in.ListAssetsQuery | schemas_in.TagsRefineQuery, +) -> tuple[list[str], list[str], list[str]]: + """Resolve legacy (include/exclude) and new (all/any/none) tag-filter + spellings into effective (all, any, none) lists. + + Combination validation applies only when the request uses at least one + new-name parameter (non-empty after normalisation); requests using only + the legacy names keep their historical behaviour, including degenerate + combinations like include_tags=a&exclude_tags=a. + """ + # model_dump, not attribute access: deprecated fields warn on every attribute read. + legacy = q.model_dump(include={"include_tags", "exclude_tags"}) + include_tags = normalize_tags(legacy["include_tags"]) + exclude_tags = normalize_tags(legacy["exclude_tags"]) + tags_all = normalize_tags(q.tags_all) + tags_any = normalize_tags(q.tags_any) + tags_none = normalize_tags(q.tags_none) + + for param_name, values in ( + ("include_tags", include_tags), + ("exclude_tags", exclude_tags), + ("tags_all", tags_all), + ("tags_any", tags_any), + ("tags_none", tags_none), + ): + if len(values) > MAX_TAG_FILTER_TAGS: + raise InvalidTagFilterError( + f"'{param_name}' lists {len(values)} tags; the maximum is " + f"{MAX_TAG_FILTER_TAGS}.", + { + "parameter": param_name, + "count": len(values), + "max": MAX_TAG_FILTER_TAGS, + }, + ) + + if not (tags_all or tags_any or tags_none): + return include_tags, [], exclude_tags + + if include_tags and tags_all: + raise InvalidTagFilterError( + "Cannot combine 'include_tags' with 'tags_all'; use 'tags_all'.", + {"parameters": ["include_tags", "tags_all"]}, + ) + if exclude_tags and tags_none: + raise InvalidTagFilterError( + "Cannot combine 'exclude_tags' with 'tags_none'; use 'tags_none'.", + {"parameters": ["exclude_tags", "tags_none"]}, + ) + + all_param, all_list = ( + ("tags_all", tags_all) if tags_all else ("include_tags", include_tags) + ) + none_param, none_list = ( + ("tags_none", tags_none) if tags_none else ("exclude_tags", exclude_tags) + ) + + conflicting = sorted(set(all_list) & set(none_list)) + if conflicting: + raise InvalidTagFilterError( + f"Query can never match: {', '.join(repr(t) for t in conflicting)} " + f"required by '{all_param}' but rejected by '{none_param}'.", + {"conflicting_tags": conflicting, "parameters": [all_param, none_param]}, + ) + + return all_list, tags_any, none_list + + def _validate_sort_field(requested: str | None) -> str: if not requested: return "created_at" @@ -217,6 +298,11 @@ async def list_assets_route(request: web.Request) -> web.Response: except ValidationError as ve: return _build_validation_error_response("INVALID_QUERY", ve) + try: + tags_all, tags_any, tags_none = _resolve_tag_filters(q) + except InvalidTagFilterError as e: + return _build_error_response(400, "INVALID_TAG_FILTER", str(e), e.details) + sort = _validate_sort_field(q.sort) order_candidate = (q.order or "desc").lower() order = order_candidate if order_candidate in {"asc", "desc"} else "desc" @@ -224,8 +310,9 @@ async def list_assets_route(request: web.Request) -> web.Response: try: result = list_assets_page( owner_id=USER_MANAGER.get_request_user_id(request), - include_tags=q.include_tags, - exclude_tags=q.exclude_tags, + include_tags=tags_all, + exclude_tags=tags_none, + any_tags=tags_any, name_contains=q.name_contains, metadata_filter=q.metadata_filter, limit=q.limit, @@ -715,10 +802,16 @@ async def get_tags_refine(request: web.Request) -> web.Response: except ValidationError as ve: return _build_validation_error_response("INVALID_QUERY", ve) + try: + tags_all, tags_any, tags_none = _resolve_tag_filters(q) + except InvalidTagFilterError as e: + return _build_error_response(400, "INVALID_TAG_FILTER", str(e), e.details) + tag_counts = list_tag_histogram( owner_id=USER_MANAGER.get_request_user_id(request), - include_tags=q.include_tags, - exclude_tags=q.exclude_tags, + include_tags=tags_all, + exclude_tags=tags_none, + any_tags=tags_any, name_contains=q.name_contains, metadata_filter=q.metadata_filter, limit=q.limit, diff --git a/app/assets/api/schemas_in.py b/app/assets/api/schemas_in.py index 38a942b7b..862700a24 100644 --- a/app/assets/api/schemas_in.py +++ b/app/assets/api/schemas_in.py @@ -50,8 +50,12 @@ class ParsedUpload: class ListAssetsQuery(BaseModel): - include_tags: list[str] = Field(default_factory=list) - exclude_tags: list[str] = Field(default_factory=list) + # Deprecated spellings: include_tags ≡ tags_all, exclude_tags ≡ tags_none. + include_tags: list[str] = Field(default_factory=list, deprecated=True) + exclude_tags: list[str] = Field(default_factory=list, deprecated=True) + tags_all: list[str] = Field(default_factory=list) + tags_any: list[str] = Field(default_factory=list) + tags_none: list[str] = Field(default_factory=list) name_contains: str | None = None # Accept either a JSON string (query param) or a dict @@ -70,7 +74,10 @@ class ListAssetsQuery(BaseModel): ) order: Literal["asc", "desc"] = "desc" - @field_validator("include_tags", "exclude_tags", mode="before") + @field_validator( + "include_tags", "exclude_tags", "tags_all", "tags_any", "tags_none", + mode="before", + ) @classmethod def _split_csv_tags(cls, v): # Accept "a,b,c" or ["a","b"] (we are liberal in what we accept) @@ -154,13 +161,20 @@ class CreateFromHashBody(BaseModel): class TagsRefineQuery(BaseModel): - include_tags: list[str] = Field(default_factory=list) - exclude_tags: list[str] = Field(default_factory=list) + # Deprecated spellings: include_tags ≡ tags_all, exclude_tags ≡ tags_none. + include_tags: list[str] = Field(default_factory=list, deprecated=True) + exclude_tags: list[str] = Field(default_factory=list, deprecated=True) + tags_all: list[str] = Field(default_factory=list) + tags_any: list[str] = Field(default_factory=list) + tags_none: list[str] = Field(default_factory=list) name_contains: str | None = None metadata_filter: dict[str, Any] | None = None limit: conint(ge=1, le=1000) = 100 - @field_validator("include_tags", "exclude_tags", mode="before") + @field_validator( + "include_tags", "exclude_tags", "tags_all", "tags_any", "tags_none", + mode="before", + ) @classmethod def _split_csv_tags(cls, v): if v is None: diff --git a/app/assets/database/queries/asset_reference.py b/app/assets/database/queries/asset_reference.py index 967b0e43a..126a0c9a4 100644 --- a/app/assets/database/queries/asset_reference.py +++ b/app/assets/database/queries/asset_reference.py @@ -268,6 +268,8 @@ def list_references_page( order: str | None = None, after_cursor_value: object | None = None, after_cursor_id: str | None = None, + # Appended last so pre-existing positional callers keep binding correctly. + any_tags: Sequence[str] | None = None, ) -> tuple[list[AssetReference], dict[str, list[str]], int]: """List references with pagination, filtering, and sorting. @@ -293,7 +295,7 @@ def list_references_page( escaped, esc = escape_sql_like_string(name_contains) base = base.where(AssetReference.name.ilike(f"%{escaped}%", escape=esc)) - base = apply_tag_filters(base, include_tags, exclude_tags) + base = apply_tag_filters(base, include_tags, exclude_tags, any_tags) base = apply_metadata_filter(base, metadata_filter) sort = (sort or "created_at").lower() @@ -345,7 +347,7 @@ def list_references_page( count_stmt = count_stmt.where( AssetReference.name.ilike(f"%{escaped}%", escape=esc) ) - count_stmt = apply_tag_filters(count_stmt, include_tags, exclude_tags) + count_stmt = apply_tag_filters(count_stmt, include_tags, exclude_tags, any_tags) count_stmt = apply_metadata_filter(count_stmt, metadata_filter) total = int(session.execute(count_stmt).scalar_one() or 0) diff --git a/app/assets/database/queries/common.py b/app/assets/database/queries/common.py index 89bb49327..7b0c211a0 100644 --- a/app/assets/database/queries/common.py +++ b/app/assets/database/queries/common.py @@ -60,10 +60,13 @@ def apply_tag_filters( stmt: sa.sql.Select, include_tags: Sequence[str] | None = None, exclude_tags: Sequence[str] | None = None, + any_tags: Sequence[str] | None = None, ) -> sa.sql.Select: - """include_tags: every tag must be present; exclude_tags: none may be present.""" + """include_tags: every tag must be present; any_tags: at least one must be + present; exclude_tags: none may be present.""" include_tags = normalize_tags(include_tags) exclude_tags = normalize_tags(exclude_tags) + any_tags = normalize_tags(any_tags) if include_tags: for tag_name in include_tags: @@ -74,6 +77,14 @@ def apply_tag_filters( ) ) + if any_tags: + stmt = stmt.where( + exists().where( + (AssetReferenceTag.asset_reference_id == AssetReference.id) + & (AssetReferenceTag.tag_name.in_(any_tags)) + ) + ) + if exclude_tags: stmt = stmt.where( ~exists().where( diff --git a/app/assets/database/queries/tags.py b/app/assets/database/queries/tags.py index 148f34801..e5f70e3df 100644 --- a/app/assets/database/queries/tags.py +++ b/app/assets/database/queries/tags.py @@ -340,6 +340,8 @@ def list_tag_counts_for_filtered_assets( name_contains: str | None = None, metadata_filter: dict | None = None, limit: int = 100, + # Appended last so pre-existing positional callers keep binding correctly. + any_tags: Sequence[str] | None = None, ) -> dict[str, int]: """Return tag counts for assets matching the given filters. @@ -359,7 +361,7 @@ def list_tag_counts_for_filtered_assets( escaped, esc = escape_sql_like_string(name_contains) ref_sq = ref_sq.where(AssetReference.name.ilike(f"%{escaped}%", escape=esc)) - ref_sq = apply_tag_filters(ref_sq, include_tags, exclude_tags) + ref_sq = apply_tag_filters(ref_sq, include_tags, exclude_tags, any_tags) ref_sq = apply_metadata_filter(ref_sq, metadata_filter) ref_sq = ref_sq.subquery() diff --git a/app/assets/services/asset_management.py b/app/assets/services/asset_management.py index a4c8b5a75..efdfd31a8 100644 --- a/app/assets/services/asset_management.py +++ b/app/assets/services/asset_management.py @@ -279,6 +279,8 @@ def list_assets_page( sort: str = "created_at", order: str = "desc", after: str | None = None, + # Appended last so pre-existing positional callers keep binding correctly. + any_tags: Sequence[str] | None = None, ) -> ListAssetsResult: """List assets with optional cursor pagination. @@ -317,6 +319,7 @@ def list_assets_page( owner_id=owner_id, include_tags=include_tags, exclude_tags=exclude_tags, + any_tags=any_tags, name_contains=name_contains, metadata_filter=metadata_filter, limit=fetch_limit, diff --git a/app/assets/services/tagging.py b/app/assets/services/tagging.py index 5fa39d26a..69c8cf39c 100644 --- a/app/assets/services/tagging.py +++ b/app/assets/services/tagging.py @@ -85,6 +85,8 @@ def list_tag_histogram( name_contains: str | None = None, metadata_filter: dict | None = None, limit: int = 100, + # Appended last so pre-existing positional callers keep binding correctly. + any_tags: Sequence[str] | None = None, ) -> dict[str, int]: with create_session() as session: return list_tag_counts_for_filtered_assets( @@ -92,6 +94,7 @@ def list_tag_histogram( owner_id=owner_id, include_tags=include_tags, exclude_tags=exclude_tags, + any_tags=any_tags, name_contains=name_contains, metadata_filter=metadata_filter, limit=limit, diff --git a/comfy/cli_args.py b/comfy/cli_args.py index ee9e1ce9f..c6660846d 100644 --- a/comfy/cli_args.py +++ b/comfy/cli_args.py @@ -149,6 +149,7 @@ attn_group.add_argument("--use-quad-cross-attention", action="store_true", help= attn_group.add_argument("--use-pytorch-cross-attention", action="store_true", help="Use the new pytorch 2.0 cross attention function.") attn_group.add_argument("--use-sage-attention", action="store_true", help="Use sage attention.") attn_group.add_argument("--use-flash-attention", action="store_true", help="Use FlashAttention.") +attn_group.add_argument("--use-ck-attention", action="store_true", help="Use Comfy Kitchen attention.") parser.add_argument("--disable-xformers", action="store_true", help="Disable xformers.") @@ -179,6 +180,7 @@ parser.add_argument("--disable-async-offload", action="store_true", help="Disabl parser.add_argument("--disable-dynamic-vram", action="store_true", help="Disable dynamic VRAM and use estimate based model loading.") parser.add_argument("--enable-dynamic-vram", action="store_true", help="Enable dynamic VRAM on systems where it's not enabled by default.") parser.add_argument("--fast-disk", action="store_true", help="Prefer disk-backed dynamic loading and offload over unpinned RAM. Can be faster for users with fast NVME disks.") +parser.add_argument("--disable-cuda-graphs", action="store_true", help="Disable CUDA graphs.") parser.add_argument("--force-non-blocking", action="store_true", help="Force ComfyUI to use non-blocking operations for all applicable tensors. This may improve performance on some non-Nvidia systems but can cause issues with some workflows.") diff --git a/comfy/clip_model.py b/comfy/clip_model.py index d7d3f994c..26cc5d7ee 100644 --- a/comfy/clip_model.py +++ b/comfy/clip_model.py @@ -314,13 +314,18 @@ class CLIPVisionModelProjection(torch.nn.Module): if "projection_dim" in config_dict: self.visual_projection = operations.Linear(config_dict["hidden_size"], config_dict["projection_dim"], bias=False) else: - self.visual_projection = lambda a: a + self.visual_projection = torch.nn.Identity() if "llava3" == config_dict.get("projector_type", None): self.multi_modal_projector = LlavaProjector(config_dict["hidden_size"], 4096, dtype, device, operations) else: self.multi_modal_projector = None + def _load_from_state_dict(self, state_dict, prefix, *args, **kwargs): + if "{}visual_projection.weight".format(prefix) not in state_dict: + self.visual_projection = torch.nn.Identity() + super()._load_from_state_dict(state_dict, prefix, *args, **kwargs) + def forward(self, *args, **kwargs): x = self.vision_model(*args, **kwargs) out = self.visual_projection(x[2]) diff --git a/comfy/latent_formats.py b/comfy/latent_formats.py index c4270022b..dc737fc7d 100644 --- a/comfy/latent_formats.py +++ b/comfy/latent_formats.py @@ -957,6 +957,11 @@ class ACEAudio15(LatentFormat): latent_dimensions = 1 temporal_downscale_ratio = 1764 +class MiniMaxMusic3(LatentFormat): + latent_channels = 128 + latent_dimensions = 1 + temporal_downscale_ratio = 512 + class ChromaRadiance(LatentFormat): latent_channels = 3 spacial_downscale_ratio = 1 diff --git a/comfy/ldm/lightricks/av_model.py b/comfy/ldm/lightricks/av_model.py index 8e360f6a8..c60148e2a 100644 --- a/comfy/ldm/lightricks/av_model.py +++ b/comfy/ldm/lightricks/av_model.py @@ -96,6 +96,8 @@ class BasicAVTransformerBlock(nn.Module): attn_precision=None, apply_gated_attention=False, cross_attention_adaln=False, + ff_bias=True, + audio_ff_bias=True, dtype=None, device=None, operations=None, @@ -178,10 +180,10 @@ class BasicAVTransformerBlock(nn.Module): ) self.ff = FeedForward( - v_dim, dim_out=v_dim, glu=True, dtype=dtype, device=device, operations=operations + v_dim, dim_out=v_dim, glu=True, ff_bias=ff_bias, dtype=dtype, device=device, operations=operations ) self.audio_ff = FeedForward( - a_dim, dim_out=a_dim, glu=True, dtype=dtype, device=device, operations=operations + a_dim, dim_out=a_dim, glu=True, ff_bias=audio_ff_bias, dtype=dtype, device=device, operations=operations ) num_ada_params = ADALN_CROSS_ATTN_PARAMS_COUNT if cross_attention_adaln else ADALN_BASE_PARAMS_COUNT @@ -413,12 +415,16 @@ class LTXAVModel(LTXVModel): apply_gated_attention=False, caption_proj_before_connector=False, cross_attention_adaln=False, + ff_bias=True, + audio_ff_bias=True, + use_prompt_adaln_single=True, dtype=None, device=None, operations=None, **kwargs, ): # Store audio-specific parameters + self.audio_ff_bias = audio_ff_bias self.audio_in_channels = audio_in_channels self.audio_cross_attention_dim = audio_cross_attention_dim self.audio_attention_head_dim = audio_attention_head_dim @@ -451,6 +457,8 @@ class LTXAVModel(LTXVModel): timestep_scale_multiplier=timestep_scale_multiplier, caption_proj_before_connector=caption_proj_before_connector, cross_attention_adaln=cross_attention_adaln, + ff_bias=ff_bias, + use_prompt_adaln_single=use_prompt_adaln_single, dtype=dtype, device=device, operations=operations, @@ -475,7 +483,7 @@ class LTXAVModel(LTXVModel): operations=self.operations, ) - if self.cross_attention_adaln: + if self.cross_attention_adaln and self.use_prompt_adaln_single: self.audio_prompt_adaln_single = AdaLayerNormSingle( self.audio_inner_dim, embedding_coefficient=2, @@ -606,6 +614,8 @@ class LTXAVModel(LTXVModel): a_context_dim=self.audio_cross_attention_dim, apply_gated_attention=self.apply_gated_attention, cross_attention_adaln=self.cross_attention_adaln, + ff_bias=self.ff_bias, + audio_ff_bias=self.audio_ff_bias, dtype=dtype, device=device, operations=self.operations, @@ -924,9 +934,15 @@ class LTXAVModel(LTXVModel): blocks_replace = patches_replace.get("dit", {}) prefetch_queue = comfy.model_prefetch.make_prefetch_queue(list(self.transformer_blocks), vx.device, transformer_options) + # Blocks whose self-attention should be perturbed to a value-passthrough (STG). + stg_self_attn_blocks = transformer_options.get("stg_self_attn_blocks", ()) + # Process transformer blocks for i, block in enumerate(self.transformer_blocks): comfy.model_prefetch.prefetch_queue_pop(prefetch_queue, vx.device, block) + block_transformer_options = transformer_options + if i in stg_self_attn_blocks: + block_transformer_options = {**transformer_options, "stg_skip_self_attn": True} if ("double_block", i) in blocks_replace: def block_wrap(args): @@ -969,7 +985,7 @@ class LTXAVModel(LTXVModel): "a_cross_scale_shift_timestep": av_ca_audio_scale_shift_timestep, "v_cross_gate_timestep": av_ca_a2v_gate_noise_timestep, "a_cross_gate_timestep": av_ca_v2a_gate_noise_timestep, - "transformer_options": transformer_options, + "transformer_options": block_transformer_options, "self_attention_mask": self_attention_mask, "v_prompt_timestep": v_prompt_timestep, "a_prompt_timestep": a_prompt_timestep, @@ -993,7 +1009,7 @@ class LTXAVModel(LTXVModel): a_cross_scale_shift_timestep=av_ca_audio_scale_shift_timestep, v_cross_gate_timestep=av_ca_a2v_gate_noise_timestep, a_cross_gate_timestep=av_ca_v2a_gate_noise_timestep, - transformer_options=transformer_options, + transformer_options=block_transformer_options, self_attention_mask=self_attention_mask, v_prompt_timestep=v_prompt_timestep, a_prompt_timestep=a_prompt_timestep, diff --git a/comfy/ldm/lightricks/duration_head.py b/comfy/ldm/lightricks/duration_head.py new file mode 100644 index 000000000..7d45d1fa7 --- /dev/null +++ b/comfy/ldm/lightricks/duration_head.py @@ -0,0 +1,81 @@ +"""LTX 2.4 DurationHead: predicts the natural shot duration (in seconds) from +the caption connector token outputs, without running the diffusion pipeline. +""" + +import torch +import torch.nn.functional as F +from torch import nn + + +class AttentionPooler(nn.Module): + """Cross-attend ``num_queries`` learnable tokens against ``tokens``.""" + + def __init__(self, hidden_dim=256, num_queries=1, num_heads=4): + super().__init__() + self.num_queries = num_queries + self.query_tokens = nn.Parameter(torch.empty(num_queries, hidden_dim)) + self.cross_attn = nn.MultiheadAttention(embed_dim=hidden_dim, num_heads=num_heads, batch_first=True) + + def forward(self, tokens): + queries = self.query_tokens.unsqueeze(0).expand(tokens.shape[0], -1, -1) + pooled, _ = self.cross_attn(queries, tokens, tokens, need_weights=False) + return pooled + + +class DurationHead(nn.Module): + """Predict duration in seconds from one or both connector outputs.""" + + def __init__( + self, + video_cross_attention_dim=4096, + audio_cross_attention_dim=2048, + pooler_hidden_dim=256, + num_queries=1, + num_pooler_heads=4, + mlp_hidden=256, + ): + super().__init__() + self.video_input_proj = nn.Linear(video_cross_attention_dim, pooler_hidden_dim) + self.video_modality_emb = nn.Parameter(torch.empty(pooler_hidden_dim)) + self.audio_input_proj = nn.Linear(audio_cross_attention_dim, pooler_hidden_dim) + self.audio_modality_emb = nn.Parameter(torch.empty(pooler_hidden_dim)) + self.attention_pooler = AttentionPooler( + hidden_dim=pooler_hidden_dim, num_queries=num_queries, num_heads=num_pooler_heads) + self.mlp_hidden = nn.Linear(pooler_hidden_dim * num_queries, mlp_hidden) + self.mlp_out = nn.Linear(mlp_hidden, 1) + + def forward(self, video_tokens=None, audio_tokens=None): + """``video_tokens``: (B, T_v, 4096), ``audio_tokens``: (B, T_a, 2048); + at least one required. Returns duration in seconds, shape (B,).""" + token_groups = [] + if video_tokens is not None: + token_groups.append(self.video_input_proj(video_tokens) + self.video_modality_emb) + if audio_tokens is not None: + token_groups.append(self.audio_input_proj(audio_tokens) + self.audio_modality_emb) + if not token_groups: + raise ValueError("DurationHead requires at least one of video_tokens / audio_tokens") + pooled = self.attention_pooler(torch.cat(token_groups, dim=1)) + pooled = pooled.reshape(pooled.shape[0], -1) + hidden = F.gelu(self.mlp_hidden(pooled), approximate="tanh") + return self.mlp_out(hidden).squeeze(-1).exp() + + +def normalize_state_dict(sd): + for prefix in ("model.diffusion_model.duration_head.", "duration_head."): + stripped = {k[len(prefix):]: v for k, v in sd.items() if k.startswith(prefix)} + if stripped: + return stripped + return sd + + +def seconds_to_num_frames(seconds, frame_rate, min_seconds, max_seconds, time_scale=8): + """Convert seconds to a frame count clamped to ``[min_seconds, max_seconds]`` + and snapped (floor) to the VAE's ``8k + 1`` causal temporal grid; snapping + that undershoots the minimum bumps up to the next grid point instead.""" + min_frames = max(1, round(min_seconds * frame_rate)) + max_frames = round(max_seconds * frame_rate) + raw_frames = max(min_frames, min(round(seconds * frame_rate), max_frames)) + frames = (raw_frames - 1) // time_scale * time_scale + 1 + if frames < min_frames: + frames = min(-(-(min_frames - 1) // time_scale) * time_scale + 1, max_frames) + return frames diff --git a/comfy/ldm/lightricks/embeddings_connector.py b/comfy/ldm/lightricks/embeddings_connector.py index 1a6ddcc8d..9c412827f 100644 --- a/comfy/ldm/lightricks/embeddings_connector.py +++ b/comfy/ldm/lightricks/embeddings_connector.py @@ -50,6 +50,7 @@ class BasicTransformerBlock1D(nn.Module): context_dim=None, attn_precision=None, apply_gated_attention=False, + ff_bias=True, dtype=None, device=None, operations=None, @@ -74,6 +75,7 @@ class BasicTransformerBlock1D(nn.Module): dim, dim_out=dim, glu=True, + ff_bias=ff_bias, dtype=dtype, device=device, operations=operations, @@ -123,6 +125,7 @@ class Embeddings1DConnector(nn.Module): causal_temporal_positioning=False, num_learnable_registers: Optional[int] = 128, apply_gated_attention=False, + connector_ff_bias=True, dtype=None, device=None, operations=None, @@ -148,6 +151,7 @@ class Embeddings1DConnector(nn.Module): attention_head_dim, context_dim=cross_attention_dim, apply_gated_attention=apply_gated_attention, + ff_bias=connector_ff_bias, dtype=dtype, device=device, operations=operations, diff --git a/comfy/ldm/lightricks/model.py b/comfy/ldm/lightricks/model.py index f80bffba7..dcbfa43ad 100644 --- a/comfy/ldm/lightricks/model.py +++ b/comfy/ldm/lightricks/model.py @@ -303,22 +303,22 @@ class NormSingleLinearTextProjection(nn.Module): class GELU_approx(nn.Module): - def __init__(self, dim_in, dim_out, dtype=None, device=None, operations=None): + def __init__(self, dim_in, dim_out, bias=True, dtype=None, device=None, operations=None): super().__init__() - self.proj = operations.Linear(dim_in, dim_out, dtype=dtype, device=device) + self.proj = operations.Linear(dim_in, dim_out, bias=bias, dtype=dtype, device=device) def forward(self, x): return torch.nn.functional.gelu(self.proj(x), approximate="tanh") class FeedForward(nn.Module): - def __init__(self, dim, dim_out, mult=4, glu=False, dropout=0.0, dtype=None, device=None, operations=None): + def __init__(self, dim, dim_out, mult=4, glu=False, dropout=0.0, ff_bias=True, dtype=None, device=None, operations=None): super().__init__() inner_dim = int(dim * mult) - project_in = GELU_approx(dim, inner_dim, dtype=dtype, device=device, operations=operations) + project_in = GELU_approx(dim, inner_dim, bias=ff_bias, dtype=dtype, device=device, operations=operations) self.net = nn.Sequential( - project_in, nn.Dropout(dropout), operations.Linear(inner_dim, dim_out, dtype=dtype, device=device) + project_in, nn.Dropout(dropout), operations.Linear(inner_dim, dim_out, bias=ff_bias, dtype=dtype, device=device) ) def forward(self, x): @@ -462,28 +462,34 @@ class CrossAttention(nn.Module): ) def forward(self, x, context=None, mask=None, pe=None, k_pe=None, transformer_options={}): + self_attn = context is None q = self.to_q(x) context = x if context is None else context k = self.to_k(context) v = self.to_v(context) - q = self.q_norm(q) - k = self.k_norm(k) - - # These norms span all heads, so the per-head RMS+RoPE kernel is not equivalent. - if pe is not None: - if k_pe is None and q.shape == k.shape: - q, k = apply_rotary_emb_qk(q, k, pe) - else: - q = apply_rotary_emb(q, pe) - k = apply_rotary_emb(k, pe if k_pe is None else k_pe) - - if mask is None: - out = comfy.ldm.modules.attention.optimized_attention(q, k, v, self.heads, attn_precision=self.attn_precision, transformer_options=transformer_options) - elif isinstance(mask, GuideAttentionMask): - out = _attention_with_guide_mask(q, k, v, self.heads, mask, attn_precision=self.attn_precision, transformer_options=transformer_options) + # Spatio-Temporal Guidance (STG) perturbation: for the flagged self-attention + # layers, the attention degrades to a passthrough of the value projection (out = V). + if self_attn and transformer_options.get("stg_skip_self_attn", False): + out = v else: - out = comfy.ldm.modules.attention.optimized_attention(q, k, v, self.heads, mask=mask, attn_precision=self.attn_precision, transformer_options=transformer_options) + q = self.q_norm(q) + k = self.k_norm(k) + + # These norms span all heads, so the per-head RMS+RoPE kernel is not equivalent. + if pe is not None: + if k_pe is None and q.shape == k.shape: + q, k = apply_rotary_emb_qk(q, k, pe) + else: + q = apply_rotary_emb(q, pe) + k = apply_rotary_emb(k, pe if k_pe is None else k_pe) + + if mask is None: + out = comfy.ldm.modules.attention.optimized_attention(q, k, v, self.heads, attn_precision=self.attn_precision, transformer_options=transformer_options) + elif isinstance(mask, GuideAttentionMask): + out = _attention_with_guide_mask(q, k, v, self.heads, mask, attn_precision=self.attn_precision, transformer_options=transformer_options) + else: + out = comfy.ldm.modules.attention.optimized_attention(q, k, v, self.heads, mask=mask, attn_precision=self.attn_precision, transformer_options=transformer_options) # Apply per-head gating if enabled if self.to_gate_logits is not None: @@ -502,7 +508,7 @@ ADALN_CROSS_ATTN_PARAMS_COUNT = 9 class BasicTransformerBlock(nn.Module): def __init__( - self, dim, n_heads, d_head, context_dim=None, attn_precision=None, cross_attention_adaln=False, dtype=None, device=None, operations=None + self, dim, n_heads, d_head, context_dim=None, attn_precision=None, cross_attention_adaln=False, ff_bias=True, dtype=None, device=None, operations=None ): super().__init__() @@ -518,7 +524,7 @@ class BasicTransformerBlock(nn.Module): device=device, operations=operations, ) - self.ff = FeedForward(dim, dim_out=dim, glu=True, dtype=dtype, device=device, operations=operations) + self.ff = FeedForward(dim, dim_out=dim, glu=True, ff_bias=ff_bias, dtype=dtype, device=device, operations=operations) self.attn2 = CrossAttention( query_dim=dim, @@ -717,6 +723,9 @@ class LTXBaseModel(torch.nn.Module, ABC): caption_proj_before_connector=False, cross_attention_adaln=False, caption_projection_first_linear=True, + ff_bias=True, + use_prompt_adaln_single=True, + use_keyframes_abs_pos_embedding=False, dtype=None, device=None, operations=None, @@ -746,6 +755,9 @@ class LTXBaseModel(torch.nn.Module, ABC): self.caption_proj_before_connector = caption_proj_before_connector self.cross_attention_adaln = cross_attention_adaln self.caption_projection_first_linear = caption_projection_first_linear + self.ff_bias = ff_bias + self.use_prompt_adaln_single = use_prompt_adaln_single + self.use_keyframes_abs_pos_embedding = use_keyframes_abs_pos_embedding # Common dimensions self.inner_dim = num_attention_heads * attention_head_dim @@ -773,12 +785,17 @@ class LTXBaseModel(torch.nn.Module, ABC): self.in_channels, self.inner_dim, bias=True, dtype=dtype, device=device ) + if self.use_keyframes_abs_pos_embedding: + self.keyframes_abs_pos_embedding = nn.Parameter(torch.zeros(1, self.inner_dim, dtype=dtype, device=device)) + else: + self.keyframes_abs_pos_embedding = None + embedding_coefficient = ADALN_CROSS_ATTN_PARAMS_COUNT if self.cross_attention_adaln else ADALN_BASE_PARAMS_COUNT self.adaln_single = AdaLayerNormSingle( self.inner_dim, embedding_coefficient=embedding_coefficient, use_additional_conditions=False, dtype=dtype, device=device, operations=self.operations ) - if self.cross_attention_adaln: + if self.cross_attention_adaln and self.use_prompt_adaln_single: self.prompt_adaln_single = AdaLayerNormSingle( self.inner_dim, embedding_coefficient=2, use_additional_conditions=False, dtype=dtype, device=device, operations=self.operations ) @@ -1070,6 +1087,7 @@ class LTXVModel(LTXBaseModel): self.attention_head_dim, context_dim=self.cross_attention_dim, cross_attention_adaln=self.cross_attention_adaln, + ff_bias=self.ff_bias, dtype=dtype, device=device, operations=self.operations, @@ -1099,6 +1117,15 @@ class LTXVModel(LTXBaseModel): grid_mask = None if keyframe_idxs is not None and keyframe_idxs.shape[2] > 0: + tokens_per_frame = self.tokens_per_latent_frame(additional_args["orig_shape"]) + if keyframe_idxs.shape[2] % tokens_per_frame != 0: + raise ValueError( + f"keyframe_idxs holds {keyframe_idxs.shape[2]} tokens, which is not a whole number of " + f"{tokens_per_frame}-token latent frames. The appended frames were recorded against a " + "different spatial resolution than the latent being sampled, so their positions would land " + "on the wrong tokens. Crop the guides and separate the generated keyframes before " + "upscaling the latent." + ) additional_args.update({ "orig_patchified_shape": list(x.shape)}) denoise_mask = self.patchifier.patchify(denoise_mask)[0] grid_mask = ~torch.any(denoise_mask < 0, dim=-1)[0] @@ -1141,8 +1168,64 @@ class LTXVModel(LTXBaseModel): additional_args["num_guide_tokens"] = keyframe_idxs.shape[2] x = self.patchify_proj(x) + x = self.apply_keyframes_abs_pos_embedding( + x, + pixel_coords, + orig_shape=additional_args["orig_shape"], + grid_mask=grid_mask, + num_guide_tokens=additional_args.get("num_guide_tokens", 0), + generated_keyframes=kwargs.get("generated_keyframes", None), + ) return x, pixel_coords, additional_args + def tokens_per_latent_frame(self, orig_shape): + """Token count of a single latent frame at the given latent shape.""" + patch_size = self.patchifier.patch_size + return (orig_shape[3] // patch_size[1]) * (orig_shape[4] // patch_size[2]) + + def keyframes_abs_pos_mask(self, pixel_coords, orig_shape, grid_mask, num_guide_tokens, generated_keyframes): + """Per-token mask selecting the latents that encode a single standalone pixel frame. + + Returns a (batch, tokens) boolean mask over the already grid-filtered token sequence. + """ + temporal_start = pixel_coords[:, 0] + if temporal_start.ndim == 3: # (batch, tokens, [start, end]) + temporal_start = temporal_start[..., 0] + mask = temporal_start == 0 + if num_guide_tokens > 0: + mask[:, -num_guide_tokens:] = False + + if generated_keyframes is not None: + # The temporal patch size is always 1, so one latent frame is one row of tokens. + tokens_per_frame = self.tokens_per_latent_frame(orig_shape) + if generated_keyframes["tokens_per_frame"] != tokens_per_frame: + raise ValueError( + f"The generated keyframes were recorded at {generated_keyframes['tokens_per_frame']} tokens " + f"per latent frame but this latent has {tokens_per_frame}. Separate the generated keyframes " + "before upscaling the latent." + ) + first_token = generated_keyframes["first_latent_frame"] * tokens_per_frame + num_slot_tokens = generated_keyframes["num_keyframes"] * tokens_per_frame + slots = torch.zeros(orig_shape[2] * tokens_per_frame, dtype=torch.bool, device=mask.device) + slots[first_token:first_token + num_slot_tokens] = True + if grid_mask is not None: + slots = slots[grid_mask] + mask = mask | slots + + return mask + + def apply_keyframes_abs_pos_embedding(self, x, pixel_coords, orig_shape, grid_mask, num_guide_tokens, generated_keyframes): + """Add the learned keyframe marker to the single-pixel-frame tokens. + + A no-op for every checkpoint built without the parameter. + """ + if self.keyframes_abs_pos_embedding is None: + return x + + mask = self.keyframes_abs_pos_mask(pixel_coords, orig_shape, grid_mask, num_guide_tokens, generated_keyframes) + embedding = self.keyframes_abs_pos_embedding.to(device=x.device, dtype=x.dtype) + return x + mask.unsqueeze(-1).to(x.dtype) * embedding + def _build_guide_self_attention_mask(self, x, transformer_options, merged_args): """Build self-attention mask for per-guide attention attenuation. diff --git a/comfy/ldm/lightricks/vae/audio_vae.py b/comfy/ldm/lightricks/vae/audio_vae.py index b4a8c7524..f5b1756d3 100644 --- a/comfy/ldm/lightricks/vae/audio_vae.py +++ b/comfy/ldm/lightricks/vae/audio_vae.py @@ -1,6 +1,5 @@ import json from dataclasses import dataclass -import math import torch import torchaudio @@ -186,7 +185,7 @@ class AudioVAE(torch.nn.Module): ) def num_of_latents_from_frames(self, frames_number: int, frame_rate: float) -> int: - return math.ceil((float(frames_number) / frame_rate) * self.latents_per_second) + return round((float(frames_number) / frame_rate) * self.latents_per_second) def run_vocoder(self, mel_spec: torch.Tensor) -> torch.Tensor: audio_channels = self.autoencoder.decoder.out_ch diff --git a/comfy/ldm/lightricks/vae/na_diffusion_decoder.py b/comfy/ldm/lightricks/vae/na_diffusion_decoder.py new file mode 100644 index 000000000..ec539c665 --- /dev/null +++ b/comfy/ldm/lightricks/vae/na_diffusion_decoder.py @@ -0,0 +1,520 @@ +"""LTX 2.4 diffusion video VAE decoder (NADiffusionDecoder). + +Port of the reference ``DiffusionVideoDecoder`` without the NATTEN dependency: +``natten.na3d`` is replaced by ``comfy_kitchen.na3d``, which reproduces +NATTEN's semantics (window of exactly ``kernel_size`` per query, shifted +inward at grid boundaries, dilation 1) and dispatches cuda/triton/eager per +device and dtype (the eager backend covers CPU and fp32). + +Stages 1-4 deterministically upsample the latent into a context volume via +NA transformer blocks + linear pixel-shuffle upsamples. Stage 5 runs +``DiffusionNABlock``s that denoise patchified noised pixels ``x_t`` guided by +that context through AdaLN-Zero scale/shift. The 2.4 checkpoint is single-step +``x0``: one forward pass yields the pixels directly, no Euler loop. + +State dict keys match the shipped checkpoints directly (fused ``attn.qkv``, +``t_embedder.mlp.{0,2}``, ``shared_adaln.proj``); no rename pass is needed. +""" + +import math + +import torch +import torch.nn.functional as F +from einops import rearrange +from torch import nn +import comfy.model_management + +from comfy.ldm.lightricks.model import get_timestep_embedding +from .causal_video_autoencoder import Encoder, processor + +import comfy_kitchen + +# Token chunk for the SwiGLU MLP (bounds the [chunk, hidden] workspace). +MLP_TOKEN_CHUNK = 65536 + + +def rms_norm(x, weight, eps=1e-6): + if hasattr(F, "rms_norm"): + return F.rms_norm(x, (x.shape[-1],), weight=weight.to(x.dtype), eps=eps) + x_f = x.float() + x_f = x_f * torch.rsqrt(x_f.pow(2).mean(-1, keepdim=True) + eps) + return (x_f * weight.float()).to(x.dtype) + + +class RMSNorm(nn.Module): + def __init__(self, dim, eps=1e-6): + super().__init__() + self.eps = eps + self.weight = nn.Parameter(torch.ones(dim)) + + def forward(self, x): + return rms_norm(x, self.weight, self.eps) + + +def patchify(x, patch_size_hw, patch_size_t=1): + if patch_size_hw == 1 and patch_size_t == 1: + return x + return rearrange(x, "b c (f p) (h q) (w r) -> b (c p r q) f h w", p=patch_size_t, q=patch_size_hw, r=patch_size_hw) + + +def unpatchify(x, patch_size_hw, patch_size_t=1): + if patch_size_hw == 1 and patch_size_t == 1: + return x + return rearrange(x, "b (c p r q) f h w -> b c (f p) (h q) (w r)", p=patch_size_t, q=patch_size_hw, r=patch_size_hw) + + +# --- Absolute per-axis RoPE (matches ltx-core rope.py numerics) --- + +def default_rope_dim_split(head_dim): + d_t = (head_dim // 4) // 2 * 2 + d_hw = (head_dim - d_t) // 2 + if d_hw % 2 != 0: + d_t -= 2 + d_hw = (head_dim - d_t) // 2 + return (d_t, d_hw, d_hw) + + +def rope_inv_freqs(dim, base=10000.0, device=None): + out_device = device + if not comfy.model_management.supports_fp64(device): + device = torch.device("cpu") + + exponents = torch.arange(0, dim, 2, dtype=torch.float64, device=device) / dim + return (1.0 / torch.pow(torch.tensor(float(base), dtype=torch.float64, device=device), exponents)).to(dtype=torch.float32, device=out_device) + + +def _rope_tables(lengths, inv_freqs, device): + """Precompute per-axis fp32 cos/sin tables for global 0-based positions.""" + tables = [] + for length, inv in zip(lengths, inv_freqs): + pos = torch.arange(length, dtype=torch.float32, device=device) + ang = pos[:, None] * inv[None, :] + tables.append((ang.cos(), ang.sin())) + return tables + + +def _rope_matrices_slice(tables, t0, t1, h, w): + """Per-token rotation matrices ``(1, ts*h*w, 1, hd/2, 2, 2)`` fp32 for + ``comfy_kitchen.rms_rope_`` (interleaved-pair convention), covering global + frames ``[t0, t1)`` of the axis-factorized tables.""" + parts = [] + for (c, s), sl in zip(tables, (slice(t0, t1), slice(None), slice(None))): + c, s = c[sl], s[sl] + parts.append(torch.stack([c, -s, s, c], dim=-1).reshape(c.shape[0], 1, 1, c.shape[1], 2, 2)) + ts = t1 - t0 + freqs = torch.cat([ + parts[0].expand(ts, h, w, -1, 2, 2), + parts[1].transpose(0, 1).expand(ts, h, w, -1, 2, 2), + parts[2].movedim(0, 2).expand(ts, h, w, -1, 2, 2), + ], dim=3) + return freqs.reshape(1, ts * h * w, 1, -1, 2, 2) + + +class NeighborhoodAttention3D(nn.Module): + """QKV (fused, matching checkpoint keys) + q/k RMSNorm + abs RoPE + NA.""" + + def __init__(self, dim, kernel_size, head_dim=64, rope_base=10000.0): + super().__init__() + self.dim = dim + self.num_heads = dim // head_dim + self.head_dim = head_dim + self.kernel_size = tuple(kernel_size) + self.scale = head_dim ** -0.5 + self.rope_split = default_rope_dim_split(head_dim) + self.rope_base = rope_base + + self.qkv = nn.Linear(dim, dim * 3, bias=True) + self.proj = nn.Linear(dim, dim, bias=True) + self.q_norm = RMSNorm(head_dim, eps=1e-6) + self.k_norm = RMSNorm(head_dim, eps=1e-6) + + def forward(self, x, pre=None, add_to=None): + """``pre`` (per-token norm/modulate) is applied slice-wise so the full + pre-attention tensor is never materialized; ``add_to`` streams the + output projection into it in place (residual add) and returns it. + Both bound peak memory without changing results.""" + batch, t, h, w, _ = x.shape + inv_freqs = tuple(rope_inv_freqs(d, self.rope_base, device=x.device) for d in self.rope_split) + tables = _rope_tables((t, h, w), inv_freqs, x.device) + shape = (batch, t, h, w, self.num_heads, self.head_dim) + q = torch.empty(shape, dtype=x.dtype, device=x.device) + k = torch.empty(shape, dtype=x.dtype, device=x.device) + v = torch.empty(shape, dtype=x.dtype, device=x.device) + q_weight = (self.q_norm.weight.detach() * self.scale).to(x.dtype) # scale commutes with the rotation + k_weight = self.k_norm.weight.detach().to(x.dtype) + chunk = max(1, (2 ** 25) // max(h * w * self.dim, 1)) + for t0 in range(0, t, chunk): + t1 = min(t0 + chunk, t) + sl = x[:, t0:t1] if pre is None else pre(x[:, t0:t1]) + qc, kc, vc = self.qkv(sl).chunk(3, dim=-1) + cshape = (batch, t1 - t0, h, w, self.num_heads, self.head_dim) + q[:, t0:t1] = qc.reshape(cshape) + k[:, t0:t1] = kc.reshape(cshape) + v[:, t0:t1] = vc.reshape(cshape) + freqs = _rope_matrices_slice(tables, t0, t1, h, w) + nt = (t1 - t0) * h * w + for b in range(batch): + comfy_kitchen.rms_rope_( + q[b, t0:t1].view(1, nt, self.num_heads, self.head_dim), + k[b, t0:t1].view(1, nt, self.num_heads, self.head_dim), + freqs, q_weight, k_weight) + out = comfy_kitchen.na3d(q, k, v, list(self.kernel_size), None, 1.0) + del q, k, v + out = out.reshape(batch, t, h, w, self.dim) + res = add_to if add_to is not None else torch.empty_like(out) + for t0 in range(0, t, chunk): + t1 = min(t0 + chunk, t) + if add_to is not None: + res[:, t0:t1] += self.proj(out[:, t0:t1]) + else: + res[:, t0:t1] = self.proj(out[:, t0:t1]) + return res + + +class SwiGLU(nn.Module): + """``w_down(silu(w_gate(x)) * w_up(x))``, chunked over tokens to bound the + ``[chunk, hidden]`` workspace.""" + + def __init__(self, dim, hidden_dim): + super().__init__() + self.w_up = nn.Linear(dim, hidden_dim, bias=False) + self.w_gate = nn.Linear(dim, hidden_dim, bias=False) + self.w_down = nn.Linear(hidden_dim, dim, bias=False) + + def forward(self, x, pre=None, add_to=None): + """``pre``/``add_to`` as in ``NeighborhoodAttention3D.forward``.""" + _, t, h, w, _ = x.shape + chunk = max(1, MLP_TOKEN_CHUNK // max(h * w, 1)) + out = add_to if add_to is not None else torch.empty_like(x) + for t0 in range(0, t, chunk): + t1 = min(t0 + chunk, t) + sl = x[:, t0:t1] if pre is None else pre(x[:, t0:t1]) + y = self.w_down(F.silu(self.w_gate(sl)) * self.w_up(sl)) + if add_to is not None: + out[:, t0:t1] += y + else: + out[:, t0:t1] = y + return out + + +class NABlock(nn.Module): + """Pre-norm transformer block: NA -> SwiGLU MLP with residual adds.""" + + def __init__(self, dim, kernel_size, head_dim=64, mlp_ratio=4.0): + super().__init__() + self.norm1 = RMSNorm(dim, eps=1e-6) + self.attn = NeighborhoodAttention3D(dim, kernel_size, head_dim=head_dim) + self.norm2 = RMSNorm(dim, eps=1e-6) + hidden = (int(dim * mlp_ratio) + 15) // 16 * 16 + self.mlp = SwiGLU(dim, hidden) + + def forward(self, x): + x = self.attn(x, pre=self.norm1, add_to=x) + return self.mlp(x, pre=self.norm2, add_to=x) + + +def modulate(x, scale, shift): + return x * (1.0 + scale) + shift + + +class AdaLNZero(nn.Module): + """``t_emb`` -> 7 (scale/shift/gate) chunks; gate slots unused (folded at export).""" + + NUM_CHUNKS = 7 + + def __init__(self, dim, t_emb_dim): + super().__init__() + self.proj = nn.Linear(t_emb_dim, self.NUM_CHUNKS * dim, bias=True) + + def forward(self, t_emb): + h = self.proj(F.silu(t_emb)) + return tuple(c[:, None, None, None, :] for c in h.chunk(self.NUM_CHUNKS, dim=-1)) + + +class DiffusionNABlock(nn.Module): + """NA + SwiGLU with shared AdaLN-Zero scale/shift (ungated residuals).""" + + def __init__(self, dim, kernel_size, context_channels, head_dim=64, mlp_ratio=4.0): + super().__init__() + self.context_proj = nn.Linear(context_channels, dim, bias=True) + self.scale_shift_table = nn.Parameter(torch.zeros(AdaLNZero.NUM_CHUNKS, dim)) + self.norm1 = RMSNorm(dim, eps=1e-6) + self.attn = NeighborhoodAttention3D(dim, kernel_size, head_dim=head_dim) + self.norm2 = RMSNorm(dim, eps=1e-6) + hidden = (int(dim * mlp_ratio) + 15) // 16 * 16 + self.mlp = SwiGLU(dim, hidden) + + def forward(self, x, latent_context, modulation): + scale_msa, shift_msa, _, scale_mlp, shift_mlp, _, _ = [ + modulation[i] + self.scale_shift_table[i].view(1, 1, 1, 1, -1) for i in range(AdaLNZero.NUM_CHUNKS) + ] + chunk = max(1, MLP_TOKEN_CHUNK // max(x.shape[2] * x.shape[3], 1)) + for t0 in range(0, x.shape[1], chunk): + x[:, t0:t0 + chunk] += self.context_proj(latent_context[:, t0:t0 + chunk]) + x = self.attn(x, pre=lambda s: modulate(self.norm1(s), scale_msa, shift_msa), add_to=x) + return self.mlp(x, pre=lambda s: modulate(self.norm2(s), scale_mlp, shift_mlp), add_to=x) + + +class LinearPixelShuffleUpsample(nn.Module): + """Linear channel-expand, then channels-last pixel shuffle.""" + + def __init__(self, in_channels, stride, out_channels_reduction_factor=1): + super().__init__() + self.stride = tuple(stride) + proj_out_channels = math.prod(stride) * in_channels // out_channels_reduction_factor + self.out_channels = proj_out_channels // math.prod(stride) + self.proj = nn.Linear(in_channels, proj_out_channels, bias=True) + + def forward(self, x, drop_leading_frame=True): + batch, t, h, w, _ = x.shape + p1, p2, p3 = self.stride + out = torch.empty((batch, t * p1, h * p2, w * p3, self.out_channels), dtype=x.dtype, device=x.device) + chunk = max(1, MLP_TOKEN_CHUNK // max(h * w, 1)) + for t0 in range(0, t, chunk): + t1 = min(t0 + chunk, t) + out[:, t0 * p1:t1 * p1] = rearrange( + self.proj(x[:, t0:t1]), "b t h w (c p1 p2 p3) -> b (t p1) (h p2) (w p3) c", + p1=p1, p2=p2, p3=p3, + ) + if p1 == 2 and drop_leading_frame: + # The causal temporal pixel-shuffle duplicates the leading frame. + out = out[:, 1:] + return out + + +class TimestepEmbedder(nn.Module): + """Sinusoidal(256) -> MLP. ``mlp.{0,2}`` naming matches the checkpoint.""" + + def __init__(self, t_emb_dim=384, freq_dim=256): + super().__init__() + self.freq_dim = freq_dim + self.mlp = nn.Sequential( + nn.Linear(freq_dim, t_emb_dim, bias=True), + nn.SiLU(), + nn.Linear(t_emb_dim, t_emb_dim, bias=True), + ) + + def forward(self, timestep, dtype): + emb = get_timestep_embedding(timestep.flatten(), self.freq_dim, flip_sin_to_cos=True, + downscale_freq_shift=0, scale=1) + return self.mlp(emb.to(dtype)) + + +class NADiffusionDecoder(nn.Module): + """Stages 1-4 (deterministic NA upsample) + stage-5 diffusion blocks. + + Input latent must already be un-normalized (the wrapper applies + ``per_channel_statistics.un_normalize``, same as the conv VAE path). + """ + + def __init__( + self, + in_channels=128, + out_channels=3, + patch_size=4, + head_dim=64, + stage_channels=(2048, 1024, 512, 512, 256), + stage_depths=(4, 6, 4, 2, 8), + stage_kernels=((3, 7, 7), (3, 7, 7), (3, 5, 5), (3, 5, 5), (11, 11, 11)), + upsamples=(((1, 2, 2), 2), ((2, 1, 1), 2), ((2, 2, 2), 1), ((2, 2, 2), 2)), + stage5_kernel=(11, 11, 11), + t_emb_dim=384, + default_num_inference_steps=1, + timestep_scale_multiplier=1000.0, + model_output_type="x0", + ): + super().__init__() + self.patch_size = patch_size + self.out_channels = out_channels + self.timestep_scale_multiplier = timestep_scale_multiplier + self.model_output_type = model_output_type + self.register_buffer( + "default_inference_timesteps", + torch.linspace(1.0, 1.0 / default_num_inference_steps, default_num_inference_steps), + persistent=False, + ) + self.temporal_upscale = math.prod(s[0] for s, _ in upsamples) + self.spatial_upscale = math.prod(s[1] for s, _ in upsamples) * patch_size + # NATTEN-style last-frame border mitigation: replicate the last latent + # frame through stages 1-4, crop the appendix off the context after. + self.trailing_pad_latent_frames = (stage_kernels[0][0] // 2) * 2 + + self.conv_in = nn.Linear(in_channels, stage_channels[0], bias=True) + + self.det_stages = nn.ModuleList() + self.upsamples = nn.ModuleList() + for stage_i in range(len(stage_channels) - 1): + c = stage_channels[stage_i] + self.det_stages.append(nn.ModuleList( + [NABlock(c, stage_kernels[stage_i], head_dim=head_dim) for _ in range(stage_depths[stage_i])] + )) + stride, reduction = upsamples[stage_i] + self.upsamples.append(LinearPixelShuffleUpsample(c, stride, out_channels_reduction_factor=reduction)) + + self.t_embedder = TimestepEmbedder(t_emb_dim=t_emb_dim) + + c5 = stage_channels[-1] + self.context_channels = c5 + noised_pixel_channels = out_channels * (patch_size ** 2) + self.conv_in_x_t = nn.Linear(noised_pixel_channels, c5, bias=True) + self.shared_adaln = AdaLNZero(c5, t_emb_dim) + self.diff_blocks = nn.ModuleList([ + DiffusionNABlock(c5, stage5_kernel, context_channels=c5, head_dim=head_dim) + for _ in range(stage_depths[-1]) + ]) + self.norm_out = RMSNorm(c5, eps=1e-6) + self.conv_out = nn.Linear(c5, noised_pixel_channels, bias=True) + + def forward_pre_diffusion(self, z, drop_leading_frame=True, pad_trailing=True): + """Stages 1-4: latent -> stage-5 context, channels-last. + + ``drop_leading_frame`` must be True only when ``z`` contains the + latent's true temporal origin (t=0); tiled callers decoding a later + temporal chunk pass False (the duplicate leading frame belongs solely + to the origin chunk). ``pad_trailing`` only for chunks containing the + latent's last frame.""" + n = self.trailing_pad_latent_frames if pad_trailing else 0 + if n > 0: + z = torch.cat([z, z[:, :, -1:].expand(-1, -1, n, -1, -1)], dim=2) + x = z.permute(0, 2, 3, 4, 1) + x = self.conv_in(x) + for stage_i, blocks in enumerate(self.det_stages): + for block in blocks: + x = block(x) + x = self.upsamples[stage_i](x, drop_leading_frame=drop_leading_frame) + if n > 0: + x = x[:, :-(n * self.temporal_upscale)] + return x + + def forward_diff_step(self, context, x_t, t): + x = patchify(x_t, patch_size_hw=self.patch_size, patch_size_t=1) + x = self.conv_in_x_t(x.permute(0, 2, 3, 4, 1)) + t_emb = self.t_embedder(self.timestep_scale_multiplier * t, dtype=x.dtype) + modulation = self.shared_adaln(t_emb) + for block in self.diff_blocks: + x = block(x, context, modulation) + x = self.norm_out(x) + x = self.conv_out(x) + x = x.permute(0, 4, 1, 2, 3) + return unpatchify(x, patch_size_hw=self.patch_size, patch_size_t=1) + + def forward(self, z, generator=None, drop_leading_frame=True, pad_trailing=True): + context = self.forward_pre_diffusion(z, drop_leading_frame=drop_leading_frame, pad_trailing=pad_trailing) + batch, t5, h5, w5, _ = context.shape + pixel_shape = (batch, self.out_channels, t5, h5 * self.patch_size, w5 * self.patch_size) + x_t = torch.randn(pixel_shape, dtype=z.dtype, device=z.device, generator=generator) + + timesteps = self.default_inference_timesteps.to(z.device) + num_steps = timesteps.shape[0] + for i in range(num_steps): + t_now = timesteps[i].expand(batch) + model_out = self.forward_diff_step(context, x_t, t_now) + if self.model_output_type == "x0": + x0 = model_out + if i == num_steps - 1: + return x0 + velocity = (x_t.float() - x0.float()) / timesteps[i] + else: # "v" + velocity = model_out.float() + if i == num_steps - 1: + return (x_t.float() - timesteps[i] * velocity).to(z.dtype) + t_next = timesteps[i + 1] if i + 1 < num_steps else torch.zeros_like(timesteps[i]) + x_t = (x_t.float() - (timesteps[i] - t_next) * velocity).to(z.dtype) + return x_t + + +LTX_24_VAE_CONFIG = { + "_class_name": "CausalDiffusionVAE", + "dims": 3, + "model_output_type": "x0", + "encoder": { + "dims": 3, + "in_channels": 3, + "out_channels": 128, + "blocks": [ + ["res_x", {"num_layers": 4}], + ["compress_space_res", {"multiplier": 2}], + ["res_x", {"num_layers": 6}], + ["compress_time_res", {"multiplier": 2}], + ["res_x", {"num_layers": 4}], + ["compress_all_res", {"multiplier": 2}], + ["res_x", {"num_layers": 2}], + ["compress_all_res", {"multiplier": 1}], + ["res_x", {"num_layers": 2}], + ], + "patch_size": 4, + "latent_log_var": "constant", + "norm_layer": "pixel_norm", + "base_channels": 128, + "spatial_padding_mode": "zeros", + }, + "decoder": { + "in_channels": 128, + "out_channels": 3, + "patch_size": 4, + "head_dim": 64, + "stage_channels": [2048, 1024, 512, 512, 256], + "stage_depths": [4, 6, 4, 2, 8], + "stage_kernels": [[3, 7, 7], [3, 7, 7], [3, 5, 5], [3, 5, 5], [11, 11, 11]], + "upsamples": [[[1, 2, 2], 2], [[2, 1, 1], 2], [[2, 2, 2], 1], [[2, 2, 2], 2]], + "stage5_kernel": [11, 11, 11], + "timestep_scale_multiplier": 1000.0, + "default_num_inference_steps": 1, + }, +} + + +class CausalDiffusionVAE(nn.Module): + """LTX 2.4 video VAE: conv encoder (shared with the 2.0 arch) + NA + diffusion decoder. Interface mirrors ``causal_video_autoencoder.VideoVAE``. + """ + + def __init__(self, config=None): + super().__init__() + if config is None: + config = LTX_24_VAE_CONFIG + self.config = config + enc = config.get("encoder", LTX_24_VAE_CONFIG["encoder"]) + dec = config.get("decoder", LTX_24_VAE_CONFIG["decoder"]) + dec_defaults = LTX_24_VAE_CONFIG["decoder"] + + self.encoder = Encoder( + dims=enc.get("dims", 3), + in_channels=enc.get("in_channels", 3), + out_channels=enc.get("out_channels", 128), + blocks=enc.get("blocks", LTX_24_VAE_CONFIG["encoder"]["blocks"]), + patch_size=enc.get("patch_size", 4), + latent_log_var=enc.get("latent_log_var", "constant"), + norm_layer=enc.get("norm_layer", "pixel_norm"), + spatial_padding_mode=enc.get("spatial_padding_mode", "zeros"), + base_channels=enc.get("base_channels", 128), + ) + + self.decoder = NADiffusionDecoder( + in_channels=dec.get("in_channels", 128), + out_channels=dec.get("out_channels", 3), + patch_size=dec.get("patch_size", 4), + head_dim=dec.get("head_dim", 64), + stage_channels=tuple(dec.get("stage_channels", dec_defaults["stage_channels"])), + stage_depths=tuple(dec.get("stage_depths", dec_defaults["stage_depths"])), + stage_kernels=tuple(tuple(k) for k in dec.get("stage_kernels", dec_defaults["stage_kernels"])), + upsamples=tuple((tuple(s), r) for s, r in dec.get("upsamples", dec_defaults["upsamples"])), + stage5_kernel=tuple(dec.get("stage5_kernel", dec_defaults["stage5_kernel"])), + t_emb_dim=dec.get("t_emb_dim", 384), + default_num_inference_steps=dec.get("default_num_inference_steps", 1), + timestep_scale_multiplier=dec.get("timestep_scale_multiplier", 1000.0), + model_output_type=config.get("model_output_type", "x0"), + ) + + self.per_channel_statistics = processor() + + def encode(self, x, device=None): + x = x[:, :, :max(1, 1 + ((x.shape[2] - 1) // 8) * 8), :, :] + means, logvar = torch.chunk(self.encoder(x, device=device), 2, dim=1) + return self.per_channel_statistics.normalize(means) + + def decode(self, x): + # Fixed-seed noise so decodes are reproducible TODO: expose? + generator = torch.Generator(device=x.device) + generator.manual_seed(0) + return self.decoder(self.per_channel_statistics.un_normalize(x), generator=generator) diff --git a/comfy/ldm/minimax/model.py b/comfy/ldm/minimax/model.py index bc06288ab..b6feb8860 100644 --- a/comfy/ldm/minimax/model.py +++ b/comfy/ldm/minimax/model.py @@ -25,7 +25,7 @@ import comfy.model_prefetch import comfy.ops import comfy.patcher_extension import comfy.quant_ops -from comfy.ldm.modules.attention import optimized_attention +from comfy.ldm.modules.attention import AttentionTensorContainer, optimized_attention FRAME_PER_TOKEN = (1, 4, 4, 4, 4) FRAME_RESCALE = 5.0 / 3.0 @@ -91,6 +91,18 @@ def _video_t_grid(n, origin): return float(origin) + torch.cat([torch.zeros(1, dtype=torch.float64), spans[:-1].cumsum(0)]) +def _ref_t_span(blk): + # time-axis span a reference block occupies ahead of the target streams + kind = blk["kind"] + if kind == "image": + return 1.0 + if kind == "audio": + return float(blk["ref_audio_t"]) + if kind in ("video", "video_audio"): + return max(float(blk["ref_audio_t"]), sum(_video_t_spans(blk["latent_t"]))) + return 0.0 + + def _audio_grid(cursor, t, w_low, w_high): # channel-major stereo rows: t advances per latent frame, w pinned to the grid extremes per stereo channel, h stays 0 g = torch.zeros(t * 2, 3, dtype=torch.float64) @@ -165,9 +177,10 @@ class Attention(nn.Module): else: q = self.q_norm(q.view(s, self.heads, self.head_dim)) k = self.k_norm(k.view(s, self.heads, self.head_dim)) - q = q.transpose(0, 1).unsqueeze(0) - k = k.transpose(0, 1).unsqueeze(0) - v = v.transpose(0, 1).unsqueeze(0) + v = v.clone() + q = AttentionTensorContainer(q.transpose(0, 1).unsqueeze(0)) + k = AttentionTensorContainer(k.transpose(0, 1).unsqueeze(0)) + v = AttentionTensorContainer(v.transpose(0, 1).unsqueeze(0)) out = optimized_attention(q, k, v, self.heads, mask=None, skip_reshape=True, transformer_options=transformer_options) return self.out_proj(out.squeeze(0)) @@ -287,7 +300,7 @@ class FinalLayer(nn.Module): class PackedLayout: """Static packed-sequence structure for one shape/conditioning signature.""" - def __init__(self, text_len, latent_t, latent_h, latent_w, audio_t, keyframes=None, refs=None, frame_count=None): + def __init__(self, text_len, latent_t, latent_h, latent_w, audio_t, keyframes=None, refs=None): frame, w_grid = _frame_grid(latent_h, latent_w) frame_rows = frame.shape[0] @@ -298,29 +311,37 @@ class PackedLayout: img_pos, img_update = [], [] audio_pos, audio_update = [], [] - cursor = text_len row = text_len - if keyframes: - # fl2va: keyframe cond rows right after text, sharing the target spatial grid - for kf in keyframes: - pixel_index = kf["resolved_frame_index"] - if pixel_index == 0: - cond_t = float(text_len) - elif frame_count is not None and pixel_index == frame_count - 1: - cond_t = float(text_len) + sum(_video_t_spans(latent_t)) - FRAME_RESCALE - else: - raise ValueError("only first/last keyframe anchors are supported") - g = torch.empty(frame_rows, 3, dtype=torch.float64) - g[:, 0] = cond_t - g[:, 1:] = frame - segments.append(("cond", frame_rows)) - pos.append(g) - img_pos.append(torch.arange(row, row + frame_rows)) - img_update.append(torch.zeros(frame_rows, dtype=torch.bool)) - row += frame_rows - target_audio_w = (float(w_grid[0]), float(w_grid[-1])) + # refs pack between text and the targets, so the target timeline starts after their spans + cursor = float(text_len) + for blk in refs or (): + cursor += _ref_t_span(blk) + + if keyframes: + # fl2va: keyframe cond rows right after text, sharing the target spatial grid; + # anchors count from the target timeline origin, FRAME_RESCALE per pixel frame, 1.0 per audio latent frame + for kf in keyframes: + cond_t = cursor + FRAME_RESCALE * kf["resolved_frame_index"] + video_latent = kf.get("latent") + if video_latent is not None: + vt = video_latent.shape[2] + n = vt * frame_rows + segments.append(("cond", n)) + pos.append(_video_grid(vt, frame, cond_t)) + img_pos.append(torch.arange(row, row + n)) + img_update.append(torch.zeros(n, dtype=torch.bool)) + row += n + audio_latent = kf.get("audio_latent") + if audio_latent is not None: + rt = audio_latent.shape[-1] + segments.append(("cond_audio", rt * 2)) + pos.append(_audio_grid(cond_t, rt, *target_audio_w)) + audio_pos.append(torch.arange(row, row + rt * 2)) + audio_update.append(torch.zeros(rt * 2, dtype=torch.bool)) + row += rt * 2 + if refs: cursor = float(text_len) for blk in refs: @@ -388,7 +409,7 @@ class PackedLayout: self.audio_update = torch.cat(audio_update) self.signature = (text_len, latent_t, latent_h, latent_w, audio_t) # contiguous segment table (start, stop, kind) - # kinds: text / cond / ref_img / ref_audio / audio / video + # kinds: text / cond / cond_audio / ref_img / ref_audio / audio / video # the packed sequence is uniform per segment in (modality tag, timestep class), # except the text span (tag runs resolved at forward time from the presentation tags) seg_abs = [] @@ -528,8 +549,7 @@ class MiniMaxH3Model(nn.Module): if layout is None or layout.signature != (text_len, latent_t, lat_h, lat_w, audio_t): layout = PackedLayout(text_len, latent_t, lat_h, lat_w, audio_t, keyframes=payload.get("keyframes"), - refs=payload.get("refs"), - frame_count=payload.get("frame_count")) + refs=payload.get("refs")) # model_base passes model_sampling.timestep(sigma) = sigma * 1000 shift_v = float(transformer_options.get("minimax_h3_sigma_shift_video", self.sigma_shift_video)) @@ -542,14 +562,14 @@ class MiniMaxH3Model(nn.Module): vis_aug = float(payload.get("visual_cond_noise_aug", VISUAL_COND_TIMESTEP)) aud_aug = float(payload.get("audio_cond_noise_aug", AUDIO_COND_TIMESTEP)) has_vis_cond = any(k in ("cond", "ref_img") for _, _, k in layout.segments) - has_aud_cond = any(k == "ref_audio" for _, _, k in layout.segments) + has_aud_cond = any(k in ("cond_audio", "ref_audio") for _, _, k in layout.segments) seg_t = {"text": t_v, "video": t_v, "audio": t_a, "cond": max(t_v, vis_aug), "ref_img": max(t_v, vis_aug), - "ref_audio": max(t_a, aud_aug)} + "cond_audio": max(t_a, aud_aug), "ref_audio": max(t_a, aud_aug)} unique_t = sorted({t_v, t_a} | ({seg_t["cond"]} if has_vis_cond else set()) | ({seg_t["ref_audio"]} if has_aud_cond else set())) t_row = {t: i for i, t in enumerate(unique_t)} - seg_tag = {"text": 1, "video": 0, "audio": 2, "cond": 0, "ref_img": 0, "ref_audio": 2} + seg_tag = {"text": 1, "video": 0, "audio": 2, "cond": 0, "ref_img": 0, "cond_audio": 2, "ref_audio": 2} text_tags = payload.get("text_token_tags") mod_segments = [] diff --git a/comfy/ldm/minimax_music/__init__.py b/comfy/ldm/minimax_music/__init__.py new file mode 100644 index 000000000..e69de29bb diff --git a/comfy/ldm/minimax_music/ar.py b/comfy/ldm/minimax_music/ar.py new file mode 100644 index 000000000..2a8935318 --- /dev/null +++ b/comfy/ldm/minimax_music/ar.py @@ -0,0 +1,343 @@ +import dataclasses +import hashlib + +import torch +from torch import nn + +import comfy.model_management +import comfy.model_prefetch +import comfy.ops +import comfy.utils +from comfy.ldm.modules.attention import optimized_attention_for_device +from comfy.text_encoders.llama import Llama2_, Qwen3_8BConfig + +from .prompt import AUDIO_CODE_OFFSET, SPECIAL_TOKEN_IDS + + +CFG_SCALE = 1.5 +CFG_TOP_K = 50 +C0_VOCAB_SIZE = 16384 +MAX_PROMPT_TOKENS = 5000 +MAX_AUDIO_FRAMES = 9000 +AUDIO_FRAMES_PER_SECOND = 25 + + +def derive_seed(seed, *parts): + digest = hashlib.blake2b(digest_size=8, person=b"minimax-ttm") + digest.update(int(seed).to_bytes(8, "little", signed=False)) + for part in parts: + value = str(part).encode("utf-8") + digest.update(len(value).to_bytes(4, "little")) + digest.update(value) + return int.from_bytes(digest.digest(), "little") & ((1 << 63) - 1) + + +def sample_topk(logits, top_k, generator): + values = torch.nan_to_num(logits.float(), nan=-1e9, posinf=1e9, neginf=-1e9) + top_k = min(top_k, values.shape[-1]) + threshold = torch.topk(values, top_k, dim=-1).values[..., -1, None] + values = values.masked_fill(values < threshold, -float("inf")) + probabilities = torch.nan_to_num(torch.softmax(values, dim=-1), nan=0.0) + probabilities = probabilities / probabilities.sum(dim=-1, keepdim=True).clamp_min(1e-12) + return torch.multinomial(probabilities, 1, generator=generator).squeeze(-1) + + +class RVQAttention(nn.Module): + def __init__(self, hidden_size, num_heads, merged_qkv, dtype, device, operations): + super().__init__() + self.num_heads = num_heads + self.head_dim = hidden_size // num_heads + self.merged_qkv = merged_qkv + if merged_qkv: + self.qkv_proj = operations.Linear(hidden_size, hidden_size * 3, bias=False, dtype=dtype, device=device) + else: + self.q_proj = operations.Linear(hidden_size, hidden_size, bias=False, dtype=dtype, device=device) + self.k_proj = operations.Linear(hidden_size, hidden_size, bias=False, dtype=dtype, device=device) + self.v_proj = operations.Linear(hidden_size, hidden_size, bias=False, dtype=dtype, device=device) + self.o_proj = operations.Linear(hidden_size, hidden_size, bias=False, dtype=dtype, device=device) + + def forward(self, x): + batch, length, hidden_size = x.shape + if self.merged_qkv: + q, k, v = self.qkv_proj(x).chunk(3, dim=-1) + else: + q = self.q_proj(x) + k = self.k_proj(x) + v = self.v_proj(x) + q = q.reshape(batch, length, self.num_heads, self.head_dim).transpose(1, 2) + k = k.reshape(batch, length, self.num_heads, self.head_dim).transpose(1, 2) + v = v.reshape(batch, length, self.num_heads, self.head_dim).transpose(1, 2) + mask = torch.full((length, length), torch.finfo(q.dtype).min, device=q.device, dtype=q.dtype).triu_(1) + attention = optimized_attention_for_device(q.device, mask=True, small_input=True) + out = attention(q, k, v, self.num_heads, mask=mask, skip_reshape=True) + return self.o_proj(out) + + +class RVQRMSNorm(nn.Module): + def __init__(self, hidden_size, dtype, device): + super().__init__() + self.weight = nn.Parameter(torch.empty(hidden_size, dtype=dtype, device=device)) + + def forward(self, x): + return torch.nn.functional.rms_norm(x, (x.shape[-1],), comfy.ops.cast_to_input(self.weight, x), 1e-6) + + +class RVQMLP(nn.Module): + def __init__(self, hidden_size, intermediate_size, merged_mlp, dtype, device, operations): + super().__init__() + self.merged_mlp = merged_mlp + if merged_mlp: + self.gate_up_proj = operations.Linear(hidden_size, intermediate_size * 2, bias=False, dtype=dtype, device=device) + else: + self.gate_proj = operations.Linear(hidden_size, intermediate_size, bias=False, dtype=dtype, device=device) + self.up_proj = operations.Linear(hidden_size, intermediate_size, bias=False, dtype=dtype, device=device) + self.down_proj = operations.Linear(intermediate_size, hidden_size, bias=False, dtype=dtype, device=device) + + def forward(self, x): + if self.merged_mlp: + return comfy.ops.linear_input_act(self.down_proj, self.gate_up_proj(x), "swiglu") + return self.down_proj(torch.nn.functional.silu(self.gate_proj(x)) * self.up_proj(x)) + + +class RVQDecoderBlock(nn.Module): + def __init__(self, hidden_size, num_heads, intermediate_size, merged_qkv, merged_mlp, dtype, device, operations): + super().__init__() + self.input_layernorm = RVQRMSNorm(hidden_size, dtype, device) + self.self_attn = RVQAttention(hidden_size, num_heads, merged_qkv, dtype, device, operations) + self.post_attention_layernorm = RVQRMSNorm(hidden_size, dtype, device) + self.mlp = RVQMLP(hidden_size, intermediate_size, merged_mlp, dtype, device, operations) + + def forward(self, x): + x = x + self.self_attn(self.input_layernorm(x)) + return x + self.mlp(self.post_attention_layernorm(x)) + + +class RVQDepthDecoder(nn.Module): + def __init__(self, config, dtype, device, operations): + super().__init__() + hidden_size = int(config["hidden_size"]) + audio_vocab_size = int(config["audio_vocab_size"]) + merged_qkv = config.get("decoder_merged_qkv", False) + merged_mlp = config.get("decoder_merged_mlp", False) + num_codebooks = int(config["audio_num_codebooks"]) + self.projection = operations.Linear(hidden_size, hidden_size, bias=False, dtype=dtype, device=device) + self.pos_embedding = operations.Embedding(16, hidden_size, dtype=dtype, device=device) + self.audio_heads = nn.ModuleList([ + operations.Linear(hidden_size, audio_vocab_size, bias=False, dtype=dtype, device=device) + for _ in range(num_codebooks - 1) + ]) + self.layers = nn.ModuleList([ + RVQDecoderBlock( + hidden_size, + int(config["decoder_num_heads"]), + int(config["decoder_intermediate_size"]), + merged_qkv, + merged_mlp, + dtype, + device, + operations, + ) + for _ in range(int(config["decoder_num_layers"])) + ]) + self.norm = RVQRMSNorm(hidden_size, dtype, device) + + def forward(self, sequence): + positions = torch.arange(sequence.shape[1], device=sequence.device) + x = sequence + self.pos_embedding(positions, out_dtype=sequence.dtype).unsqueeze(0) + for layer in self.layers: + x = layer(x) + return self.norm(x) + + +class MiniMaxMusic3AR(nn.Module): + def __init__(self, config, dtype, device, operations): + super().__init__() + config_fields = {field.name for field in dataclasses.fields(Qwen3_8BConfig)} + qwen_config = Qwen3_8BConfig(**{key: value for key, value in config.items() if key in config_fields}) + qwen_config.lm_head = False + qwen_config.fixed_kv = True + self.model = Llama2_(qwen_config, device=device, dtype=dtype, ops=operations) + self.model.prefetch_dynamic_vbars = True + self.model.graph_dynamic_vbar_blocks = True + self.model.lm_head = operations.Linear(qwen_config.hidden_size, qwen_config.vocab_size, bias=False, dtype=dtype, device=device) + self.model.lm_head_pruned = operations.Linear(qwen_config.hidden_size, C0_VOCAB_SIZE + 1, bias=False, dtype=dtype, device=device) + self.model.embed_tokens_prefill = operations.Embedding(AUDIO_CODE_OFFSET, qwen_config.hidden_size, dtype=dtype, device=device) + self.model.embed_tokens_audio = operations.Embedding(C0_VOCAB_SIZE, qwen_config.hidden_size, dtype=dtype, device=device) + self.model.pruned_lm_head = None + self.model.pruned_embedding = None + self.model.audio_extra_embedding = operations.Embedding( + int(config["audio_vocab_size"]) * (int(config["audio_num_codebooks"]) - 1), + qwen_config.hidden_size, + dtype=dtype, + device=device, + ) + self.model.audio_decoder = RVQDepthDecoder(config, dtype, device, operations) + self.audio_vocab_size = int(config["audio_vocab_size"]) + self.num_codebooks = int(config["audio_num_codebooks"]) + self.embedding_scale = self.num_codebooks ** -0.5 + + def _guided_c0(self, logits, cfg_scale, top_k): + conditioned = logits[0:1].float() + unconditioned = logits[1:2].float() + guided = unconditioned + (conditioned - unconditioned) * cfg_scale + threshold = torch.topk(conditioned, top_k, dim=-1).values[..., -1, None] + return guided.masked_fill(conditioned < threshold, -float("inf")) + + def _depth_codes(self, hidden, c0, c0_embed, generator, execution_dtype, cfg_scale, top_k): + decoder = self.model.audio_decoder + sequence = [decoder.projection(hidden).unsqueeze(1)] + sequence.append(decoder.projection(c0_embed).unsqueeze(1)) + codes = [c0] + hidden_parts = [] + for index in range(1, self.num_codebooks): + out = decoder(torch.cat(sequence, dim=1))[:, -1] + hidden_parts.append(out[:1].detach()) + logits = decoder.audio_heads[index - 1](out) + conditioned = logits[:1].float() + unconditioned = logits[1:2].float() + code = sample_topk(unconditioned + (conditioned - unconditioned) * cfg_scale, top_k, generator).repeat(2) + codes.append(code) + if index < self.num_codebooks - 1: + embedding = self.model.audio_extra_embedding( + code + (index - 1) * self.audio_vocab_size, + out_dtype=execution_dtype, + ) + sequence.append(decoder.projection(embedding).unsqueeze(1)) + return torch.stack(codes, dim=1), torch.cat(hidden_parts, dim=-1) + + def _embed_c0(self, codes, execution_dtype): + if self.model.pruned_embedding: + return self.model.embed_tokens_audio(codes, out_dtype=execution_dtype) + return self.model.embed_tokens(codes + AUDIO_CODE_OFFSET, out_dtype=execution_dtype) + + def _embed_audio_frame(self, codes, execution_dtype): + c0 = self._embed_c0(codes[:, 0], execution_dtype) + offsets = torch.arange(self.num_codebooks - 1, device=codes.device) * self.audio_vocab_size + extra = self.model.audio_extra_embedding(codes[:, 1:] + offsets.unsqueeze(0), out_dtype=execution_dtype).sum(dim=1) + return ((c0 + extra) * self.embedding_scale).unsqueeze(1) + + def _sample_c0(self, hidden, cfg_scale, top_k, generator, vocab_mask): + if self.model.pruned_lm_head: + guided = self._guided_c0(self.model.lm_head_pruned(hidden).float(), cfg_scale, top_k) + code = sample_topk(guided, top_k, generator) + stop_token = 0 + offset = 1 + else: + logits = self.model.lm_head(hidden).float() + stop_token = SPECIAL_TOKEN_IDS["<|audio_end|>"] + logits = logits.masked_fill(vocab_mask, -float("inf")) + guided = self._guided_c0(logits, cfg_scale, top_k).masked_fill(vocab_mask, -float("inf")) + code = sample_topk(guided, top_k, generator) + offset = AUDIO_CODE_OFFSET + return torch.where(code == stop_token, 0, code - offset), code, stop_token + + def generate(self, input_ids, seed, max_audio_frames, device, cfg_scale=CFG_SCALE, top_k=CFG_TOP_K): + prompt_tokens = int(input_ids.shape[1]) + if prompt_tokens > MAX_PROMPT_TOKENS: + raise ValueError(f"MiniMax Music3 prompt has {prompt_tokens} tokens; maximum is {MAX_PROMPT_TOKENS}") + + input_ids = input_ids.to(device) + if comfy.model_management.should_use_bf16(device): + execution_dtype = torch.bfloat16 + else: + execution_dtype = torch.float32 + unconditioned = input_ids.clone() + unconditioned[:, 1:-2] = SPECIAL_TOKEN_IDS["<|audio_cfg|>"] + text_ids = torch.cat((input_ids, unconditioned), dim=0) + if self.model.pruned_embedding: + text_embeds = self.model.embed_tokens_prefill(text_ids, out_dtype=execution_dtype) + else: + text_embeds = self.model.embed_tokens(text_ids, out_dtype=execution_dtype) + decode_limit = min(int(max_audio_frames), MAX_AUDIO_FRAMES) + past = self.model.init_kv_cache(2, prompt_tokens + decode_limit + 1, device, execution_dtype) + output = self.model(None, embeds=text_embeds, past_key_values=past, dtype=execution_dtype) + last_hidden = output[0][:, -1] + past = output[2] + + generator = torch.Generator(device=device).manual_seed(derive_seed(seed, "ar")) + decoder = self.model.audio_decoder + depth_io = { + "hidden": torch.empty_like(last_hidden), + "c0": torch.empty((last_hidden.shape[0],), dtype=torch.long, device=device), + "c0_embed": torch.empty_like(last_hidden), + "codes": torch.empty((last_hidden.shape[0], self.num_codebooks), dtype=torch.long, device=device), + "depth_hidden": torch.empty((1, last_hidden.shape[-1] * (self.num_codebooks - 1)), dtype=execution_dtype, device=device), + } + decoder._comfy_cross_step_state = depth_io + comfy.model_management._register_cross_step(decoder) + hidden_frames = [] + pending_code = None + stop_token = None + pending_event = None + pending_hidden = None + progress = comfy.utils.ProgressBar(decode_limit) + cuda_device = torch.device(device).type == "cuda" + vocab_mask = None + if not self.model.pruned_lm_head: + vocab_mask = torch.ones(self.model.vocab_size, dtype=torch.bool, device=device) + vocab_mask[AUDIO_CODE_OFFSET:AUDIO_CODE_OFFSET + C0_VOCAB_SIZE] = False + vocab_mask[SPECIAL_TOKEN_IDS["<|audio_end|>"]] = False + + for frame_index in comfy.utils.model_trange(decode_limit + 1, desc="AR sampling"): + comfy.model_management.throw_exception_if_processing_interrupted() + if pending_code is not None: + if pending_event is not None: + pending_event.synchronize() + if int(pending_code.item()) == stop_token: + pending_hidden = None + break + if pending_hidden is not None: + hidden_frames.append(pending_hidden) + progress.update_absolute(len(hidden_frames)) + if len(hidden_frames) >= decode_limit: + break + + c0, code_or_stop, stop_token = self._sample_c0(last_hidden, cfg_scale, top_k, generator, vocab_mask) + if pending_code is None: + pending_code = torch.empty_like(code_or_stop, device="cpu", pin_memory=cuda_device) + if cuda_device: + pending_event = torch.cuda.Event() + pending_code.copy_(code_or_stop, non_blocking=cuda_device) + if pending_event is not None: + pending_event.record() + + c0 = c0.repeat(2) + c0_embed = self._embed_c0(c0, execution_dtype) + depth_io["hidden"].copy_(last_hidden) + depth_io["c0"].copy_(c0) + depth_io["c0_embed"].copy_(c0_embed) + + def depth_core(): + codes, depth_hidden = self._depth_codes( + depth_io["hidden"], depth_io["c0"], depth_io["c0_embed"], generator, execution_dtype, cfg_scale, top_k + ) + depth_io["codes"].copy_(codes) + depth_io["depth_hidden"].copy_(depth_hidden) + + depth_queue = comfy.model_prefetch.make_prefetch_queue( + [[decoder, self.model.audio_extra_embedding]], device, {"prefetch_dynamic_vbars": True} + ) + comfy.model_prefetch.prefetch_queue_pop( + depth_queue, device, decoder, execution_dtype, core=depth_core, enable_graph=True, generator=generator + ) + comfy.model_prefetch.prefetch_queue_pop(depth_queue, device, None) + feedback_codes = depth_io["codes"] + depth_hidden = depth_io["depth_hidden"] + frame_hidden = torch.cat((last_hidden[:1].detach(), depth_hidden), dim=-1) + if frame_index > 0: + pending_hidden = frame_hidden[0].clone() + + feedback = self._embed_audio_frame(feedback_codes, execution_dtype) + output = self.model(None, embeds=feedback, past_key_values=past, dtype=execution_dtype) + last_hidden = output[0][:, -1] + past = output[2] + + if pending_hidden is not None and len(hidden_frames) < decode_limit: + if pending_event is not None: + pending_event.synchronize() + if int(pending_code.item()) != stop_token: + hidden_frames.append(pending_hidden) + + if not hidden_frames: + raise ValueError("MiniMax Music3 generated zero audio frames") + return torch.stack(hidden_frames).to(device="cpu") diff --git a/comfy/ldm/minimax_music/dav.py b/comfy/ldm/minimax_music/dav.py new file mode 100644 index 000000000..d442559f4 --- /dev/null +++ b/comfy/ldm/minimax_music/dav.py @@ -0,0 +1,137 @@ +import math + +import torch +from torch import nn + +import comfy.ops + + +def snake(x, alpha): + shape = x.shape + flat = x.reshape(shape[0], shape[1], -1) + alpha = comfy.ops.cast_to_input(alpha, flat) + flat = flat + (alpha + 1e-9).reciprocal() * torch.sin(alpha * flat).pow(2) + return flat.reshape(shape) + + +class Snake1d(nn.Module): + def __init__(self, channels, dtype, device): + super().__init__() + self.alpha = nn.Parameter(torch.empty(1, channels, 1, dtype=dtype, device=device)) + + def forward(self, x): + return snake(x, self.alpha) + + +def _weight_norm_conv(operations, *args, **kwargs): + return nn.utils.parametrizations.weight_norm(operations.Conv1d(*args, **kwargs)) + + +def _weight_norm_conv_transpose(operations, *args, **kwargs): + return nn.utils.parametrizations.weight_norm(operations.ConvTranspose1d(*args, **kwargs)) + + +class ResidualUnit(nn.Module): + def __init__(self, dim, dilation, dtype, device, operations): + super().__init__() + padding = 3 * dilation + self.block = nn.Sequential( + Snake1d(dim, dtype, device), + _weight_norm_conv( + operations, + dim, + dim, + kernel_size=7, + dilation=dilation, + padding=padding, + dtype=dtype, + device=device, + ), + Snake1d(dim, dtype, device), + _weight_norm_conv(operations, dim, dim, kernel_size=1, dtype=dtype, device=device), + ) + + def forward(self, x): + residual = self.block(x) + if residual.shape[-1] != x.shape[-1]: + padding = (x.shape[-1] - residual.shape[-1]) // 2 + x = x[..., padding:x.shape[-1] - padding] + return x + residual + + +class DecoderBlock(nn.Module): + def __init__(self, input_dim, output_dim, stride, dtype, device, operations): + super().__init__() + self.block = nn.Sequential( + Snake1d(input_dim, dtype, device), + _weight_norm_conv_transpose( + operations, + input_dim, + output_dim, + kernel_size=2 * stride, + stride=stride, + padding=math.ceil(stride / 2), + dtype=dtype, + device=device, + ), + ResidualUnit(output_dim, 1, dtype, device, operations), + ResidualUnit(output_dim, 3, dtype, device, operations), + ResidualUnit(output_dim, 9, dtype, device, operations), + ) + + def forward(self, x): + return self.block(x) + + +class Decoder(nn.Module): + def __init__(self, dtype, device, operations): + super().__init__() + layers = [ + _weight_norm_conv( + operations, + 1024, + 1536, + kernel_size=7, + padding=3, + dtype=dtype, + device=device, + ) + ] + channels = 1536 + output_dim = channels + for index, stride in enumerate((8, 8, 4, 2)): + input_dim = channels // (2 ** index) + output_dim = channels // (2 ** (index + 1)) + layers.append(DecoderBlock(input_dim, output_dim, stride, dtype, device, operations)) + layers.extend(( + Snake1d(output_dim, dtype, device), + _weight_norm_conv( + operations, + output_dim, + 1, + kernel_size=7, + padding=3, + dtype=dtype, + device=device, + ), + nn.Tanh(), + )) + self.model = nn.Sequential(*layers) + + def forward(self, x): + return self.model(x) + + +class MiniMaxMusic3DAV(nn.Module): + def __init__(self, dtype=None, device=None, operations=None): + super().__init__() + self.dec_in_proj = operations.Conv1d(64, 1024, kernel_size=1, dtype=dtype, device=device) + self.decoder = Decoder(dtype, device, operations) + + def decode(self, latent): + batch, _, frames = latent.shape + folded = latent.reshape(batch * 2, 64, frames) + waveform = self.decoder(self.dec_in_proj(folded)) + return waveform.reshape(batch, 2, -1) + + forward = decode diff --git a/comfy/ldm/minimax_music/dit.py b/comfy/ldm/minimax_music/dit.py new file mode 100644 index 000000000..211e0d7db --- /dev/null +++ b/comfy/ldm/minimax_music/dit.py @@ -0,0 +1,213 @@ +import math + +import torch +from torch import nn + +import comfy.model_management +import comfy.ops +import comfy.quant_ops +from comfy.ldm.modules.attention import optimized_attention_for_device + + +MAX_CONDITION_FRAMES = 200 +CONDITION_HOP_FRAMES = 100 + + +def latent_length(audio_frames): + return max(1, int(audio_frames * 44100 / 24000 * 960 / 512)) + + +class FourierFeatures(nn.Module): + def __init__(self, in_features, out_features, dtype, device): + super().__init__() + self.weight = nn.Parameter(torch.empty(out_features // 2, in_features, dtype=dtype, device=device)) + + def forward(self, value): + weight = comfy.ops.cast_to_input(self.weight, value) + features = 2.0 * math.pi * value @ weight.T + return torch.cat((features.cos(), features.sin()), dim=-1) + + +class LayerNorm(nn.Module): + def __init__(self, dim, dtype, device): + super().__init__() + self.gamma = nn.Parameter(torch.empty(dim, dtype=dtype, device=device)) + self.register_buffer("beta", torch.empty(dim, dtype=dtype, device=device)) + + def forward(self, x): + return torch.nn.functional.layer_norm( + x, + (x.shape[-1],), + comfy.ops.cast_to_input(self.gamma, x), + comfy.ops.cast_to_input(self.beta, x), + ) + + +class RotaryEmbedding(nn.Module): + def __init__(self, dim, dtype, device): + super().__init__() + self.register_buffer("inv_freq", torch.empty(dim // 2, dtype=dtype, device=device)) + + def forward_from_seq_len(self, length, device, dtype): + positions = torch.arange(length, device=device, dtype=torch.float32) + frequencies = torch.outer(positions, comfy.ops.cast_to_input(self.inv_freq, positions)) + frequencies = frequencies.to(dtype) + cos, sin = frequencies.cos(), frequencies.sin() + return torch.stack((cos, -sin, sin, cos), dim=-1).reshape(1, 1, length, frequencies.shape[-1], 2, 2) + + +def _apply_rope(x, rotation_matrix): + x_dtype = x.dtype + x = x.reshape(*x.shape[:-1], 2, -1).movedim(-2, -1).unsqueeze(-2).to(rotation_matrix.dtype) + x = rotation_matrix[..., 0] * x[..., 0] + rotation_matrix[..., 1] * x[..., 1] + return x.movedim(-1, -2).flatten(-2).to(x_dtype) + + +class Attention(nn.Module): + def __init__(self, dim, dim_heads, dtype, device, operations): + super().__init__() + self.num_heads = dim // dim_heads + self.dim_heads = dim_heads + self.to_qkv = operations.Linear(dim, dim * 3, bias=False, dtype=dtype, device=device) + self.to_out = operations.Linear(dim, dim, bias=False, dtype=dtype, device=device) + + def forward(self, x, rotation_matrix): + batch, length, dim = x.shape + q, k, v = self.to_qkv(x).chunk(3, dim=-1) + q = q.reshape(batch, length, self.num_heads, self.dim_heads).transpose(1, 2) + k = k.reshape(batch, length, self.num_heads, self.dim_heads).transpose(1, 2) + v = v.reshape(batch, length, self.num_heads, self.dim_heads).transpose(1, 2) + rotary_dims = rotation_matrix.shape[-3] * 2 + if comfy.model_management.in_training: + q = torch.cat((_apply_rope(q[..., :rotary_dims], rotation_matrix), q[..., rotary_dims:]), dim=-1) + k = torch.cat((_apply_rope(k[..., :rotary_dims], rotation_matrix), k[..., rotary_dims:]), dim=-1) + else: + rotated_q, rotated_k = comfy.quant_ops.ck.apply_rope_split_half(q[..., :rotary_dims], k[..., :rotary_dims], rotation_matrix) + q = torch.cat((rotated_q, q[..., rotary_dims:]), dim=-1) + k = torch.cat((rotated_k, k[..., rotary_dims:]), dim=-1) + attention = optimized_attention_for_device(q.device) + out = attention(q, k, v, self.num_heads, skip_reshape=True) + return self.to_out(out) + + +class GLU(nn.Module): + def __init__(self, dim, inner_dim, dtype, device, operations): + super().__init__() + self.proj = operations.Linear(dim, inner_dim * 2, dtype=dtype, device=device) + + def forward(self, x): + value, gate = self.proj(x).chunk(2, dim=-1) + return value * torch.nn.functional.silu(gate) + + +class FeedForward(nn.Module): + def __init__(self, dim, inner_dim, dtype, device, operations): + super().__init__() + self.ff = nn.Sequential( + GLU(dim, inner_dim, dtype, device, operations), + nn.Identity(), + operations.Linear(inner_dim, dim, dtype=dtype, device=device), + ) + + def forward(self, x): + return self.ff(x) + + +class TransformerBlock(nn.Module): + def __init__(self, dim, dim_heads, inner_dim, dtype, device, operations): + super().__init__() + self.pre_norm = LayerNorm(dim, dtype, device) + self.self_attn = Attention(dim, dim_heads, dtype, device, operations) + self.ff_norm = LayerNorm(dim, dtype, device) + self.ff = FeedForward(dim, inner_dim, dtype, device, operations) + + def forward(self, x, rotation_matrix): + x = x + self.self_attn(self.pre_norm(x), rotation_matrix) + return x + self.ff(self.ff_norm(x)) + + +class ContinuousTransformer(nn.Module): + def __init__(self, dtype, device, operations): + super().__init__() + self.project_in = operations.Linear(2304, 2048, bias=False, dtype=dtype, device=device) + self.project_out = operations.Linear(2048, 128, bias=False, dtype=dtype, device=device) + self.rotary_pos_emb = RotaryEmbedding(32, dtype, device) + self.layers = nn.ModuleList([ + TransformerBlock(2048, 64, 8192, dtype, device, operations) + for _ in range(36) + ]) + + def forward(self, x, timestep_embedding): + x = self.project_in(x) + x = torch.cat((timestep_embedding.unsqueeze(1), x), dim=1) + rotation_matrix = self.rotary_pos_emb.forward_from_seq_len(x.shape[1], x.device, x.dtype) + for layer in self.layers: + x = layer(x, rotation_matrix) + return self.project_out(x[:, 1:]) + + +class DiffusionTransformer(nn.Module): + def __init__(self, dtype, device, operations): + super().__init__() + self.transformer = ContinuousTransformer(dtype, device, operations) + self.timestep_features = FourierFeatures(1, 256, dtype, device) + self.to_timestep_embed = nn.Sequential( + operations.Linear(256, 2048, dtype=dtype, device=device), + nn.SiLU(), + operations.Linear(2048, 2048, dtype=dtype, device=device), + ) + self.preprocess_conv = operations.Conv1d(2304, 2304, 1, bias=False, dtype=dtype, device=device) + self.postprocess_conv = operations.Conv1d(128, 128, 1, bias=False, dtype=dtype, device=device) + + def forward(self, x, timestep, condition): + full = torch.cat((x, torch.zeros_like(x), condition), dim=1) + full = self.preprocess_conv(full) + full + timestep_features = self.timestep_features(timestep[:, None]).to(dtype=x.dtype) + timestep_embedding = self.to_timestep_embed(timestep_features) + out = self.transformer(full.transpose(1, 2), timestep_embedding).transpose(1, 2) + return self.postprocess_conv(out) + out + + +class MiniMaxMusic3DiT(nn.Module): + def __init__(self, dtype=None, device=None, operations=None, **kwargs): + super().__init__() + self.dtype = dtype + self.latent_conditioners = nn.Sequential( + operations.Conv1d(4096, 2048, kernel_size=3, padding=1, dtype=dtype, device=device) + ) + self.diffusion_transformer = DiffusionTransformer(dtype, device, operations) + self.cond_layer_logits = nn.Parameter(torch.empty(8, dtype=dtype, device=device)) + self.cond_layer_scale = nn.Parameter(torch.empty(1, dtype=dtype, device=device)) + + def aligned_condition(self, hidden): + frames = hidden.shape[1] + hidden = hidden.transpose(1, 2).reshape(hidden.shape[0], 8, 4096, frames) + weights = torch.softmax(comfy.ops.cast_to_input(self.cond_layer_logits, hidden), dim=0) + hidden = torch.einsum("blht,l->bht", hidden, weights) + hidden = comfy.ops.cast_to_input(self.cond_layer_scale, hidden) * hidden + condition = self.latent_conditioners(hidden) + return torch.nn.functional.interpolate(condition, size=latent_length(frames), mode="nearest") + + def forward(self, x, timestep, context, conditioning_scale, **kwargs): + condition = self.aligned_condition(context) + condition = condition * conditioning_scale[:, :1, :1] + if condition.shape[-1] < x.shape[-1]: + condition = torch.nn.functional.pad(condition, (0, x.shape[-1] - condition.shape[-1])) + else: + condition = condition[..., :x.shape[-1]] + window = latent_length(MAX_CONDITION_FRAMES) + if x.shape[-1] <= window: + return -self.diffusion_transformer(x, timestep, condition) + + output = torch.zeros_like(x) + count = torch.zeros((1, 1, x.shape[-1]), device=x.device, dtype=x.dtype) + hop = latent_length(CONDITION_HOP_FRAMES) + start = 0 + while start < x.shape[-1]: + end = min(start + window, x.shape[-1]) + output[..., start:end] -= self.diffusion_transformer(x[..., start:end], timestep, condition[..., start:end]) + count[..., start:end] += 1 + if end == x.shape[-1]: + break + start += hop + return output / count diff --git a/comfy/ldm/minimax_music/prompt.py b/comfy/ldm/minimax_music/prompt.py new file mode 100644 index 000000000..5f197ee12 --- /dev/null +++ b/comfy/ldm/minimax_music/prompt.py @@ -0,0 +1,70 @@ +import re + + +SPECIAL_TOKEN_IDS = { + "<|im_start|>": 151644, + "<|im_end|>": 151645, + "<|audio_cfg|>": 151654, + "<|audio_start|>": 151669, + "<|audio_end|>": 151670, + "<|caption_start|>": 151671, + "<|caption_end|>": 151672, + "<|lyrics_start|>": 151673, + "<|lyrics_end|>": 151674, +} +AUDIO_CODE_OFFSET = 151675 + +_SPECIAL_TAG_RE = re.compile(r"<\|([^|]*)\|>") +_LYRIC_TAG_RE = re.compile(r"\s*(\[[^\]]+\])\s*") + + +def _remove_markdown_format(text): + lines = [] + for raw_line in text.splitlines(): + line = re.sub(r"^\s{0,3}#{1,6}\s+", "", raw_line) + line = re.sub(r"^\s*[*+-]\s+", "", line) + while "**" in line: + updated = re.sub(r"\*\*([^*]+)\*\*", r"\1", line) + if updated == line: + break + line = updated + line = re.sub(r"(?<|caption_start|>" + f"{clean_caption(caption)}" + "<|caption_end|><|lyrics_start|>" + f"{normalize_lyrics(lyrics)}" + "<|lyrics_end|><|im_end|><|audio_start|>" + ) + + +def validate_tokenizer(tokenizer): + for token, expected in SPECIAL_TOKEN_IDS.items(): + token_id = tokenizer.convert_tokens_to_ids(token) + if token_id != expected: + raise ValueError(f"MiniMax Music3 tokenizer mismatch for {token}: expected {expected}, got {token_id}") diff --git a/comfy/ldm/modules/attention.py b/comfy/ldm/modules/attention.py index 2c549e095..b22d03d77 100644 --- a/comfy/ldm/modules/attention.py +++ b/comfy/ldm/modules/attention.py @@ -10,6 +10,8 @@ from typing import Optional, Any, Callable, Union import logging import functools +import comfy_kitchen + from .diffusionmodules.util import AlphaBlender, timestep_embedding from .sub_quadratic_attention import efficient_dot_product_attention @@ -49,6 +51,8 @@ except ImportError: logging.error(f"\n\nTo use the `--use-flash-attention` feature, the `flash-attn` package must be installed first.\ncommand:\n\t{sys.executable} -m pip install flash-attn") exit(-1) +COMFY_KITCHEN_INT8_ATTENTION_IS_AVAILABLE = comfy_kitchen.int8_attention_is_available() + REGISTERED_ATTENTION_FUNCTIONS = {} def register_attention_function(name: str, func: Callable): # avoid replacing existing functions @@ -145,9 +149,34 @@ def Normalize(in_channels, dtype=None, device=None): return torch.nn.GroupNorm(num_groups=32, num_channels=in_channels, eps=1e-6, affine=True, dtype=dtype, device=device) +class AttentionTensorContainer: + """Single-owner tensor input consumed by an optimized attention backend.""" + + __slots__ = ("tensor",) + + def __init__(self, tensor: torch.Tensor): + self.tensor: torch.Tensor | None = tensor + + def peek(self) -> torch.Tensor: + if self.tensor is None: + raise RuntimeError("attention tensor container has already been consumed") + return self.tensor + + def take(self) -> torch.Tensor: + tensor = self.peek() + self.tensor = None + return tensor + + def wrap_attn(func): @functools.wraps(func) def wrapper(*args, **kwargs): + containers = None + if len(args) >= 3 and isinstance(args[0], AttentionTensorContainer): + if not isinstance(args[1], AttentionTensorContainer) or not isinstance(args[2], AttentionTensorContainer): + raise TypeError("q, k, and v must all be attention tensor containers") + containers = args[:3] + remove_attn_wrapper_key = False try: if "_inside_attn_wrapper" not in kwargs: @@ -156,11 +185,22 @@ def wrap_attn(func): kwargs["_inside_attn_wrapper"] = True if transformer_options is not None: if "optimized_attention_override" in transformer_options: - return transformer_options["optimized_attention_override"](func, *args, **kwargs) + optimized_attention_override = transformer_options["optimized_attention_override"] + if containers is not None: + if hasattr(optimized_attention_override, "container_function"): + return optimized_attention_override.container_function(*args, **kwargs) + args = tuple(container.take() for container in containers) + args[3:] + return optimized_attention_override(func, *args, **kwargs) + + if containers is not None: + if wrapper.container_function is not None: + return wrapper.container_function(*args, **kwargs) + args = tuple(container.take() for container in containers) + args[3:] return func(*args, **kwargs) finally: if remove_attn_wrapper_key: del kwargs["_inside_attn_wrapper"] + wrapper.container_function = None return wrapper @wrap_attn @@ -545,6 +585,63 @@ def attention_pytorch(q, k, v, heads, mask=None, attn_precision=None, skip_resha ).transpose(1, 2).reshape(-1, q.shape[2], heads * dim_head) return out +def _comfy_kitchen_int8_inputs(q, k, v, heads, mask, skip_reshape, enable_gqa): + dim_head = q.shape[-1] if skip_reshape else q.shape[-1] // heads + b = q.shape[0] + if not skip_reshape: + q, k, v = _reshape_qkv_to_heads(q, k, v, b, heads, dim_head, enable_gqa, expand_kv=False) + q, k, v = map(lambda t: t.transpose(1, 2), (q, k, v)) + + if mask is not None: + if mask.ndim == 2: + mask = mask.unsqueeze(0) + if mask.ndim == 3: + mask = mask.unsqueeze(1) + + return q, k, v, mask, b, dim_head + + +@wrap_attn +def attention_comfy_kitchen_int8(q, k, v, heads, mask=None, attn_precision=None, skip_reshape=False, skip_output_reshape=False, **kwargs): + q, k, v, mask, b, dim_head = _comfy_kitchen_int8_inputs( + q, k, v, heads, mask, skip_reshape, kwargs.get("enable_gqa", False) + ) + out = comfy_kitchen.int8_attention( + q, + k, + v, + scale=kwargs.get("scale", None), + attn_mask=mask, + ) + if not skip_output_reshape: + out = out.transpose(1, 2).reshape(b, -1, heads * dim_head) + return out + + +def _attention_comfy_kitchen_int8_containers(q, k, v, heads, mask=None, attn_precision=None, skip_reshape=False, skip_output_reshape=False, **kwargs): + q = q.take() + k = k.take() + v = v.take() + q, k, v, mask, b, dim_head = _comfy_kitchen_int8_inputs( + q, k, v, heads, mask, skip_reshape, kwargs.get("enable_gqa", False) + ) + quantized = comfy_kitchen.prequantize_int8_attention( + q, + k, + v, + scale=kwargs.get("scale", None), + attn_mask=mask, + ) + del q, k, v + out = comfy_kitchen.int8_attention_from_prequantized(quantized) + if not skip_output_reshape: + out = out.transpose(1, 2).reshape(b, -1, heads * dim_head) + return out + + +attention_comfy_kitchen_int8.container_function = _attention_comfy_kitchen_int8_containers + + @wrap_attn def attention_sage(q, k, v, heads, mask=None, attn_precision=None, skip_reshape=False, skip_output_reshape=False, **kwargs): if kwargs.get("low_precision_attention", True) is False or (mask is not None and not SAGE_ATTENTION_SUPPORTS_MASK): @@ -775,10 +872,20 @@ else: logging.info("Using sub quadratic optimization for attention, if you have memory or speed issues try using: --use-split-cross-attention") optimized_attention = attention_sub_quad +if model_management.comfy_kitchen_attention_enabled(): + if COMFY_KITCHEN_INT8_ATTENTION_IS_AVAILABLE: + logging.info("Using Comfy Kitchen attention") + optimized_attention = attention_comfy_kitchen_int8 + else: + logging.error("Comfy Kitchen attention is unavailable. Install a Comfy Kitchen build with attention support to use --use-ck-attention.") + exit(-1) + optimized_attention_masked = optimized_attention # register core-supported attention functions +if COMFY_KITCHEN_INT8_ATTENTION_IS_AVAILABLE: + register_attention_function("comfy_kitchen_int8", attention_comfy_kitchen_int8) if SAGE_ATTENTION_IS_AVAILABLE: register_attention_function("sage", attention_sage) if SAGE_ATTENTION3_IS_AVAILABLE: diff --git a/comfy/model_base.py b/comfy/model_base.py index 469d301ea..6705eb6c3 100644 --- a/comfy/model_base.py +++ b/comfy/model_base.py @@ -22,6 +22,7 @@ import torch import logging import comfy.ldm.lightricks.av_model import comfy.ldm.minimax.model +import comfy.ldm.minimax_music.dit import comfy.nested_tensor import comfy.ldm.lightricks.symmetric_patchifier import comfy.context_windows @@ -1153,6 +1154,10 @@ class LTXV(BaseModel): if guide_attention_entries is not None: out['guide_attention_entries'] = comfy.conds.CONDConstant(guide_attention_entries) + generated_keyframes = kwargs.get("generated_keyframes", None) + if generated_keyframes is not None: + out['generated_keyframes'] = comfy.conds.CONDConstant(generated_keyframes) + return out def process_timestep(self, timestep, x, denoise_mask=None, **kwargs): @@ -1213,6 +1218,10 @@ class LTXAV(BaseModel): if ref_audio is not None: out['ref_audio'] = comfy.conds.CONDConstant(ref_audio) + generated_keyframes = kwargs.get("generated_keyframes", None) + if generated_keyframes is not None: + out['generated_keyframes'] = comfy.conds.CONDConstant(generated_keyframes) + return out def process_timestep(self, timestep, x, denoise_mask=None, audio_denoise_mask=None, **kwargs): @@ -2156,13 +2165,13 @@ class MiniMaxH3(BaseModel): keyframes = kwargs.get("minimax_keyframes", None) if keyframes is not None: payload["keyframes"] = keyframes - payload["frame_count"] = kwargs.get("minimax_frame_count", None) - payload["cond_video_latents"] = [kf["latent"] for kf in keyframes] + payload["cond_video_latents"] = [kf["latent"] for kf in keyframes if kf.get("latent") is not None] + payload["cond_audio_latents"] = [kf["audio_latent"] for kf in keyframes if kf.get("audio_latent") is not None] refs = kwargs.get("minimax_refs", None) if refs is not None: payload["refs"] = refs - payload["cond_video_latents"] = [r["latent"] for r in refs if "latent" in r] - payload["cond_audio_latents"] = [r["audio_latent"] for r in refs if r.get("audio_latent") is not None] + payload["cond_video_latents"] = payload.get("cond_video_latents", []) + [r["latent"] for r in refs if "latent" in r] + payload["cond_audio_latents"] = payload.get("cond_audio_latents", []) + [r["audio_latent"] for r in refs if r.get("audio_latent") is not None] if kwargs.get("minimax_visual_cond_noise_aug", None) is not None: payload["visual_cond_noise_aug"] = kwargs["minimax_visual_cond_noise_aug"] if kwargs.get("minimax_audio_cond_noise_aug", None) is not None: @@ -2176,7 +2185,7 @@ class MiniMaxH3(BaseModel): payload["layout"] = comfy.ldm.minimax.model.PackedLayout( cross_attn.shape[1], vs[2], (vs[3] + 1) // 2 * 2, (vs[4] + 1) // 2 * 2, latent_shapes[1][-1], keyframes=payload.get("keyframes"), - refs=payload.get("refs"), frame_count=payload.get("frame_count")) + refs=payload.get("refs")) out['minimax_payload'] = comfy.conds.CONDConstant(payload) return out @@ -2329,6 +2338,18 @@ class ACEStep15(BaseModel): out['refer_audio'] = comfy.conds.CONDRegular(refer_audio) return out +class MiniMaxMusic3(BaseModel): + def __init__(self, model_config, model_type=ModelType.FLOW, device=None): + super().__init__(model_config, model_type, device=device, unet_model=comfy.ldm.minimax_music.dit.MiniMaxMusic3DiT) + + def process_timestep(self, timestep, **kwargs): + return 1.0 - timestep + + def extra_conds(self, **kwargs): + out = super().extra_conds(**kwargs) + out["conditioning_scale"] = comfy.conds.CONDRegular(kwargs["conditioning_scale"]) + return out + class Omnigen2(BaseModel): def __init__(self, model_config, model_type=ModelType.FLOW, device=None): super().__init__(model_config, model_type, device=device, unet_model=comfy.ldm.omnigen.omnigen2.OmniGen2Transformer2DModel) diff --git a/comfy/model_detection.py b/comfy/model_detection.py index 103680fd1..e4bf30b78 100644 --- a/comfy/model_detection.py +++ b/comfy/model_detection.py @@ -44,6 +44,13 @@ def calculate_transformer_depth(prefix, state_dict_keys, state_dict): def detect_unet_config(state_dict, key_prefix, metadata=None): state_dict_keys = list(state_dict.keys()) + if ( + '{}cond_layer_logits'.format(key_prefix) in state_dict_keys + and '{}latent_conditioners.0.weight'.format(key_prefix) in state_dict_keys + and '{}diffusion_transformer.transformer.layers.0.self_attn.to_qkv.weight'.format(key_prefix) in state_dict_keys + ): + return {"audio_model": "minimax_music3"} + if '{}joint_blocks.0.context_block.attn.qkv.weight'.format(key_prefix) in state_dict_keys: #mmdit model unet_config = {} unet_config["in_channels"] = state_dict['{}x_embedder.proj.weight'.format(key_prefix)].shape[1] @@ -397,6 +404,7 @@ def detect_unet_config(state_dict, key_prefix, metadata=None): dit_config["cross_attention_dim"] = shape[1] if metadata is not None and "config" in metadata: dit_config.update(json.loads(metadata["config"]).get("transformer", {})) + dit_config["use_keyframes_abs_pos_embedding"] = '{}keyframes_abs_pos_embedding'.format(key_prefix) in state_dict_keys return dit_config if '{}genre_embedder.weight'.format(key_prefix) in state_dict_keys: #ACE-Step model @@ -829,11 +837,10 @@ def detect_unet_config(state_dict, key_prefix, metadata=None): dit_config["use_adaln_lora"] = True dit_config["adaln_lora_dim"] = 256 + dit_config["num_blocks"] = count_blocks(state_dict_keys, '{}blocks.'.format(key_prefix) + '{}.') if dit_config["model_channels"] == 2048: - dit_config["num_blocks"] = 28 dit_config["num_heads"] = 16 elif dit_config["model_channels"] == 5120: - dit_config["num_blocks"] = 36 dit_config["num_heads"] = 40 if dit_config["in_channels"] == 16: diff --git a/comfy/model_management.py b/comfy/model_management.py index 9f8e7f07b..ff963eb8e 100644 --- a/comfy/model_management.py +++ b/comfy/model_management.py @@ -490,28 +490,36 @@ try: except: rocm_version = (6, -1) - def aotriton_supported(gpu_arch): - path = torch.__path__[0] - path = os.path.join(os.path.join(path, "lib"), "aotriton.images") - gfx = set(map(lambda a: a[4:], filter(lambda a: a.startswith("amd-gfx"), os.listdir(path)))) - if gpu_arch in gfx: - return True - if "{}x".format(gpu_arch[:-1]) in gfx: - return True - if "{}xx".format(gpu_arch[:-2]) in gfx: - return True - return False + def aotriton_supported(): + """Whether pytorch reports flash attention as usable on this gpu. + + can_use_flash_attention() evaluates runtime eligibility for the given + parameters; on a ROCm build that includes checking the gpu arch against the + kernel images AOTriton was compiled for. Querying it avoids assuming where + those images live inside the torch install. The probe tensor is shaped and + typed to pass the unrelated SDPA checks, so False means no hardware support + rather than a rejected shape. + """ + try: + if not torch.backends.cuda.is_flash_attention_available(): # not built with flash attention + return False + q = torch.empty((1, 1, 8, 64), dtype=torch.float16, device=get_torch_device()) + params = torch.backends.cuda.SDPAParams(q, q, q, None, 0.0, False, False) + return torch.backends.cuda.can_use_flash_attention(params, False) + except (AttributeError, RuntimeError, TypeError) as e: + logging.warning("Could not query aotriton support: {}".format(e)) + return False logging.info("AMD arch: {}".format(arch)) logging.info("ROCm version: {}".format(rocm_version)) if args.use_split_cross_attention == False and args.use_quad_cross_attention == False: - if aotriton_supported(arch): # AMD efficient attention implementation depends on aotriton. + if aotriton_supported(): # AMD efficient attention implementation depends on aotriton. if torch_version_numeric >= (2, 7): # works on 2.6 but doesn't actually seem to improve much if any((a in arch) for a in ["gfx90a", "gfx942", "gfx950", "gfx1100", "gfx1101", "gfx1150", "gfx1151"]): # TODO: more arches, TODO: gfx950 ENABLE_PYTORCH_ATTENTION = True if rocm_version >= (7, 0): - if any((a in arch) for a in ["gfx1200", "gfx1201"]): - ENABLE_PYTORCH_ATTENTION = True + if any((a in arch) for a in ["gfx1200", "gfx1201"]): + ENABLE_PYTORCH_ATTENTION = True if torch_version_numeric >= (2, 7) and rocm_version >= (6, 4): if any((a in arch) for a in ["gfx1200", "gfx1201", "gfx950"]): # TODO: more arches, "gfx942" gives error on pytorch nightly 2.10 1013 rocm7.0 SUPPORT_FP8_OPS = True @@ -1360,9 +1368,14 @@ STREAM_CAST_BUFFERS = {} LARGEST_CASTED_WEIGHT = (None, 0) STREAM_AIMDO_CAST_BUFFERS = {} LARGEST_AIMDO_CASTED_WEIGHT = (None, 0) +CROSS_STEP_STATE = weakref.WeakSet() DEFAULT_AIMDO_CAST_BUFFER_RESERVATION_SIZE = 16 * 1024 ** 3 +# NOTE: devs/agents: this is temporary and will be removed in a future comfy. Not supported for custom node use. +def _register_cross_step(module): + CROSS_STEP_STATE.add(module) + def get_cast_buffer(offload_stream, device, size, ref): global LARGEST_CASTED_WEIGHT @@ -1417,6 +1430,10 @@ def reset_cast_buffers(): mmap_obj.bounce() DIRTY_MMAPS.clear() + for module in CROSS_STEP_STATE: + del module._comfy_cross_step_state + CROSS_STEP_STATE.clear() + for loaded_model in current_loaded_models: model = loaded_model.model if model is not None and model.is_dynamic(): @@ -1658,6 +1675,9 @@ def unpin_memory(tensor): def sage_attention_enabled(): return args.use_sage_attention +def comfy_kitchen_attention_enabled(): + return args.use_ck_attention + def flash_attention_enabled(): return args.use_flash_attention diff --git a/comfy/model_patcher.py b/comfy/model_patcher.py index ae3f0191d..72942aa04 100644 --- a/comfy/model_patcher.py +++ b/comfy/model_patcher.py @@ -685,6 +685,14 @@ class ModelPatcher: def set_model_attn2_output_patch(self, patch): self.set_model_patch(patch, "attn2_output_patch") + def set_model_optimized_attention(self, optimized_attention): + def optimized_attention_override(_, *args, **kwargs): + return optimized_attention(*args, **kwargs) + + if hasattr(optimized_attention, "container_function") and optimized_attention.container_function is not None: + optimized_attention_override.container_function = optimized_attention.container_function + self.model_options["transformer_options"]["optimized_attention_override"] = optimized_attention_override + def set_model_input_block_patch(self, patch): self.set_model_patch(patch, "input_block_patch") @@ -1879,8 +1887,29 @@ class ModelPatcherDynamic(ModelPatcher): loading = self._load_list(for_dynamic=True, default_device=device_to) loading.sort() + get_units = getattr(self.model, "get_dynamic_vram__units", None) + dynamic_units, last_dynamic_units = get_units() if get_units is not None else ([], []) + dynamic_units = list(dynamic_units) + last_dynamic_units = list(last_dynamic_units) + loading_by_module = {entry[-2]: entry for entry in loading} + loading = [] + for unit in dynamic_units: + unit_modules = unit if isinstance(unit, (list, tuple)) else (unit,) + modules = [module for root in unit_modules for module in root.modules() if module in loading_by_module] + for index, module in enumerate(modules): + loading.append((*loading_by_module.pop(module), unit if index == len(modules) - 1 else None)) + last_loading = [] + for unit in last_dynamic_units: + unit_modules = unit if isinstance(unit, (list, tuple)) else (unit,) + modules = [module for root in unit_modules for module in root.modules() if module in loading_by_module] + for index, module in enumerate(modules): + last_loading.append((*loading_by_module.pop(module), unit if index == len(modules) - 1 else None)) + loading.extend((*entry, None) for entry in loading_by_module.values()) + loading.extend(last_loading) + v_block = None + for x in loading: - *_, module_mem, n, m, params = x + *_, module_mem, n, m, params, end_of_block = x def set_dirty(item, dirty): if dirty or not hasattr(item, "_v_signature"): @@ -1973,6 +2002,13 @@ class ModelPatcherDynamic(ModelPatcher): move_weight_functions(m, device_to) + if hasattr(m, "_v"): + v_block = m._v if v_block is None else (v_block[0], v_block[1], max(v_block[2], m._v[1] + m._v[2] - v_block[1])) + if end_of_block is not None: + unit = end_of_block + (unit[0] if isinstance(unit, (list, tuple)) else unit)._v_block = v_block + v_block = None + for key, buf in self.model.named_buffers(recurse=True): if key not in self.backup_buffers: self.backup_buffers[key] = buf diff --git a/comfy/model_prefetch.py b/comfy/model_prefetch.py index aa6d22d77..bdde5137a 100644 --- a/comfy/model_prefetch.py +++ b/comfy/model_prefetch.py @@ -1,11 +1,19 @@ +import torch +import warnings +import weakref + import comfy_aimdo.model_vbar +from comfy.cli_args import args import comfy.memory_management import comfy.model_management import comfy.ops PREFETCH_QUEUES = [] +GRAPH_MODULES = weakref.WeakSet() +GRAPH_WARMED_MODULES = weakref.WeakSet() +GRAPH_CAPTURE_STREAMS = {} -def cleanup_prefetched_modules(comfy_modules): +def cleanup_prefetched_modules(module, comfy_modules): for s in comfy_modules: prefetch = getattr(s, "_prefetch", None) if prefetch is None: @@ -17,39 +25,86 @@ def cleanup_prefetched_modules(comfy_modules): if prefetch["signature"] is not None: comfy_aimdo.model_vbar.vbar_unpin(s._v) delattr(s, "_prefetch") + if getattr(module, "_v_block_faulted", False): + comfy_aimdo.model_vbar.vbar_unpin(module._v_block) + del module._v_block_faulted + +def _drop_graph(module): + graph = getattr(module, "_comfy_graph", None) + if graph is None: + return + # reset() through the bound method surfaces the allocator's benign + # "uncaptured free of a captured allocation" as catchable Python warnings; + # a plain del frees from the C++ dealloc path and spams stderr instead + with warnings.catch_warnings(): + warnings.simplefilter("ignore") + graph["graph"].reset() + del module._comfy_graph def cleanup_prefetch_queues(): - global PREFETCH_QUEUES + global PREFETCH_QUEUES, GRAPH_CAPTURE_STREAMS for queue in PREFETCH_QUEUES: for entry in queue: if entry is None or not isinstance(entry, tuple): continue _, prefetch_state = entry - comfy_modules = prefetch_state[1] + prefetched_module, comfy_modules = prefetch_state if comfy_modules is not None: - cleanup_prefetched_modules(comfy_modules) + cleanup_prefetched_modules(prefetched_module, comfy_modules) PREFETCH_QUEUES = [] + for module in GRAPH_MODULES: + _drop_graph(module) + GRAPH_MODULES.clear() + GRAPH_WARMED_MODULES.clear() + GRAPH_CAPTURE_STREAMS = {} -def prefetch_queue_pop(queue, device, module): +def prefetch_queue_pop(queue, device, module, dtype=None, core=None, enable_graph=False, generator=None): + enable_graph = enable_graph and not args.disable_cuda_graphs and comfy.model_management.is_device_cuda(device) and getattr(module, "_v_block", None) is not None if queue is None: + if core is not None: + core() return + capture_stream = None + if enable_graph: + capture_stream = GRAPH_CAPTURE_STREAMS.get(device) + if capture_stream is None: + capture_stream = torch.cuda.Stream(device=device) + GRAPH_CAPTURE_STREAMS[device] = capture_stream + + signature = None + graph_hit = False + graph = getattr(module, "_comfy_graph", None) if enable_graph else None + if graph is not None: + signature = comfy_aimdo.model_vbar.vbar_fault(module._v_block) + if signature is not None: + module._v_block_faulted = True + graph_hit = comfy_aimdo.model_vbar.vbar_signature_compare(signature, graph["signature"]) + consumed = queue.pop(0) if consumed is not None: offload_stream, prefetch_state = consumed if offload_stream is not None: offload_stream.wait_stream(comfy.model_management.current_stream(device)) - _, comfy_modules = prefetch_state + prefetched_module, comfy_modules = prefetch_state if comfy_modules is not None: - cleanup_prefetched_modules(comfy_modules) + cleanup_prefetched_modules(prefetched_module, comfy_modules) + if graph_hit: + queue[0] = (None, (module, [])) + graph["graph"].replay() + return + + fully_faulted = False prefetch = queue[0] if prefetch is not None: comfy_modules = [] - for s in prefetch.modules(): - if hasattr(s, "_v"): - comfy_modules.append(s) + prefetch_modules = prefetch if isinstance(prefetch, (list, tuple)) else (prefetch,) + for root in prefetch_modules: + for s in root.modules(): + if hasattr(s, "_v"): + comfy_modules.append(s) registerable_size = 0 for s in comfy_modules: @@ -59,11 +114,42 @@ def prefetch_queue_pop(queue, device, module): if lowvram_fn is not None: registerable_size += lowvram_fn.memory_required() - offload_stream = comfy.ops.cast_modules_with_vbar(comfy_modules, None, device, None, True) + offload_stream, fully_faulted = comfy.ops.cast_modules_with_vbar(comfy_modules, None, device, None, True, return_faulted=True) if not comfy.model_management.args.fast_disk: comfy.model_management.ensure_pin_registerable(registerable_size) comfy.model_management.sync_stream(device, offload_stream) - queue[0] = (offload_stream, (prefetch, comfy_modules)) + if fully_faulted and dtype is not None: + for comfy_module in comfy_modules: + comfy.ops.resolve_cast_module_with_vbar(comfy_module, dtype, device, dtype, None, False, return_weights=False) + queue[0] = (offload_stream, (module, comfy_modules)) + + if core is not None: + if enable_graph and fully_faulted and module in GRAPH_WARMED_MODULES: + if signature is None: + signature = comfy_aimdo.model_vbar.vbar_fault(module._v_block) + if signature is not None: + module._v_block_faulted = True + if signature is not None: + _drop_graph(module) + graph = torch.cuda.CUDAGraph() + if generator is not None: + graph.register_generator_state(generator) + capture_stream.wait_stream(comfy.model_management.current_stream(device)) + with torch.cuda.graph(graph, stream=capture_stream, capture_error_mode="thread_local"): + core() + comfy.model_management.current_stream(device).wait_stream(capture_stream) + graph.replay() + module._comfy_graph = {"graph": graph, "signature": signature} + GRAPH_MODULES.add(module) + return + if capture_stream is None: + core() + else: + capture_stream.wait_stream(comfy.model_management.current_stream(device)) + with torch.cuda.stream(capture_stream): + core() + comfy.model_management.current_stream(device).wait_stream(capture_stream) + GRAPH_WARMED_MODULES.add(module) def make_prefetch_queue(queue, device, transformer_options): if (not transformer_options.get("prefetch_dynamic_vbars", False) diff --git a/comfy/ops.py b/comfy/ops.py index 9ec44cfa2..ff64aad59 100644 --- a/comfy/ops.py +++ b/comfy/ops.py @@ -123,10 +123,12 @@ def materialize_meta_param(s, param_keys): # FIXME: add n=1 cache hit fast path -def cast_modules_with_vbar(comfy_modules, dtype, device, bias_dtype, non_blocking): +def cast_modules_with_vbar(comfy_modules, dtype, device, bias_dtype, non_blocking, return_faulted=False): offload_stream = None cast_buffer = None cast_buffer_offset = 0 + if return_faulted: + fully_faulted = all(not getattr(s, param_key + "_function", []) for s in comfy_modules for param_key in ("weight", "bias")) def ensure_offload_stream(module, required_size, check_largest): nonlocal offload_stream @@ -163,6 +165,8 @@ def cast_modules_with_vbar(comfy_modules, dtype, device, bias_dtype, non_blockin for s in comfy_modules: signature = comfy_aimdo.model_vbar.vbar_fault(s._v) resident = comfy_aimdo.model_vbar.vbar_signature_compare(signature, s._v_signature) + if return_faulted and (signature is None or not resident): + fully_faulted = False prefetch = { "signature": signature, "resident": resident, @@ -255,10 +259,12 @@ def cast_modules_with_vbar(comfy_modules, dtype, device, bias_dtype, non_blockin prefetch["needs_cast"] = needs_cast s._prefetch = prefetch + if return_faulted: + return offload_stream, fully_faulted return offload_stream -def resolve_cast_module_with_vbar(s, dtype, device, bias_dtype, compute_dtype, want_requant): +def resolve_cast_module_with_vbar(s, dtype, device, bias_dtype, compute_dtype, want_requant, return_weights=True): prefetch = getattr(s, "_prefetch", None) @@ -298,7 +304,7 @@ def resolve_cast_module_with_vbar(s, dtype, device, bias_dtype, compute_dtype, w tensor = tensor.dequantize() return tensor - if orig.dtype != dtype or len(fns) > 0: + if (return_weights and orig.dtype != dtype) or len(fns) > 0: x = to_dequant(x, dtype) if not resident and lowvram_fn is not None: x = to_dequant(x, dtype if compute_dtype is None else compute_dtype) @@ -325,7 +331,7 @@ def resolve_cast_module_with_vbar(s, dtype, device, bias_dtype, compute_dtype, w if prefetch["signature"] is not None: prefetch["resident"] = True - return weight, bias + return (weight, bias) if return_weights else None def cast_bias_weight(s, input=None, dtype=None, device=None, bias_dtype=None, offloadable=False, compute_dtype=None, want_requant=False): @@ -1633,7 +1639,7 @@ def mixed_precision_ops(quant_config={}, compute_dtype=torch.bfloat16, full_prec self.norm_type, self.scale_grad_by_freq, self.sparse) target_dtype = out_dtype if out_dtype is not None else weight._params.orig_dtype x = x.to(dtype=target_dtype) - if scale is not None and scale != 1.0: + if scale is not None: x = x * scale.to(dtype=target_dtype) return x diff --git a/comfy/quant_ops.py b/comfy/quant_ops.py index 6d9112dbb..18fd2d613 100644 --- a/comfy/quant_ops.py +++ b/comfy/quant_ops.py @@ -40,7 +40,7 @@ try: cuda_version = tuple(map(int, str(torch.version.cuda).split('.'))) if cuda_version < (13,): ck.registry.disable("cuda") - logging.warning("WARNING: You need pytorch with cu130 or higher to use optimized CUDA operations.") + logging.warning("WARNING: You need pytorch with cu130 or higher to use optimized CUDA operations.\nWARNING WARNING WARNING\nIf you are on nvidia 20 series and above it is required that you update your pytorch to cu130 or higher.\n") # On ROCm/AMD the CUDA backend is unavailable, so Triton is the only accelerated # comfy-kitchen backend. Enable it by default there, but only on Triton >= 3.7 AND a diff --git a/comfy/sample.py b/comfy/sample.py index 2be0cae5f..617816882 100644 --- a/comfy/sample.py +++ b/comfy/sample.py @@ -37,6 +37,11 @@ def prepare_noise(latent_image, seed, noise_inds=None): return noises +def prepare_empty_noise(latent_image): + if latent_image.is_nested: + return comfy.nested_tensor.NestedTensor([torch.zeros_like(t, device="cpu") for t in latent_image.unbind()]) + return torch.zeros_like(latent_image, device="cpu") + def fix_empty_latent_channels(model, latent_image, downscale_ratio_spacial=None, downscale_ratio_temporal=None): if latent_image.is_nested: return latent_image diff --git a/comfy/sd.py b/comfy/sd.py index 5fed4ca9a..4bdaa978c 100644 --- a/comfy/sd.py +++ b/comfy/sd.py @@ -11,6 +11,7 @@ from .ldm.cascade.stage_c_coder import StageC_coder from .ldm.audio.autoencoder import AudioOobleckVAE import comfy.ldm.genmo.vae.model import comfy.ldm.lightricks.vae.causal_video_autoencoder +import comfy.ldm.lightricks.vae.na_diffusion_decoder import comfy.ldm.lightricks.vae.audio_vae import comfy.ldm.cosmos.vae import comfy.ldm.wan.vae @@ -24,6 +25,7 @@ import comfy.ldm.cogvideo.vae import comfy.ldm.hunyuan_video.vae import comfy.ldm.mmaudio.vae.autoencoder import comfy.ldm.audio.vae_sa3 +import comfy.ldm.minimax_music.dav import comfy.pixel_space_convert import comfy.weight_adapter import yaml @@ -31,6 +33,7 @@ import math import os import comfy.utils +import comfy.ops from . import clip_vision from . import gligen @@ -73,6 +76,7 @@ import comfy.text_encoders.longcat_image import comfy.text_encoders.qwen35 import comfy.text_encoders.qwen3vl import comfy.text_encoders.minimax +import comfy.text_encoders.minimax_music import comfy.ldm.minimax.vae import comfy.ldm.minimax.audio_vae import comfy.text_encoders.boogu @@ -514,7 +518,22 @@ class VAE: self.audio_sample_rate = 44100 if config is None: - if "decoder.mid.block_1.mix_factor" in sd: + if "dec_in_proj.weight" in sd and "decoder.model.0.weight_g" in sd: # MiniMax Music3 DAV + self.first_stage_model = comfy.ldm.minimax_music.dav.MiniMaxMusic3DAV(operations=comfy.ops.disable_weight_init) + self.latent_channels = 128 + self.output_channels = 2 + self.upscale_ratio = 512 + self.downscale_ratio = 512 + self.latent_dim = 1 + self.process_output = lambda audio: audio + self.process_input = lambda audio: audio + self.working_dtypes = [torch.float32] + self.disable_offload = True + self.memory_used_decode = lambda shape, dtype: (shape[-1] * 512 * 1400 + 800_000_000) * model_management.dtype_size(dtype) + def _no_encode(*args, **kwargs): + raise RuntimeError("MiniMax Music3 DAV cannot encode audio") + self.memory_used_encode = _no_encode + elif "decoder.mid.block_1.mix_factor" in sd: encoder_config = {'double_z': True, 'z_channels': 4, 'resolution': 256, 'in_channels': 3, 'out_ch': 3, 'ch': 128, 'ch_mult': [1, 2, 4, 4], 'num_res_blocks': 2, 'attn_resolutions': [], 'dropout': 0.0} decoder_config = encoder_config.copy() decoder_config["video_kernel_size"] = [3, 1, 1] @@ -583,6 +602,22 @@ class VAE: self.working_dtypes = [torch.bfloat16, torch.float32] self.memory_used_encode = lambda shape, dtype: (400 * shape[2] * shape[3]) * model_management.dtype_size(dtype) self.memory_used_decode = lambda shape, dtype: (1000 * shape[2] * shape[3] * 16 * 16) * model_management.dtype_size(dtype) + elif "decoder.conv_in_x_t.weight" in sd: # lightricks LTX 2.4 diffusion VAE decoder + vae_config = None + if metadata is not None and "config" in metadata: + vae_config = json.loads(metadata["config"]).get("vae", None) + self.first_stage_model = comfy.ldm.lightricks.vae.na_diffusion_decoder.CausalDiffusionVAE(config=vae_config) + self.latent_channels = sd["decoder.conv_in.weight"].shape[1] + self.latent_dim = 3 + self.disable_offload = True + self.crop_input = False # generic crop would narrow the frame axis by the 32x spatial ratio + self.memory_used_decode = lambda shape, dtype: (1700 * shape[2] * shape[3] * shape[4] * (8 * 8 * 8)) * model_management.dtype_size(dtype) + self.memory_used_encode = lambda shape, dtype: (80 * max(shape[2], 7) * shape[3] * shape[4]) * model_management.dtype_size(dtype) + self.upscale_ratio = (lambda a: max(0, a * 8 - 7), 32, 32) + self.upscale_index_formula = (8, 32, 32) + self.downscale_ratio = (lambda a: max(0, math.floor((a + 7) / 8)), 32, 32) + self.downscale_index_formula = (8, 32, 32) + self.working_dtypes = [torch.bfloat16, torch.float32] elif "decoder.conv_in.weight" in sd: if sd['decoder.conv_in.weight'].shape[1] == 64: ddconfig = {"block_out_channels": [128, 256, 512, 512, 1024, 1024], "in_channels": 3, "out_channels": 3, "num_res_blocks": 2, "ffactor_spatial": 32, "downsample_match_channel": True, "upsample_match_channel": True} @@ -1222,16 +1257,48 @@ class VAE: tile = 256 // self.spacial_compression_decode() overlap = tile // 4 if self.handles_tiling: + memory_used = self.memory_used_decode(self._tile_bounded_shape(samples_in.shape, tile, tile, None), self.vae_dtype) + model_management.load_models_gpu([self.patcher], memory_required=memory_used, force_full_load=self.disable_offload) pixel_samples = self._decode_tiled_owned(samples_in, tile_x=tile, tile_y=tile, overlap=overlap) else: - pixel_samples = self.decode_tiled_3d(samples_in, tile_x=tile, tile_y=tile, overlap=(1, overlap, overlap)) + # Reserve as much as an untiled decode could use (capped by what the device can provide), then size the tiles to fill that reservation: + # shrink the temporal tile until one tile fits, then grow the spatial tile while it still fits. + budget = min(memory_used, int(model_management.get_total_memory(self.device) * 0.8)) + model_management.load_models_gpu([self.patcher], memory_required=budget, force_full_load=self.disable_offload) + tile_t = samples_in.shape[2] + est = lambda tt, txy: self.memory_used_decode(self._tile_bounded_shape(samples_in.shape, txy, txy, tt), self.vae_dtype) + while tile_t > 2 and est(tile_t, tile) > budget: + tile_t = -(-tile_t // 2) + while tile * 2 <= max(samples_in.shape[3], samples_in.shape[4]) and est(tile_t, tile * 2) <= budget: + tile *= 2 + overlap = tile // 4 + pixel_samples = self.decode_tiled_3d(samples_in, tile_t=tile_t, tile_x=tile, tile_y=tile, overlap=(1, overlap, overlap)) pixel_samples = pixel_samples.to(self.output_device).movedim(1,-1) return pixel_samples + def _tile_bounded_shape(self, shape, tile_x, tile_y, tile_t): + """Clamp a latent shape to one tile for memory estimates: peak memory of a tiled decode is per-tile. Only caller-provided tile dims are clamped.""" + s = list(shape) + if len(s) == 5: + if tile_t is not None: + s[2] = min(s[2], tile_t) + if tile_y is not None: + s[3] = min(s[3], tile_y) + if tile_x is not None: + s[4] = min(s[4], tile_x) + elif len(s) == 4 and self.extra_1d_channel is None: + if tile_y is not None: + s[2] = min(s[2], tile_y) + if tile_x is not None: + s[3] = min(s[3], tile_x) + elif tile_x is not None: + s[-1] = min(s[-1], tile_x) + return tuple(s) + def decode_tiled(self, samples, tile_x=None, tile_y=None, overlap=None, tile_t=None, overlap_t=None): self.throw_exception_if_invalid() - memory_used = self.memory_used_decode(samples.shape, self.vae_dtype) #TODO: calculate mem required for tile + memory_used = self.memory_used_decode(self._tile_bounded_shape(samples.shape, tile_x, tile_y, tile_t), self.vae_dtype) model_management.load_models_gpu([self.patcher], memory_required=memory_used, force_full_load=self.disable_offload) dims = samples.ndim - 2 args = {} @@ -1643,7 +1710,16 @@ def load_text_encoder_state_dicts(state_dicts=[], embedding_directory=None, clip clip_target.params = {} if len(clip_data) == 1: te_model = detect_te_model(clip_data[0]) - if te_model == TEModel.CLIP_G: + if clip_type == CLIPType.MINIMAX and "model.audio_decoder.projection.weight" in clip_data[0]: + tokenizer_data["tokenizer_json"] = clip_data[0].pop("tokenizer_json", None) + quant = comfy.utils.detect_layer_quantization(clip_data[0], "") + if quant is not None: + model_options = model_options.copy() + model_options["quantization_metadata"] = quant + clip_target.params["projection_config"] = comfy.text_encoders.minimax_music.detect_merged_config(clip_data[0]) + clip_target.clip = comfy.text_encoders.minimax_music.MiniMaxMusic3TEModel + clip_target.tokenizer = comfy.text_encoders.minimax_music.MiniMaxMusic3Tokenizer + elif te_model == TEModel.CLIP_G: if clip_type == CLIPType.STABLE_CASCADE: clip_target.clip = sdxl_clip.StableCascadeClipModel clip_target.tokenizer = sdxl_clip.StableCascadeTokenizer @@ -1702,12 +1778,21 @@ def load_text_encoder_state_dicts(state_dicts=[], embedding_directory=None, clip clip_target.tokenizer = comfy.text_encoders.sa3.SAT5GemmaTokenizer tokenizer_data["spiece_model"] = clip_data[0].get("spiece_model", None) elif te_model in (TEModel.GEMMA_4_E4B, TEModel.GEMMA_4_E2B, TEModel.GEMMA_4_31B, TEModel.GEMMA_4_12B): - variant = {TEModel.GEMMA_4_E4B: comfy.text_encoders.gemma4.Gemma4_E4B, - TEModel.GEMMA_4_E2B: comfy.text_encoders.gemma4.Gemma4_E2B, - TEModel.GEMMA_4_31B: comfy.text_encoders.gemma4.Gemma4_31B, - TEModel.GEMMA_4_12B: comfy.text_encoders.gemma4.Gemma4_12B}[te_model] - clip_target.clip = comfy.text_encoders.gemma4.gemma4_te(**llama_detect(clip_data), model_class=variant) - clip_target.tokenizer = variant.tokenizer + if te_model == TEModel.GEMMA_4_12B and "text_embedding_projection.video_aggregate_embed.weight" in clip_data[0]: + clip_target.clip = comfy.text_encoders.lt.ltxav_te( + **llama_detect(clip_data), + **comfy.text_encoders.lt.sd_detect(clip_data), + text_encoder_model=comfy.text_encoders.gemma4.gemma4_text_encoder_model(comfy.text_encoders.gemma4.Gemma4_12B), + text_encoder_key="gemma4", + ) + clip_target.tokenizer = comfy.text_encoders.lt.ltxav_gemma4_tokenizer(comfy.text_encoders.gemma4.Gemma4_12B.tokenizer) + else: + variant = {TEModel.GEMMA_4_E4B: comfy.text_encoders.gemma4.Gemma4_E4B, + TEModel.GEMMA_4_E2B: comfy.text_encoders.gemma4.Gemma4_E2B, + TEModel.GEMMA_4_31B: comfy.text_encoders.gemma4.Gemma4_31B, + TEModel.GEMMA_4_12B: comfy.text_encoders.gemma4.Gemma4_12B}[te_model] + clip_target.clip = comfy.text_encoders.gemma4.gemma4_te(**llama_detect(clip_data), model_class=variant) + clip_target.tokenizer = variant.tokenizer tokenizer_data["tokenizer_json"] = clip_data[0].get("tokenizer_json", None) elif te_model == TEModel.GEMMA_2_2B: if clip_type == CLIPType.PIXELDIT: @@ -1875,9 +1960,30 @@ def load_text_encoder_state_dicts(state_dicts=[], embedding_directory=None, clip clip_target.clip = comfy.text_encoders.kandinsky5.te(**llama_detect(clip_data)) clip_target.tokenizer = comfy.text_encoders.kandinsky5.Kandinsky5TokenizerImage elif clip_type == CLIPType.LTXV: - clip_target.clip = comfy.text_encoders.lt.ltxav_te(**llama_detect(clip_data), **comfy.text_encoders.lt.sd_detect(clip_data)) - clip_target.tokenizer = comfy.text_encoders.lt.LTXAVGemmaTokenizer - tokenizer_data["spiece_model"] = clip_data[0].get("spiece_model", None) + te_models = [detect_te_model(sd) for sd in clip_data] + gemma4_models = { + TEModel.GEMMA_4_E4B: comfy.text_encoders.gemma4.Gemma4_E4B, + TEModel.GEMMA_4_E2B: comfy.text_encoders.gemma4.Gemma4_E2B, + TEModel.GEMMA_4_31B: comfy.text_encoders.gemma4.Gemma4_31B, + TEModel.GEMMA_4_12B: comfy.text_encoders.gemma4.Gemma4_12B, + } + gemma4_type = next((model for model in te_models if model in gemma4_models), None) + if gemma4_type is None: + clip_target.clip = comfy.text_encoders.lt.ltxav_te(**llama_detect(clip_data), **comfy.text_encoders.lt.sd_detect(clip_data)) + clip_target.tokenizer = comfy.text_encoders.lt.LTXAVGemmaTokenizer + gemma_sd = clip_data[te_models.index(TEModel.GEMMA_3_12B)] if TEModel.GEMMA_3_12B in te_models else clip_data[0] + tokenizer_data["spiece_model"] = gemma_sd.get("spiece_model", None) + else: + variant = gemma4_models[gemma4_type] + clip_target.clip = comfy.text_encoders.lt.ltxav_te( + **llama_detect(clip_data), + **comfy.text_encoders.lt.sd_detect(clip_data), + text_encoder_model=comfy.text_encoders.gemma4.gemma4_text_encoder_model(variant), + text_encoder_key="gemma4", + ) + clip_target.tokenizer = comfy.text_encoders.lt.ltxav_gemma4_tokenizer(variant.tokenizer) + gemma_sd = clip_data[te_models.index(gemma4_type)] + tokenizer_data["tokenizer_json"] = gemma_sd.get("tokenizer_json", None) elif clip_type == CLIPType.NEWBIE: clip_target.clip = comfy.text_encoders.newbie.te(**llama_detect(clip_data)) clip_target.tokenizer = comfy.text_encoders.newbie.NewBieTokenizer diff --git a/comfy/supported_models.py b/comfy/supported_models.py index b9952db55..33c378435 100644 --- a/comfy/supported_models.py +++ b/comfy/supported_models.py @@ -16,6 +16,7 @@ import comfy.text_encoders.genmo import comfy.text_encoders.lt import comfy.text_encoders.hunyuan_video import comfy.text_encoders.minimax +import comfy.text_encoders.minimax_music import comfy.text_encoders.cosmos import comfy.text_encoders.lumina2 import comfy.text_encoders.wan @@ -2200,6 +2201,28 @@ class ACEStep15(supported_models_base.BASE): return supported_models_base.ClipTarget(comfy.text_encoders.ace15.ACE15Tokenizer, comfy.text_encoders.ace15.te(**detect)) +class MiniMaxMusic3(supported_models_base.BASE): + unet_config = { + "audio_model": "minimax_music3", + } + + latent_format = comfy.latent_formats.MiniMaxMusic3 + memory_usage_factor = 2.0 + supported_inference_dtypes = [torch.float16, torch.bfloat16, torch.float32] + sampling_settings = {"multiplier": 1.0} + + def get_model(self, state_dict, prefix="", device=None): + return model_base.MiniMaxMusic3(self, device=device) + + def model_type(self, state_dict, prefix=""): + return model_base.ModelType.FLOW + + def clip_target(self, state_dict={}): + detect = comfy.text_encoders.minimax_music.detect_merged_config(state_dict, self.text_encoder_key_prefix[0]) + target = supported_models_base.ClipTarget(comfy.text_encoders.minimax_music.MiniMaxMusic3Tokenizer, comfy.text_encoders.minimax_music.MiniMaxMusic3TEModel) + target.params["projection_config"] = detect + return target + class LongCatImage(supported_models_base.BASE): unet_config = { @@ -2494,6 +2517,7 @@ models = [ ChromaRadiance, ACEStep, ACEStep15, + MiniMaxMusic3, Omnigen2, Boogu, MageFlow, diff --git a/comfy/text_encoders/bpe_tokenizer.py b/comfy/text_encoders/bpe_tokenizer.py new file mode 100644 index 000000000..e49e36ca0 --- /dev/null +++ b/comfy/text_encoders/bpe_tokenizer.py @@ -0,0 +1,333 @@ +""" +Pure-Python byte-level BPE tokenizer. +Supports loading from HuggingFace tokenizer.json (LLaMA-style) +and from Mistral tekken JSON blobs. +No dependency on the `transformers`, `tokenizers`, or `regex` packages. +""" +import base64 +import json +import os +import re +import unicodedata + + +# This is also the default pattern used by the previous MistralConverter path. +_LLAMA_PATTERN = r"""(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\r\n\p{L}\p{N}]?\p{L}+|\p{N}{1,3}| ?[^\s\p{L}\p{N}]+[\r\n]*|\s*[\r\n]+|\s+(?!\S)|\s+""" +_CONTRACTIONS = ("'re", "'ve", "'ll", "'s", "'t", "'m", "'d") + + +def _is_letter(c): + return unicodedata.category(c)[0] == "L" + + +def _is_number(c): + return unicodedata.category(c)[0] == "N" + + +def _is_whitespace(c): + return c in " \t\n\r\v\f\x85\u2028\u2029" or unicodedata.category(c) == "Zs" + + +def _split_llama(text): + pieces = [] + i = 0 + while i < len(text): + contraction = None + if text[i] == "'": + for suffix in _CONTRACTIONS: + if text[i:i + len(suffix)].casefold() == suffix: + contraction = text[i:i + len(suffix)] + break + if contraction is not None: + pieces.append(contraction) + i += len(contraction) + continue + + j = i + if text[j] not in "\r\n" and not _is_letter(text[j]) and not _is_number(text[j]): + j += 1 + if j < len(text) and _is_letter(text[j]): + j += 1 + while j < len(text) and _is_letter(text[j]): + j += 1 + pieces.append(text[i:j]) + i = j + continue + + if _is_number(text[i]): + j = i + 1 + while j < len(text) and j - i < 3 and _is_number(text[j]): + j += 1 + pieces.append(text[i:j]) + i = j + continue + + j = i + if text[j] == " ": + j += 1 + punct_start = j + while j < len(text) and not _is_whitespace(text[j]) and not _is_letter(text[j]) and not _is_number(text[j]): + j += 1 + if j > punct_start: + while j < len(text) and text[j] in "\r\n": + j += 1 + pieces.append(text[i:j]) + i = j + continue + + if _is_whitespace(text[i]): + j = i + 1 + while j < len(text) and _is_whitespace(text[j]): + j += 1 + last_newline = max(text.rfind("\r", i, j), text.rfind("\n", i, j)) + if last_newline >= i: + j = last_newline + 1 + elif j < len(text) and j - i > 1: + j -= 1 + pieces.append(text[i:j]) + i = j + continue + + pieces.append(text[i]) + i += 1 + return pieces + + +def _make_split_pattern(pattern_str): + if pattern_str != _LLAMA_PATTERN: + raise ValueError(f"Unsupported tokenizer split pattern: {pattern_str}") + return _split_llama + + +def _bytes_to_unicode(): + bs = (list(range(ord("!"), ord("~") + 1)) + + list(range(ord("¡"), ord("¬") + 1)) + + list(range(ord("®"), ord("ÿ") + 1))) + cs = bs[:] + n = 0 + for b in range(2**8): + if b not in bs: + bs.append(b) + cs.append(2**8 + n) + n += 1 + cs = [chr(n) for n in cs] + return dict(zip(bs, cs)) + + +class BPETokenizer: + """Byte-level BPE tokenizer with optional BOS prepending.""" + + def __init__(self, vocab, merges_by_pair, special_token_ids, pattern_str, + byte_encoder, byte_decoder, bos_id=None): + self._vocab = vocab # str -> int + self._inv_vocab = {v: k for k, v in vocab.items()} + self._merges = merges_by_pair # (str, str) -> priority int + self._special_token_ids = special_token_ids # str -> int + self._special_ids = set(special_token_ids.values()) + self._byte_encoder = byte_encoder + self._byte_decoder = byte_decoder + self._bos_id = bos_id + + self._split = _make_split_pattern(pattern_str) + sorted_specials = sorted(special_token_ids.keys(), key=len, reverse=True) + if sorted_specials: + self._special_split = re.compile( + '(' + '|'.join(re.escape(s) for s in sorted_specials) + ')' + ) + else: + self._special_split = None + + def _bpe_encode_piece(self, chars): + if len(chars) <= 1: + return chars + while True: + min_rank = float('inf') + best_pair = None + for i in range(len(chars) - 1): + r = self._merges.get((chars[i], chars[i + 1]), float('inf')) + if r < min_rank: + min_rank = r + best_pair = (chars[i], chars[i + 1]) + if best_pair is None: + break + merged = best_pair[0] + best_pair[1] + new_chars = [] + i = 0 + while i < len(chars): + if i < len(chars) - 1 and chars[i] == best_pair[0] and chars[i + 1] == best_pair[1]: + new_chars.append(merged) + i += 2 + else: + new_chars.append(chars[i]) + i += 1 + chars = new_chars + if len(chars) == 1: + break + return chars + + def _encode_raw(self, text): + ids = [] + parts = self._special_split.split(text) if self._special_split else [text] + for part in parts: + if not part: + continue + if part in self._special_token_ids: + ids.append(self._special_token_ids[part]) + else: + for piece in self._split(part): + byte_chars = [self._byte_encoder[b] for b in piece.encode('utf-8')] + for tok in self._bpe_encode_piece(byte_chars): + ids.append(self._vocab[tok]) + return ids + + def __call__(self, text): + ids = self._encode_raw(text) + if self._bos_id is not None: + ids = [self._bos_id] + ids + return {"input_ids": ids} + + def get_vocab(self): + return dict(self._vocab) + + def decode(self, token_ids, skip_special_tokens=True): + buf = bytearray() + for tid in token_ids: + s = self._inv_vocab.get(tid, '') + if tid in self._special_ids: + if not skip_special_tokens: + buf.extend(s.encode('utf-8')) + else: + for c in s: + buf.append(self._byte_decoder[c]) + return buf.decode('utf-8', errors='replace') + + +def _extract_pattern(pretok): + if pretok.get('type') == 'Sequence': + for sub in pretok.get('pretokenizers', []): + if sub.get('type') == 'Split': + pat = sub.get('pattern', {}) + if 'Regex' in pat: + return pat['Regex'] + elif pretok.get('type') == 'Split': + pat = pretok.get('pattern', {}) + if 'Regex' in pat: + return pat['Regex'] + return None + + +def _extract_bos_id(post_processor, special_token_ids): + if post_processor.get('type') == 'TemplateProcessing': + single = post_processor.get('single', []) + if single and 'SpecialToken' in single[0]: + bos_str = single[0]['SpecialToken']['id'] + return special_token_ids.get(bos_str) + return None + + +def from_tokenizer_json(path): + """Load a BPETokenizer from a directory containing tokenizer.json.""" + tok_file = os.path.join(path, 'tokenizer.json') + with open(tok_file, encoding='utf-8') as f: + data = json.load(f) + + vocab = dict(data['model']['vocab']) # str -> int + + merges_by_pair = {} + for i, merge_str in enumerate(data['model'].get('merges', [])): + a, b = merge_str.split(' ', 1) + if (a, b) not in merges_by_pair: + merges_by_pair[(a, b)] = i + + special_token_ids = {} + for tok in data.get('added_tokens', []): + special_token_ids[tok['content']] = tok['id'] + vocab[tok['content']] = tok['id'] # include in vocab for inv_vocab decode + + pattern = _extract_pattern(data.get('pre_tokenizer', {})) + if pattern is None: + raise ValueError(f"Could not extract regex pattern from {tok_file}") + + bos_id = _extract_bos_id(data.get('post_processor', {}), special_token_ids) + + byte_encoder = _bytes_to_unicode() + byte_decoder = {v: k for k, v in byte_encoder.items()} + + return BPETokenizer(vocab, merges_by_pair, special_token_ids, pattern, + byte_encoder, byte_decoder, bos_id=bos_id) + + +def from_tekken_json(data): + """Build a BPETokenizer from a Mistral tekken JSON blob (bytes or str).""" + mistral_vocab = json.loads(data) + config = mistral_vocab["config"] + + byte_encoder = _bytes_to_unicode() + byte_decoder = {v: k for k, v in byte_encoder.items()} + + def tbts(b): + return "".join(byte_encoder[ord(c)] for c in b.decode("latin-1")) + + special_token_offset = config["default_num_special_tokens"] + max_vocab = config["default_vocab_size"] - special_token_offset + + raw_vocab = {} + for w in mistral_vocab["vocab"]: + r = w["rank"] + if r >= max_vocab: + continue + raw_vocab[base64.b64decode(w["token_bytes"])] = r + special_token_offset + + special_tokens_dict = {} + for w in mistral_vocab["special_tokens"]: + if "token_bytes" in w: + special_tokens_dict[base64.b64decode(w["token_bytes"])] = w["rank"] + else: + special_tokens_dict[w["token_str"]] = w["rank"] + + all_special = list(special_tokens_dict.keys()) + combined = dict(special_tokens_dict) + combined.update(raw_vocab) + + bpe_vocab = {} + merge_triples = [] + for token, rank in combined.items(): + if token not in all_special: + bpe_vocab[tbts(token)] = rank + if len(token) == 1: + continue + local = [] + for i in range(1, len(token)): + pl, pr = token[:i], token[i:] + if pl in combined and pr in combined and (pl + pr) in combined: + local.append((pl, pr, rank)) + local.sort(key=lambda x: (combined[x[0]], combined[x[1]])) + merge_triples.extend(local) + else: + tok_str = token.decode("utf-8", errors="replace") if isinstance(token, bytes) else token + bpe_vocab[tok_str] = rank + + merge_triples.sort(key=lambda v: v[2]) + + merges_by_pair = {} + for i, (pl, pr, _) in enumerate(merge_triples): + pair = (tbts(pl), tbts(pr)) + if pair not in merges_by_pair: + merges_by_pair[pair] = i + + special_str_ids = {} + for tok in all_special: + tok_str = tok.decode("utf-8", errors="replace") if isinstance(tok, bytes) else tok + if tok_str in bpe_vocab: + special_str_ids[tok_str] = bpe_vocab[tok_str] + + return BPETokenizer(bpe_vocab, merges_by_pair, special_str_ids, _LLAMA_PATTERN, + byte_encoder, byte_decoder, bos_id=None) + + +class LlamaTokenizerFast: + """Drop-in replacement for transformers.LlamaTokenizerFast (read-only use).""" + + @staticmethod + def from_pretrained(path, **kwargs): + return from_tokenizer_json(path) diff --git a/comfy/text_encoders/flux.py b/comfy/text_encoders/flux.py index d5eb91dcb..fbdb1d13a 100644 --- a/comfy/text_encoders/flux.py +++ b/comfy/text_encoders/flux.py @@ -3,11 +3,10 @@ import comfy.text_encoders.t5 import comfy.text_encoders.sd3_clip import comfy.text_encoders.llama import comfy.model_management -from transformers import T5TokenizerFast, LlamaTokenizerFast, Qwen2Tokenizer +from transformers import T5TokenizerFast, Qwen2Tokenizer +from .bpe_tokenizer import from_tekken_json import torch import os -import json -import base64 class T5XXLTokenizer(sd1_clip.SDTokenizer): def __init__(self, embedding_directory=None, tokenizer_data={}): @@ -75,45 +74,13 @@ def flux_clip(dtype_t5=None, t5_quantization_metadata=None): def load_mistral_tokenizer(data): if torch.is_tensor(data): data = data.numpy().tobytes() + return {"tokenizer_object": from_tekken_json(data)} - try: - from transformers.integrations.mistral import MistralConverter - except ModuleNotFoundError: - from transformers.models.pixtral.convert_pixtral_weights_to_hf import MistralConverter - - mistral_vocab = json.loads(data) - - special_tokens = {} - vocab = {} - - max_vocab = mistral_vocab["config"]["default_vocab_size"] - max_vocab -= len(mistral_vocab["special_tokens"]) - - for w in mistral_vocab["vocab"]: - r = w["rank"] - if r >= max_vocab: - continue - - vocab[base64.b64decode(w["token_bytes"])] = r - - for w in mistral_vocab["special_tokens"]: - if "token_bytes" in w: - special_tokens[base64.b64decode(w["token_bytes"])] = w["rank"] - else: - special_tokens[w["token_str"]] = w["rank"] - - all_special = [] - for v in special_tokens: - all_special.append(v) - - special_tokens.update(vocab) - vocab = special_tokens - return {"tokenizer_object": MistralConverter(vocab=vocab, additional_special_tokens=all_special).converted(), "legacy": False} class MistralTokenizerClass: @staticmethod - def from_pretrained(path, **kwargs): - return LlamaTokenizerFast(**kwargs) + def from_pretrained(path, tokenizer_object=None, **kwargs): + return tokenizer_object class Mistral3Tokenizer(sd1_clip.SDTokenizer): def __init__(self, embedding_directory=None, embedding_size=5120, embedding_key='mistral3_24b', tokenizer_data={}): diff --git a/comfy/text_encoders/gemma4.py b/comfy/text_encoders/gemma4.py index 5163c1676..61bc3a3f0 100644 --- a/comfy/text_encoders/gemma4.py +++ b/comfy/text_encoders/gemma4.py @@ -9,10 +9,12 @@ import math from comfy import sd1_clip import comfy.model_management +import comfy.model_prefetch import comfy.ops +import comfy.quant_ops from comfy.ldm.modules.attention import optimized_attention_for_device from comfy.rmsnorm import rms_norm -from comfy.text_encoders.llama import RMSNorm, MLP, BaseLlama, BaseGenerate, _make_scaled_embedding +from comfy.text_encoders.llama import RMSNorm, MLP, BaseLlama, BaseGenerate, FixedKV, _make_scaled_embedding # Intentional minor divergences from transformers -reference implementation: @@ -109,7 +111,28 @@ class Gemma4_12B_Config(Gemma4Config): suppress_tokens = [258883, 258882] -# unfused RoPE as addcmul_ RoPE diverges from reference code +class RingKV(FixedKV): + # sliding-window ring: writes wrap at capacity, validity saturates + def prepare(self, num_tokens): + capacity = self.key.shape[2] + self.position.fill_(self.index % capacity) + self.seqlen.fill_(min(self.index + num_tokens, capacity)) + + +def _fixed_kv_decode_mask(mask, cache, min_val): + capacity = cache.key.shape[2] + valid = min(cache.index + 1, capacity) + output = mask.new_full((*mask.shape[:-1], capacity), min_val) + if isinstance(cache, RingKV): + positions = torch.arange(cache.index + 1 - valid, cache.index + 1, device=mask.device) % capacity + output.index_copy_(-1, positions, mask[..., -valid:]) + else: + output[..., :valid] = mask[..., :valid] + return output + + +# unfused RoPE as addcmul_ RoPE diverges from reference code (vision only; text +# layers use the kitchen split-half kernel, bitwise-equal to this with bf16 freqs) def _apply_rotary_pos_emb(x, freqs_cis): cos, sin = freqs_cis[0], freqs_cis[1] half = x.shape[-1] // 2 @@ -140,6 +163,23 @@ class Gemma4Attention(nn.Module): if config.k_norm == "gemma3": self.k_norm = RMSNorm(head_dim, eps=config.rms_norm_eps, device=device, dtype=dtype) + def _decode_attention(self, xq, cache, bias): + if bias is None: + # eager decode: slice the cache to the valid length (python-side index, + # no mask needed; a full ring is order-invariant under softmax) + n = min(cache.index + 1, cache.key.shape[2]) + gqa_kwargs = {"enable_gqa": True} if self.num_heads != self.num_kv_heads else {} + attention = optimized_attention_for_device(xq.device, mask=False, small_input=True) + return attention(xq, cache.key[:, :, :n], cache.value[:, :, :n], self.num_heads, skip_reshape=True, scale=1.0, **gqa_kwargs) + # graph capture: fixed-length masked attention over the full capacity, explicit + # math (SDPA leaves its fast path on broadcast-bias + GQA and costs ~0.5ms/layer) + batch_size = xq.shape[0] + groups = self.num_heads // self.num_kv_heads + q = xq.reshape(batch_size, self.num_kv_heads, groups, self.head_dim) + scores = q @ cache.key.transpose(-1, -2) + bias + probs = torch.softmax(scores.float(), dim=-1).to(xq.dtype) + return (probs @ cache.value).reshape(batch_size, 1, self.inner_size) + def forward( self, hidden_states: torch.Tensor, @@ -156,10 +196,16 @@ class Gemma4Attention(nn.Module): if self.q_norm is not None: xq = self.q_norm(xq) + if isinstance(shared_kv, FixedKV): + # decode on a KV-shared layer: attend the source layer's fixed cache + xq = comfy.quant_ops.ck.apply_rope_split_half1(xq, freqs_cis) + output = self._decode_attention(xq, shared_kv, attention_mask) + return self.o_proj(output), None, None + if shared_kv is not None: xk, xv = shared_kv # Apply RoPE to Q only (K already has RoPE from source layer) - xq = _apply_rotary_pos_emb(xq, freqs_cis) + xq = comfy.quant_ops.ck.apply_rope_split_half1(xq, freqs_cis) present_key_value = None shareable_kv = None else: @@ -173,11 +219,39 @@ class Gemma4Attention(nn.Module): xv = rms_norm(xv) xk = xk.transpose(1, 2) xv = xv.transpose(1, 2) - xq = _apply_rotary_pos_emb(xq, freqs_cis) - xk = _apply_rotary_pos_emb(xk, freqs_cis) + xq = comfy.quant_ops.ck.apply_rope_split_half1(xq, freqs_cis) + xk = comfy.quant_ops.ck.apply_rope_split_half1(xk, freqs_cis) present_key_value = None - if past_key_value is not None: + fixed_cache = past_key_value if isinstance(past_key_value, FixedKV) else None + if fixed_cache is not None: + if seq_length == 1: + # CUDA-graphable decode: write at the device-side ring/linear position + fixed_cache.key.index_copy_(2, fixed_cache.position, xk) + fixed_cache.value.index_copy_(2, fixed_cache.position, xv) + output = self._decode_attention(xq, fixed_cache, attention_mask) + return self.o_proj(output), fixed_cache, None + + # prefill: attend the local sequence, persist the tail into the cache + capacity = fixed_cache.key.shape[2] + index = fixed_cache.index + if index + seq_length <= capacity: + fixed_cache.key[:, :, index:index + seq_length] = xk + fixed_cache.value[:, :, index:index + seq_length] = xv + if index > 0: + xk = fixed_cache.key[:, :, :index + seq_length] + xv = fixed_cache.value[:, :, :index + seq_length] + elif index == 0: + # prefill longer than the sliding ring: attend the full local K/V + # (per-query windows come from the prefill sliding mask), cache only + # the last `capacity` keys at their wrapped slots (position % capacity) + slots = torch.arange(seq_length - capacity, seq_length, device=xk.device) % capacity + fixed_cache.key.index_copy_(2, slots, xk[:, :, -capacity:]) + fixed_cache.value.index_copy_(2, slots, xv[:, :, -capacity:]) + else: + raise RuntimeError("gemma4: chunked prefill past the sliding window is not supported") + present_key_value = fixed_cache + elif past_key_value is not None: cumulative_len = 0 if len(past_key_value) > 0: past_key, past_value, cumulative_len = past_key_value @@ -245,6 +319,7 @@ class TransformerBlockGemma4(nn.Module): self.register_buffer("layer_scalar", torch.empty(1, device=device, dtype=dtype)) def forward(self, x, attention_mask=None, freqs_cis=None, past_key_value=None, per_layer_input=None, shared_kv=None): + output = x sliding_window = None if self.sliding_attention: sliding_window = self.sliding_attention @@ -281,7 +356,8 @@ class TransformerBlockGemma4(nn.Module): x = self.post_per_layer_input_norm(x) x = residual + x - x = x * comfy.ops.cast_to_input(self.layer_scalar, x) + # in-place into the input buffer so CUDA-graph replays land in the static x + x = torch.mul(x, comfy.ops.cast_to_input(self.layer_scalar, x), out=output) return x, present_key_value, shareable_kv @@ -290,6 +366,9 @@ class Gemma4Transformer(nn.Module): def __init__(self, config, device=None, dtype=None, ops=None): super().__init__() self.config = config + self.fixed_kv = True + self.prefetch_dynamic_vbars = True + self.graph_dynamic_vbar_blocks = True self.embed_tokens = _make_scaled_embedding(ops, config.vocab_size, config.hidden_size, config.hidden_size ** 0.5, device, dtype) @@ -298,6 +377,19 @@ class Gemma4Transformer(nn.Module): for i in range(config.num_hidden_layers) ]) + # KV-shared layers never run k_proj/v_proj/k_norm: their never-resolved vbar + # signatures would block layer graph capture, so prefetch only what executes + first_kv_shared = config.num_hidden_layers - config.num_kv_shared_layers if config.num_kv_shared_layers > 0 else config.num_hidden_layers + self._prefetch_units = [] + for i, layer in enumerate(self.layers): + if i >= first_kv_shared: + dead = {layer.self_attn.k_proj, layer.self_attn.v_proj, layer.self_attn.k_norm} + self._prefetch_units.append([ + m for m in layer.modules() if next(m.children(), None) is None and m not in dead + ]) + else: + self._prefetch_units.append(layer) + self.norm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps, device=device, dtype=dtype) if config.final_norm else None # Precompute RoPE inv_freq on CPU to match reference code's exact value @@ -311,6 +403,9 @@ class Gemma4Transformer(nn.Module): sliding_inv = 1.0 / (config.rope_theta[1] ** (torch.arange(0, config.head_dim, 2).float() / config.head_dim)) self.register_buffer("_sliding_inv_freq", sliding_inv, persistent=False) + if config.suppress_tokens: + self.register_buffer("_suppress_tokens", torch.tensor(config.suppress_tokens, dtype=torch.long), persistent=False) + # Per-layer input mechanism self.hidden_size_per_layer_input = config.hidden_size_per_layer_input if self.hidden_size_per_layer_input: @@ -322,19 +417,26 @@ class Gemma4Transformer(nn.Module): self.hidden_size_per_layer_input, eps=config.rms_norm_eps, device=device, dtype=dtype) + def get_dynamic_vram__units(self): + return (list(self.layers), []) if self.graph_dynamic_vbar_blocks else ([], []) + def get_past_len(self, past_key_values): for kv in past_key_values: + if isinstance(kv, FixedKV): + return kv.index if len(kv) >= 3: return kv[2] return 0 def _freqs_from_inv(self, inv_freq, position_ids, device, dtype): - """Compute cos/sin from stored inv_freq""" + """Compute per-pair 2x2 rotation matrices [B, 1, S, d/2, 2, 2] from stored inv_freq""" inv_exp = inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1).to(device) pos_exp = position_ids[:, None, :].float() freqs = (inv_exp @ pos_exp).transpose(1, 2) - emb = torch.cat((freqs, freqs), dim=-1) - return emb.cos().unsqueeze(1).to(dtype), emb.sin().unsqueeze(1).to(dtype) + cos, sin = freqs.cos(), freqs.sin() + mat = torch.stack((torch.stack((cos, -sin), dim=-1), + torch.stack((sin, cos), dim=-1)), dim=-2) + return mat.unsqueeze(1).to(dtype) def compute_freqs_cis(self, position_ids, device, dtype=None): global_freqs = self._freqs_from_inv(self._global_inv_freq, position_ids, device, dtype) @@ -401,6 +503,71 @@ class Gemma4Transformer(nn.Module): first_kv_shared = self.config.num_hidden_layers - num_kv_shared if num_kv_shared > 0 else self.config.num_hidden_layers shared_sliding_kv = None # KV from last non-shared sliding layer shared_global_kv = None # KV from last non-shared global layer + share_source = {} + if num_kv_shared > 0: + for i in range(first_kv_shared): + share_source[bool(self.layers[i].sliding_attention)] = i + + prefetch_queue = comfy.model_prefetch.make_prefetch_queue( + list(self._prefetch_units), x.device, + {"prefetch_dynamic_vbars": self.prefetch_dynamic_vbars and past_key_values is not None}) + + fixed_kv = (past_key_values is not None and len(past_key_values) > 0 + and isinstance(past_key_values[0], FixedKV)) + decode = fixed_kv and seq_len == 1 + # mirror the conditions under which prefetch_queue_pop can actually capture, so + # eager fallbacks keep the sliced decode path instead of the full-capacity one + enable_graph = (decode and mask is None and self.graph_dynamic_vbar_blocks + and prefetch_queue is not None + and hasattr(self.layers[0], "_v_block") + and not comfy.model_management.args.disable_cuda_graphs + and comfy.model_management.is_device_cuda(x.device)) + decode_bias = None + decode_masks = None + if decode: + prepared = set() + for kv in past_key_values: + if isinstance(kv, FixedKV) and id(kv.position) not in prepared: + kv.prepare(seq_len) + prepared.add(id(kv.position)) + if mask is not None: + decode_masks = {} + for kv in past_key_values: + if isinstance(kv, FixedKV) and id(kv.position) not in decode_masks: + decode_masks[id(kv.position)] = _fixed_kv_decode_mask(mask, kv, min_val) + if enable_graph: + # static buffers + per-capacity attention biases: layer graphs replay against + # stable storage, refreshed eagerly each step + capacities = tuple(sorted({kv.key.shape[2] for kv in past_key_values if isinstance(kv, FixedKV)})) + state_key = (x.shape, x.dtype, x.device, tuple(t.shape for t in freqs_cis), capacities, + None if per_layer_inputs is None else per_layer_inputs.shape) + state = getattr(self, "_comfy_cross_step_state", None) + if state is None or state["key"] != state_key: + state = {"key": state_key, + "x": torch.empty_like(x), + "freqs_cis": [torch.empty_like(t) for t in freqs_cis], + "bias": {c: torch.empty((1, 1, 1, c), dtype=x.dtype, device=x.device) for c in capacities}, + "per_layer": None if per_layer_inputs is None else torch.empty_like(per_layer_inputs), + "bias_valid": -1} + self._comfy_cross_step_state = state + comfy.model_management._register_cross_step(self) + state["x"].copy_(x) + for source, target in zip(freqs_cis, state["freqs_cis"]): + target.copy_(source) + x = state["x"] + freqs_cis = state["freqs_cis"] + if per_layer_inputs is not None: + state["per_layer"].copy_(per_layer_inputs) + per_layer_inputs = state["per_layer"] + valid = past_len + 1 + for capacity, bias in state["bias"].items(): + if state["bias_valid"] != past_len: + bias.fill_(min_val) + bias[..., :min(valid, capacity)] = 0 + elif past_len < capacity: + bias[..., past_len:valid] = 0 + state["bias_valid"] = valid + decode_bias = state["bias"] intermediate = None all_intermediate = None @@ -429,12 +596,36 @@ class Gemma4Transformer(nn.Module): is_sliding = hasattr(layer, 'sliding_attention') and layer.sliding_attention if i >= first_kv_shared and num_kv_shared > 0: - shared = shared_sliding_kv if is_sliding else shared_global_kv - if shared is not None: - layer_kwargs['shared_kv'] = shared + if decode: + layer_kwargs['shared_kv'] = past_key_values[share_source[bool(is_sliding)]] + else: + shared = shared_sliding_kv if is_sliding else shared_global_kv + if shared is not None: + layer_kwargs['shared_kv'] = shared - x, current_kv, shareable_kv = layer(x=x, attention_mask=mask, freqs_cis=freqs_cis, past_key_value=past_kv, **layer_kwargs) + if enable_graph: + bias_cache = layer_kwargs.get('shared_kv', past_kv) + layer_mask = decode_bias[bias_cache.key.shape[2]] + elif decode: + bias_cache = layer_kwargs.get('shared_kv', past_kv) + layer_mask = None if decode_masks is None else decode_masks[id(bias_cache.position)] + else: + layer_mask = mask + result = [] + + def core(): + nonlocal x + x, current_kv, shareable_kv = layer(x=x, attention_mask=layer_mask, freqs_cis=freqs_cis, past_key_value=past_kv, **layer_kwargs) + result.append((current_kv, shareable_kv)) + + comfy.model_prefetch.prefetch_queue_pop(prefetch_queue, x.device, layer, x.dtype, core=core, enable_graph=enable_graph) + + if result: + current_kv, shareable_kv = result[0] + else: + # graph replay: the cache already holds this step's write + current_kv, shareable_kv = past_kv, None next_key_values.append(current_kv if current_kv is not None else ()) # Only track the last sliding/global before the sharing boundary @@ -447,6 +638,14 @@ class Gemma4Transformer(nn.Module): if i == intermediate_output: intermediate = x.clone() + if prefetch_queue is not None: + comfy.model_prefetch.prefetch_queue_pop(prefetch_queue, x.device, None) + + if fixed_kv: + for kv in past_key_values: + if isinstance(kv, FixedKV): + kv.advance(seq_len) + if self.norm is not None: x = self.norm(x) @@ -481,14 +680,37 @@ class Gemma4Base(BaseLlama, BaseGenerate, torch.nn.Module): if cap: logits = cap * torch.tanh(logits / cap) if self.model.config.suppress_tokens: - logits[..., self.model.config.suppress_tokens] = torch.finfo(logits.dtype).min + logits.index_fill_(-1, self.model._suppress_tokens, torch.finfo(logits.dtype).min) return logits def init_kv_cache(self, batch, max_cache_len, device, execution_dtype): - past_key_values = [] - for _ in range(self.model.config.num_hidden_layers): - past_key_values.append(()) - return past_key_values + cfg = self.model.config + num_layers = cfg.num_hidden_layers + if not self.model.fixed_kv: + return [() for _ in range(num_layers)] + first_shared = num_layers - cfg.num_kv_shared_layers if cfg.num_kv_shared_layers > 0 else num_layers + # position/seqlen device tensors are shared per cache geometry and filled once per step + trackers = {} + caches = [] + for i in range(num_layers): + if i >= first_shared: + caches.append(()) + continue + sliding = cfg.sliding_attention[i % len(cfg.sliding_attention)] if cfg.sliding_attention else False + head_dim = cfg.head_dim if sliding else cfg.global_head_dim + k_eq_v = cfg.attention_k_eq_v and not sliding + kv_heads = cfg.num_global_key_value_heads if k_eq_v else cfg.num_key_value_heads + length = min(sliding, max_cache_len) if sliding else max_cache_len + cache_cls = RingKV if sliding else FixedKV + tracker = trackers.get((cache_cls, length)) + if tracker is None: + tracker = (torch.empty((1,), device=device, dtype=torch.int64), + torch.empty((batch,), device=device, dtype=torch.int32)) + trackers[(cache_cls, length)] = tracker + # zero-init: decode attends full capacity with masked tails, 0*0 stays finite + key = torch.zeros((batch, kv_heads, length, head_dim), device=device, dtype=execution_dtype) + caches.append(cache_cls(key, torch.zeros_like(key), 0, tracker[0], tracker[1])) + return caches def preprocess_embed(self, embed, device): if embed["type"] == "image": @@ -1183,6 +1405,7 @@ def _get_aspect_ratio_preserving_size(height, width, patch_size, max_patches, po class Gemma4_Tokenizer(): tokenizer_json_data = None + prime_empty_thought = False def state_dict(self): if self.tokenizer_json_data is not None: @@ -1333,8 +1556,8 @@ class Gemma4_Tokenizer(): num_samples = int(waveform.shape[-1] * 16000 / sample_rate) if sample_rate != 16000 else waveform.shape[-1] n_audio_tokens = self._audio_token_count(num_samples) media += "<|audio>" + "<|audio|>" * n_audio_tokens + "" - # Non-thinking mode primes an empty thought channel so the model answers directly. - model_open = "" if thinking else "<|channel>thought\n" + # 12B/31B prime a closed thought block for non-thinking mode, E2B/E4B must not: it cues them into reasoning inline. + model_open = "<|channel>thought\n" if self.prime_empty_thought and not thinking else "" llama_text = f"{system}<|turn>user\n{text}{media}\n<|turn>model\n{model_open}" text_tokens = super().tokenize_with_weights(llama_text, return_word_ids) @@ -1418,6 +1641,7 @@ class Gemma4Tokenizer(sd1_clip.SD1Tokenizer): class Gemma4UnifiedSDTokenizer(Gemma4SDTokenizer): """Encoder-free (gemma4_unified) audio: raw 16kHz waveform frames instead of mel spectrogram.""" embedding_size = 3840 + prime_empty_thought = True def _extract_audio_features(self, waveform, sample_rate): audio = self._resample_16k(waveform, sample_rate) @@ -1443,14 +1667,14 @@ class Gemma4UnifiedTokenizer(Gemma4Tokenizer): class Gemma4Model(sd1_clip.SDClipModel): model_class = None def __init__(self, device="cpu", layer="all", layer_idx=None, dtype=None, attention_mask=True, model_options={}): + llama_quantization_metadata = model_options.get("llama_quantization_metadata", None) + if llama_quantization_metadata is not None: + model_options = model_options.copy() + model_options["quantization_metadata"] = llama_quantization_metadata self.dtypes = set() self.dtypes.add(dtype) super().__init__(device=device, layer=layer, layer_idx=layer_idx, textmodel_json_config={}, dtype=dtype, special_tokens={"start": 2, "pad": 0}, layer_norm_hidden_state=False, model_class=self.model_class, enable_attention_masks=attention_mask, return_attention_masks=attention_mask, model_options=model_options) - def process_tokens(self, tokens, device): - embeds, _, _, _ = super().process_tokens(tokens, device) - return embeds - def generate(self, tokens, do_sample, max_length, temperature, top_k, top_p, min_p, repetition_penalty, seed, presence_penalty=0.0): if isinstance(tokens, dict): tokens = next(iter(tokens.values())) @@ -1474,8 +1698,19 @@ class Gemma4Model(sd1_clip.SDClipModel): return self.transformer.generate(embeds, do_sample, max_length, temperature, top_k, top_p, min_p, repetition_penalty, seed, initial_tokens=initial_token_ids[0], presence_penalty=presence_penalty, initial_input_ids=input_ids, embeds_info=embeds_info) +def gemma4_clip_model(model_class): + return type('Gemma4Model_', (Gemma4Model,), {'model_class': model_class}) + + +def gemma4_text_encoder_model(model_class): + return type('Gemma4TextEncoderModel_', (Gemma4Model,), { + 'model_class': model_class, + 'process_tokens': sd1_clip.SDClipModel.process_tokens, + }) + + def gemma4_te(dtype_llama=None, llama_quantization_metadata=None, model_class=None): - clip_model = type('Gemma4Model_', (Gemma4Model,), {'model_class': model_class}) + clip_model = gemma4_clip_model(model_class) class Gemma4TEModel_(sd1_clip.SD1ClipModel): def __init__(self, device="cpu", dtype=None, model_options={}): if llama_quantization_metadata is not None: @@ -1484,12 +1719,15 @@ def gemma4_te(dtype_llama=None, llama_quantization_metadata=None, model_class=No if dtype_llama is not None: dtype = dtype_llama super().__init__(device=device, dtype=dtype, name="gemma4", clip_model=clip_model, model_options=model_options) + + def get_dynamic_vram__units(self): + return getattr(self, self.clip).transformer.model.get_dynamic_vram__units() return Gemma4TEModel_ # Variants -def _make_variant(config_cls): +def _make_variant(config_cls, prime_empty_thought=False): audio = config_cls.audio_config is not None bases = (Gemma4AudioMixin, Gemma4Base) if audio else (Gemma4Base,) class Variant(*bases): @@ -1499,8 +1737,8 @@ def _make_variant(config_cls): if audio: self._init_audio(self.model.config, dtype, device, operations) embedding_size = config_cls.hidden_size - if embedding_size != Gemma4SDTokenizer.embedding_size: - tok_cls = type('T', (Gemma4SDTokenizer,), {'embedding_size': embedding_size}) + if embedding_size != Gemma4SDTokenizer.embedding_size or prime_empty_thought: + tok_cls = type('T', (Gemma4SDTokenizer,), {'embedding_size': embedding_size, 'prime_empty_thought': prime_empty_thought}) class Tokenizer(Gemma4Tokenizer): tokenizer_class = tok_cls Variant.tokenizer = Tokenizer @@ -1510,7 +1748,7 @@ def _make_variant(config_cls): Gemma4_E4B = _make_variant(Gemma4Config) Gemma4_E2B = _make_variant(Gemma4_E2B_Config) -Gemma4_31B = _make_variant(Gemma4_31B_Config) +Gemma4_31B = _make_variant(Gemma4_31B_Config, prime_empty_thought=True) # Gemma4 12B Unified: encoder-free multimodal, distinct base/tokenizer (not via _make_variant). diff --git a/comfy/text_encoders/hunyuan_video.py b/comfy/text_encoders/hunyuan_video.py index 2ddb4da60..932a3d49b 100644 --- a/comfy/text_encoders/hunyuan_video.py +++ b/comfy/text_encoders/hunyuan_video.py @@ -2,7 +2,7 @@ from comfy import sd1_clip import comfy.model_management import comfy.text_encoders.llama from .hunyuan_image import HunyuanImageTokenizer -from transformers import LlamaTokenizerFast +from .bpe_tokenizer import LlamaTokenizerFast import torch import os import numbers diff --git a/comfy/text_encoders/llama.py b/comfy/text_encoders/llama.py index 371ec1bbc..f182e5147 100644 --- a/comfy/text_encoders/llama.py +++ b/comfy/text_encoders/llama.py @@ -5,15 +5,33 @@ from typing import Optional, Any, Tuple import math from tqdm import tqdm import comfy.utils +import comfy_kitchen from comfy.ldm.modules.attention import optimized_attention_for_device import comfy.model_management +import comfy.model_prefetch import comfy.ops import comfy.ldm.common_dit import comfy.clip_model from . import qwen_vl + +@dataclass +class FixedKV: + key: torch.Tensor + value: torch.Tensor + index: int + position: torch.Tensor + seqlen: torch.Tensor + + def prepare(self, num_tokens): + self.position.fill_(self.index) + self.seqlen.fill_(self.index + num_tokens) + + def advance(self, num_tokens): + self.index += num_tokens + @dataclass class Llama2Config: vocab_size: int = 128320 @@ -249,6 +267,9 @@ class Qwen3_8BConfig: rope_scale = None final_norm: bool = True lm_head: bool = True + fixed_kv: bool = False + merged_qkv: bool = False + merged_mlp: bool = False stop_tokens = [151643, 151645] @dataclass @@ -498,9 +519,14 @@ class Attention(nn.Module): self.inner_size = self.num_heads * self.head_dim ops = ops or nn - self.q_proj = ops.Linear(config.hidden_size, self.inner_size, bias=config.qkv_bias, device=device, dtype=dtype) - self.k_proj = ops.Linear(config.hidden_size, self.num_kv_heads * self.head_dim, bias=config.qkv_bias, device=device, dtype=dtype) - self.v_proj = ops.Linear(config.hidden_size, self.num_kv_heads * self.head_dim, bias=config.qkv_bias, device=device, dtype=dtype) + self.kv_size = self.num_kv_heads * self.head_dim + self.merged_qkv = getattr(config, "merged_qkv", False) + if self.merged_qkv: + self.qkv_proj = ops.Linear(config.hidden_size, self.inner_size + self.kv_size * 2, bias=config.qkv_bias, device=device, dtype=dtype) + else: + self.q_proj = ops.Linear(config.hidden_size, self.inner_size, bias=config.qkv_bias, device=device, dtype=dtype) + self.k_proj = ops.Linear(config.hidden_size, self.kv_size, bias=config.qkv_bias, device=device, dtype=dtype) + self.v_proj = ops.Linear(config.hidden_size, self.kv_size, bias=config.qkv_bias, device=device, dtype=dtype) self.o_proj = ops.Linear(self.inner_size, config.hidden_size, bias=False, device=device, dtype=dtype) self.q_norm = None @@ -522,9 +548,12 @@ class Attention(nn.Module): ): batch_size, seq_length, _ = hidden_states.shape - xq = self.q_proj(hidden_states) - xk = self.k_proj(hidden_states) - xv = self.v_proj(hidden_states) + if self.merged_qkv: + xq, xk, xv = self.qkv_proj(hidden_states).split((self.inner_size, self.kv_size, self.kv_size), dim=-1) + else: + xq = self.q_proj(hidden_states) + xk = self.k_proj(hidden_states) + xv = self.v_proj(hidden_states) xq = xq.view(batch_size, seq_length, self.num_heads, self.head_dim).transpose(1, 2) xk = xk.view(batch_size, seq_length, self.num_kv_heads, self.head_dim).transpose(1, 2) @@ -537,8 +566,29 @@ class Attention(nn.Module): xq, xk = apply_rope(xq, xk, freqs_cis=freqs_cis) - present_key_value = None - if past_key_value is not None: + fixed_cache = past_key_value if isinstance(past_key_value, FixedKV) else None + if fixed_cache is not None: + xq = xq.transpose(1, 2) + xk = xk.transpose(1, 2) + xv = xv.transpose(1, 2) + if seq_length == 1: + # CUDA-graphable decode path. + fixed_cache.key.index_copy_(1, fixed_cache.position, xk) + fixed_cache.value.index_copy_(1, fixed_cache.position, xv) + output = comfy_kitchen.flash_attention_decode(xq, fixed_cache.key, fixed_cache.value, fixed_cache.seqlen) + return self.o_proj(output.view(batch_size, seq_length, self.inner_size)), fixed_cache + + fixed_cache.key[:, fixed_cache.index:fixed_cache.index + seq_length].copy_(xk) + fixed_cache.value[:, fixed_cache.index:fixed_cache.index + seq_length].copy_(xv) + xk = fixed_cache.key[:, :fixed_cache.index + seq_length] + xv = fixed_cache.value[:, :fixed_cache.index + seq_length] + + xq = xq.transpose(1, 2) + xk = xk.transpose(1, 2) + xv = xv.transpose(1, 2) + + present_key_value = fixed_cache + if fixed_cache is None and past_key_value is not None: index = 0 num_tokens = xk.shape[2] if len(past_key_value) > 0: @@ -569,15 +619,27 @@ class MLP(nn.Module): def __init__(self, config: Llama2Config, device=None, dtype=None, ops: Any = None, intermediate_size=None): super().__init__() intermediate_size = intermediate_size or config.intermediate_size - self.gate_proj = ops.Linear(config.hidden_size, intermediate_size, bias=False, device=device, dtype=dtype) - self.up_proj = ops.Linear(config.hidden_size, intermediate_size, bias=False, device=device, dtype=dtype) + self.merged_mlp = getattr(config, "merged_mlp", False) + if self.merged_mlp: + self.gate_up_proj = ops.Linear(config.hidden_size, intermediate_size * 2, bias=False, device=device, dtype=dtype) + else: + self.gate_proj = ops.Linear(config.hidden_size, intermediate_size, bias=False, device=device, dtype=dtype) + self.up_proj = ops.Linear(config.hidden_size, intermediate_size, bias=False, device=device, dtype=dtype) self.down_proj = ops.Linear(intermediate_size, config.hidden_size, bias=False, device=device, dtype=dtype) if config.mlp_activation == "silu": self.activation = torch.nn.functional.silu + self.merged_input_act = "swiglu" elif config.mlp_activation == "gelu_pytorch_tanh": self.activation = lambda a: torch.nn.functional.gelu(a, approximate="tanh") + self.merged_input_act = None def forward(self, x): + if self.merged_mlp: + x = self.gate_up_proj(x) + if self.merged_input_act is not None: + return comfy.ops.linear_input_act(self.down_proj, x, self.merged_input_act) + gate, up = x.chunk(2, dim=-1) + return self.down_proj(self.activation(gate) * up) return self.down_proj(self.activation(self.gate_proj(x)) * self.up_proj(x)) class TransformerBlock(nn.Module): @@ -596,6 +658,7 @@ class TransformerBlock(nn.Module): optimized_attention=None, past_key_value: Optional[Tuple[torch.Tensor, torch.Tensor]] = None, ): + output = x # Self Attention residual = x x = self.input_layernorm(x) @@ -612,7 +675,7 @@ class TransformerBlock(nn.Module): residual = x x = self.post_attention_layernorm(x) x = self.mlp(x) - x = residual + x + x = torch.add(residual, x, out=output) return x, present_key_value @@ -641,6 +704,7 @@ class TransformerBlockGemma2(nn.Module): optimized_attention=None, past_key_value: Optional[Tuple[torch.Tensor, torch.Tensor]] = None, ): + output = x sliding_window = None if self.transformer_type == 'gemma3': if self.sliding_attention: @@ -676,7 +740,7 @@ class TransformerBlockGemma2(nn.Module): x = self.pre_feedforward_layernorm(x) x = self.mlp(x) x = self.post_feedforward_layernorm(x) - x = residual + x + x = torch.add(residual, x, out=output) return x, present_key_value @@ -688,9 +752,14 @@ def _make_scaled_embedding(ops, vocab_size, hidden_size, scale, device, dtype): class Llama2_(nn.Module): + fixed_kv = False + graph_dynamic_vbar_blocks = False + def __init__(self, config, device=None, dtype=None, ops=None): super().__init__() self.config = config + self.fixed_kv = getattr(config, "fixed_kv", False) + self.graph_dynamic_vbar_blocks = False self.vocab_size = config.vocab_size if self.config.transformer_type == "gemma2" or self.config.transformer_type == "gemma3": @@ -713,8 +782,27 @@ class Llama2_(nn.Module): if config.lm_head: self.lm_head = ops.Linear(config.hidden_size, config.vocab_size, bias=False, device=device, dtype=dtype) + def get_dynamic_vram__units(self): + return (list(self.layers), []) if self.graph_dynamic_vbar_blocks else ([], []) + def get_past_len(self, past_key_values): - return past_key_values[0][2] + first = past_key_values[0] + return first.index if isinstance(first, FixedKV) else first[2] + + def init_kv_cache(self, batch, capacity, device, dtype): + caches = [] + fixed_kv = self.fixed_kv and comfy_kitchen.flash_attention_decode_is_available(device) + for _ in range(self.config.num_hidden_layers): + if fixed_kv: + key = torch.empty((batch, capacity, self.config.num_key_value_heads, self.config.head_dim), device=device, dtype=dtype) + value = torch.empty_like(key) + position = torch.empty((1,), device=device, dtype=torch.int64) + seqlen = torch.empty((batch,), device=device, dtype=torch.int32) + caches.append(FixedKV(key, value, 0, position, seqlen)) + else: + key = torch.empty((batch, self.config.num_key_value_heads, capacity, self.config.head_dim), device=device, dtype=dtype) + caches.append((key, torch.empty_like(key), 0)) + return caches def compute_freqs_cis(self, position_ids, device): return precompute_freqs_cis(self.config.head_dim, @@ -756,6 +844,33 @@ class Llama2_(nn.Module): optimized_attention = optimized_attention_for_device(x.device, mask=mask is not None, small_input=True) + fixed_kv = past_key_values is not None and len(past_key_values) > 0 and isinstance(past_key_values[0], FixedKV) + enable_graph = self.graph_dynamic_vbar_blocks and fixed_kv and seq_len == 1 and mask is None + if enable_graph: + freqs_cis_groups = freqs_cis if isinstance(freqs_cis, list) else [freqs_cis] + cross_step_state_key = [(x.shape, x.stride(), x.dtype, x.device)] + for group in freqs_cis_groups: + for tensor in group: + cross_step_state_key.append((tensor.shape, tensor.stride(), tensor.dtype, tensor.device)) + cross_step_state_key = tuple(cross_step_state_key) + cross_step_state = getattr(self, "_comfy_cross_step_state", None) + if cross_step_state is None or cross_step_state["key"] != cross_step_state_key: + static_freqs_cis = [] + for group in freqs_cis_groups: + static_freqs_cis.append(tuple(torch.empty_like(tensor) for tensor in group)) + if not isinstance(freqs_cis, list): + static_freqs_cis = static_freqs_cis[0] + cross_step_state = {"key": cross_step_state_key, "x": torch.empty_like(x), "freqs_cis": static_freqs_cis} + self._comfy_cross_step_state = cross_step_state + comfy.model_management._register_cross_step(self) + cross_step_state["x"].copy_(x) + static_freqs_cis_groups = cross_step_state["freqs_cis"] if isinstance(freqs_cis, list) else [cross_step_state["freqs_cis"]] + for source_group, target_group in zip(freqs_cis_groups, static_freqs_cis_groups): + for source, target in zip(source_group, target_group): + target.copy_(source) + x = cross_step_state["x"] + freqs_cis = cross_step_state["freqs_cis"] + intermediate = None all_intermediate = None only_layers = None @@ -769,7 +884,8 @@ class Llama2_(nn.Module): elif intermediate_output < 0: intermediate_output = len(self.layers) + intermediate_output - next_key_values = [] + prefetch_queue = comfy.model_prefetch.make_prefetch_queue(list(self.layers), x.device, {"prefetch_dynamic_vbars": getattr(self, "prefetch_dynamic_vbars", False)}) + next_key_values = list(past_key_values) if past_key_values is not None else [] for i, layer in enumerate(self.layers): if all_intermediate is not None: if only_layers is None or (i in only_layers): @@ -779,16 +895,24 @@ class Llama2_(nn.Module): if past_key_values is not None: past_kv = past_key_values[i] if len(past_key_values) > 0 else [] - x, current_kv = layer( - x=x, - attention_mask=mask, - freqs_cis=freqs_cis, - optimized_attention=optimized_attention, - past_key_value=past_kv, - ) + if fixed_kv: + past_kv.prepare(seq_len) - if current_kv is not None: - next_key_values.append(current_kv) + def core(): + nonlocal x + x, current_kv = layer( + x=x, + attention_mask=mask, + freqs_cis=freqs_cis, + optimized_attention=optimized_attention, + past_key_value=past_kv, + ) + if next_key_values: + next_key_values[i] = current_kv + + comfy.model_prefetch.prefetch_queue_pop(prefetch_queue, x.device, layer, x.dtype, core=core, enable_graph=enable_graph) + if fixed_kv: + next_key_values[i].advance(seq_len) # DeepStack: add per-layer visual features into the first len() decoder layers at image positions (Qwen3-VL) if deepstack_embeds is not None and i < len(deepstack_embeds): @@ -797,6 +921,9 @@ class Llama2_(nn.Module): if i == intermediate_output: intermediate = x.clone() + if prefetch_queue is not None: + comfy.model_prefetch.prefetch_queue_pop(prefetch_queue, x.device, None) + if self.norm is not None: x = self.norm(x) @@ -810,7 +937,7 @@ class Llama2_(nn.Module): if intermediate is not None and final_layer_norm_intermediate and self.norm is not None: intermediate = self.norm(intermediate) - if len(next_key_values) > 0: + if next_key_values: return x, intermediate, next_key_values else: return x, intermediate @@ -874,12 +1001,7 @@ class BaseGenerate: return torch.nn.functional.linear(input, weight, None) def init_kv_cache(self, batch, max_cache_len, device, execution_dtype): - model_config = self.model.config - past_key_values = [] - for x in range(model_config.num_hidden_layers): - past_key_values.append((torch.empty([batch, model_config.num_key_value_heads, max_cache_len, model_config.head_dim], device=device, dtype=execution_dtype), - torch.empty([batch, model_config.num_key_value_heads, max_cache_len, model_config.head_dim], device=device, dtype=execution_dtype), 0)) - return past_key_values + return self.model.init_kv_cache(batch, max_cache_len, device, execution_dtype) def generate(self, embeds=None, do_sample=True, max_length=256, temperature=1.0, top_k=50, top_p=0.9, min_p=0.0, repetition_penalty=1.0, seed=42, stop_tokens=None, initial_tokens=[], execution_dtype=None, min_tokens=0, presence_penalty=0.0, initial_input_ids=None, position_ids=None, deepstack_embeds=None, visual_pos_masks=None, embeds_info=None): device = embeds.device diff --git a/comfy/text_encoders/lt.py b/comfy/text_encoders/lt.py index bc5cbae28..c512a7d48 100644 --- a/comfy/text_encoders/lt.py +++ b/comfy/text_encoders/lt.py @@ -81,6 +81,17 @@ class LTXAVGemmaTokenizer(sd1_clip.SD1Tokenizer): super().__init__(embedding_directory=embedding_directory, tokenizer_data=tokenizer_data, name="gemma3_12b", tokenizer=Gemma3_12BTokenizer) +def ltxav_gemma4_tokenizer(tokenizer): + class LTXAVGemma4Tokenizer(tokenizer): + def __init__(self, embedding_directory=None, tokenizer_data={}): + super().__init__(embedding_directory=embedding_directory, tokenizer_data=tokenizer_data) + gemma_tokenizer = getattr(self, self.clip) + if gemma_tokenizer.min_length == 1: + gemma_tokenizer.min_length = 1024 + + return LTXAVGemma4Tokenizer + + class Gemma3_12BModel(sd1_clip.SDClipModel): def __init__(self, device="cpu", layer="all", layer_idx=None, dtype=None, attention_mask=True, model_options={}): llama_quantization_metadata = model_options.get("llama_quantization_metadata", None) @@ -97,10 +108,10 @@ class Gemma3_12BModel(sd1_clip.SDClipModel): return self.transformer.generate(embeds, do_sample, max_length, temperature, top_k, top_p, min_p, repetition_penalty, seed, stop_tokens=[106], presence_penalty=presence_penalty) # 106 is class DualLinearProjection(torch.nn.Module): - def __init__(self, in_dim, out_dim_video, out_dim_audio, dtype=None, device=None, operations=None): + def __init__(self, in_dim, out_dim_video, out_dim_audio, video_bias=True, audio_bias=True, dtype=None, device=None, operations=None): super().__init__() - self.audio_aggregate_embed = operations.Linear(in_dim, out_dim_audio, bias=True, dtype=dtype, device=device) - self.video_aggregate_embed = operations.Linear(in_dim, out_dim_video, bias=True, dtype=dtype, device=device) + self.audio_aggregate_embed = operations.Linear(in_dim, out_dim_audio, bias=audio_bias, dtype=dtype, device=device) + self.video_aggregate_embed = operations.Linear(in_dim, out_dim_video, bias=video_bias, dtype=dtype, device=device) def forward(self, x): source_dim = x.shape[-1] @@ -112,22 +123,28 @@ class DualLinearProjection(torch.nn.Module): return torch.cat((video, audio), dim=-1) class LTXAVTEModel(torch.nn.Module): - def __init__(self, dtype_llama=None, device="cpu", dtype=None, text_projection_type="single_linear", model_options={}): + def __init__(self, dtype_llama=None, device="cpu", dtype=None, text_projection_type="single_linear", text_encoder_model=Gemma3_12BModel, text_encoder_key="gemma3_12b", video_projection_dim=3840, audio_projection_dim=2048, video_projection_bias=None, audio_projection_bias=True, model_options={}): super().__init__() self.dtypes = set() self.dtypes.add(dtype) self.compat_mode = False self.text_projection_type = text_projection_type + self.text_encoder_key = text_encoder_key + self.execution_device = None - self.gemma3_12b = Gemma3_12BModel(device=device, dtype=dtype_llama, model_options=model_options, layer="all", layer_idx=None) + self.gemma3_12b = text_encoder_model(device=device, dtype=dtype_llama, model_options=model_options, layer="all", layer_idx=None) self.dtypes.add(dtype_llama) operations = self.gemma3_12b.operations # TODO + text_encoder_config = self.gemma3_12b.transformer.model.config + projection_in_dim = text_encoder_config.hidden_size * (text_encoder_config.num_hidden_layers + 1) + if video_projection_bias is None: + video_projection_bias = self.text_projection_type == "dual_linear" if self.text_projection_type == "single_linear": - self.text_embedding_projection = operations.Linear(3840 * 49, 3840, bias=False, dtype=dtype, device=device) + self.text_embedding_projection = operations.Linear(projection_in_dim, video_projection_dim, bias=video_projection_bias, dtype=dtype, device=device) elif self.text_projection_type == "dual_linear": - self.text_embedding_projection = DualLinearProjection(3840 * 49, 4096, 2048, dtype=dtype, device=device, operations=operations) + self.text_embedding_projection = DualLinearProjection(projection_in_dim, video_projection_dim, audio_projection_dim, video_bias=video_projection_bias, audio_bias=audio_projection_bias, dtype=dtype, device=device, operations=operations) def enable_compat_mode(self): # TODO: remove @@ -161,7 +178,7 @@ class LTXAVTEModel(torch.nn.Module): self.execution_device = None def encode_token_weights(self, token_weight_pairs): - token_weight_pairs = token_weight_pairs["gemma3_12b"] + token_weight_pairs = token_weight_pairs[self.text_encoder_key] out, pooled, extra = self.gemma3_12b.encode_token_weights(token_weight_pairs) out = out[:, :, -torch.sum(extra["attention_mask"]).item():] @@ -189,51 +206,54 @@ class LTXAVTEModel(torch.nn.Module): return out.to(device=out_device, dtype=torch.float), pooled, extra def generate(self, tokens, do_sample, max_length, temperature, top_k, top_p, min_p, repetition_penalty, seed, presence_penalty): - return self.gemma3_12b.generate(tokens["gemma3_12b"], do_sample, max_length, temperature, top_k, top_p, min_p, repetition_penalty, seed, presence_penalty) + return self.gemma3_12b.generate(tokens[self.text_encoder_key], do_sample, max_length, temperature, top_k, top_p, min_p, repetition_penalty, seed, presence_penalty) def load_sd(self, sd): - if "model.layers.47.self_attn.q_norm.weight" in sd: - return self.gemma3_12b.load_sd(sd) - else: - sdo = comfy.utils.state_dict_prefix_replace(sd, {"text_embedding_projection.aggregate_embed.weight": "text_embedding_projection.weight", "text_embedding_projection.": "text_embedding_projection."}, filter_keys=True) - if len(sdo) == 0: - sdo = sd + missing_all = [] + unexpected_all = [] - missing_all = [] - unexpected_all = [] + if "model.layers.0.self_attn.q_norm.weight" in sd: + gemma_sd = {k: v for k, v in sd.items() if not k.startswith("text_embedding_projection.")} + missing, unexpected = self.gemma3_12b.load_sd(gemma_sd) + missing_all.extend(missing) + unexpected_all.extend(unexpected) - for prefix, component in [("text_embedding_projection.", self.text_embedding_projection)]: - component_sd = {k.replace(prefix, ""): v for k, v in sdo.items() if k.startswith(prefix)} - if component_sd: - missing, unexpected = component.load_state_dict(component_sd, strict=False, assign=getattr(self, "can_assign_sd", False)) - missing_all.extend([f"{prefix}{k}" for k in missing]) - unexpected_all.extend([f"{prefix}{k}" for k in unexpected]) + sdo = comfy.utils.state_dict_prefix_replace(sd, {"text_embedding_projection.aggregate_embed.": "text_embedding_projection.", "text_embedding_projection.": "text_embedding_projection."}, filter_keys=True) + if len(sdo) == 0: + sdo = sd - if "model.diffusion_model.audio_embeddings_connector.transformer_1d_blocks.2.attn1.to_q.bias" not in sd: # TODO: remove - ww = sd.get("model.diffusion_model.audio_embeddings_connector.transformer_1d_blocks.0.attn1.to_q.bias", None) - if ww is not None: - if ww.shape[0] == 3840: - self.enable_compat_mode() - sdv = comfy.utils.state_dict_prefix_replace(sd, {"model.diffusion_model.video_embeddings_connector.": ""}, filter_keys=True) - self.video_embeddings_connector.load_state_dict(sdv, strict=False, assign=getattr(self, "can_assign_sd", False)) - sda = comfy.utils.state_dict_prefix_replace(sd, {"model.diffusion_model.audio_embeddings_connector.": ""}, filter_keys=True) - self.audio_embeddings_connector.load_state_dict(sda, strict=False, assign=getattr(self, "can_assign_sd", False)) + for prefix, component in [("text_embedding_projection.", self.text_embedding_projection)]: + component_sd = {k.replace(prefix, ""): v for k, v in sdo.items() if k.startswith(prefix)} + if component_sd: + missing, unexpected = component.load_state_dict(component_sd, strict=False, assign=getattr(self, "can_assign_sd", False)) + missing_all.extend([f"{prefix}{k}" for k in missing]) + unexpected_all.extend([f"{prefix}{k}" for k in unexpected]) - return (missing_all, unexpected_all) + if "model.diffusion_model.audio_embeddings_connector.transformer_1d_blocks.2.attn1.to_q.bias" not in sd: # TODO: remove + ww = sd.get("model.diffusion_model.audio_embeddings_connector.transformer_1d_blocks.0.attn1.to_q.bias", None) + if ww is not None: + if ww.shape[0] == 3840: + self.enable_compat_mode() + sdv = comfy.utils.state_dict_prefix_replace(sd, {"model.diffusion_model.video_embeddings_connector.": ""}, filter_keys=True) + self.video_embeddings_connector.load_state_dict(sdv, strict=False, assign=getattr(self, "can_assign_sd", False)) + sda = comfy.utils.state_dict_prefix_replace(sd, {"model.diffusion_model.audio_embeddings_connector.": ""}, filter_keys=True) + self.audio_embeddings_connector.load_state_dict(sda, strict=False, assign=getattr(self, "can_assign_sd", False)) + + return (missing_all, unexpected_all) def memory_estimation_function(self, token_weight_pairs, device=None): constant = 6.0 if comfy.model_management.should_use_bf16(device): constant /= 2.0 - token_weight_pairs = token_weight_pairs.get("gemma3_12b", []) + token_weight_pairs = token_weight_pairs.get(self.text_encoder_key, []) m = min([sum(1 for _ in itertools.takewhile(lambda x: x[0] == 0, sub)) for sub in token_weight_pairs]) num_tokens = sum(map(lambda a: len(a), token_weight_pairs)) - m num_tokens = max(num_tokens, 642) return num_tokens * constant * 1024 * 1024 -def ltxav_te(dtype_llama=None, llama_quantization_metadata=None, text_projection_type="single_linear"): +def ltxav_te(dtype_llama=None, llama_quantization_metadata=None, text_projection_type="single_linear", text_encoder_model=Gemma3_12BModel, text_encoder_key="gemma3_12b", video_projection_dim=3840, audio_projection_dim=2048, video_projection_bias=None, audio_projection_bias=True): class LTXAVTEModel_(LTXAVTEModel): def __init__(self, device="cpu", dtype=None, model_options={}): if llama_quantization_metadata is not None: @@ -241,16 +261,29 @@ def ltxav_te(dtype_llama=None, llama_quantization_metadata=None, text_projection model_options["llama_quantization_metadata"] = llama_quantization_metadata if dtype_llama is not None: dtype = dtype_llama - super().__init__(dtype_llama=dtype_llama, device=device, dtype=dtype, text_projection_type=text_projection_type, model_options=model_options) + super().__init__(dtype_llama=dtype_llama, device=device, dtype=dtype, text_projection_type=text_projection_type, text_encoder_model=text_encoder_model, text_encoder_key=text_encoder_key, video_projection_dim=video_projection_dim, audio_projection_dim=audio_projection_dim, video_projection_bias=video_projection_bias, audio_projection_bias=audio_projection_bias, model_options=model_options) return LTXAVTEModel_ def sd_detect(state_dict_list, prefix=""): for sd in state_dict_list: - if "{}text_embedding_projection.audio_aggregate_embed.bias".format(prefix) in sd: - return {"text_projection_type": "dual_linear"} - if "{}text_embedding_projection.weight".format(prefix) in sd or "{}text_embedding_projection.aggregate_embed.weight".format(prefix) in sd: - return {"text_projection_type": "single_linear"} + video_key = "{}text_embedding_projection.video_aggregate_embed.weight".format(prefix) + audio_key = "{}text_embedding_projection.audio_aggregate_embed.weight".format(prefix) + if video_key in sd and audio_key in sd: + return { + "text_projection_type": "dual_linear", + "video_projection_dim": sd[video_key].shape[0], + "audio_projection_dim": sd[audio_key].shape[0], + "video_projection_bias": "{}text_embedding_projection.video_aggregate_embed.bias".format(prefix) in sd, + "audio_projection_bias": "{}text_embedding_projection.audio_aggregate_embed.bias".format(prefix) in sd, + } + for key in ("{}text_embedding_projection.weight".format(prefix), "{}text_embedding_projection.aggregate_embed.weight".format(prefix)): + if key in sd: + return { + "text_projection_type": "single_linear", + "video_projection_dim": sd[key].shape[0], + "video_projection_bias": key.removesuffix("weight") + "bias" in sd, + } return {} diff --git a/comfy/text_encoders/lumina2.py b/comfy/text_encoders/lumina2.py index b1f1dbb9f..e44920203 100644 --- a/comfy/text_encoders/lumina2.py +++ b/comfy/text_encoders/lumina2.py @@ -49,10 +49,6 @@ class Gemma3_4B_Vision_Model(sd1_clip.SDClipModel): super().__init__(device=device, layer=layer, layer_idx=layer_idx, textmodel_json_config={}, dtype=dtype, special_tokens={"start": 2, "pad": 0}, layer_norm_hidden_state=False, model_class=comfy.text_encoders.llama.Gemma3_4B_Vision, enable_attention_masks=attention_mask, return_attention_masks=attention_mask, model_options=model_options) - def process_tokens(self, tokens, device): - embeds, _, _, _ = super().process_tokens(tokens, device) - return embeds - class LuminaModel(sd1_clip.SD1ClipModel): def __init__(self, device="cpu", dtype=None, model_options={}, name="gemma2_2b", clip_model=Gemma2_2BModel): super().__init__(device=device, dtype=dtype, name=name, clip_model=clip_model, model_options=model_options) diff --git a/comfy/text_encoders/minimax_music.py b/comfy/text_encoders/minimax_music.py new file mode 100644 index 000000000..c88d463cc --- /dev/null +++ b/comfy/text_encoders/minimax_music.py @@ -0,0 +1,117 @@ +import torch +from tokenizers import Tokenizer + +import comfy.ops +from comfy.ldm.minimax_music.ar import CFG_SCALE, CFG_TOP_K, MAX_AUDIO_FRAMES, MiniMaxMusic3AR +from comfy.ldm.minimax_music.prompt import SPECIAL_TOKEN_IDS, build_prompt + + +MODEL_CONFIG = { + "vocab_size": 200000, + "hidden_size": 4096, + "intermediate_size": 12288, + "num_hidden_layers": 36, + "num_attention_heads": 32, + "num_key_value_heads": 8, + "max_position_embeddings": 10240, + "rms_norm_eps": 1e-6, + "rope_theta": 1000000.0, + "head_dim": 128, + "audio_vocab_size": 1024, + "audio_num_codebooks": 8, + "decoder_num_heads": 16, + "decoder_intermediate_size": 6144, + "decoder_num_layers": 4, +} + + +def detect_merged_config(state_dict, prefix=""): + return { + "merged_qkv": "{}model.layers.0.self_attn.qkv_proj.weight".format(prefix) in state_dict, + "merged_mlp": "{}model.layers.0.mlp.gate_up_proj.weight".format(prefix) in state_dict, + "decoder_merged_qkv": "{}model.audio_decoder.layers.0.self_attn.qkv_proj.weight".format(prefix) in state_dict, + "decoder_merged_mlp": "{}model.audio_decoder.layers.0.mlp.gate_up_proj.weight".format(prefix) in state_dict, + } + + +class MiniMaxMusic3Tokenizer: + def __init__(self, embedding_directory=None, tokenizer_data={}): + tokenizer_json = tokenizer_data.get("tokenizer_json") + if tokenizer_json is None: + raise ValueError("MiniMax Music3 text encoder checkpoint is missing tokenizer_json") + if torch.is_tensor(tokenizer_json): + tokenizer_json = tokenizer_json.detach().cpu().numpy().tobytes() + self.tokenizer_json = tokenizer_json + self.tokenizer = Tokenizer.from_str(tokenizer_json.decode("utf-8")) + for token, expected in SPECIAL_TOKEN_IDS.items(): + if self.tokenizer.token_to_id(token) != expected: + raise ValueError(f"MiniMax Music3 tokenizer mismatch for {token}") + + def tokenize_with_weights(self, text, return_word_ids=False, **kwargs): + prompt = build_prompt(text, kwargs.get("lyrics", "")) + token_ids = self.tokenizer.encode(prompt, add_special_tokens=False).ids + return { + "minimax_music3": [[(token, 1.0) for token in token_ids]], + "seed": int(kwargs.get("seed", 0)), + "max_audio_frames": int(kwargs.get("max_audio_frames", MAX_AUDIO_FRAMES)), + "cfg_scale": float(kwargs.get("cfg_scale", CFG_SCALE)), + "top_k": int(kwargs.get("top_k", CFG_TOP_K)), + } + + def state_dict(self): + return {"tokenizer_json": torch.frombuffer(bytearray(self.tokenizer_json), dtype=torch.uint8)} + + def decode(self, token_ids, skip_special_tokens=True): + return self.tokenizer.decode(token_ids, skip_special_tokens=skip_special_tokens) + + +class MiniMaxMusic3TEModel(MiniMaxMusic3AR): + def __init__(self, device="cpu", dtype=None, model_options={}, projection_config=None): + dtype = torch.bfloat16 + quant_config = model_options.get("quantization_metadata", None) + operations = model_options.get("custom_operations", None) + if operations is None: + operations = comfy.ops.mixed_precision_ops(quant_config, dtype) if quant_config is not None else comfy.ops.manual_cast + super().__init__({**MODEL_CONFIG, **(projection_config or {})}, dtype, device, operations) + self.dtypes = {dtype} + self.execution_device = device + + def set_clip_options(self, options): + self.execution_device = options.get("execution_device", self.execution_device) + + def reset_clip_options(self): + pass + + def get_dynamic_vram__units(self): + units, last_units = self.model.get_dynamic_vram__units() + if self.model.pruned_embedding: + last_units = [*last_units, self.model.embed_tokens_prefill] + return [(self.model.audio_decoder, self.model.audio_extra_embedding), *units], last_units + + def encode_token_weights(self, token_weight_pairs): + token_ids = [token for token, _ in token_weight_pairs["minimax_music3"][0]] + input_ids = torch.tensor([token_ids], dtype=torch.long) + seed = token_weight_pairs["seed"] + max_audio_frames = token_weight_pairs["max_audio_frames"] + cfg_scale = token_weight_pairs["cfg_scale"] + top_k = token_weight_pairs["top_k"] + hidden = self.generate(input_ids, seed, max_audio_frames, self.execution_device, cfg_scale, top_k) + return hidden.unsqueeze(0), None, {} + + def load_state_dict(self, state_dict, strict=True, assign=False): + if self.model.pruned_embedding is None: + self.model.pruned_embedding = "model.embed_tokens_prefill.weight" in state_dict + if self.model.pruned_embedding: + del self.model.embed_tokens + else: + del self.model.embed_tokens_prefill, self.model.embed_tokens_audio + if self.model.pruned_lm_head is None: + self.model.pruned_lm_head = "model.lm_head_pruned.weight" in state_dict + if self.model.pruned_lm_head: + del self.model.lm_head + else: + del self.model.lm_head_pruned + return super().load_state_dict(state_dict, strict=strict, assign=assign) + + def load_sd(self, state_dict): + return self.load_state_dict(state_dict, strict=False, assign=getattr(self, "can_assign_sd", False)) diff --git a/comfy_api_nodes/apis/bria.py b/comfy_api_nodes/apis/bria.py index 7a98428c3..f55de74bc 100644 --- a/comfy_api_nodes/apis/bria.py +++ b/comfy_api_nodes/apis/bria.py @@ -57,6 +57,81 @@ class BriaRemoveBackgroundRequest(BaseModel): seed: int = Field(...) +class BriaGenFillRequest(BaseModel): + image: str = Field(...) + mask: str = Field( + ..., + description="Binary mask defining the region to fill: white (255) pixels are generated, " + "black (0) pixels are preserved. Must have the same aspect ratio as the image.", + ) + prompt: str = Field(...) + negative_prompt: str | None = Field(None) + refine_prompt: bool = Field(True) + seed: int = Field(...) + prompt_content_moderation: bool = Field(False, description="If true, returns 422 on prompt moderation failure.") + visual_input_content_moderation: bool = Field( + False, description="If true, returns 422 on image or mask moderation failure." + ) + visual_output_content_moderation: bool = Field( + False, description="If true, returns 422 on visual output moderation failure." + ) + + +class BriaEraseRequest(BaseModel): + image: str = Field(...) + mask: str = Field( + ..., + description="Binary mask defining the region to erase: white (255) pixels are removed, " + "black (0) pixels are preserved. Must have the same aspect ratio as the image.", + ) + mask_type: str = Field("manual", description="'manual' for hand-drawn masks, 'automatic' for segmentation masks.") + visual_input_content_moderation: bool = Field( + False, description="If true, returns 422 on image or mask moderation failure." + ) + visual_output_content_moderation: bool = Field( + False, description="If true, returns 422 on visual output moderation failure." + ) + + +class BriaExpandRequest(BaseModel): + image: str = Field(...) + aspect_ratio: str | float | None = Field( + None, + description="Target ratio: a preset string (1:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9) " + "or a float between 0.5 and 3.0. When set, the canvas/placement fields are ignored.", + ) + canvas_size: list[int] | None = Field(None, description="Output canvas [width, height]; area up to 5000x5000.") + original_image_size: list[int] | None = Field( + None, description="Size [width, height] of the original image inside the canvas." + ) + original_image_location: list[int] | None = Field( + None, + description="Top-left corner [x, y] of the original image inside the canvas; " + "values may fall outside the canvas, cropping the image.", + ) + prompt: str | None = Field(None, description="If omitted, Bria auto-generates a prompt from the image.") + negative_prompt: str | None = Field(None) + seed: int = Field(...) + prompt_content_moderation: bool = Field(False, description="If true, returns 422 on prompt moderation failure.") + visual_input_content_moderation: bool = Field( + False, description="If true, returns 422 on image moderation failure." + ) + visual_output_content_moderation: bool = Field( + False, description="If true, returns 422 on visual output moderation failure." + ) + + +class BriaIncreaseResolutionRequest(BaseModel): + image: str = Field(...) + desired_increase: int = Field(..., description="Resolution multiplier, 2 or 4.") + visual_input_content_moderation: bool = Field( + False, description="If true, returns 422 on image moderation failure." + ) + visual_output_content_moderation: bool = Field( + False, description="If true, returns 422 on visual output moderation failure." + ) + + class BriaStatusResponse(BaseModel): request_id: str = Field(...) status_url: str = Field(...) @@ -72,6 +147,26 @@ class BriaRemoveBackgroundResponse(BaseModel): result: BriaRemoveBackgroundResult | None = Field(None) +class BriaImageResult(BaseModel): + image_url: str = Field(...) + + +class BriaImageResultResponse(BaseModel): + status: str = Field(...) + result: BriaImageResult | None = Field(None) + + +class BriaExpandResult(BaseModel): + image_url: str = Field(...) + prompt: str | None = Field(None) + seed: int | None = Field(None) + + +class BriaExpandResponse(BaseModel): + status: str = Field(...) + result: BriaExpandResult | None = Field(None) + + class BriaImageEditResult(BaseModel): structured_prompt: str = Field(...) image_url: str = Field(...) diff --git a/comfy_api_nodes/apis/grok.py b/comfy_api_nodes/apis/grok.py index 526d8c8ab..82dfc8b82 100644 --- a/comfy_api_nodes/apis/grok.py +++ b/comfy_api_nodes/apis/grok.py @@ -9,6 +9,7 @@ class ImageGenerationRequest(BaseModel): seed: int = Field(...) response_format: str = Field("url") resolution: str = Field(...) + quality: str | None = Field(None) class InputUrlObject(BaseModel): @@ -28,6 +29,7 @@ class ImageEditRequest(BaseModel): seed: int = Field(...) response_format: str = Field("url") aspect_ratio: str | None = Field(...) + quality: str | None = Field(None) class VideoGenerationRequest(BaseModel): diff --git a/comfy_api_nodes/apis/minimax.py b/comfy_api_nodes/apis/minimax.py index bac4572d4..12a0853ac 100644 --- a/comfy_api_nodes/apis/minimax.py +++ b/comfy_api_nodes/apis/minimax.py @@ -161,12 +161,30 @@ class Hailuo03TaskCreationRequest(BaseModel): ..., min_length=1 ) resolution: str = Field(...) - duration: int = Field(..., ge=5, le=15) + duration: int = Field(..., ge=4, le=15) ratio: str | None = Field(None) seed: int | None = Field(None, ge=0, le=4294967295) aigc_watermark: bool | None = Field(None) +class Hailuo03ContextIRRequest(BaseModel): + model: str = Field(...) + content: list[Hailuo03TextContent | Hailuo03ImageContent | Hailuo03VideoContent | Hailuo03AudioContent] = Field( + ..., min_length=1 + ) + duration: int = Field(..., ge=4, le=15) + ratio: str | None = Field(None) + + +class Hailuo03RegenerationRequest(BaseModel): + model: str = Field(...) + content: list[Hailuo03TextContent | Hailuo03ImageContent | Hailuo03VideoContent | Hailuo03AudioContent] = Field( + ..., min_length=1 + ) + resolution: str = Field(...) + aigc_watermark: bool | None = Field(None) + + class Hailuo03TaskCreationResponse(BaseModel): task_id: str = Field(...) @@ -178,6 +196,7 @@ class Hailuo03TaskError(BaseModel): class Hailuo03TaskContent(BaseModel): url: str | None = Field(None) + prompt: str | None = Field(None) class Hailuo03TaskUsage(BaseModel): diff --git a/comfy_api_nodes/apis/qwen.py b/comfy_api_nodes/apis/qwen.py new file mode 100644 index 000000000..90b68dee8 --- /dev/null +++ b/comfy_api_nodes/apis/qwen.py @@ -0,0 +1,46 @@ +from pydantic import BaseModel, Field + + +class QwenImageContentItem(BaseModel): + image: str | None = Field(None) + text: str | None = Field(None) + + +class QwenImageMessage(BaseModel): + role: str = Field("user") + content: list[QwenImageContentItem] = Field(...) + + +class QwenImageInputField(BaseModel): + messages: list[QwenImageMessage] = Field(...) + + +class QwenImageParametersField(BaseModel): + size: str | None = Field(None, description="Output resolution as 'width*height'; omit for the model default.") + n: int = Field(1, ge=1, le=6) + seed: int = Field(..., ge=0, le=2147483647) + prompt_extend: bool = Field(True) + watermark: bool = Field(False) + negative_prompt: str | None = Field(None) + + +class QwenImageGenerationRequest(BaseModel): + model: str = Field(...) + input: QwenImageInputField = Field(...) + parameters: QwenImageParametersField = Field(...) + + +class QwenImageChoice(BaseModel): + finish_reason: str | None = Field(None) + message: QwenImageMessage | None = Field(None) + + +class QwenImageOutputField(BaseModel): + choices: list[QwenImageChoice] = Field(default_factory=list) + + +class QwenImageGenerationResponse(BaseModel): + output: QwenImageOutputField | None = Field(None) + request_id: str = Field(...) + code: str | None = Field(None, description="Error code for the failed request.") + message: str | None = Field(None, description="Details about the failed request.") diff --git a/comfy_api_nodes/nodes_bria.py b/comfy_api_nodes/nodes_bria.py index 77f780a3b..90cade2d0 100644 --- a/comfy_api_nodes/nodes_bria.py +++ b/comfy_api_nodes/nodes_bria.py @@ -6,7 +6,13 @@ from typing_extensions import override from comfy_api.latest import IO, ComfyExtension, Input from comfy_api_nodes.apis.bria import ( BriaEditImageRequest, + BriaEraseRequest, + BriaExpandRequest, + BriaExpandResponse, + BriaGenFillRequest, BriaImageEditResponse, + BriaImageResultResponse, + BriaIncreaseResolutionRequest, BriaRemoveBackgroundRequest, BriaRemoveBackgroundResponse, BriaRemoveVideoBackgroundRequest, @@ -21,13 +27,30 @@ from comfy_api_nodes.util import ( convert_mask_to_image, download_url_to_image_tensor, download_url_to_video_output, + downscale_image_tensor_by_max_side, + get_image_dimensions, poll_op, sync_op, upload_image_to_comfyapi, upload_video_to_comfyapi, + validate_string, validate_video_duration, ) +BRIA_MAX_OUTPUT_SIDE = 8192 +BRIA_MIN_RATIO = 0.5 +BRIA_MAX_RATIO = 3.0 +BRIA_MIN_SHORT_SIDE = 224 + + +def _upscaled_output_side(height: int, width: int, multiplier: int) -> int: + prescale = max(1.0, BRIA_MIN_SHORT_SIDE / min(height, width)) + return round(max(height, width) * prescale * multiplier) + + +def _smallest_output_side(height: int, width: int, multiplier: int) -> int: + return round(max(height, width) / min(height, width) * BRIA_MIN_SHORT_SIDE * multiplier) + class BriaImageEditNode(IO.ComfyNode): @@ -243,6 +266,503 @@ class BriaRemoveImageBackground(IO.ComfyNode): return IO.NodeOutput(await download_url_to_image_tensor(response.result.image_url)) +def _mask_to_binary_image(mask: Input.Image, action: str) -> torch.Tensor: + binary = (mask > 0.5).float() + if not binary.any(): + raise ValueError( + f"The mask is empty, so there is nothing to {action}. Masks are binarized at 50%: " + f"areas painted at less than half opacity are ignored." + ) + return convert_mask_to_image(binary) + + +def _validate_mask_aspect_ratio(image: Input.Image, mask: Input.Image) -> None: + ih, iw = image.shape[1], image.shape[2] + mh, mw = mask.shape[-2], mask.shape[-1] + if abs(iw * mh - ih * mw) > 0.01 * ih * mw: + raise ValueError(f"Mask must have the same aspect ratio as the image: image is {iw}x{ih}, mask is {mw}x{mh}.") + + +class BriaGenFill(IO.ComfyNode): + + @classmethod + def define_schema(cls): + return IO.Schema( + node_id="BriaGenFill", + display_name="Bria Generative Fill", + category="partner/image/Bria", + description="Generate objects or scenery inside a masked region of an image using Bria.", + inputs=[ + IO.Image.Input("image"), + IO.Mask.Input( + "mask", + tooltip="White areas are filled with generated content, black areas are preserved. " + "The mask is binarized before sending, so partially painted areas count as white. " + "Must have the same aspect ratio as the image.", + ), + IO.String.Input( + "prompt", + multiline=True, + default="", + tooltip="Description of what to generate inside the masked region.", + ), + IO.String.Input("negative_prompt", multiline=True, default=""), + IO.Boolean.Input( + "refine_prompt", + default=True, + tooltip="Automatically adjust the prompt for better results; " + "disable to use the prompt exactly as written.", + ), + IO.Int.Input( + "seed", + default=42, + min=1, + max=2147483647, + step=1, + display_mode=IO.NumberDisplay.number, + control_after_generate=True, + ), + IO.DynamicCombo.Input( + "moderation", + options=[ + IO.DynamicCombo.Option("false", []), + IO.DynamicCombo.Option( + "true", + [ + IO.Boolean.Input("prompt_content_moderation", default=False), + IO.Boolean.Input("visual_input_moderation", default=False), + IO.Boolean.Input("visual_output_moderation", default=False), + ], + ), + ], + tooltip="Moderation settings", + ), + ], + outputs=[IO.Image.Output()], + hidden=[ + IO.Hidden.auth_token_comfy_org, + IO.Hidden.api_key_comfy_org, + IO.Hidden.unique_id, + ], + is_api_node=True, + price_badge=IO.PriceBadge( + expr="""{"type":"usd","usd":0.0429}""", + ), + ) + + @classmethod + async def execute( + cls, + image: Input.Image, + mask: Input.Image, + prompt: str, + negative_prompt: str, + refine_prompt: bool, + seed: int, + moderation: InputModerationSettings, + ) -> IO.NodeOutput: + validate_string(prompt, min_length=1) + _validate_mask_aspect_ratio(image, mask) + mask_image = _mask_to_binary_image(mask, "fill") + response = await sync_op( + cls, + ApiEndpoint(path="/proxy/bria/v2/image/edit/gen_fill", method="POST"), + data=BriaGenFillRequest( + image=await upload_image_to_comfyapi(cls, image, total_pixels=None, wait_label="Uploading image"), + mask=await upload_image_to_comfyapi( + cls, mask_image, total_pixels=None, wait_label="Uploading mask" + ), + prompt=prompt, + negative_prompt=negative_prompt if negative_prompt else None, + refine_prompt=refine_prompt, + seed=seed, + prompt_content_moderation=moderation.get("prompt_content_moderation", False), + visual_input_content_moderation=moderation.get("visual_input_moderation", False), + visual_output_content_moderation=moderation.get("visual_output_moderation", False), + ), + response_model=BriaStatusResponse, + ) + response = await poll_op( + cls, + ApiEndpoint(path=f"/proxy/bria/v2/status/{response.request_id}"), + status_extractor=lambda r: r.status, + response_model=BriaImageResultResponse, + ) + return IO.NodeOutput(await download_url_to_image_tensor(response.result.image_url)) + + +class BriaEraser(IO.ComfyNode): + + @classmethod + def define_schema(cls): + return IO.Schema( + node_id="BriaEraser", + display_name="Bria Eraser", + category="partner/image/Bria", + description="Remove objects or areas outlined by a mask from an image using Bria.", + inputs=[ + IO.Image.Input("image"), + IO.Mask.Input( + "mask", + tooltip="White areas are erased, black areas are preserved. " + "The mask is binarized before sending, so partially painted areas count as white. " + "Must have the same aspect ratio as the image.", + ), + IO.Combo.Input( + "mask_type", + options=["manual", "automatic"], + tooltip="manual for hand-drawn or brush masks, " + "automatic for masks produced by segmentation models such as SAM.", + ), + IO.DynamicCombo.Input( + "moderation", + options=[ + IO.DynamicCombo.Option("false", []), + IO.DynamicCombo.Option( + "true", + [ + IO.Boolean.Input("visual_input_moderation", default=False), + IO.Boolean.Input("visual_output_moderation", default=False), + ], + ), + ], + tooltip="Moderation settings", + ), + ], + outputs=[IO.Image.Output()], + hidden=[ + IO.Hidden.auth_token_comfy_org, + IO.Hidden.api_key_comfy_org, + IO.Hidden.unique_id, + ], + is_api_node=True, + price_badge=IO.PriceBadge( + expr="""{"type":"usd","usd":0.0286}""", + ), + ) + + @classmethod + async def execute( + cls, + image: Input.Image, + mask: Input.Image, + mask_type: str, + moderation: dict, + ) -> IO.NodeOutput: + _validate_mask_aspect_ratio(image, mask) + mask_image = _mask_to_binary_image(mask, "erase") + response = await sync_op( + cls, + ApiEndpoint(path="/proxy/bria/v2/image/edit/erase", method="POST"), + data=BriaEraseRequest( + image=await upload_image_to_comfyapi(cls, image, total_pixels=None, wait_label="Uploading image"), + mask=await upload_image_to_comfyapi( + cls, mask_image, total_pixels=None, wait_label="Uploading mask" + ), + mask_type=mask_type, + visual_input_content_moderation=moderation.get("visual_input_moderation", False), + visual_output_content_moderation=moderation.get("visual_output_moderation", False), + ), + response_model=BriaStatusResponse, + ) + response = await poll_op( + cls, + ApiEndpoint(path=f"/proxy/bria/v2/status/{response.request_id}"), + status_extractor=lambda r: r.status, + response_model=BriaImageResultResponse, + ) + return IO.NodeOutput(await download_url_to_image_tensor(response.result.image_url)) + + +class BriaExpandImage(IO.ComfyNode): + + @classmethod + def define_schema(cls): + return IO.Schema( + node_id="BriaExpandImage", + display_name="Bria Expand Image", + category="partner/image/Bria", + description="Expand an image beyond its borders with generated content using Bria.", + inputs=[ + IO.Image.Input("image"), + IO.DynamicCombo.Input( + "expand_mode", + options=[ + *[IO.DynamicCombo.Option(ratio, []) for ratio in + ["1:1", "2:3", "3:2", "3:4", "4:3", "4:5", "5:4", "9:16", "16:9"]], + IO.DynamicCombo.Option( + "custom_ratio", + [ + IO.Int.Input( + "ratio_width", + default=21, + min=1, + max=100, + tooltip="Width side of the target ratio: 21 and 9 give 21:9.", + ), + IO.Int.Input( + "ratio_height", + default=9, + min=1, + max=100, + tooltip="Height side of the target ratio: 21 and 9 give 21:9. " + f"Bria only accepts width/height between {BRIA_MIN_RATIO} and " + f"{BRIA_MAX_RATIO}, so anything taller than 1:2 needs the manual mode.", + ), + ], + ), + IO.DynamicCombo.Option( + "manual", + [ + IO.Int.Input("canvas_width", default=1000, min=64, max=5000), + IO.Int.Input("canvas_height", default=1000, min=64, max=5000), + IO.Int.Input( + "image_width", + default=500, + min=1, + max=5000, + tooltip="Width of the original image inside the canvas.", + ), + IO.Int.Input( + "image_height", + default=500, + min=1, + max=5000, + tooltip="Height of the original image inside the canvas.", + ), + IO.Int.Input( + "image_x", + default=250, + min=-5000, + max=5000, + tooltip="X position of the image's top-left corner inside the canvas; " + "may fall outside the canvas, cropping the image.", + ), + IO.Int.Input( + "image_y", + default=250, + min=-5000, + max=5000, + tooltip="Y position of the image's top-left corner inside the canvas; " + "may fall outside the canvas, cropping the image.", + ), + ], + ), + ], + tooltip="Target shape of the expanded image: a preset aspect ratio, a custom ratio, " + "or manual placement of the original image on a canvas. " + "Manual is the only mode that can reach a canvas taller than 1:2.", + ), + IO.String.Input( + "prompt", + multiline=True, + default="", + tooltip="Optional description of the expanded scene; " + "when empty, Bria generates one from the image.", + ), + IO.String.Input("negative_prompt", multiline=True, default=""), + IO.Int.Input( + "seed", + default=42, + min=1, + max=2147483647, + step=1, + display_mode=IO.NumberDisplay.number, + control_after_generate=True, + ), + IO.DynamicCombo.Input( + "moderation", + options=[ + IO.DynamicCombo.Option("false", []), + IO.DynamicCombo.Option( + "true", + [ + IO.Boolean.Input("prompt_content_moderation", default=False), + IO.Boolean.Input("visual_input_moderation", default=False), + IO.Boolean.Input("visual_output_moderation", default=False), + ], + ), + ], + tooltip="Moderation settings", + ), + ], + outputs=[ + IO.Image.Output(), + IO.String.Output(display_name="prompt", tooltip="The prompt used for the expansion; " + "auto-generated by Bria when the prompt input is empty."), + ], + hidden=[ + IO.Hidden.auth_token_comfy_org, + IO.Hidden.api_key_comfy_org, + IO.Hidden.unique_id, + ], + is_api_node=True, + price_badge=IO.PriceBadge( + expr="""{"type":"usd","usd":0.0286}""", + ), + ) + + @classmethod + async def execute( + cls, + image: Input.Image, + expand_mode: dict, + prompt: str, + negative_prompt: str, + seed: int, + moderation: InputModerationSettings, + ) -> IO.NodeOutput: + mode = expand_mode["expand_mode"] + aspect_ratio = canvas_size = original_image_size = original_image_location = None + if mode == "manual": + canvas_size = [expand_mode["canvas_width"], expand_mode["canvas_height"]] + original_image_size = [expand_mode["image_width"], expand_mode["image_height"]] + original_image_location = [expand_mode["image_x"], expand_mode["image_y"]] + elif mode == "custom_ratio": + ratio_width, ratio_height = expand_mode["ratio_width"], expand_mode["ratio_height"] + aspect_ratio = ratio_width / ratio_height + if not BRIA_MIN_RATIO <= aspect_ratio <= BRIA_MAX_RATIO: + raise ValueError( + f"Bria accepts a width-to-height ratio between {BRIA_MIN_RATIO} and {BRIA_MAX_RATIO}: " + f"{ratio_width}:{ratio_height} is {aspect_ratio:.4f}. " + f"Use the manual expand mode to reach a canvas of any shape." + ) + else: + aspect_ratio = mode + response = await sync_op( + cls, + ApiEndpoint(path="/proxy/bria/v2/image/edit/expand", method="POST"), + data=BriaExpandRequest( + image=await upload_image_to_comfyapi(cls, image, total_pixels=None, wait_label="Uploading image"), + aspect_ratio=aspect_ratio, + canvas_size=canvas_size, + original_image_size=original_image_size, + original_image_location=original_image_location, + prompt=prompt if prompt else None, + negative_prompt=negative_prompt if negative_prompt else None, + seed=seed, + prompt_content_moderation=moderation.get("prompt_content_moderation", False), + visual_input_content_moderation=moderation.get("visual_input_moderation", False), + visual_output_content_moderation=moderation.get("visual_output_moderation", False), + ), + response_model=BriaStatusResponse, + ) + response = await poll_op( + cls, + ApiEndpoint(path=f"/proxy/bria/v2/status/{response.request_id}"), + status_extractor=lambda r: r.status, + response_model=BriaExpandResponse, + ) + return IO.NodeOutput( + await download_url_to_image_tensor(response.result.image_url), + response.result.prompt or "", + ) + + +class BriaIncreaseResolution(IO.ComfyNode): + + @classmethod + def define_schema(cls): + return IO.Schema( + node_id="BriaIncreaseResolution", + display_name="Bria Increase Resolution", + category="partner/image/Bria", + description="Upscale an image by 2x or 4x using Bria, preserving the original content.", + inputs=[ + IO.Image.Input("image"), + IO.Combo.Input( + "desired_increase", + options=["2", "4"], + tooltip="Resolution multiplier. The output must fit within 8192 pixels on each side.", + ), + IO.Boolean.Input( + "auto_downscale", + default=False, + tooltip="Automatically lower the multiplier, and downscale the input image if that is " + "still not enough, when the output would exceed the limit.", + ), + IO.DynamicCombo.Input( + "moderation", + options=[ + IO.DynamicCombo.Option("false", []), + IO.DynamicCombo.Option( + "true", + [ + IO.Boolean.Input("visual_input_moderation", default=False), + IO.Boolean.Input("visual_output_moderation", default=False), + ], + ), + ], + tooltip="Moderation settings", + ), + ], + outputs=[IO.Image.Output()], + hidden=[ + IO.Hidden.auth_token_comfy_org, + IO.Hidden.api_key_comfy_org, + IO.Hidden.unique_id, + ], + is_api_node=True, + price_badge=IO.PriceBadge( + expr="""{"type":"usd","usd":0.0286}""", + ), + ) + + @classmethod + async def execute( + cls, + image: Input.Image, + desired_increase: str, + auto_downscale: bool, + moderation: dict, + ) -> IO.NodeOutput: + multiplier = int(desired_increase) + height, width = get_image_dimensions(image) + if _upscaled_output_side(height, width, multiplier) > BRIA_MAX_OUTPUT_SIDE: + candidates = [c for c in (4, 2) if c <= multiplier] + if not auto_downscale: + predicted = _upscaled_output_side(height, width, multiplier) + raise ValueError( + f"Bria can upscale up to a maximum output dimension of {BRIA_MAX_OUTPUT_SIDE} pixels: " + f"input is {width}x{height}, x{multiplier} would be {predicted} pixels on the long side. " + f"Enable auto_downscale, or use a smaller input image or a lower multiplier." + ) + fitted = next( + (c for c in candidates if _upscaled_output_side(height, width, c) <= BRIA_MAX_OUTPUT_SIDE), None + ) + if fitted is not None: + multiplier = fitted + else: + shrinkable = next((c for c in sorted(candidates) if _smallest_output_side(height, width, c) + <= BRIA_MAX_OUTPUT_SIDE), None) + if shrinkable is None: + raise ValueError( + f"This image cannot be upscaled by Bria at any multiplier: it is {width}x{height}, and " + f"Bria first enlarges the short side to {BRIA_MIN_SHORT_SIDE} pixels, which pushes the " + f"long side past the {BRIA_MAX_OUTPUT_SIDE} pixel limit. Crop it to a squarer shape first." + ) + multiplier = shrinkable + image = downscale_image_tensor_by_max_side(image, max_side=BRIA_MAX_OUTPUT_SIDE // multiplier) + response = await sync_op( + cls, + ApiEndpoint(path="/proxy/bria/v2/image/edit/increase_resolution", method="POST"), + data=BriaIncreaseResolutionRequest( + image=await upload_image_to_comfyapi(cls, image, total_pixels=None, wait_label="Uploading image"), + desired_increase=multiplier, + visual_input_content_moderation=moderation.get("visual_input_moderation", False), + visual_output_content_moderation=moderation.get("visual_output_moderation", False), + ), + response_model=BriaStatusResponse, + ) + response = await poll_op( + cls, + ApiEndpoint(path=f"/proxy/bria/v2/status/{response.request_id}"), + status_extractor=lambda r: r.status, + response_model=BriaImageResultResponse, + ) + return IO.NodeOutput(await download_url_to_image_tensor(response.result.image_url)) + + class BriaRemoveVideoBackground(IO.ComfyNode): @classmethod @@ -572,6 +1092,10 @@ class BriaExtension(ComfyExtension): return [ BriaImageEditNode, BriaRemoveImageBackground, + BriaGenFill, + BriaEraser, + BriaExpandImage, + BriaIncreaseResolution, BriaRemoveVideoBackground, BriaVideoGreenScreen, BriaVideoReplaceBackground, diff --git a/comfy_api_nodes/nodes_grok.py b/comfy_api_nodes/nodes_grok.py index 672a3e537..c8f0f20d5 100644 --- a/comfy_api_nodes/nodes_grok.py +++ b/comfy_api_nodes/nodes_grok.py @@ -36,6 +36,26 @@ _GROK_VIDEO_MODEL_API_IDS = { "grok-imagine-video-1.5": "grok-imagine-video-1.5", } +_GROK_IMAGE_MODEL_API_IDS = { + "grok-imagine-image-2.0": "grok-imagine-image-2.0", +} + +_GROK_IMAGE_QUALITY_MODELS = {"grok-imagine-image-2.0"} + +_GROK_IMAGE_QUALITY_OPTIONS = ["medium", "low"] + +_GROK_IMAGE_EDIT_MAX_IMAGES = { + "grok-imagine-image-2.0": 3, + "grok-imagine-image-pro": 1, + "grok-imagine-image-quality": 3, + "grok-imagine-image": 3, +} + +_GROK_IMAGE_EDIT_ASPECT_RATIO_NEEDS_MULTIPLE = { + "grok-imagine-image-quality", + "grok-imagine-image", +} + _GROK_VOICE_OPTIONS = [ "none", "ara", @@ -132,6 +152,7 @@ class GrokImageNode(IO.ComfyNode): IO.Combo.Input( "model", options=[ + "grok-imagine-image-2.0", "grok-imagine-image-quality", "grok-imagine-image-pro", "grok-imagine-image", @@ -181,6 +202,12 @@ class GrokImageNode(IO.ComfyNode): "actual results are nondeterministic regardless of seed.", ), IO.Combo.Input("resolution", options=["1K", "2K"], optional=True), + IO.Combo.Input( + "quality", + options=_GROK_IMAGE_QUALITY_OPTIONS, + optional=True, + tooltip="Quality level, supported only by the grok-imagine-image-2.0 model.", + ), ], outputs=[ IO.Image.Output(), @@ -192,12 +219,15 @@ class GrokImageNode(IO.ComfyNode): ], is_api_node=True, price_badge=IO.PriceBadge( - depends_on=IO.PriceBadgeDepends(widgets=["model", "number_of_images", "resolution"]), + depends_on=IO.PriceBadgeDepends(widgets=["model", "number_of_images", "resolution", "quality"]), expr=""" ( - $rate := widgets.model = "grok-imagine-image-quality" - ? (widgets.resolution = "1k" ? 0.05 : 0.07) - : ($contains(widgets.model, "pro") ? 0.07 : 0.02); + $is1k := widgets.resolution = "1k"; + $rate := widgets.model = "grok-imagine-image-2.0" + ? (widgets.quality = "low" ? ($is1k ? 0.04 : 0.06) : ($is1k ? 0.06 : 0.08)) + : (widgets.model = "grok-imagine-image-quality" + ? ($is1k ? 0.05 : 0.07) + : ($contains(widgets.model, "pro") ? 0.07 : 0.02)); {"type":"usd","usd": $rate * widgets.number_of_images} ) """, @@ -213,18 +243,20 @@ class GrokImageNode(IO.ComfyNode): number_of_images: int, seed: int, resolution: str = "1K", + quality: str = "medium", ) -> IO.NodeOutput: validate_string(prompt, strip_whitespace=True, min_length=1) response = await sync_op( cls, ApiEndpoint(path="/proxy/xai/v1/images/generations", method="POST"), data=ImageGenerationRequest( - model=model, + model=_GROK_IMAGE_MODEL_API_IDS.get(model, model), prompt=prompt, aspect_ratio=aspect_ratio, n=number_of_images, seed=seed, resolution=resolution.lower(), + quality=quality if model in _GROK_IMAGE_QUALITY_MODELS else None, ), response_model=ImageGenerationResponse, ) @@ -255,7 +287,9 @@ _GROK_IMAGE_EDIT_ASPECT_RATIO_OPTIONS = [ ] -def _grok_image_edit_model_inputs(*, max_ref_images: int, with_aspect_ratio: bool): +def _grok_image_edit_model_inputs( + *, max_ref_images: int, with_aspect_ratio: bool, with_quality: bool = False, aspect_ratio_needs_multiple: bool = True +): inputs = [ IO.Autogrow.Input( "images", @@ -281,12 +315,18 @@ def _grok_image_edit_model_inputs(*, max_ref_images: int, with_aspect_ratio: boo display_mode=IO.NumberDisplay.number, ), ] + if with_quality: + inputs.append(IO.Combo.Input("quality", options=_GROK_IMAGE_QUALITY_OPTIONS)) if with_aspect_ratio: inputs.append( IO.Combo.Input( "aspect_ratio", options=_GROK_IMAGE_EDIT_ASPECT_RATIO_OPTIONS, - tooltip="Only allowed when multiple images are connected.", + tooltip=( + "Only allowed when multiple images are connected." + if aspect_ratio_needs_multiple + else "Aspect ratio of the edited image." + ), ) ) return inputs @@ -451,6 +491,15 @@ class GrokImageEditNodeV2(IO.ComfyNode): IO.DynamicCombo.Input( "model", options=[ + IO.DynamicCombo.Option( + "grok-imagine-image-2.0", + _grok_image_edit_model_inputs( + max_ref_images=3, + with_aspect_ratio=True, + with_quality=True, + aspect_ratio_needs_multiple=False, + ), + ), IO.DynamicCombo.Option( "grok-imagine-image-quality", _grok_image_edit_model_inputs(max_ref_images=3, with_aspect_ratio=True), @@ -488,18 +537,23 @@ class GrokImageEditNodeV2(IO.ComfyNode): is_api_node=True, price_badge=IO.PriceBadge( depends_on=IO.PriceBadgeDepends( - widgets=["model", "model.resolution", "model.number_of_images"], + widgets=["model", "model.resolution", "model.number_of_images", "model.quality"], ), expr=""" ( - $isQualityModel := widgets.model = "grok-imagine-image-quality"; + $is20 := widgets.model = "grok-imagine-image-2.0"; $isPro := $contains(widgets.model, "pro"); $res := $lookup(widgets, "model.resolution"); $n := $lookup(widgets, "model.number_of_images"); - $rate := $isQualityModel - ? ($res = "1k" ? 0.05 : 0.07) - : ($isPro ? 0.07 : 0.02); - $base := $isQualityModel ? 0.01 : 0.002; + $is1k := $res = "1k"; + $rate := $is20 + ? ($lookup(widgets, "model.quality") = "low" + ? ($is1k ? 0.04 : 0.06) + : ($is1k ? 0.06 : 0.08)) + : (widgets.model = "grok-imagine-image-quality" + ? ($is1k ? 0.05 : 0.07) + : ($isPro ? 0.07 : 0.02)); + $base := ($is20 or widgets.model = "grok-imagine-image-quality") ? 0.01 : 0.002; $output := $rate * $n; $isPro ? {"type":"usd","usd": $base + $output} @@ -525,13 +579,15 @@ class GrokImageEditNodeV2(IO.ComfyNode): image_tensors: list[Input.Image] = [t for t in images_dict.values() if t is not None] n_images = sum(get_number_of_images(t) for t in image_tensors) + max_images = _GROK_IMAGE_EDIT_MAX_IMAGES.get(model_id, 3) if n_images < 1: raise ValueError("At least one image is required for editing.") - if model_id == "grok-imagine-image-pro" and n_images > 1: - raise ValueError("The pro model supports only 1 input image.") - if model_id != "grok-imagine-image-pro" and n_images > 3: - raise ValueError("A maximum of 3 input images is supported.") - if aspect_ratio != "auto" and n_images == 1: + if n_images > max_images: + raise ValueError( + f"The {model_id} model supports at most {max_images} input " + f"image{'s' if max_images > 1 else ''}; {n_images} are connected." + ) + if aspect_ratio != "auto" and model_id in _GROK_IMAGE_EDIT_ASPECT_RATIO_NEEDS_MULTIPLE and n_images == 1: raise ValueError( "Custom aspect ratio is only allowed when multiple images are connected to the image input." ) @@ -547,7 +603,7 @@ class GrokImageEditNodeV2(IO.ComfyNode): cls, ApiEndpoint(path="/proxy/xai/v1/images/edits", method="POST"), data=ImageEditRequest( - model=model_id, + model=_GROK_IMAGE_MODEL_API_IDS.get(model_id, model_id), images=[ InputUrlObject(url=f"data:image/png;base64,{tensor_to_base64_string(i)}") for i in flat_tensors ], @@ -556,6 +612,7 @@ class GrokImageEditNodeV2(IO.ComfyNode): n=number_of_images, seed=seed, aspect_ratio=None if aspect_ratio == "auto" else aspect_ratio, + quality=model.get("quality") if model_id in _GROK_IMAGE_QUALITY_MODELS else None, ), response_model=ImageGenerationResponse, ) diff --git a/comfy_api_nodes/nodes_ltxv.py b/comfy_api_nodes/nodes_ltxv.py index 878e04b4e..44723dde2 100644 --- a/comfy_api_nodes/nodes_ltxv.py +++ b/comfy_api_nodes/nodes_ltxv.py @@ -6,8 +6,12 @@ from typing_extensions import override from comfy_api.latest import IO, ComfyExtension, Input, InputImpl from comfy_api_nodes.util import ( ApiEndpoint, + download_url_to_video_output, get_number_of_images, + poll_op, + sync_op, sync_op_raw, + upload_audio_to_comfyapi, upload_images_to_comfyapi, validate_string, ) @@ -17,6 +21,11 @@ MODELS_MAP = { "LTX-2 (Fast)": "ltx-2-fast", } +V25_MODELS_MAP = { + "LTX-2.5 (Fast)": "ltx-2-5-fast", + "LTX-2.5 (Pro)": "ltx-2-5-pro", +} + class ExecuteTaskRequest(BaseModel): prompt: str = Field(...) @@ -26,6 +35,48 @@ class ExecuteTaskRequest(BaseModel): fps: int | None = Field(25) generate_audio: bool | None = Field(True) image_uri: str | None = Field(None) + last_frame_uri: str | None = Field(None) + + +class AudioToVideoRequest(BaseModel): + prompt: str = Field(...) + model: str = Field(...) + resolution: str = Field(...) + audio_uri: str = Field(...) + image_uri: str | None = Field(None) + + +class Ltx25SubmitResponse(BaseModel): + id: str = Field(...) + + +class Ltx25JobResult(BaseModel): + video_url: str | None = Field(None) + + +class Ltx25JobStatusResponse(BaseModel): + id: str = Field(...) + status: str = Field(...) + result: Ltx25JobResult | None = Field(None) + + +async def _v25_submit_and_poll(cls: type[IO.ComfyNode], route: str, data: BaseModel) -> IO.NodeOutput: + submit = await sync_op( + cls, + ApiEndpoint(f"/proxy/ltx/v2/{route}", "POST"), + response_model=Ltx25SubmitResponse, + data=data, + max_retries=1, + ) + job = await poll_op( + cls, + ApiEndpoint(f"/proxy/ltx/v2/{route}/{submit.id}"), + response_model=Ltx25JobStatusResponse, + status_extractor=lambda r: r.status, + ) + if not job.result or not job.result.video_url: + raise RuntimeError(f"LTX job {job.id} completed without a video URL.") + return IO.NodeOutput(await download_url_to_video_output(job.result.video_url, cls=cls)) PRICE_BADGE = IO.PriceBadge( @@ -43,6 +94,128 @@ PRICE_BADGE = IO.PriceBadge( """, ) +V25_PRICE_BADGE = IO.PriceBadge( + depends_on=IO.PriceBadgeDepends(widgets=["model", "model.duration", "model.resolution"]), + expr=""" + ( + $prices := { + "ltx-2.5 (fast)": { + "1280x720":0.1287,"720x1280":0.1287, + "1920x1080":0.1859,"1080x1920":0.1859, + "2560x1440":0.2717,"1440x2560":0.2717, + "3840x2160":0.429,"2160x3840":0.429 + }, + "ltx-2.5 (pro)": { + "1280x720":0.1716,"720x1280":0.1716, + "1920x1080":0.2431,"1080x1920":0.2431 + } + }; + $model := $lookup(widgets, "model"); + $table := $type($model) = "string" ? $lookup($prices, $model) : undefined; + $res := $lookup(widgets, "model.resolution"); + $pps := $type($table) = "object" and $type($res) = "string" ? $lookup($table, $res) : undefined; + $durRaw := $lookup(widgets, "model.duration"); + $dur := $type($durRaw) in ["string", "number"] ? $number($durRaw) : undefined; + $type($pps) = "number" and $type($dur) = "number" + ? {"type":"usd","usd": $pps * $dur} + : undefined + ) + """, +) + +V25_A2V_PRICE_BADGE = IO.PriceBadge( + depends_on=IO.PriceBadgeDepends(widgets=["model"]), + expr=""" + ( + $rates := {"ltx-2.5 (fast)":0.1859, "ltx-2.5 (pro)":0.2431}; + $model := $lookup(widgets, "model"); + $rate := $type($model) = "string" ? $lookup($rates, $model) : undefined; + $type($rate) = "number" + ? {"type":"usd","usd": $rate, "format":{"suffix":"/second"}} + : undefined + ) + """, +) + + +def _v25_generation_inputs( + durations: list[str], resolutions: list[str], fps_options: list[str], tooltip: str | None +) -> list: + return [ + IO.Combo.Input( + "duration", + options=durations, + default="8", + tooltip=tooltip, + ), + IO.Combo.Input( + "resolution", + options=resolutions, + default="1920x1080", + ), + IO.Combo.Input("fps", options=fps_options, default="25"), + IO.Boolean.Input( + "generate_audio", + default=True, + tooltip="When true, the generated video will include AI-generated audio matching the scene.", + advanced=True, + ), + ] + + +def _v25_model_combo() -> IO.DynamicCombo.Input: + return IO.DynamicCombo.Input( + "model", + options=[ + IO.DynamicCombo.Option( + "LTX-2.5 (Fast)", + _v25_generation_inputs( + ["2", "3", "4", "5", "6", "8", "10", "12", "14", "16", "18", "20"], + [ + "1280x720", + "720x1280", + "1920x1080", + "1080x1920", + "2560x1440", + "1440x2560", + "3840x2160", + "2160x3840", + ], + ["24", "25", "48", "50"], + "Video duration in seconds. Durations over 10s require a 720p/1080p resolution and 24/25 FPS.", + ), + ), + IO.DynamicCombo.Option( + "LTX-2.5 (Pro)", + _v25_generation_inputs( + ["2", "3", "4", "5", "6", "8", "10"], + ["1280x720", "720x1280", "1920x1080", "1080x1920"], + ["24", "25", "50"], + "Video duration in seconds.", + ), + ), + ], + ) + + +def _v25_seed_input() -> IO.Int.Input: + return IO.Int.Input( + "seed", + default=42, + min=0, + max=0xFFFFFFFF, + control_after_generate=True, + tooltip="Seed to determine if node should re-run; " + "actual results are nondeterministic regardless of seed.", + ) + + +def _v25_validate_settings(model: dict) -> None: + if int(model["duration"]) > 10 and ( + int(model["fps"]) > 25 or model["resolution"] in ("2560x1440", "1440x2560", "3840x2160", "2160x3840") + ): + raise ValueError("Durations over 10s require a 720p or 1080p resolution and 24/25 FPS.") + class TextToVideoNode(IO.ComfyNode): @classmethod @@ -86,6 +259,7 @@ class TextToVideoNode(IO.ComfyNode): IO.Hidden.unique_id, ], is_api_node=True, + is_deprecated=True, price_badge=PRICE_BADGE, ) @@ -164,6 +338,7 @@ class ImageToVideoNode(IO.ComfyNode): IO.Hidden.unique_id, ], is_api_node=True, + is_deprecated=True, price_badge=PRICE_BADGE, ) @@ -203,12 +378,217 @@ class ImageToVideoNode(IO.ComfyNode): return IO.NodeOutput(InputImpl.VideoFromFile(BytesIO(response))) +class Ltx25TextToVideoNode(IO.ComfyNode): + @classmethod + def define_schema(cls): + return IO.Schema( + node_id="LtxApi25TextToVideo", + display_name="LTX 2.5 Text To Video", + category="partner/video/LTXV", + description="Professional-quality videos with customizable duration and resolution.", + inputs=[ + _v25_model_combo(), + IO.String.Input( + "prompt", + multiline=True, + default="", + ), + _v25_seed_input(), + ], + outputs=[ + IO.Video.Output(), + ], + hidden=[ + IO.Hidden.auth_token_comfy_org, + IO.Hidden.api_key_comfy_org, + IO.Hidden.unique_id, + ], + is_api_node=True, + price_badge=V25_PRICE_BADGE, + ) + + @classmethod + async def execute( + cls, + model: dict, + prompt: str, + seed: int = 42, + ) -> IO.NodeOutput: + validate_string(prompt, min_length=1, max_length=10000) + _v25_validate_settings(model) + return await _v25_submit_and_poll( + cls, + "text-to-video", + ExecuteTaskRequest( + prompt=prompt, + model=V25_MODELS_MAP[model["model"]], + duration=int(model["duration"]), + resolution=model["resolution"], + fps=int(model["fps"]), + generate_audio=model["generate_audio"], + ), + ) + + +class Ltx25ImageToVideoNode(IO.ComfyNode): + @classmethod + def define_schema(cls): + return IO.Schema( + node_id="LtxApi25ImageToVideo", + display_name="LTX 2.5 Image To Video", + category="partner/video/LTXV", + description="Professional-quality videos with customizable duration and resolution based on start image.", + inputs=[ + IO.Image.Input("image", tooltip="First frame to be used for the video."), + _v25_model_combo(), + IO.String.Input( + "prompt", + multiline=True, + default="", + ), + _v25_seed_input(), + IO.Image.Input( + "last_frame", + optional=True, + tooltip="Last frame to be used for the video.", + ), + ], + outputs=[ + IO.Video.Output(), + ], + hidden=[ + IO.Hidden.auth_token_comfy_org, + IO.Hidden.api_key_comfy_org, + IO.Hidden.unique_id, + ], + is_api_node=True, + price_badge=V25_PRICE_BADGE, + ) + + @classmethod + async def execute( + cls, + image: Input.Image, + model: dict, + prompt: str, + seed: int = 42, + last_frame: Input.Image | None = None, + ) -> IO.NodeOutput: + validate_string(prompt, min_length=1, max_length=10000) + _v25_validate_settings(model) + if get_number_of_images(image) != 1: + raise ValueError("Currently only one input image is supported.") + last_frame_uri = None + if last_frame is not None: + if get_number_of_images(last_frame) != 1: + raise ValueError("Currently only one last frame image is supported.") + last_frame_uri = (await upload_images_to_comfyapi(cls, last_frame, max_images=1, mime_type="image/png"))[0] + return await _v25_submit_and_poll( + cls, + "image-to-video", + ExecuteTaskRequest( + image_uri=(await upload_images_to_comfyapi(cls, image, max_images=1, mime_type="image/png"))[0], + last_frame_uri=last_frame_uri, + prompt=prompt, + model=V25_MODELS_MAP[model["model"]], + duration=int(model["duration"]), + resolution=model["resolution"], + fps=int(model["fps"]), + generate_audio=model["generate_audio"], + ), + ) + + +class Ltx25AudioToVideoNode(IO.ComfyNode): + @classmethod + def define_schema(cls): + return IO.Schema( + node_id="LtxApi25AudioToVideo", + display_name="LTX 2.5 Audio To Video", + category="partner/video/LTXV", + description="Generate a video driven by an audio track, with an optional first frame image.", + inputs=[ + IO.Audio.Input( + "audio", + tooltip="Audio track driving the video. Its length (2-20 seconds) sets the video duration.", + ), + IO.DynamicCombo.Input( + "model", + options=[ + IO.DynamicCombo.Option( + "LTX-2.5 (Fast)", + [IO.Combo.Input("resolution", options=["1920x1080", "1080x1920"])], + ), + IO.DynamicCombo.Option( + "LTX-2.5 (Pro)", + [IO.Combo.Input("resolution", options=["1920x1080", "1080x1920"])], + ), + ], + ), + IO.String.Input( + "prompt", + multiline=True, + default="", + ), + _v25_seed_input(), + IO.Image.Input( + "image", + optional=True, + tooltip="Optional first frame to be used for the video.", + ), + ], + outputs=[ + IO.Video.Output(), + ], + hidden=[ + IO.Hidden.auth_token_comfy_org, + IO.Hidden.api_key_comfy_org, + IO.Hidden.unique_id, + ], + is_api_node=True, + price_badge=V25_A2V_PRICE_BADGE, + ) + + @classmethod + async def execute( + cls, + audio: Input.Audio, + model: dict, + prompt: str, + seed: int = 42, + image: Input.Image | None = None, + ) -> IO.NodeOutput: + validate_string(prompt, min_length=1, max_length=10000) + audio_duration = audio["waveform"].shape[-1] / audio["sample_rate"] + if not 2 <= audio_duration <= 20: + raise ValueError(f"Audio duration must be between 2 and 20 seconds, got {audio_duration:.1f}s.") + image_uri = None + if image is not None: + if get_number_of_images(image) != 1: + raise ValueError("Currently only one input image is supported.") + image_uri = (await upload_images_to_comfyapi(cls, image, max_images=1, mime_type="image/png"))[0] + return await _v25_submit_and_poll( + cls, + "audio-to-video", + AudioToVideoRequest( + prompt=prompt, + model=V25_MODELS_MAP[model["model"]], + resolution=model["resolution"], + audio_uri=await upload_audio_to_comfyapi(cls, audio), + image_uri=image_uri, + ), + ) + + class LtxvApiExtension(ComfyExtension): @override async def get_node_list(self) -> list[type[IO.ComfyNode]]: return [ TextToVideoNode, ImageToVideoNode, + Ltx25TextToVideoNode, + Ltx25ImageToVideoNode, + Ltx25AudioToVideoNode, ] diff --git a/comfy_api_nodes/nodes_minimax.py b/comfy_api_nodes/nodes_minimax.py index 3c1d29257..de3895221 100644 --- a/comfy_api_nodes/nodes_minimax.py +++ b/comfy_api_nodes/nodes_minimax.py @@ -3,12 +3,14 @@ from typing import Optional import torch from typing_extensions import override -from comfy_api.latest import IO, ComfyExtension +from comfy_api.latest import IO, ComfyExtension, Input from comfy_api_nodes.apis.minimax import ( Hailuo03AudioContent, Hailuo03AudioContentUrl, + Hailuo03ContextIRRequest, Hailuo03ImageContent, Hailuo03ImageContentUrl, + Hailuo03RegenerationRequest, Hailuo03TaskCreationRequest, Hailuo03TaskCreationResponse, Hailuo03TaskQueryResponse, @@ -456,6 +458,9 @@ HAILUO_03_QUERY_ENDPOINT = "/proxy/minimax/v2/query/video_generation" # + /{tas HAILUO_03_MODELS = {"MiniMax H3": "MiniMax-H3"} HAILUO_03_FAILED_STATUSES = ["failed", "cancelled", "expired"] +HAILUO_03_CONTEXT_IR_ENDPOINT = "/proxy/minimax/v2/h3_context_ir" +HAILUO_03_REGENERATION_ENDPOINT = "/proxy/minimax/v2/video_regeneration" + def _hailuo03_model_inputs(include_ratio: bool = True, allow_adaptive: bool = True): inputs = [ @@ -487,10 +492,10 @@ def _hailuo03_model_inputs(include_ratio: bool = True, allow_adaptive: bool = Tr IO.Int.Input( "duration", default=5, - min=5, + min=4, max=15, step=1, - tooltip="Duration of the output video in seconds (5-15).", + tooltip="Duration of the output video in seconds (4-15).", display_mode=IO.NumberDisplay.slider, ) ) @@ -939,6 +944,592 @@ class MinimaxHailuo03ReferenceNode(IO.ComfyNode): ) +class MinimaxHailuo03ContextIRNode(IO.ComfyNode): + @classmethod + def define_schema(cls): + return IO.Schema( + node_id="MinimaxHailuo03ContextIRNode", + display_name="MiniMax H3 Context IR (Prompt Enhancer)", + category="partner/video/MiniMax", + description="Analyze text and media context with MiniMax H3 Context IR and produce an enhanced, " + "structured video prompt. Feed the output into the prompt of a MiniMax H3 video node and attach " + "the same media there in the same order, because the enhanced prompt refers to the attached " + "media by position.", + inputs=[ + IO.DynamicCombo.Input( + "model", + options=[ + IO.DynamicCombo.Option( + "MiniMax H3", + [ + IO.String.Input( + "prompt", + multiline=True, + default="", + tooltip="Description of the video you intend to generate.", + ), + IO.Int.Input( + "duration", + default=5, + min=4, + max=15, + step=1, + tooltip="Duration of the video you intend to generate, in seconds (4-15).", + display_mode=IO.NumberDisplay.slider, + ), + IO.Combo.Input( + "ratio", + options=["adaptive", "16:9", "4:3", "1:1", "3:4", "9:16", "21:9"], + default="adaptive", + tooltip="Aspect ratio of the video you intend to generate. 'adaptive' " + "requires at least one image, video, or audio input.", + ), + IO.Autogrow.Input( + "reference_images", + template=IO.Autogrow.TemplateNames( + IO.Image.Input("reference_image"), + names=[ + "image_1", + "image_2", + "image_3", + "image_4", + "image_5", + "image_6", + "image_7", + "image_8", + "image_9", + ], + min=0, + ), + tooltip="Subject or style reference images, referred to in the prompt " + "as 'Image 1'..'Image 9' in connection order. Up to 9 images.", + ), + IO.Autogrow.Input( + "reference_videos", + template=IO.Autogrow.TemplateNames( + IO.Video.Input("reference_video"), + names=["video_1", "video_2", "video_3"], + min=0, + ), + tooltip="Motion or scene reference videos, referred to in the prompt " + "as 'Video 1'..'Video 3' in connection order. Up to 3 videos, " + "2-15 seconds each, 15 seconds in total.", + ), + IO.Autogrow.Input( + "reference_audios", + template=IO.Autogrow.TemplateNames( + IO.Audio.Input("reference_audio"), + names=["audio_1", "audio_2", "audio_3"], + min=0, + ), + tooltip="Audio references, referred to in the prompt as " + "'Audio 1'..'Audio 3' in connection order. Up to 3 clips, " + "2-15 seconds each, 15 seconds in total. Cannot be used without " + "a reference image or video.", + ), + ], + ) + ], + tooltip="Model to use for prompt enhancement.", + ), + IO.Image.Input( + "first_frame", + tooltip="First frame of the video you intend to generate. Cannot be combined with " + "reference media.", + optional=True, + ), + IO.Image.Input( + "last_frame", + tooltip="Last frame of the video you intend to generate. Cannot be combined with " + "reference media.", + optional=True, + ), + ], + outputs=[ + IO.String.Output(), + ], + hidden=[ + IO.Hidden.auth_token_comfy_org, + IO.Hidden.api_key_comfy_org, + IO.Hidden.unique_id, + ], + is_api_node=True, + price_badge=IO.PriceBadge( + depends_on=IO.PriceBadgeDepends( + inputs=["first_frame", "last_frame"], + input_groups=["model.reference_images", "model.reference_videos", "model.reference_audios"], + ), + expr=""" + ( + $imgsRaw := $lookup(inputGroups, "model.reference_images"); + $imgs := $imgsRaw ? $imgsRaw : 0; + $vidsRaw := $lookup(inputGroups, "model.reference_videos"); + $vids := $vidsRaw ? $vidsRaw : 0; + $audsRaw := $lookup(inputGroups, "model.reference_audios"); + $auds := $audsRaw ? $audsRaw : 0; + $frames := (inputs.first_frame.connected ? 1 : 0) + (inputs.last_frame.connected ? 1 : 0); + ($imgs + $vids + $auds) > 0 + ? {"type": "range_usd", "min_usd": 0.06, "max_usd": 0.11, "format": {"approximate": true}} + : $frames > 0 + ? {"type": "usd", "usd": 0.05, "format": {"approximate": true}} + : {"type": "usd", "usd": 0.02, "format": {"approximate": true}} + ) + """, + ), + ) + + @classmethod + async def execute( + cls, + model: dict, + first_frame: torch.Tensor | None = None, + last_frame: torch.Tensor | None = None, + ) -> IO.NodeOutput: + validate_string(model["prompt"], strip_whitespace=True, min_length=1) + + reference_images = {k: v for k, v in (model.get("reference_images") or {}).items() if v is not None} + reference_videos = {k: v for k, v in (model.get("reference_videos") or {}).items() if v is not None} + reference_audios = {k: v for k, v in (model.get("reference_audios") or {}).items() if v is not None} + has_frames = first_frame is not None or last_frame is not None + has_references = bool(reference_images) or bool(reference_videos) or bool(reference_audios) + if has_frames and has_references: + raise ValueError( + "First/last frame and reference media are mutually exclusive. Use frames for an " + "image-to-video prompt, or reference media for a reference-to-video prompt." + ) + if reference_audios and not reference_images and not reference_videos: + raise ValueError("Reference audio cannot be used without a reference image or video.") + if not has_frames and not has_references and model["ratio"] == "adaptive": + raise ValueError( + "Ratio 'adaptive' is not supported for text-only requests; select an explicit aspect ratio." + ) + + for frame in (first_frame, last_frame): + if frame is not None: + validate_image_aspect_ratio(frame, (2, 5), (5, 2), strict=False) # 0.4 to 2.5 + validate_image_dimensions(frame, min_width=256, min_height=256) + for image in reference_images.values(): + validate_image_aspect_ratio(image, (2, 5), (5, 2), strict=False) # 0.4 to 2.5 + validate_image_dimensions(image, min_width=256, min_height=256) + + total_video_duration = 0.0 + for i, video in enumerate(reference_videos.values(), 1): + try: + fps = float(video.get_frame_rate()) + except Exception: + fps = 0.0 + if fps and not (23.9 <= fps <= 60.5): + raise ValueError(f"Reference video {i} is {fps:.2f} FPS. Supported range is 23.976-60 FPS.") + try: + dur = video.get_duration() + except Exception: + continue + if dur < 1.8: + raise ValueError(f"Reference video {i} is too short: {dur:.1f}s. Minimum duration is 2 seconds.") + total_video_duration += dur + if total_video_duration > 15.1: + raise ValueError( + f"Total reference video duration is {total_video_duration:.1f}s. Maximum is 15 seconds." + ) + + total_audio_duration = 0.0 + for i, audio in enumerate(reference_audios.values(), 1): + dur = int(audio["waveform"].shape[-1]) / int(audio["sample_rate"]) + if dur < 1.8: + raise ValueError(f"Reference audio {i} is too short: {dur:.1f}s. Minimum duration is 2 seconds.") + total_audio_duration += dur + if total_audio_duration > 15.1: + raise ValueError( + f"Total reference audio duration is {total_audio_duration:.1f}s. Maximum is 15 seconds." + ) + + content: list = [Hailuo03TextContent(text=model["prompt"])] + if first_frame is not None: + content.append( + Hailuo03ImageContent( + image_url=Hailuo03ImageContentUrl( + url=( + await upload_images_to_comfyapi( + cls, first_frame, max_images=1, wait_label="Uploading first frame" + ) + )[0], + ), + role="first_frame", + ) + ) + if last_frame is not None: + content.append( + Hailuo03ImageContent( + image_url=Hailuo03ImageContentUrl( + url=( + await upload_images_to_comfyapi( + cls, last_frame, max_images=1, wait_label="Uploading last frame" + ) + )[0], + ), + role="last_frame", + ) + ) + for i, image in enumerate(reference_images.values(), 1): + content.append( + Hailuo03ImageContent( + image_url=Hailuo03ImageContentUrl( + url=( + await upload_images_to_comfyapi( + cls, image, max_images=1, wait_label=f"Uploading image {i}" + ) + )[0], + ), + role="reference_image", + ) + ) + for i, video in enumerate(reference_videos.values(), 1): + content.append( + Hailuo03VideoContent( + video_url=Hailuo03VideoContentUrl( + url=await upload_video_to_comfyapi(cls, video, wait_label=f"Uploading video {i}"), + ), + ) + ) + for audio in reference_audios.values(): + content.append( + Hailuo03AudioContent( + audio_url=Hailuo03AudioContentUrl( + url=await upload_audio_to_comfyapi( + cls, + audio, + container_format="mp3", + codec_name="libmp3lame", + mime_type="audio/mpeg", + ), + ), + ) + ) + + response = await sync_op( + cls, + ApiEndpoint(path=HAILUO_03_CONTEXT_IR_ENDPOINT, method="POST"), + response_model=Hailuo03TaskCreationResponse, + data=Hailuo03ContextIRRequest( + model=HAILUO_03_MODELS[model["model"]], + content=content, + duration=model["duration"], + ratio=None if model["ratio"] == "adaptive" else model["ratio"], + ), + ) + task_result = await poll_op( + cls, + ApiEndpoint(path=f"{HAILUO_03_QUERY_ENDPOINT}/{response.task_id}"), + response_model=Hailuo03TaskQueryResponse, + status_extractor=lambda r: r.task.status, + failed_statuses=HAILUO_03_FAILED_STATUSES, + poll_interval=5, + ) + prompt = task_result.task.content.prompt if task_result.task.content else None + if not prompt: + raise Exception(f"No enhanced prompt in the response: {task_result.model_dump()}") + return IO.NodeOutput(prompt) + + +class MinimaxHailuo03RegenerateNode(IO.ComfyNode): + @classmethod + def define_schema(cls): + return IO.Schema( + node_id="MinimaxHailuo03RegenerateNode", + display_name="MiniMax H3 Regenerate to 2K", + category="partner/video/MiniMax", + description="Re-render a MiniMax H3 768P output at 2K resolution. Connect the unmodified 768P " + "video and the exact prompt used to generate it; if the original generation used first/last " + "frames or reference media, attach the same inputs.", + inputs=[ + IO.DynamicCombo.Input( + "model", + options=[ + IO.DynamicCombo.Option( + "MiniMax H3", + [ + IO.String.Input( + "prompt", + multiline=True, + default="", + tooltip="The exact prompt used to generate the source video.", + ), + IO.Combo.Input( + "resolution", + options=["2K"], + tooltip="Resolution to re-render the source video at.", + ), + IO.Autogrow.Input( + "reference_images", + template=IO.Autogrow.TemplateNames( + IO.Image.Input("reference_image"), + names=[ + "image_1", + "image_2", + "image_3", + "image_4", + "image_5", + "image_6", + "image_7", + "image_8", + "image_9", + ], + min=0, + ), + tooltip="Reference images from the original generation, in the same " + "order. Up to 9 images.", + ), + IO.Autogrow.Input( + "reference_videos", + template=IO.Autogrow.TemplateNames( + IO.Video.Input("reference_video"), + names=["video_1", "video_2", "video_3"], + min=0, + ), + tooltip="Reference videos from the original generation, in the same " + "order. Up to 3 videos, 2-15 seconds each, 15 seconds in total.", + ), + IO.Autogrow.Input( + "reference_audios", + template=IO.Autogrow.TemplateNames( + IO.Audio.Input("reference_audio"), + names=["audio_1", "audio_2", "audio_3"], + min=0, + ), + tooltip="Audio references from the original generation, in the same " + "order. Up to 3 clips, 2-15 seconds each, 15 seconds in total. " + "Cannot be used without a reference image or video.", + ), + ], + ) + ], + tooltip="Model to use for video regeneration.", + ), + IO.Video.Input( + "video", + tooltip="The MiniMax H3 768P output video to re-render. Connect the unmodified output " + "of a MiniMax H3 video node (24 FPS, 4-15 seconds). 2K outputs cannot be used.", + ), + IO.Image.Input( + "first_frame", + tooltip="First frame image from the original generation, if one was used.", + optional=True, + ), + IO.Image.Input( + "last_frame", + tooltip="Last frame image from the original generation, if one was used.", + optional=True, + ), + IO.Boolean.Input( + "watermark", + default=False, + tooltip="Whether to add an AIGC watermark to the video.", + advanced=True, + ), + ], + outputs=[ + IO.Video.Output(), + ], + hidden=[ + IO.Hidden.auth_token_comfy_org, + IO.Hidden.api_key_comfy_org, + IO.Hidden.unique_id, + ], + is_api_node=True, + price_badge=IO.PriceBadge( + expr="""{"type": "usd", "usd": 0.0715, "format": {"suffix": "/second"}}""", + ), + ) + + @classmethod + async def execute( + cls, + model: dict, + video: Input.Video, + watermark: bool, + first_frame: torch.Tensor | None = None, + last_frame: torch.Tensor | None = None, + ) -> IO.NodeOutput: + validate_string(model["prompt"], strip_whitespace=True, min_length=1) + + try: + fps = float(video.get_frame_rate()) + except Exception: + fps = 0.0 + if fps and not (23.9 <= fps <= 24.1): + raise ValueError( + f"The source video is {fps:.2f} FPS. Regeneration accepts unmodified MiniMax H3 768P " + "outputs, which are 24 FPS." + ) + try: + width, height = video.get_dimensions() + except Exception: + width = height = 0 + if width and height and (width % 32 or height % 32 or width * height > 1_032_192): + raise ValueError( + f"The source video is {width}x{height}. Regeneration accepts MiniMax H3 768P outputs " + "(width and height divisible by 32, at most 1,032,192 total pixels); 2K outputs cannot " + "be used as a source." + ) + try: + frame_count = video.get_frame_count() + except Exception: + frame_count = 0 + if frame_count and (frame_count < 107 or frame_count > 362 or (frame_count - 107) % 17): + raise ValueError( + f"The source video has {frame_count} frames. Regeneration accepts unmodified " + "MiniMax H3 outputs, whose length is 107 to 362 frames in steps of 17 " + "(4 to 15 seconds at 24 FPS)." + ) + + reference_images = {k: v for k, v in (model.get("reference_images") or {}).items() if v is not None} + reference_videos = {k: v for k, v in (model.get("reference_videos") or {}).items() if v is not None} + reference_audios = {k: v for k, v in (model.get("reference_audios") or {}).items() if v is not None} + if (first_frame is not None or last_frame is not None) and ( + reference_images or reference_videos or reference_audios + ): + raise ValueError( + "First/last frame and reference media are mutually exclusive. Use frames for an " + "image-to-video prompt, or reference media for a reference-to-video prompt." + ) + if reference_audios and not reference_images and not reference_videos: + raise ValueError("Reference audio cannot be used without a reference image or video.") + + for frame in (first_frame, last_frame): + if frame is not None: + validate_image_aspect_ratio(frame, (2, 5), (5, 2), strict=False) # 0.4 to 2.5 + validate_image_dimensions(frame, min_width=256, min_height=256) + for image in reference_images.values(): + validate_image_aspect_ratio(image, (2, 5), (5, 2), strict=False) # 0.4 to 2.5 + validate_image_dimensions(image, min_width=256, min_height=256) + + total_video_duration = 0.0 + for i, ref_video in enumerate(reference_videos.values(), 1): + try: + ref_fps = float(ref_video.get_frame_rate()) + except Exception: + ref_fps = 0.0 + if ref_fps and not (23.9 <= ref_fps <= 60.5): + raise ValueError(f"Reference video {i} is {ref_fps:.2f} FPS. Supported range is 23.976-60 FPS.") + try: + dur = ref_video.get_duration() + except Exception: + continue + if dur < 1.8: + raise ValueError(f"Reference video {i} is too short: {dur:.1f}s. Minimum duration is 2 seconds.") + total_video_duration += dur + if total_video_duration > 15.1: + raise ValueError( + f"Total reference video duration is {total_video_duration:.1f}s. Maximum is 15 seconds." + ) + + total_audio_duration = 0.0 + for i, audio in enumerate(reference_audios.values(), 1): + dur = int(audio["waveform"].shape[-1]) / int(audio["sample_rate"]) + if dur < 1.8: + raise ValueError(f"Reference audio {i} is too short: {dur:.1f}s. Minimum duration is 2 seconds.") + total_audio_duration += dur + if total_audio_duration > 15.1: + raise ValueError( + f"Total reference audio duration is {total_audio_duration:.1f}s. Maximum is 15 seconds." + ) + + content: list = [ + Hailuo03VideoContent( + video_url=Hailuo03VideoContentUrl( + url=await upload_video_to_comfyapi(cls, video, wait_label="Uploading source video"), + ), + role="base_video", + ), + Hailuo03TextContent(text=model["prompt"]), + ] + if first_frame is not None: + content.append( + Hailuo03ImageContent( + image_url=Hailuo03ImageContentUrl( + url=( + await upload_images_to_comfyapi( + cls, first_frame, max_images=1, wait_label="Uploading first frame" + ) + )[0], + ), + role="first_frame", + ) + ) + if last_frame is not None: + content.append( + Hailuo03ImageContent( + image_url=Hailuo03ImageContentUrl( + url=( + await upload_images_to_comfyapi( + cls, last_frame, max_images=1, wait_label="Uploading last frame" + ) + )[0], + ), + role="last_frame", + ) + ) + for i, image in enumerate(reference_images.values(), 1): + content.append( + Hailuo03ImageContent( + image_url=Hailuo03ImageContentUrl( + url=( + await upload_images_to_comfyapi( + cls, image, max_images=1, wait_label=f"Uploading image {i}" + ) + )[0], + ), + role="reference_image", + ) + ) + for i, ref_video in enumerate(reference_videos.values(), 1): + content.append( + Hailuo03VideoContent( + video_url=Hailuo03VideoContentUrl( + url=await upload_video_to_comfyapi(cls, ref_video, wait_label=f"Uploading video {i}"), + ), + ) + ) + for audio in reference_audios.values(): + content.append( + Hailuo03AudioContent( + audio_url=Hailuo03AudioContentUrl( + url=await upload_audio_to_comfyapi( + cls, + audio, + container_format="mp3", + codec_name="libmp3lame", + mime_type="audio/mpeg", + ), + ), + ) + ) + + response = await sync_op( + cls, + ApiEndpoint(path=HAILUO_03_REGENERATION_ENDPOINT, method="POST"), + response_model=Hailuo03TaskCreationResponse, + data=Hailuo03RegenerationRequest( + model=HAILUO_03_MODELS[model["model"]], + content=content, + resolution=model["resolution"], + aigc_watermark=watermark, + ), + ) + task_result = await poll_op( + cls, + ApiEndpoint(path=f"{HAILUO_03_QUERY_ENDPOINT}/{response.task_id}"), + response_model=Hailuo03TaskQueryResponse, + status_extractor=lambda r: r.task.status, + failed_statuses=HAILUO_03_FAILED_STATUSES, + poll_interval=10, + ) + video_url = task_result.task.content.url if task_result.task.content else None + if not video_url: + raise Exception(f"No video URL in the response: {task_result.model_dump()}") + return IO.NodeOutput(await download_url_to_video_output(video_url)) + + class MinimaxExtension(ComfyExtension): @override async def get_node_list(self) -> list[type[IO.ComfyNode]]: @@ -950,6 +1541,8 @@ class MinimaxExtension(ComfyExtension): MinimaxHailuo03TextToVideoNode, MinimaxHailuo03FirstLastFrameNode, MinimaxHailuo03ReferenceNode, + MinimaxHailuo03ContextIRNode, + MinimaxHailuo03RegenerateNode, ] diff --git a/comfy_api_nodes/nodes_qwen.py b/comfy_api_nodes/nodes_qwen.py new file mode 100644 index 000000000..3b6c5023c --- /dev/null +++ b/comfy_api_nodes/nodes_qwen.py @@ -0,0 +1,442 @@ +import math +import re + +import torch +from typing_extensions import override + +from comfy_api.latest import IO, ComfyExtension +from comfy_api_nodes.apis.qwen import ( + QwenImageContentItem, + QwenImageGenerationRequest, + QwenImageGenerationResponse, + QwenImageInputField, + QwenImageMessage, + QwenImageParametersField, +) +from comfy_api_nodes.util import ( + ApiEndpoint, + download_url_to_image_tensor, + sync_op, + tensor_to_base64_string, + validate_string, +) + +GENERATION_PATH = "/proxy/qwen/api/v1/services/aigc/multimodal-generation/generation" +QWEN_IMAGE_MODELS = ["qwen-image-3.0-pro", "qwen-image-3.0"] +MIN_AREA = 262144 # 512*512 +MAX_AREA = 6553600 # 2560*2560 +MAX_ASPECT = 8 # the API allows aspect ratios from 1:8 to 8:1 +MAX_INPUT_BYTES = 10 * 1024 * 1024 # the API rejects decoded input images over 10MB + +_IMAGE_REF_RE = re.compile(r"@image(?P\d*)(?!\w)", re.IGNORECASE | re.ASCII) + + +def _resolve_image_refs(prompt: str, total_images: int) -> str: + """Rewrite @Image1-style references (shared partner-node syntax, 1-based; an unnumbered + @image means the first image) into the plain 'Image N' wording the model resolves + natively. A tag counts only at a word boundary or right after a previous tag, so + adjacent tags like '@Image1@Image2' all resolve while addresses like user@image1.com + pass through untouched.""" + parts = [] + pos = 0 + prev_end = -1 + for match in _IMAGE_REF_RE.finditer(prompt): + start = match.start() + if start > 0 and start != prev_end and (prompt[start - 1].isalnum() or prompt[start - 1] == "_"): + continue + idx = int(match.group("idx") or 1) + if not 1 <= idx <= total_images: + raise ValueError( + f"The prompt references @Image{idx}, but only {total_images} reference images " + f"are connected (a batched input counts once per image)." + ) + parts.append(prompt[pos:start]) + parts.append(f"Image {idx}") + pos = match.end() + prev_end = match.end() + parts.append(prompt[pos:]) + return "".join(parts) + + +def _validate_size(width: int, height: int) -> None: + if not MIN_AREA <= width * height <= MAX_AREA: + raise ValueError( + f"Image area must be between {MIN_AREA} (512x512) and {MAX_AREA} (2560x2560) pixels; " + f"got {width}x{height} = {width * height}." + ) + if width > MAX_ASPECT * height or height > MAX_ASPECT * width: + raise ValueError(f"Aspect ratio must be between 1:8 and 8:1; got {width}x{height}.") + + +def _fit_to_size(width: int, height: int) -> tuple[int, int]: + """Scale dimensions into the supported pixel area and 1:8..8:1 aspect range, preserving + the aspect ratio where possible.""" + if width > MAX_ASPECT * height: + height = math.ceil(width / MAX_ASPECT) + elif height > MAX_ASPECT * width: + width = math.ceil(height / MAX_ASPECT) + area = width * height + if area < MIN_AREA: + scale = math.sqrt(MIN_AREA / area) + width, height = math.ceil(width * scale), math.ceil(height * scale) + elif area > MAX_AREA: + scale = math.sqrt(MAX_AREA / area) + width, height = math.floor(width * scale), math.floor(height * scale) + # rounding can push the ratio a hair past the limit; trimming only ever shrinks the area + return min(width, MAX_ASPECT * height), min(height, MAX_ASPECT * width) + + +def _image_data_uri(image: torch.Tensor) -> str: + """PNG data URI of an RGB view of the image, downscaled to <=2048x2048; falls back to + JPEG when the PNG exceeds the API's decoded-size cap (e.g. noisy, incompressible images).""" + image = image[..., :3] + b64 = tensor_to_base64_string(image, total_pixels=2048 * 2048) + if len(b64) * 3 > MAX_INPUT_BYTES * 4: + return "data:image/jpeg;base64," + tensor_to_base64_string( + image, total_pixels=2048 * 2048, mime_type="image/jpeg" + ) + return "data:image/png;base64," + b64 + + +async def _download_result_images(response: QwenImageGenerationResponse) -> torch.Tensor: + if not response.output: + raise Exception(f"An unknown error occurred: {response.code} - {response.message}") + urls = [ + item.image + for choice in response.output.choices + if choice.message + for item in choice.message.content + if item.image + ] + if not urls: + raise Exception(f"The response contains no images: {response.code} - {response.message}") + return torch.cat([await download_url_to_image_tensor(url) for url in urls]) + + +def _size_inputs() -> list[IO.Int.Input]: + return [ + IO.Int.Input( + "width", + default=1024, + min=256, + max=2560, + step=16, + tooltip="The total pixel area must be between 512x512 and 2560x2560; " + "any aspect ratio within that area works.", + ), + IO.Int.Input( + "height", + default=1024, + min=256, + max=2560, + step=16, + tooltip="The total pixel area must be between 512x512 and 2560x2560; " + "any aspect ratio within that area works.", + ), + ] + + +def _t2i_model_option(model_id: str) -> IO.DynamicCombo.Option: + return IO.DynamicCombo.Option( + model_id, + [ + IO.String.Input( + "prompt", + multiline=True, + default="", + tooltip="Prompt describing the image. Supports English and Chinese.", + ), + IO.String.Input( + "negative_prompt", + multiline=True, + default="", + tooltip="Negative prompt describing what to avoid.", + ), + *_size_inputs(), + ], + ) + + +def _edit_model_option(model_id: str) -> IO.DynamicCombo.Option: + return IO.DynamicCombo.Option( + model_id, + [ + IO.Autogrow.Input( + "images", + template=IO.Autogrow.TemplateNames( + IO.Image.Input("image"), + names=["image_1", "image_2", "image_3"], + min=1, + ), + tooltip="1-3 reference images. Refer to them in the prompt as @Image1, @Image2, " + "@Image3, numbered in input order; a batched input counts once per image.", + ), + IO.String.Input( + "prompt", + multiline=True, + default="", + tooltip="Editing instructions. Supports English and Chinese, " + "and @Image1-style references to the input images.", + ), + IO.String.Input( + "negative_prompt", + multiline=True, + default="", + tooltip="Negative prompt describing what to avoid.", + ), + ], + ) + + +class QwenImageTextToImageApi(IO.ComfyNode): + @classmethod + def define_schema(cls): + return IO.Schema( + node_id="QwenImageTextToImageApi", + display_name="Qwen Image 3 Text to Image", + category="partner/image/Qwen", + description="Generates images from a text prompt using the Qwen-Image 3.0 models.", + inputs=[ + IO.DynamicCombo.Input( + "model", + options=[_t2i_model_option(model_id) for model_id in QWEN_IMAGE_MODELS], + tooltip="Model to use.", + ), + IO.Int.Input( + "n", + default=1, + min=1, + max=6, + display_mode=IO.NumberDisplay.number, + tooltip="Number of images to generate, returned as a batch.", + ), + IO.Int.Input( + "seed", + default=42, + min=0, + max=2147483647, + step=1, + display_mode=IO.NumberDisplay.number, + control_after_generate=True, + tooltip="Seed to use for generation.", + ), + IO.Boolean.Input( + "prompt_extend", + default=True, + tooltip="Whether to enhance the prompt with AI assistance.", + advanced=True, + ), + IO.Boolean.Input( + "watermark", + default=False, + tooltip="Whether to add an AI-generated watermark to the result.", + advanced=True, + ), + ], + outputs=[ + IO.Image.Output(), + ], + hidden=[ + IO.Hidden.auth_token_comfy_org, + IO.Hidden.api_key_comfy_org, + IO.Hidden.unique_id, + ], + is_api_node=True, + price_badge=IO.PriceBadge( + depends_on=IO.PriceBadgeDepends(widgets=["model", "model.width", "model.height", "n"]), + expr=""" + ( + $isPro := widgets.model = "qwen-image-3.0-pro"; + $area := $lookup(widgets, "model.width") * $lookup(widgets, "model.height"); + $rate := $isPro ? ($area > 2250000 ? 0.10725 : 0.0572) : 0.0429; + {"type":"usd","usd": $rate * widgets.n} + ) + """, + ), + ) + + @classmethod + async def execute( + cls, + model: dict, + n: int = 1, + seed: int = 42, + prompt_extend: bool = True, + watermark: bool = False, + ): + validate_string(model["prompt"], strip_whitespace=False, min_length=1) + width, height = model["width"], model["height"] + _validate_size(width, height) + response = await sync_op( + cls, + ApiEndpoint(path=GENERATION_PATH, method="POST"), + response_model=QwenImageGenerationResponse, + data=QwenImageGenerationRequest( + model=model["model"], + input=QwenImageInputField( + messages=[QwenImageMessage(content=[QwenImageContentItem(text=model["prompt"])])], + ), + parameters=QwenImageParametersField( + size=f"{width}*{height}", + n=n, + seed=seed, + prompt_extend=prompt_extend, + watermark=watermark, + negative_prompt=model["negative_prompt"] or None, + ), + ), + ) + return IO.NodeOutput(await _download_result_images(response)) + + +class QwenImageEditApi(IO.ComfyNode): + @classmethod + def define_schema(cls): + return IO.Schema( + node_id="QwenImageEditApi", + display_name="Qwen Image 3 Edit", + category="partner/image/Qwen", + description="Edits or combines up to 3 reference images guided by a text prompt " + "using the Qwen-Image 3.0 models.", + inputs=[ + IO.DynamicCombo.Input( + "model", + options=[_edit_model_option(model_id) for model_id in QWEN_IMAGE_MODELS], + tooltip="Model to use.", + ), + IO.DynamicCombo.Input( + "size", + options=[ + IO.DynamicCombo.Option("match input", []), + IO.DynamicCombo.Option("auto", []), + IO.DynamicCombo.Option("custom", _size_inputs()), + ], + tooltip="Output resolution. 'match input' reuses the first reference image's size, " + "'auto' lets the model pick a size with the same aspect ratio, " + "'custom' sets an explicit width and height.", + ), + IO.Int.Input( + "n", + default=1, + min=1, + max=6, + display_mode=IO.NumberDisplay.number, + tooltip="Number of images to generate, returned as a batch.", + ), + IO.Int.Input( + "seed", + default=42, + min=0, + max=2147483647, + step=1, + display_mode=IO.NumberDisplay.number, + control_after_generate=True, + tooltip="Seed to use for generation.", + ), + IO.Boolean.Input( + "prompt_extend", + default=True, + tooltip="Whether to enhance the prompt with AI assistance.", + advanced=True, + ), + IO.Boolean.Input( + "watermark", + default=False, + tooltip="Whether to add an AI-generated watermark to the result.", + advanced=True, + ), + ], + outputs=[ + IO.Image.Output(), + ], + hidden=[ + IO.Hidden.auth_token_comfy_org, + IO.Hidden.api_key_comfy_org, + IO.Hidden.unique_id, + ], + is_api_node=True, + price_badge=IO.PriceBadge( + depends_on=IO.PriceBadgeDepends( + widgets=["model", "size", "size.width", "size.height", "n"], + input_groups=["model.images"], + ), + expr=""" + ( + $isPro := widgets.model = "qwen-image-3.0-pro"; + $mode := widgets.size; + $count := $max([$lookup(inputGroups, "model.images"), 1]); + $inputCost := 0.00429 * $count; + $area := $mode = "custom" + ? $lookup(widgets, "size.width") * $lookup(widgets, "size.height") : 0; + $customRate := $area > 2250000 ? 0.10725 : 0.0572; + $isPro and $mode != "custom" + ? {"type":"range_usd", + "min_usd": 0.0572 * widgets.n + $inputCost, + "max_usd": 0.10725 * widgets.n + $inputCost} + : {"type":"usd", + "usd": ($isPro ? $customRate : 0.0429) * widgets.n + $inputCost} + ) + """, + ), + ) + + @classmethod + async def execute( + cls, + model: dict, + size: dict, + n: int = 1, + seed: int = 42, + prompt_extend: bool = True, + watermark: bool = False, + ): + validate_string(model["prompt"], strip_whitespace=False, min_length=1) + reference_images = [image for key in model["images"] for image in model["images"][key]] + if len(reference_images) > 3: + raise ValueError( + f"A maximum of 3 reference images is supported; got {len(reference_images)} " + f"(a batched input counts once per image)." + ) + prompt = _resolve_image_refs(model["prompt"], len(reference_images)) + if size["size"] == "custom": + _validate_size(size["width"], size["height"]) + size_str = f"{size['width']}*{size['height']}" + elif size["size"] == "match input": + height, width = reference_images[0].shape[0], reference_images[0].shape[1] + width, height = _fit_to_size(width, height) + size_str = f"{width}*{height}" + else: # auto: the API picks a size preserving the input aspect ratio (1.9-4.2 MP) + size_str = None + content = [QwenImageContentItem(image=_image_data_uri(image)) for image in reference_images] + content.append(QwenImageContentItem(text=prompt)) + response = await sync_op( + cls, + ApiEndpoint(path=GENERATION_PATH, method="POST"), + response_model=QwenImageGenerationResponse, + data=QwenImageGenerationRequest( + model=model["model"], + input=QwenImageInputField(messages=[QwenImageMessage(content=content)]), + parameters=QwenImageParametersField( + size=size_str, + n=n, + seed=seed, + prompt_extend=prompt_extend, + watermark=watermark, + negative_prompt=model["negative_prompt"] or None, + ), + ), + ) + return IO.NodeOutput(await _download_result_images(response)) + + +class QwenApiExtension(ComfyExtension): + @override + async def get_node_list(self) -> list[type[IO.ComfyNode]]: + return [ + QwenImageTextToImageApi, + QwenImageEditApi, + ] + + +async def comfy_entrypoint() -> QwenApiExtension: + return QwenApiExtension() diff --git a/comfy_extras/nodes_custom_sampler.py b/comfy_extras/nodes_custom_sampler.py index d5aa730d2..c73a8f6dc 100644 --- a/comfy_extras/nodes_custom_sampler.py +++ b/comfy_extras/nodes_custom_sampler.py @@ -718,15 +718,7 @@ class Noise_EmptyNoise: self.seed = 0 def generate_noise(self, input_latent): - latent_image = input_latent["samples"] - if latent_image.is_nested: - tensors = latent_image.unbind() - zeros = [] - for t in tensors: - zeros.append(torch.zeros(t.shape, dtype=t.dtype, layout=t.layout, device="cpu")) - return comfy.nested_tensor.NestedTensor(zeros) - else: - return torch.zeros(latent_image.shape, dtype=latent_image.dtype, layout=latent_image.layout, device="cpu") + return comfy.sample.prepare_empty_noise(input_latent["samples"]) class Noise_RandomNoise: diff --git a/comfy_extras/nodes_lt.py b/comfy_extras/nodes_lt.py index 8c85c92b1..a6e5c5d27 100644 --- a/comfy_extras/nodes_lt.py +++ b/comfy_extras/nodes_lt.py @@ -2,11 +2,14 @@ import nodes import node_helpers import torch import torchaudio +import comfy.ldm.lightricks.duration_head import comfy.model_management import comfy.model_sampling import comfy.samplers import comfy.utils +import logging import math +import re import numpy as np import av from io import BytesIO @@ -934,6 +937,243 @@ class LTXVReferenceAudio(io.ComfyNode): return io.NodeOutput(m, positive, negative) +class LTXVSpatioTemporalGuidance(io.ComfyNode): + @classmethod + def define_schema(cls): + return io.Schema( + node_id="LTXVSpatioTemporalGuidance", + display_name="LTXV Spatio-Temporal Guidance (STG)", + category="advanced/guidance", + description="Runs one extra pass per step with the self-attention of the selected blocks degraded to a value-passthrough, " + "then guides away from it - improving spatial detail and motion coherence.", + inputs=[ + io.Model.Input("model"), + io.Float.Input("scale", default=1.0, min=0.0, max=100.0, step=0.01, round=0.01), + io.String.Input("blocks", default="29", tooltip="Comma-separated transformer block indices to perturb."), + io.Float.Input("start_percent", default=0.0, min=0.0, max=1.0, step=0.001, advanced=True), + io.Float.Input("end_percent", default=1.0, min=0.0, max=1.0, step=0.001, advanced=True), + ], + outputs=[io.Model.Output()], + ) + + @classmethod + def execute(cls, model, scale, blocks, start_percent, end_percent) -> io.NodeOutput: + block_set = frozenset(int(b) for b in re.findall(r"\d+", blocks)) + + m = model.clone() + model_sampling = m.get_model_object("model_sampling") + sigma_start = model_sampling.percent_to_sigma(start_percent) + sigma_end = model_sampling.percent_to_sigma(end_percent) + + def post_cfg_function(args): + if scale == 0 or not block_set: + return args["denoised"] + + sigma_ = args["sigma"][0].item() + if sigma_ > sigma_start or sigma_ < sigma_end: + return args["denoised"] + + cond_pred = args["cond_denoised"] + cond = args["cond"] + cfg_result = args["denoised"] + x = args["input"] + + model_options = args["model_options"].copy() + transformer_options = model_options.get("transformer_options", {}).copy() + transformer_options["stg_self_attn_blocks"] = block_set + model_options["transformer_options"] = transformer_options + + (perturbed,) = comfy.samplers.calc_cond_batch(args["model"], [cond], x, args["sigma"], model_options) + + return cfg_result + (cond_pred - perturbed) * scale + + m.set_model_sampler_post_cfg_function(post_cfg_function) + return io.NodeOutput(m) + + +class LTXVModalityGuidance(io.ComfyNode): + @classmethod + def define_schema(cls): + return io.Schema( + node_id="LTXVModalityGuidance", + display_name="LTXV Modality Guidance (A/V coupling)", + category="advanced/guidance", + description="Cross-modal (audio-video) guidance for LTXV-AV. Runs one extra forward " + "pass per step with the a2v/v2a cross-attention severed, then pushes the " + "result toward the coupled prediction - strengthening audio-visual sync " + "(e.g. lip-sync). Reference default modality_scale is 3.0. Stacks with the " + "dual-CFG guider and STG. Set to 1.0 to disable (no extra pass).", + inputs=[ + io.Model.Input("model"), + io.Float.Input("modality_scale", default=3.0, min=1.0, max=100.0, step=0.1, round=0.01), + io.Float.Input("start_percent", default=0.0, min=0.0, max=1.0, step=0.001, advanced=True), + io.Float.Input("end_percent", default=1.0, min=0.0, max=1.0, step=0.001, advanced=True), + ], + outputs=[io.Model.Output()], + ) + + @classmethod + def execute(cls, model, modality_scale, start_percent, end_percent) -> io.NodeOutput: + m = model.clone() + model_sampling = m.get_model_object("model_sampling") + sigma_start = model_sampling.percent_to_sigma(start_percent) + sigma_end = model_sampling.percent_to_sigma(end_percent) + + def post_cfg_function(args): + if math.isclose(modality_scale, 1.0): + return args["denoised"] + + sigma_ = args["sigma"][0].item() + if sigma_ > sigma_start or sigma_ < sigma_end: + return args["denoised"] + + cond_pred = args["cond_denoised"] + cond = args["cond"] + cfg_result = args["denoised"] + x = args["input"] + + # Extra pass with audio-video cross-attention severed (both directions) + model_options = args["model_options"].copy() + transformer_options = model_options.get("transformer_options", {}).copy() + transformer_options["a2v_cross_attn"] = False + transformer_options["v2a_cross_attn"] = False + model_options["transformer_options"] = transformer_options + + (mod_pred,) = comfy.samplers.calc_cond_batch( + args["model"], [cond], x, args["sigma"], model_options + ) + + # (modality_scale - 1) * (cond - uncond_modality), per the reference guider. + return cfg_result + (cond_pred - mod_pred) * (modality_scale - 1.0) + + m.set_model_sampler_post_cfg_function(post_cfg_function) + return io.NodeOutput(m) + + +class Guider_LTXAVDualCFG(comfy.samplers.CFGGuider): + """CFG guider that applies separate guidance scales to the video and audio + modalities of a packed LTXV-AV latent. + """ + + def set_conds(self, positive, negative): + self.inner_set_conds({"positive": positive, "negative": negative}) + + def set_cfg(self, video_cfg, audio_cfg): + self.video_cfg = video_cfg + self.audio_cfg = audio_cfg + self.cfg = max(video_cfg, audio_cfg) + + def sample(self, noise, latent_image, *args, **kwargs): + # Capture the video/audio split from the nested latent before it is packed. + self._v_numel = None + if getattr(latent_image, "is_nested", False): + parts = latent_image.unbind() + if len(parts) >= 2: + self._v_numel = math.prod(parts[0].shape[1:]) + return super().sample(noise, latent_image, *args, **kwargs) + + def predict_noise(self, x, timestep, model_options={}, seed=None): + v = getattr(self, "_v_numel", None) + if v is None or math.isclose(self.video_cfg, self.audio_cfg): + # Not an AV latent, or equal scales: fall back to standard single-CFG. + self.cfg = self.video_cfg + return super().predict_noise(x, timestep, model_options, seed) + + video_cfg, audio_cfg = self.video_cfg, self.audio_cfg + + def dual_cfg(args): + # Noise-space: cond = x - cond_pred, uncond = x - uncond_pred; the + # returned tensor is subtracted from x by cfg_function. + cond, uncond = args["cond"], args["uncond"] + out = uncond + (cond - uncond) * video_cfg + out[..., v:] = uncond[..., v:] + (cond[..., v:] - uncond[..., v:]) * audio_cfg + return out + + # disable_cfg1_optimization so the uncond pass always runs even if one of the two scales is 1.0. + model_options = {**model_options, "sampler_cfg_function": dual_cfg, "disable_cfg1_optimization": True} + return super().predict_noise(x, timestep, model_options, seed) + + +class LTXVDualCFGGuider(io.ComfyNode): + @classmethod + def define_schema(cls): + return io.Schema( + node_id="LTXVDualCFGGuider", + display_name="LTXV Dual CFG Guider", + category="model/sampling/guiders", + description="Separate CFG scales for the video and audio modalities of a packed LTXV-AV latent.", + inputs=[ + io.Model.Input("model"), + io.Conditioning.Input("positive"), + io.Conditioning.Input("negative"), + io.Float.Input("video_cfg", default=3.0, min=0.0, max=100.0, step=0.1, round=0.01), + io.Float.Input("audio_cfg", default=7.0, min=0.0, max=100.0, step=0.1, round=0.01), + ], + outputs=[io.Guider.Output()], + ) + + @classmethod + def execute(cls, model, positive, negative, video_cfg, audio_cfg) -> io.NodeOutput: + guider = Guider_LTXAVDualCFG(model) + guider.set_conds(positive, negative) + guider.set_cfg(video_cfg, audio_cfg) + return io.NodeOutput(guider) + + +class LTXVDurationPredictor(io.ComfyNode): + @classmethod + def define_schema(cls): + return io.Schema( + node_id="LTXVDurationPredictor", + display_name="LTXV Duration Predictor", + category="conditioning/video_models", + description="Predicts the natural shot duration for a prompt using the LTX 2.4 duration " + "head (loaded with ModelPatchLoader), and snaps it to the VAE's 8k+1 frame grid.", + search_aliases=["auto duration", "duration head", "num_frames"], + inputs=[ + io.Model.Input("model"), + io.Conditioning.Input("positive"), + io.Custom("MODEL_PATCH").Input("duration_head", + tooltip="LTX 2.4 duration head loaded with ModelPatchLoader."), + io.Float.Input("frame_rate", default=24.0, min=1.0, max=120.0, step=0.01), + io.Float.Input("min_seconds", default=1.0, min=0.5, max=120.0, step=0.1), + io.Float.Input("max_seconds", default=20.0, min=0.5, max=120.0, step=0.1), + ], + outputs=[ + io.Int.Output(display_name="num_frames"), + io.Float.Output(display_name="seconds", tooltip="Raw (unclamped) predicted duration."), + ], + ) + + @classmethod + def execute(cls, model, positive, duration_head, frame_rate, min_seconds, max_seconds) -> io.NodeOutput: + dm = model.model.diffusion_model + head = duration_head.model + if not isinstance(head, comfy.ldm.lightricks.duration_head.DurationHead): + raise ValueError("The connected model_patch is not an LTX duration head.") + + context = positive[0][0] + meta = positive[0][1] + if context.shape[0] != 1: + context = context[:1] + + # Run the caption connectors exactly the way sampling does. + comfy.model_management.load_models_gpu([model, duration_head]) + device = model.load_device + head = head.to(device) + with torch.no_grad(): + context = context.to(device=device, dtype=model.model.get_dtype_inference()) + processed = dm.preprocess_text_embeds(context, unprocessed=meta.get("unprocessed_ltxav_embeds", False)) + video_tokens = processed[..., :dm.cross_attention_dim].float() + audio_tokens = processed[..., dm.cross_attention_dim:].float() + seconds = float(head(video_tokens, audio_tokens)[0]) + + num_frames = comfy.ldm.lightricks.duration_head.seconds_to_num_frames( + seconds, frame_rate, min_seconds, max_seconds) + logging.info("LTXV duration head predicted %.2fs -> %d frames @ %.2f fps", seconds, num_frames, frame_rate) + return io.NodeOutput(num_frames, seconds) + + class LtxvExtension(ComfyExtension): @override async def get_node_list(self) -> list[type[io.ComfyNode]]: @@ -951,6 +1191,10 @@ class LtxvExtension(ComfyExtension): LTXVConcatAVLatent, LTXVSeparateAVLatent, LTXVReferenceAudio, + LTXVDualCFGGuider, + LTXVModalityGuidance, + LTXVSpatioTemporalGuidance, + LTXVDurationPredictor, ] diff --git a/comfy_extras/nodes_lt_audio.py b/comfy_extras/nodes_lt_audio.py index 3ff18d8d4..0924f3e9e 100644 --- a/comfy_extras/nodes_lt_audio.py +++ b/comfy_extras/nodes_lt_audio.py @@ -173,7 +173,7 @@ class LTXAVTextEncoderLoader(io.ComfyNode): node_id="LTXAVTextEncoderLoader", display_name="Load LTXV Audio Text Encoder", category="model/loaders", - description="Recipes:\nltxav: gemma 3 12B", + description="Recipes:\nltxav: gemma 3 12B or matching gemma 4 model", inputs=[ io.Combo.Input( "text_encoder", diff --git a/comfy_extras/nodes_minimax_h3.py b/comfy_extras/nodes_minimax_h3.py index 0b1840e85..0a08f185f 100644 --- a/comfy_extras/nodes_minimax_h3.py +++ b/comfy_extras/nodes_minimax_h3.py @@ -20,6 +20,7 @@ import comfy.model_sampling import comfy.nested_tensor import comfy.utils import node_helpers +from comfy.ldm.minimax.model import FRAME_PER_TOKEN, FRAME_RESCALE from comfy_api.latest import ComfyExtension, io CANVAS_MULTIPLE = 32 @@ -67,6 +68,16 @@ def _resize(image, width, height, crop): return samples.movedim(1, -1) +def _encode_ref_audio(audio_vae, audio): + waveform = audio["waveform"] # [B, C, L] + sr = audio["sample_rate"] + vae_sr = getattr(audio_vae, "audio_sample_rate", 32000) + if sr != vae_sr: + waveform = torchaudio.functional.resample(waveform, sr, vae_sr) + z = audio_vae.encode(waveform[:1].movedim(1, -1)) # [1, 32, 2, T] + return z, z.shape[-1] + + def _empty_av_latent(width, height, length, batch_size=1): frame_count, latent_t, audio_t = temporal_shape(length) video = torch.zeros([batch_size, 24, latent_t, height // 16, width // 16], @@ -144,13 +155,87 @@ class MiniMaxH3ImageToVideo(io.ComfyNode): if keyframes: for kf in keyframes: kf["latent"] = vae.encode(kf.pop("image")) - cond = node_helpers.conditioning_set_values(cond, { - "minimax_keyframes": keyframes, - "minimax_frame_count": frame_count, - }) + cond = node_helpers.conditioning_set_values(cond, {"minimax_keyframes": keyframes}) return io.NodeOutput(cond, latent) +class MiniMaxH3AddGuide(io.ComfyNode): + """Anchor image and/or audio guides at an arbitrary pixel frame of the target video.""" + + @classmethod + def define_schema(cls): + return io.Schema( + node_id="MiniMaxH3AddGuide", + display_name="Add Guide for MiniMax H3", + category="model/conditioning/minimax", + description="Anchor an image, a short clip, audio, or a clip with its soundtrack at any frame of a MiniMax H3 video. Chain several nodes to anchor several frames.", + inputs=[ + io.Conditioning.Input("positive"), + io.Vae.Input("vae", optional=True, tooltip="Video VAE, needed when an image is connected."), + io.Vae.Input("audio_vae", optional=True, tooltip="Audio VAE, needed when an audio is connected."), + io.Latent.Input("latent"), + io.Image.Input("image", optional=True, tooltip="Image or video frames to anchor. Multi-frame batches are anchored as a clip and cropped down to the model's valid clip lengths: 5, 22, 39... (17k + 5) frames. Batches shorter than 5 frames use only the first image."), + io.Audio.Input("audio", optional=True, + tooltip="Soundtrack to anchor starting at the same frame index, cropped to the video's remaining duration."), + io.Int.Input("frame_idx", default=0, min=-9999, max=9999, + tooltip="Frame index to anchor the image or the clip's first frame at. Negative values are counted from the end of the video."), + ], + outputs=[io.Conditioning.Output(display_name="positive")], + ) + + @classmethod + def execute(cls, positive, latent, frame_idx, vae=None, audio_vae=None, image=None, audio=None) -> io.NodeOutput: + samples = latent["samples"] + if not samples.is_nested or len(samples.tensors) != 2 or samples.tensors[0].ndim != 5 or samples.tensors[0].shape[1] != 24: + raise ValueError("MiniMaxH3AddGuide expects a MiniMax H3 AV latent") + if image is None and audio is None: + raise ValueError("MiniMaxH3AddGuide needs an image or an audio to anchor") + video = samples.tensors[0] + height = video.shape[3] * 16 + width = video.shape[4] * 16 + frame_count = sum(FRAME_PER_TOKEN[k % 5] for k in range(video.shape[2])) + + guide_frames = 1 + if image is not None: + if vae is None: + raise ValueError("anchoring guide frames needs the vae input") + guide_frames = image.shape[0] + if guide_frames < 5: + guide_frames = 1 + else: + while guide_frames % 17 != 5: + guide_frames -= 1 + + resolved_frame_index = frame_idx if frame_idx >= 0 else frame_count + frame_idx + if resolved_frame_index < 0 or resolved_frame_index + guide_frames > frame_count: + if guide_frames == 1: + raise ValueError("frame_idx {} is outside the video's {} frames".format(frame_idx, frame_count)) + raise ValueError("a {} frame guide clip at frame_idx {} does not fit in the video's {} frames".format( + guide_frames, frame_idx, frame_count)) + + keyframe = {"resolved_frame_index": resolved_frame_index} + if image is not None: + frames = _resize(image[:guide_frames], width, height, "center") + keyframe["latent"] = vae.encode(frames) + + if audio is not None: + if audio_vae is None: + raise ValueError("anchoring guide audio needs the audio_vae input") + audio_latent, audio_rt = _encode_ref_audio(audio_vae, audio) + # the streams share one time axis: FRAME_RESCALE per pixel frame, 1.0 per audio latent frame + max_rt = math.floor(samples.tensors[1].shape[-1] - FRAME_RESCALE * resolved_frame_index) + if max_rt < 1: + raise ValueError("frame_idx {} is past the end of the video's audio track".format(frame_idx)) + if audio_rt > max_rt: + audio_latent = audio_latent[..., :max_rt].clone() + keyframe["audio_latent"] = audio_latent + + keyframes = list(positive[0][1].get("minimax_keyframes", [])) + keyframes.append(keyframe) + positive = node_helpers.conditioning_set_values(positive, {"minimax_keyframes": keyframes}) + return io.NodeOutput(positive) + + class MiniMaxH3ReferenceToVideo(io.ComfyNode): """ref2va: prompt + reference images / videos / audio -> conditioning + AV latent. @@ -197,16 +282,6 @@ class MiniMaxH3ReferenceToVideo(io.ComfyNode): outputs=[io.Conditioning.Output(display_name="positive"), io.Latent.Output()], ) - @staticmethod - def _encode_ref_audio(audio_vae, audio): - waveform = audio["waveform"] # [B, C, L] - sr = audio["sample_rate"] - vae_sr = getattr(audio_vae, "audio_sample_rate", 32000) - if sr != vae_sr: - waveform = torchaudio.functional.resample(waveform, sr, vae_sr) - z = audio_vae.encode(waveform[:1].movedim(1, -1)) # [1, 32, 2, T] - return z, z.shape[-1] - @classmethod def execute(cls, clip, vae, audio_vae, prompt, width, height, length, ref_image_size="match", ref_images=None, ref_videos=None, ref_video_audios=None, ref_audios=None) -> io.NodeOutput: @@ -254,7 +329,7 @@ class MiniMaxH3ReferenceToVideo(io.ComfyNode): z = vae.encode(frames) audio_latent, ref_audio_t = (None, 0) if soundtrack is not None: - audio_latent, ref_audio_t = cls._encode_ref_audio(audio_vae, soundtrack) + audio_latent, ref_audio_t = _encode_ref_audio(audio_vae, soundtrack) # the soundtrack gets its own