Remove unused grid_sizes assignment in CameraWanModel, WanModel_S2V, HumoWanModel, and AnimateWanModel; simplify patch size initialization in WanUni3CControlnet; refine sigmas check in WanUni3CCnetPatch.

comfy/ldm/wan/model.py

@@ -908,7 +908,6 @@ class CameraWanModel(WanModel):
         if self.control_adapter is not None and camera_conditions is not None:
             x = x + self.control_adapter(camera_conditions).to(x.dtype)
         grid_sizes = x.shape[2:]
-        transformer_options["grid_sizes"] = grid_sizes
         x = x.flatten(2).transpose(1, 2)
 
         # time embeddings
@@ -1368,7 +1367,6 @@ class WanModel_S2V(WanModel):
             t = t.unsqueeze(1).repeat(1, x.shape[2])
 
         grid_sizes = x.shape[2:]
-        transformer_options["grid_sizes"] = grid_sizes
         x = x.flatten(2).transpose(1, 2)
         seq_len = x.size(1)
 
@@ -1635,7 +1633,6 @@ class HumoWanModel(WanModel):
         x_input = x
         x = self.patch_embedding(x.float()).to(x.dtype)
         grid_sizes = x.shape[2:]
-        transformer_options["grid_sizes"] = grid_sizes
         x = x.flatten(2).transpose(1, 2)
 
         # time embeddings

comfy/ldm/wan/model_animate.py

@@ -497,7 +497,6 @@ class AnimateWanModel(WanModel):
         x = self.patch_embedding(x.float()).to(x.dtype)
         x, motion_vec = self.after_patch_embedding(x, pose_latents, face_pixel_values)
         grid_sizes = x.shape[2:]
-        transformer_options["grid_sizes"] = grid_sizes
         x = x.flatten(2).transpose(1, 2)
 
         # time embeddings

comfy/ldm/wan/uni3c.py

@@ -97,13 +97,11 @@ class WanUni3CControlnet(nn.Module):
         device=None, dtype=None, operations=None
     ):
         super().__init__()
-        self.dtype = dtype
-        self.patch_size = (1, 2, 2)
-        self.in_channels = in_channels
+        patch_size = (1, 2, 2)
         self.num_layers = num_layers
 
         self.controlnet_patch_embedding = operations.Conv3d(
-            in_channels, conv_out_dim, kernel_size=self.patch_size, stride=self.patch_size, device=device, dtype=torch.float32)
+            in_channels, conv_out_dim, kernel_size=patch_size, stride=patch_size, device=device, dtype=torch.float32)
         self.controlnet_mask_embedding = MaskCamEmbed(add_channels, mid_channels, conv_out_dim, device=device, dtype=dtype, operations=operations)
 
         if conv_out_dim != dim:
@@ -152,11 +150,3 @@ class WanUni3CControlnet(nn.Module):
         hidden = self.controlnet_blocks[block_index](hidden, temb, freqs, transformer_options=transformer_options)
         residual = self.proj_out[block_index](hidden)
         return hidden, residual
-
-    def forward(self, control_input, temb, render_mask=None, camera_embedding=None, transformer_options={}):
-        hidden, freqs = self.process_input(control_input, render_mask=render_mask, camera_embedding=camera_embedding)
-        controlnet_states = []
-        for i in range(self.num_layers):
-            hidden, residual = self.forward_block(i, hidden, temb, freqs, transformer_options=transformer_options)
-            controlnet_states.append(residual)
-        return controlnet_states

comfy_extras/nodes_model_patch.py

@@ -572,7 +572,7 @@ class WanUni3CCnetPatch:
             self.temp_data = None
             active = True
             sigmas = transformer_options.get("sigmas", None)
-            if sigmas is not None and sigmas.numel() > 0:
+            if sigmas is not None:
                 sigma = sigmas[0].item()
                 if sigma > self.sigma_start or sigma < self.sigma_end:
                     active = False