Minor cleanup

comfy/model_base.py

@@ -1920,6 +1920,7 @@ class WAN21_SCAIL(WAN21):
         pose_latents = kwargs.get("pose_video_latent", None)
         if pose_latents is not None:
             out['pose_latents'] = [pose_latents.shape[0], 20, *pose_latents.shape[2:]]
+
         return out
 
 class WAN21_SCAIL2(WAN21_SCAIL):

comfy/sd.py

@@ -546,13 +546,11 @@ class VAE:
                 self.latent_channels = 16
             elif "shape_dec.blocks.1.16.to_subdiv.weight" in sd: # trellis2 shape vae (struct_dec + shape_dec)
                 self.working_dtypes = [torch.float16, torch.bfloat16, torch.float32]
-                # TODO
                 self.memory_used_decode = lambda shape, dtype: (2500 * shape[2] * shape[3]) * model_management.dtype_size(dtype)
                 self.memory_used_encode = lambda shape, dtype: (2500 * shape[2] * shape[3]) * model_management.dtype_size(dtype)
                 self.first_stage_model = comfy.ldm.trellis2.vae.ShapeVae()
             elif "txt_dec.blocks.3.4.conv2.weight" in sd: # trellis2 texture vae
                 self.working_dtypes = [torch.float16, torch.bfloat16, torch.float32]
-                # TODO
                 self.memory_used_decode = lambda shape, dtype: (2500 * shape[2] * shape[3]) * model_management.dtype_size(dtype)
                 self.memory_used_encode = lambda shape, dtype: (2500 * shape[2] * shape[3]) * model_management.dtype_size(dtype)
                 self.first_stage_model = comfy.ldm.trellis2.vae.TextureVae()

comfy/supported_models.py

@@ -2046,7 +2046,6 @@ class Kandinsky5Image(Kandinsky5):
         return supported_models_base.ClipTarget(comfy.text_encoders.kandinsky5.Kandinsky5TokenizerImage, comfy.text_encoders.kandinsky5.te(**hunyuan_detect))
 
 
-
 class ACEStep15(supported_models_base.BASE):
     unet_config = {
         "audio_model": "ace1.5",
@@ -2086,6 +2085,7 @@ class ACEStep15(supported_models_base.BASE):
 
         return supported_models_base.ClipTarget(comfy.text_encoders.ace15.ACE15Tokenizer, comfy.text_encoders.ace15.te(**detect))
 
+
 class LongCatImage(supported_models_base.BASE):
     unet_config = {
         "image_model": "flux",
@@ -2180,7 +2180,6 @@ class ErnieImage(supported_models_base.BASE):
         return supported_models_base.ClipTarget(comfy.text_encoders.ernie.ErnieTokenizer, comfy.text_encoders.ernie.te(**hunyuan_detect))
 
 
-
 class SAM3(supported_models_base.BASE):
     unet_config = {"image_model": "SAM3"}
     supported_inference_dtypes = [torch.float16, torch.bfloat16, torch.float32]
@@ -2300,6 +2299,7 @@ class CogVideoX_Inpaint(CogVideoX_T2V):
         out = model_base.CogVideoX(self, image_to_video=True, device=device)
         return out
 
+
 models = [
     LotusD,
     Stable_Zero123,

comfy_extras/mesh3d/postprocess/remesh.py

@@ -402,7 +402,7 @@ def _dual_contour(voxel_coords: torch.Tensor, corner_udf: torch.Tensor,
             b.add_(centroid_verts, alpha=reg)
             try:
                 qef_solution = torch.linalg.solve(A, b.unsqueeze(-1)).squeeze(-1)
-            except Exception:
+            except torch.linalg.LinAlgError:
                 qef_solution = centroid_verts
 
             # Clamp QEF output to the voxel bbox

comfy_extras/nodes_mesh_postprocess.py

@@ -13,7 +13,6 @@ from comfy_extras.mesh3d.uv_unwrap import mesh as _uv_mesh
 from comfy_extras.mesh3d.uv_unwrap import segment as _uv_seg
 from comfy_extras.mesh3d.uv_unwrap import parameterize as _uv_param
 from comfy_extras.mesh3d.uv_unwrap import pack as _uv_pack
-import warnings
 import logging
 from tqdm import tqdm
 from scipy.sparse import csr_matrix
@@ -93,7 +92,7 @@ def paint_mesh_with_voxels(mesh, voxel_coords, voxel_colors, resolution):
     if v_colors.shape[-1] > 3:
         v_colors = v_colors[:, :3]
 
-    srgb_colors = v_colors.clamp(0, 1)#(v_colors * 0.5 + 0.5).clamp(0, 1)
+    srgb_colors = v_colors.clamp(0, 1)
 
     # to Linear RGB (required for GLTF)
     linear_colors = torch.pow(srgb_colors, 2.2)
@@ -2404,8 +2403,8 @@ def _uv_unwrap(positions, indices, segmenter, resolution, padding, weld_distance
     mesh = _uv_mesh.build_mesh(v_in, f_in)
     ff = mesh.face_face
     if ff.numel() and float((ff >= 0).float().mean().item()) < 0.25:
-        warnings.warn("[uv_unwrap] mesh face-adjacency < 25% — vertices appear un-welded "
-                      "(triangle soup); UV charts will be per-face. Raise weld_distance.")
+        logging.warning("[uv_unwrap] mesh face-adjacency < 25% — vertices appear un-welded "
+                        "(triangle soup); UV charts will be per-face. Raise weld_distance.")
 
     if segmenter == "pec":
         if mesh.faces.device.type != "cuda":

comfy_extras/nodes_trellis2.py

@@ -467,7 +467,6 @@ class Trellis2Conditioning(IO.ComfyNode):
             ]
         )
 
-    @classmethod
     @classmethod
     def execute(cls, clip_vision_model, image, mask) -> IO.NodeOutput:
         # Normalize to batched form so per-image conditioning loop below is uniform.