Don't depend on torchvision.

comfy/utils.py

 import torch
-import torchvision
 import math
 import struct
 import comfy.checkpoint_pickle
 import safetensors.torch
+import numpy as np
 from PIL import Image
 
 def load_torch_file(ckpt, safe_load=False, device=None):
@@ -349,9 +349,9 @@ def bislerp(samples, width, height):
     return result
 
 def lanczos(samples, width, height):
-    images = [torchvision.transforms.functional.to_pil_image(image) for image in samples]
+    images = [Image.fromarray(np.clip(255. * image.movedim(0, -1).cpu().numpy(), 0, 255).astype(np.uint8)) for image in samples]
     images = [image.resize((width, height), resample=Image.Resampling.LANCZOS) for image in images]
-    images = [torchvision.transforms.functional.to_tensor(image) for image in images]
+    images = [torch.from_numpy(np.array(image).astype(np.float32) / 255.0).movedim(-1, 0) for image in images]
     result = torch.stack(images)
     return result