modifying requirements.txt

inference/cli_demo_camera_i2v_pcd.py

@@ -26,7 +26,7 @@ from PIL import Image
 import numpy as np
 import torchvision.transforms as transforms
 import cv2
-
+import torchvision.transforms as T
 import cv2
 import numpy as np
 import torch
@@ -368,6 +368,14 @@ def generate_video(
 
         # ++++++++++++++++++++++++++++++++++++++
         latents = video_generate_all # This is a latent
+        # Fix the nested list structure
+        if isinstance(latents, list) and len(latents) == 1 and isinstance(latents[0], list):
+            latents = latents[0]  # ✅ Unwrap the inner list of PIL images
+
+        transform = T.ToTensor()  # Converts PIL image to torch.FloatTensor in [0,1]
+
+        latents = torch.stack([transform(img) for img in latents])  # Shape: [B, C, H, W]
+
         print(f"Type of latents: {type(latents)}")
         print(f"Length of latents: {len(latents)}")
 
@@ -387,8 +395,6 @@ def generate_video(
                 print(f"  Mode: {item.mode}")
             else:
                 print(f"  Value: {item}")
-        # Only works if all elements are tensors of the same shape
-        latents = torch.stack(latents)
 
         scale_status = True
         rife_status = True