Enhance app.py UI with new logo and updated title; add demo assets to .gitattributes. Remove unused main function from navigation.py and streamline .gitignore.

.gitattributes

@@ -42,3 +42,5 @@ test_samples/living_room_2.jpeg filter=lfs diff=lfs merge=lfs -text
 test_samples/living_room.jpg filter=lfs diff=lfs merge=lfs -text
 test_samples/oxford.jpg filter=lfs diff=lfs merge=lfs -text
 test_samples/arc_de_tromphe.jpeg filter=lfs diff=lfs merge=lfs -text
+assets/demo_teaser.gif filter=lfs diff=lfs merge=lfs -text
+assets/title_logo.png filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -1,4 +1,3 @@
-assets/*
 pycache/*
 __pycache__/*
 .DS_Store

app.py

@@ -807,9 +807,10 @@ with gr.Blocks(theme=gr.themes.Base(primary_hue="blue")) as demo:
     demo_idx = gr.State(value=3)
 
     with gr.Sidebar():
-        gr.Markdown("# VMem: Consistent Video Scene Generation with Surfel-Indexed View Memory", elem_id="page-title")
+        gr.Image("assets/title_logo.png", width=60, height=60, show_label=False, show_download_button=False, container=False, interactive=False, show_fullscreen_button=False)
+        gr.Markdown("# Consistent Interactive Video Scene Generation with Surfel-Indexed View Memory", elem_id="page-title")
         gr.Markdown(
-            "### Official Interactive Demo for [_VMem_](https://arxiv.org/abs/2502.06764) that enables interactive consistent video scene generation."
+            "### Interactive Demo for [_VMem_](https://arxiv.org/abs/2502.06764) that enables interactive consistent video scene generation."
         )
         gr.Markdown("---")
         gr.Markdown("#### Links ↓")
@@ -868,8 +869,8 @@ with gr.Blocks(theme=gr.themes.Base(primary_hue="blue")) as demo:
                 
 
 if __name__ == "__main__":
-    demo.launch(debug=True,
+    demo.launch(debug=False,
                 share=True,
-                max_threads=1,  # Limit concurrent processing
-                show_error=True,  # Show detailed error messages
+                max_threads=1,
+                show_error=False,
                 )

navigation.py

@@ -428,51 +428,3 @@ class Navigator:
     
    
 
-
-def main():
-    parser = argparse.ArgumentParser(description="Interactive navigation in VMem")
-    parser.add_argument("--config", type=str, default="configs/inference/inference.yaml", help="Path to config file")
-    parser.add_argument("--step_size", type=float, default=0.1, help="Forward step size")
-    parser.add_argument("--interpolation_frames", type=int, default=4, help="Number of frames for each movement")
-    parser.add_argument("--commands", type=str, default="a,a,a,a,a,d,d,d,d,d,d,w,w,w,w,a,a,a,a,d,d,d,d,s,s,s,s", help="Comma-separated commands to execute (w,a,s,d,c,q) where c is circulate")
-    # parser.add_argument("--commands", type=str, default="d,d,d,d,w,w,w,d,d,d,d,d,a,a,a,a,a,s,s", help="Comma-separated commands to execute (w,a,s,d,c,q) where c is circulate")
-    parser.add_argument("--output_dir", type=str, default="./visualization/navigation_frames", help="Directory to save output frames")
-    parser.add_argument("--save_poses", type=str, default="./visualization/transforms.json", help="Path to save camera poses in NeRF format")
-    args = parser.parse_args()
-    
-    # Load configuration
-    config = OmegaConf.load(args.config)
-    
-    # Initialize the pipeline
-    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-    pipeline = VMemPipeline(config, device=device)
-    
-    # Create the navigator
-    navigator = Navigator(pipeline, step_size=args.step_size, num_interpolation_frames=args.interpolation_frames)
-    
-    # Load episode data
-    frame_path = "test_samples/arc_de_tromphe.jpeg"
-    image, _ = load_img_and_K(frame_path, None, K=None, device=device)
-    image, _ = transform_img_and_K(image, (config.model.height, config.model.width), mode="crop", K=None)
-    ori_K = np.array(get_default_intrinsics()[0])
-    initial_pose = np.eye(4)
-    
-    # Initialize the navigator with the first frame using pipeline's initialize method
-    initial_frame = navigator.initialize(image, initial_pose, ori_K)
-    
-    # Create output directory if needed
-    if args.output_dir:
-        os.makedirs(args.output_dir, exist_ok=True)
-        initial_frame.save(os.path.join(args.output_dir, "initial.png"))
-    
-    # If commands are provided, execute them in sequence
-    commands = args.commands.split(',')
-    all_frames_lists = navigator.navigate(commands)
-    
-
-    # Save camera poses
-    if args.save_poses:
-        navigator.save_camera_poses(args.save_poses)
-
-if __name__ == "__main__":
-    main()