fix image_to_3d

app.py

@@ -1,7 +1,7 @@
 import gradio as gr
 import spaces
 from gradio_litmodel3d import LitModel3D
-
+import json
 import os
 import shutil
 os.environ['SPCONV_ALGO'] = 'native'
@@ -119,7 +119,7 @@ def image_to_3d(
     slat_sampling_steps: int,
     multiimage_algo: Literal["multidiffusion", "stochastic"],
     req: gr.Request,
-) -> Tuple[dict, dict, str]:
+) -> Tuple[dict, str, str]:
     """
     Convert an image (or multiple images) into a 3D model and return its state and video.
 
@@ -135,9 +135,10 @@ def image_to_3d(
         multiimage_algo (str): Multi-image algorithm to use.
 
     Returns:
-        dict: Packed state (Gaussian + Mesh) for later usage (e.g., extract_glb).
-        dict: Gradio-compatible video dictionary {"video": ..., "subtitles": None}.
-        str: Path to raw video file (used by Gradio Client or download logic).
+        dict: The information of the generated 3D model.
+        str: The path to the video of the 3D model.
+        str: serialized JSON of state
+
     """
     user_dir = os.path.join(TMP_DIR, str(req.session_hash))
     os.makedirs(user_dir, exist_ok=True)
@@ -187,7 +188,7 @@ def image_to_3d(
     # Pack state for downstream use
     state = pack_state(outputs['gaussian'][0], outputs['mesh'][0])
     torch.cuda.empty_cache()
-    return state, {"video": video_path, "subtitles": None}, video_path
+    return state, video_path, json.dumps(state)
 
 
 
@@ -324,7 +325,7 @@ with gr.Blocks(delete_cache=(600, 600)) as demo:
     
     is_multiimage = gr.State(False)
     output_buf = gr.State()
-    video_file_path = gr.Textbox(visible=False, label="Video Path")
+    state_textbox = gr.Textbox(visible=False, label="Serialized State")
 
     # Example images at the bottom of the page
     with gr.Row() as single_image_example:
@@ -384,7 +385,7 @@ with gr.Blocks(delete_cache=(600, 600)) as demo:
             ss_guidance_strength, ss_sampling_steps,
             slat_guidance_strength, slat_sampling_steps, multiimage_algo
         ],
-        outputs=[output_buf, video_output, video_file_path],  # multi output
+        outputs=[output_buf, video_output, state_textbox],  # multi output
     ).then(
         lambda: tuple([gr.Button(interactive=True), gr.Button(interactive=True)]),
         outputs=[extract_glb_btn, extract_gs_btn],