Update app.py

app.py

@@ -7,7 +7,7 @@ import torch
 import rembg
 from datetime import datetime
 import subprocess
-import gradio as gr
+import argparse
 
 try:
     # running on Hugging Face Spaces
@@ -137,83 +137,32 @@ def process_3d(input_image, num_steps=50, cfg_scale=7, grid_res=384, seed=42, si
     # export the whole mesh
     mesh.export(output_glb_path)
 
+    print(f"Saved 3D model to {output_glb_path}")
     return output_glb_path
 
-# gradio UI
-
-_TITLE = '''PartPacker: Efficient Part-level 3D Object Generation via Dual Volume Packing'''
-
-_DESCRIPTION = '''
-<div>
-<a style="display:inline-block" href="https://research.nvidia.com/labs/dir/partpacker/"><img src='https://img.shields.io/badge/public_website-8A2BE2'></a>
-<a style="display:inline-block; margin-left: .5em" href="https://github.com/NVlabs/PartPacker"><img src='https://img.shields.io/github/stars/NVlabs/PartPacker?style=social'/></a>
-</div>
-
-* Each part is visualized with a random color, and can be separated in the GLB file.
-* If the output is not satisfactory, please try different random seeds!
-'''
-
-block = gr.Blocks(title=_TITLE).queue()
-with block:
-    with gr.Row():
-        with gr.Column():
-            gr.Markdown('# ' + _TITLE)
-    gr.Markdown(_DESCRIPTION)
-
-    with gr.Row():
-        with gr.Column(scale=1):
-            with gr.Row():
-                # input image
-                input_image = gr.Image(label="Input Image", type="filepath") # use file_path and load manually
-                seg_image = gr.Image(label="Segmentation Result", type="numpy", interactive=False, image_mode="RGBA")
-            with gr.Accordion("Settings", open=True):
-                # inference steps
-                num_steps = gr.Slider(label="Inference steps", minimum=1, maximum=100, step=1, value=50)
-                # cfg scale
-                cfg_scale = gr.Slider(label="CFG scale", minimum=2, maximum=10, step=0.1, value=7.0)
-                # grid resolution
-                input_grid_res = gr.Slider(label="Grid resolution", minimum=256, maximum=512, step=1, value=384)
-                # random seed
-                with gr.Row():
-                    randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
-                    seed = gr.Slider(label="Seed", minimum=0, maximum=MAX_SEED, step=1, value=0)
-                # simplify mesh
-                with gr.Row():
-                    simplify_mesh = gr.Checkbox(label="Simplify mesh", value=False)
-                    target_num_faces = gr.Slider(label="Face number", minimum=10000, maximum=1000000, step=1000, value=100000)
-                # gen button
-                button_gen = gr.Button("Generate")
-
-        with gr.Column(scale=1):
-            # glb file
-            output_model = gr.Model3D(label="Geometry", height=512)
-
-            
-    with gr.Row():
-        gr.Examples(
-            examples=[
-                ["examples/rabbit.png"],
-                ["examples/robot.png"],
-                ["examples/teapot.png"],
-                ["examples/barrel.png"],
-                ["examples/cactus.png"],
-                ["examples/cyan_car.png"],
-                ["examples/pickup.png"],
-                ["examples/swivelchair.png"],
-                ["examples/warhammer.png"],
-            ],
-            fn=process_image,  # still need to click button_gen to get the 3d
-            inputs=[input_image],
-            outputs=[seg_image],
-            cache_examples=False,
-        )
-
-    button_gen.click(
-        process_image, inputs=[input_image], outputs=[seg_image]
-    ).then(
-        get_random_seed, inputs=[randomize_seed, seed], outputs=[seed]
-    ).then(
-        process_3d, inputs=[seg_image, num_steps, cfg_scale, input_grid_res, seed, simplify_mesh, target_num_faces], outputs=[output_model]
-    )
-
-block.launch(ssr_mode=False)
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--image_path', type=str, default='examples/robot.png', help='Path to input image')
+    parser.add_argument('--num_steps', type=int, default=50, help='Inference steps')
+    parser.add_argument('--cfg_scale', type=float, default=7.0, help='CFG scale')
+    parser.add_argument('--grid_res', type=int, default=384, help='Grid resolution')
+    parser.add_argument('--seed', type=int, default=42, help='Random seed')
+    parser.add_argument('--randomize_seed', action='store_true', help='Randomize seed')
+    parser.add_argument('--simplify_mesh', action='store_true', help='Simplify mesh')
+    parser.add_argument('--target_num_faces', type=int, default=100000, help='Target number of faces for mesh simplification')
+    
+    args = parser.parse_args()
+
+    if args.randomize_seed:
+        args.seed = get_random_seed(args.randomize_seed, args.seed)
+
+    processed_image = process_image(args.image_path)
+    process_3d(
+        input_image=processed_image,
+        num_steps=args.num_steps,
+        cfg_scale=args.cfg_scale,
+        grid_res=args.grid_res,
+        seed=args.seed,
+        simplify_mesh=args.simplify_mesh,
+        target_num_faces=args.target_num_faces
+    )