Delete app.py

app.py

@@ -1,255 +0,0 @@
-import gradio as gr
-import os
-import sys
-import numpy as np
-import torch
-from pathlib import Path
-import tempfile
-import zipfile
-
-# Add utils to path
-sys.path.append('utils')
-
-# Import your utility modules
-from model_loader import load_mdm_model, generate_motion
-from visualization import create_stick_figure_video, create_3d_visualization
-from blender_export import prepare_blender_export
-
-# Initialize models dictionary
-MODELS = {
-    "HumanML3D (50-steps Fast)": "humanml_trans_enc_512_50steps",
-    "HumanML3D (Original)": "humanml_trans_enc_512",
-    "KIT Dataset": "kit_encoder_512",
-    "HumanAct12": "humanact12",
-    "UESTC": "uestc",
-    "Unconstrained": "humanact12_unconstrained"
-}
-
-# Model loading function
-def load_selected_model(model_name):
-    model_path = f"./save/{MODELS[model_name]}"
-    return load_mdm_model(model_name, model_path)
-
-# Motion generation function
-def generate_motion_interface(
-    model_name,
-    text_prompt,
-    action_name,
-    num_samples,
-    motion_length,
-    seed,
-    temperature
-):
-    """
-    Unified interface for motion generation
-    """
-    try:
-        # Load model
-        model = load_selected_model(model_name)
-        
-        # Generate based on model type
-        if "HumanML3D" in model_name or "KIT" in model_name:
-            # Text-to-motion
-            results = generate_motion(
-                model=model,
-                text_prompt=text_prompt,
-                num_samples=num_samples,
-                motion_length=motion_length,
-                seed=seed
-            )
-        else:
-            # Action-to-motion or unconstrained
-            results = generate_motion(
-                model=model,
-                action_name=action_name,
-                num_samples=num_samples,
-                seed=seed
-            )
-        
-        # Create visualization
-        video_path = create_stick_figure_video(results)
-        plot_path = create_3d_visualization(results)
-        
-        # Prepare export data
-        export_data = prepare_blender_export(results)
-        
-        return video_path, plot_path, export_data, " Generation successful!"
-        
-    except Exception as e:
-        return None, None, None, f" Error: {str(e)}"
-
-# Blender export function
-def export_to_blender(export_data, format_type="obj"):
-    """
-    Prepare files for Blender import
-    """
-    if export_data is None:
-        return None, "No motion data to export"
-    
-    # Create temporary directory
-    with tempfile.TemporaryDirectory() as tmpdir:
-        # Prepare files based on format
-        if format_type == "obj":
-            # Save OBJ sequences
-            export_path = export_data.save_as_obj_sequence(tmpdir)
-        elif format_type == "smpl_params":
-            # Save SMPL parameters
-            export_path = export_data.save_smpl_parameters(tmpdir)
-        elif format_type == "fbx":
-            # Convert to FBX (would need additional conversion)
-            export_path = export_data.convert_to_fbx(tmpdir)
-        
-        # Create ZIP archive
-        zip_path = f"{tmpdir}/blender_export.zip"
-        with zipfile.ZipFile(zip_path, 'w') as zipf:
-            for root, dirs, files in os.walk(export_path):
-                for file in files:
-                    zipf.write(
-                        os.path.join(root, file),
-                        os.path.relpath(os.path.join(root, file), tmpdir)
-                    )
-        
-        return zip_path, " Export ready! Download and extract for Blender."
-
-# Gradio Interface
-with gr.Blocks(
-    title="Motion Diffusion Model",
-    theme=gr.themes.Soft(),
-    css=".gradio-container {max-width: 1200px !important;}"
-) as demo:
-    
-    gr.Markdown("#  Motion Diffusion Model")
-    gr.Markdown("Generate 3D human motions from text or actions")
-    
-    with gr.Row():
-        with gr.Column(scale=1):
-            # Model selection
-            model_dropdown = gr.Dropdown(
-                choices=list(MODELS.keys()),
-                value="HumanML3D (50-steps Fast)",
-                label="Select Model",
-                info="HumanML3D-50steps is 20x faster"
-            )
-            
-            # Common parameters
-            num_samples = gr.Slider(
-                minimum=1, maximum=10, value=3, step=1,
-                label="Number of Samples"
-            )
-            seed = gr.Number(value=42, label="Random Seed")
-            motion_length = gr.Slider(
-                minimum=1.0, maximum=9.8, value=3.0, step=0.5,
-                label="Motion Length (seconds)",
-                visible=True  # Only for text-to-motion
-            )
-        
-        with gr.Column(scale=2):
-            # Text input for text-to-motion models
-            text_input = gr.Textbox(
-                label="Text Prompt",
-                placeholder="the person walked forward and is picking up his toolbox...",
-                lines=3,
-                value="a person is walking slowly and then starts running"
-            )
-            
-            # Action input for action-to-motion models
-            action_input = gr.Textbox(
-                label="Action Name",
-                placeholder="walk, run, jump, drink, ...",
-                visible=False  # Shown based on model selection
-            )
-    
-    # Generate button
-    generate_btn = gr.Button(" Generate Motion", variant="primary")
-    
-    # Results area
-    with gr.Row():
-        video_output = gr.Video(
-            label="Stick Figure Animation",
-            format="mp4"
-        )
-        plot_output = gr.Plot(
-            label="3D Motion Visualization"
-        )
-    
-    # Status message
-    status_msg = gr.Markdown("")
-    
-    # Export section
-    with gr.Accordion(" Export to Blender", open=False):
-        with gr.Row():
-            export_format = gr.Radio(
-                choices=["OBJ Sequence", "SMPL Parameters", "FBX"],
-                value="OBJ Sequence",
-                label="Export Format"
-            )
-            export_btn = gr.Button("Prepare Blender Export")
-        
-        export_download = gr.File(label="Download Export Package")
-        export_status = gr.Markdown("")
-    
-    # Model-specific UI updates
-    def update_ui(model_name):
-        """Show/hide relevant inputs based on model type"""
-        is_text_model = "HumanML3D" in model_name or "KIT" in model_name
-        return [
-            gr.Textbox(visible=is_text_model),  # text_input
-            gr.Textbox(visible=not is_text_model),  # action_input
-            gr.Slider(visible=is_text_model)  # motion_length
-        ]
-    
-    # Connect events
-    model_dropdown.change(
-        update_ui,
-        inputs=[model_dropdown],
-        outputs=[text_input, action_input, motion_length]
-    )
-    
-    generate_btn.click(
-        generate_motion_interface,
-        inputs=[
-            model_dropdown,
-            text_input,
-            action_input,
-            num_samples,
-            motion_length,
-            seed,
-            gr.Slider(minimum=0.1, maximum=2.0, value=1.0, label="Temperature", visible=False)
-        ],
-        outputs=[video_output, plot_output, gr.State(), status_msg]
-    )
-    
-    export_btn.click(
-        export_to_blender,
-        inputs=[gr.State(), export_format],
-        outputs=[export_download, export_status]
-    )
-    
-    # Examples
-    gr.Examples(
-        examples=[
-            ["HumanML3D (50-steps Fast)", "a person is doing jumping jacks", None, 2, 3.0, 123],
-            ["HumanML3D (50-steps Fast)", "someone walks, stops to tie their shoe, then continues", None, 1, 5.0, 456],
-            ["HumanAct12", None, "walk", 3, 3.0, 789],
-            ["UESTC", None, "kick", 2, 2.5, 101],
-        ],
-        inputs=[model_dropdown, text_input, action_input, num_samples, motion_length, seed],
-        outputs=[video_output, plot_output],
-        fn=generate_motion_interface,
-        cache_examples=False
-    )
-
-# Launch for Hugging Face Spaces
-if __name__ == "__main__":
-    # Check and download models on first run
-    if not os.path.exists("./save"):
-        print(" Models not found. Running download script...")
-        os.system("python download_models.py --lite")
-    
-    # Launch with Space-appropriate settings
-    demo.launch(
-        server_name="0.0.0.0",
-        server_port=int(os.getenv("PORT", 7860)),
-        share=False,
-        debug=False
-    )