Update app.py

app.py

@@ -1,280 +1,315 @@
 import gradio as gr
-import torch
-import numpy as np
+import spaces
+from gradio_litmodel3d import LitModel3D
 import os
 import shutil
+os.environ['SPCONV_ALGO'] = 'native'
+from typing import *
+import torch
+import numpy as np
 import imageio
+from easydict import EasyDict as edict
 from PIL import Image
+from trellis.pipelines import TrellisImageTo3DPipeline
+from trellis.representations import Gaussian, MeshExtractResult
+from trellis.utils import render_utils, postprocessing_utils
 
-# Ensure imports are available
-try:
-    from trellis.pipelines import TrellisImageTo3DPipeline
-    from trellis.representations import Gaussian, MeshExtractResult
-    from trellis.utils import render_utils, postprocessing_utils
-    from easydict import EasyDict as edict
-except ImportError as e:
-    print(f"Error importing required libraries: {e}")
-    print("Please install the following libraries:")
-    print("- trellis-ai")
-    print("- easydict")
-    TrellisImageTo3DPipeline = None
-
-# Constants
 MAX_SEED = np.iinfo(np.int32).max
 TMP_DIR = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'tmp')
 os.makedirs(TMP_DIR, exist_ok=True)
 
-class ImageTo3DConverter:
-    def __init__(self):
-        # Initialize the pipeline with error handling
-        try:
-            self.pipeline = TrellisImageTo3DPipeline()
-        except Exception as e:
-            print(f"Failed to initialize pipeline: {e}")
-            self.pipeline = None
-
-    def validate_input(self, image, is_multiimage):
-        """Validate input images before processing"""
-        if not self.pipeline:
-            raise ValueError("Pipeline not initialized. Check library installation.")
-        
-        if is_multiimage:
-            # Handle multi-image input
-            if not image or len(image) == 0:
-                raise ValueError("No images provided for multi-image processing")
-            # Ensure images are PIL Image objects
-            valid_images = [img[0] if isinstance(img, list) else img for img in image]
-            return valid_images
-        else:
-            # Handle single image input
-            if image is None:
-                raise ValueError("No image provided")
-            return image
-
-    def preprocess_image(self, image):
-        """Safely preprocess a single image"""
-        try:
-            return self.pipeline.preprocess_image(image)
-        except Exception as e:
-            print(f"Image preprocessing error: {e}")
-            return image
-
-    def process_image(self, 
-                      image, 
-                      multiimages, 
-                      is_multiimage, 
-                      seed, 
-                      ss_guidance_strength, 
-                      ss_sampling_steps,
-                      slat_guidance_strength, 
-                      slat_sampling_steps,
-                      multiimage_algo):
-        """Main image to 3D conversion method"""
-        # Validate and preprocess input
-        try:
-            processed_input = self.validate_input(image if not is_multiimage else multiimages, is_multiimage)
-        except ValueError as e:
-            print(f"Input validation error: {e}")
-            return None, None
-
-        # Determine processing method based on input type
-        try:
-            if not is_multiimage:
-                outputs = self.pipeline.run(
-                    processed_input,
-                    seed=seed,
-                    formats=["gaussian", "mesh"],
-                    preprocess_image=False,
-                    sparse_structure_sampler_params={
-                        "steps": ss_sampling_steps,
-                        "cfg_strength": ss_guidance_strength,
-                    },
-                    slat_sampler_params={
-                        "steps": slat_sampling_steps,
-                        "cfg_strength": slat_guidance_strength,
-                    },
-                )
-            else:
-                outputs = self.pipeline.run_multi_image(
-                    processed_input,
-                    seed=seed,
-                    formats=["gaussian", "mesh"],
-                    preprocess_image=False,
-                    sparse_structure_sampler_params={
-                        "steps": ss_sampling_steps,
-                        "cfg_strength": ss_guidance_strength,
-                    },
-                    slat_sampler_params={
-                        "steps": slat_sampling_steps,
-                        "cfg_strength": slat_guidance_strength,
-                    },
-                    mode=multiimage_algo,
-                )
-        except Exception as e:
-            print(f"3D conversion error: {e}")
-            return None, None
-
-        # Generate video
-        try:
-            video = render_utils.render_video(outputs['gaussian'][0], num_frames=120)['color']
-            video_geo = render_utils.render_video(outputs['mesh'][0], num_frames=120)['normal']
-            video = [np.concatenate([video[i], video_geo[i]], axis=1) for i in range(len(video))]
-            
-            # Save video
-            user_dir = os.path.join(TMP_DIR, 'temp_session')
-            os.makedirs(user_dir, exist_ok=True)
-            video_path = os.path.join(user_dir, 'sample.mp4')
-            imageio.mimsave(video_path, video, fps=15)
-
-            # Pack and return state
-            state = {
-                'gaussian': {
-                    **outputs['gaussian'][0].init_params,
-                    '_xyz': outputs['gaussian'][0]._xyz.cpu().numpy(),
-                    '_features_dc': outputs['gaussian'][0]._features_dc.cpu().numpy(),
-                    '_scaling': outputs['gaussian'][0]._scaling.cpu().numpy(),
-                    '_rotation': outputs['gaussian'][0]._rotation.cpu().numpy(),
-                    '_opacity': outputs['gaussian'][0]._opacity.cpu().numpy(),
-                },
-                'mesh': {
-                    'vertices': outputs['mesh'][0].vertices.cpu().numpy(),
-                    'faces': outputs['mesh'][0].faces.cpu().numpy(),
-                },
-            }
-
-            return state, video_path
+def start_session(req: gr.Request):
+    user_dir = os.path.join(TMP_DIR, str(req.session_hash))
+    os.makedirs(user_dir, exist_ok=True)
+    
+def end_session(req: gr.Request):
+    user_dir = os.path.join(TMP_DIR, str(req.session_hash))
+    shutil.rmtree(user_dir)
 
-        except Exception as e:
-            print(f"Video generation error: {e}")
-            return None, None
+def preprocess_image(image: Image.Image) -> Image.Image:
+    """
+    Preprocess the input image.
+    Args:
+        image (Image.Image): The input image.
+    Returns:
+        Image.Image: The preprocessed image.
+    """
+    processed_image = pipeline.preprocess_image(image)
+    return processed_image
 
-    def extract_glb(self, state, mesh_simplify=0.95, texture_size=1024):
-        """Extract GLB from the processed state"""
-        try:
-            # Reconstruct Gaussian and Mesh from state
-            gs = Gaussian(
-                aabb=state['gaussian']['aabb'],
-                sh_degree=state['gaussian']['sh_degree'],
-                mininum_kernel_size=state['gaussian']['mininum_kernel_size'],
-                scaling_bias=state['gaussian'].get('scaling_bias', 0.1),
-                opacity_bias=state['gaussian'].get('opacity_bias', 0.1),
-                scaling_activation=state['gaussian'].get('scaling_activation', 'softplus'),
-            )
-            gs._xyz = torch.tensor(state['gaussian']['_xyz'], device='cuda')
-            gs._features_dc = torch.tensor(state['gaussian']['_features_dc'], device='cuda')
-            gs._scaling = torch.tensor(state['gaussian']['_scaling'], device='cuda')
-            gs._rotation = torch.tensor(state['gaussian']['_rotation'], device='cuda')
-            gs._opacity = torch.tensor(state['gaussian']['_opacity'], device='cuda')
-
-            mesh = edict(
-                vertices=torch.tensor(state['mesh']['vertices'], device='cuda'),
-                faces=torch.tensor(state['mesh']['faces'], device='cuda'),
-            )
-
-            # Convert mesh
-            mesh.vertices, mesh.faces = postprocessing_utils.remesh_to_quads(
-                vertices=mesh.vertices.cpu().numpy(),
-                faces=mesh.faces.cpu().numpy(),
-                simplify=mesh_simplify
-            )
-
-            # Generate GLB
-            glb = postprocessing_utils.to_glb(gs, mesh, simplify=mesh_simplify, texture_size=texture_size, verbose=False)
-            
-            # Save GLB
-            user_dir = os.path.join(TMP_DIR, 'temp_session')
-            os.makedirs(user_dir, exist_ok=True)
-            glb_path = os.path.join(user_dir, 'sample.glb')
-            glb.export(glb_path)
+def preprocess_images(images: List[Tuple[Image.Image, str]]) -> List[Image.Image]:
+    """
+    Preprocess a list of input images.
+    
+    Args:
+        images (List[Tuple[Image.Image, str]]): The input images.
+        
+    Returns:
+        List[Image.Image]: The preprocessed images.
+    """
+    images = [image[0] for image in images]
+    processed_images = [pipeline.preprocess_image(image) for image in images]
+    return processed_images
 
-            return glb_path
+def pack_state(gs: Gaussian, mesh: MeshExtractResult) -> dict:
+    return {
+        'gaussian': {
+            **gs.init_params,
+            '_xyz': gs._xyz.cpu().numpy(),
+            '_features_dc': gs._features_dc.cpu().numpy(),
+            '_scaling': gs._scaling.cpu().numpy(),
+            '_rotation': gs._rotation.cpu().numpy(),
+            '_opacity': gs._opacity.cpu().numpy(),
+        },
+        'mesh': {
+            'vertices': mesh.vertices.cpu().numpy(),
+            'faces': mesh.faces.cpu().numpy(),
+        },
+    }
+    
+def unpack_state(state: dict) -> Tuple[Gaussian, edict, str]:
+    gs = Gaussian(
+        aabb=state['gaussian']['aabb'],
+        sh_degree=state['gaussian']['sh_degree'],
+        mininum_kernel_size=state['gaussian']['mininum_kernel_size'],
+        scaling_bias=state['gaussian']['scaling_bias'],
+        opacity_bias=state['gaussian']['opacity_bias'],
+        scaling_activation=state['gaussian']['scaling_activation'],
+    )
+    gs._xyz = torch.tensor(state['gaussian']['_xyz'], device='cuda')
+    gs._features_dc = torch.tensor(state['gaussian']['_features_dc'], device='cuda')
+    gs._scaling = torch.tensor(state['gaussian']['_scaling'], device='cuda')
+    gs._rotation = torch.tensor(state['gaussian']['_rotation'], device='cuda')
+    gs._opacity = torch.tensor(state['gaussian']['_opacity'], device='cuda')
+    
+    mesh = edict(
+        vertices=torch.tensor(state['mesh']['vertices'], device='cuda'),
+        faces=torch.tensor(state['mesh']['faces'], device='cuda'),
+    )
+    
+    return gs, mesh
 
-        except Exception as e:
-            print(f"GLB extraction error: {e}")
-            return None
+def get_seed(randomize_seed: bool, seed: int) -> int:
+    """
+    Get the random seed.
+    """
+    return np.random.randint(0, MAX_SEED) if randomize_seed else seed
 
-# Gradio Interface Setup
-def create_gradio_interface():
-    converter = ImageTo3DConverter()
+@spaces.GPU
+def image_to_3d(
+    image: Image.Image,
+    multiimages: List[Tuple[Image.Image, str]],
+    is_multiimage: bool,
+    seed: int,
+    ss_guidance_strength: float,
+    ss_sampling_steps: int,
+    slat_guidance_strength: float,
+    slat_sampling_steps: int,
+    multiimage_algo: Literal["multidiffusion", "stochastic"],
+    req: gr.Request,
+) -> Tuple[dict, str]:
+    """
+    Convert an image to a 3D model.
+    Args:
+        image (Image.Image): The input image.
+        multiimages (List[Tuple[Image.Image, str]]): The input images in multi-image mode.
+        is_multiimage (bool): Whether is in multi-image mode.
+        seed (int): The random seed.
+        ss_guidance_strength (float): The guidance strength for sparse structure generation.
+        ss_sampling_steps (int): The number of sampling steps for sparse structure generation.
+        slat_guidance_strength (float): The guidance strength for structured latent generation.
+        slat_sampling_steps (int): The number of sampling steps for structured latent generation.
+        multiimage_algo (Literal["multidiffusion", "stochastic"]): The algorithm for multi-image generation.
+    Returns:
+        dict: The information of the generated 3D model.
+        str: The path to the video of the 3D model.
+    """
+    user_dir = os.path.join(TMP_DIR, str(req.session_hash))
+    if not is_multiimage:
+        outputs = pipeline.run(
+            image,
+            seed=seed,
+            formats=["gaussian", "mesh"],
+            preprocess_image=False,
+            sparse_structure_sampler_params={
+                "steps": ss_sampling_steps,
+                "cfg_strength": ss_guidance_strength,
+            },
+            slat_sampler_params={
+                "steps": slat_sampling_steps,
+                "cfg_strength": slat_guidance_strength,
+            },
+        )
+    else:
+        outputs = pipeline.run_multi_image(
+            [image[0] for image in multiimages],
+            seed=seed,
+            formats=["gaussian", "mesh"],
+            preprocess_image=False,
+            sparse_structure_sampler_params={
+                "steps": ss_sampling_steps,
+                "cfg_strength": ss_guidance_strength,
+            },
+            slat_sampler_params={
+                "steps": slat_sampling_steps,
+                "cfg_strength": slat_guidance_strength,
+            },
+            mode=multiimage_algo,
+        )
+    video = render_utils.render_video(outputs['gaussian'][0], num_frames=120)['color']
+    video_geo = render_utils.render_video(outputs['mesh'][0], num_frames=120)['normal']
+    video = [np.concatenate([video[i], video_geo[i]], axis=1) for i in range(len(video))]
+    video_path = os.path.join(user_dir, 'sample.mp4')
+    imageio.mimsave(video_path, video, fps=15)
+    state = pack_state(outputs['gaussian'][0], outputs['mesh'][0])
+    torch.cuda.empty_cache()
+    return state, video_path
 
-    with gr.Blocks() as demo:
-        # Input components
-        with gr.Row():
-            with gr.Column():
-                with gr.Tabs() as input_tabs:
-                    with gr.Tab("Single Image"):
-                        single_image = gr.Image(label="Single Image Input")
-                    with gr.Tab("Multiple Images"):
-                        multi_images = gr.Gallery(label="Multiple Image Input")
-                
-                # Generation settings
-                with gr.Accordion("Generation Settings"):
-                    seed = gr.Slider(0, MAX_SEED, label="Seed", value=0)
-                    randomize_seed = gr.Checkbox(label="Randomize Seed", value=True)
-                    
-                    with gr.Row():
-                        ss_guidance = gr.Slider(0, 10, label="Sparse Guidance Strength", value=7.5)
-                        ss_steps = gr.Slider(1, 50, label="Sparse Sampling Steps", value=12)
-                    
-                    with gr.Row():
-                        slat_guidance = gr.Slider(0, 10, label="Latent Guidance Strength", value=3.0)
-                        slat_steps = gr.Slider(1, 50, label="Latent Sampling Steps", value=12)
-                    
-                    multi_algo = gr.Radio(["stochastic", "multidiffusion"], label="Multi-image Algorithm", value="stochastic")
+@spaces.GPU(duration=90)
+def extract_glb(
+    state: dict,
+    mesh_simplify: float,
+    texture_size: int,
+    req: gr.Request,
+) -> Tuple[str, str]:
+    """
+    Extract a GLB file from the 3D model.
+    Args:
+        state (dict): The state of the generated 3D model.
+        mesh_simplify (float): The mesh simplification factor.
+        texture_size (int): The texture resolution.
+    Returns:
+        str: The path to the extracted GLB file.
+    """
+    user_dir = os.path.join(TMP_DIR, str(req.session_hash))
+    gs, mesh = unpack_state(state)
+    glb = postprocessing_utils.to_glb(gs, mesh, simplify=mesh_simplify, texture_size=texture_size, verbose=False)
+    glb_path = os.path.join(user_dir, 'sample.glb')
+    glb.export(glb_path)
+    torch.cuda.empty_cache()
+    return glb_path, glb_path
 
-                # Buttons
-                generate_btn = gr.Button("Generate 3D Model")
-                
-                # GLB Extraction
-                with gr.Accordion("GLB Extraction"):
-                    mesh_simplify = gr.Slider(0.9, 0.98, label="Mesh Simplify", value=0.95)
-                    texture_size = gr.Slider(512, 2048, label="Texture Size", value=1024)
-                    extract_glb_btn = gr.Button("Extract GLB")
+@spaces.GPU
+def extract_gaussian(state: dict, req: gr.Request) -> Tuple[str, str]:
+    """
+    Extract a Gaussian file from the 3D model.
+    Args:
+        state (dict): The state of the generated 3D model.
+    Returns:
+        str: The path to the extracted Gaussian file.
+    """
+    user_dir = os.path.join(TMP_DIR, str(req.session_hash))
+    gs, _ = unpack_state(state)
+    gaussian_path = os.path.join(user_dir, 'sample.ply')
+    gs.save_ply(gaussian_path)
+    torch.cuda.empty_cache()
+    return gaussian_path, gaussian_path
 
-        # Output components
+with gr.Blocks(theme=gr.themes.Default(), delete_cache=(600, 600)) as demo:
+    with gr.Row():
         with gr.Column():
-            video_output = gr.Video(label="Generated 3D Asset Preview")
-            glb_output = gr.File(label="Extracted GLB")
+            with gr.Tabs() as input_tabs:
+                with gr.Tab(label="Single Image", id=0) as single_image_input_tab:
+                    image_prompt = gr.Image(label="Image Prompt", format="png", image_mode="RGBA", type="pil", height=300)
+                with gr.Tab(label="Multiple Images", id=1) as multiimage_input_tab:
+                    multiimage_prompt = gr.Gallery(label="Image Prompt", format="png", type="pil", height=300, columns=3)
+            
+            with gr.Accordion(label="Generation Settings", open=False):
+                seed = gr.Slider(0, MAX_SEED, label="Seed", value=0, step=1)
+                randomize_seed = gr.Checkbox(label="Randomize Seed", value=True)
+                with gr.Row():
+                    ss_guidance_strength = gr.Slider(0.0, 10.0, label="Guidance Strength", value=7.5, step=0.1)
+                    ss_sampling_steps = gr.Slider(1, 50, label="Sampling Steps", value=12, step=1)
+                with gr.Row():
+                    slat_guidance_strength = gr.Slider(0.0, 10.0, label="Guidance Strength", value=3.0, step=0.1)
+                    slat_sampling_steps = gr.Slider(1, 50, label="Sampling Steps", value=12, step=1)
+                multiimage_algo = gr.Radio(["stochastic", "multidiffusion"], label="Multi-image Algorithm", value="stochastic")
 
-        # Event handlers
-        def generate_3d(image, multi_image, seed, ss_guidance, ss_steps, 
-                        slat_guidance, slat_steps, multi_algo):
-            # Determine if it's multi-image mode
-            is_multi = isinstance(multi_image, list) and len(multi_image) > 0
+            generate_btn = gr.Button("Generate", variant="primary")
             
-            # Randomize seed if selected
-            if randomize_seed:
-                seed = np.random.randint(0, MAX_SEED)
+            with gr.Accordion(label="GLB Extraction Settings", open=False):
+                mesh_simplify = gr.Slider(0.9, 0.98, label="Simplify", value=0.95, step=0.01)
+                texture_size = gr.Slider(512, 2048, label="Texture Size", value=1024, step=512)
             
-            # Process image
-            state, video = converter.process_image(
-                image, multi_image, is_multi, seed, 
-                ss_guidance, ss_steps, 
-                slat_guidance, slat_steps, 
-                multi_algo
-            )
+            with gr.Row():
+                extract_glb_btn = gr.Button("Extract GLB", interactive=False)
+                extract_gs_btn = gr.Button("Extract Gaussian", interactive=False)
+
+        with gr.Column():
+            video_output = gr.Video(label="Generated 3D Asset", autoplay=True, loop=True, height=300)
+            model_output = LitModel3D(label="Extracted GLB/Gaussian", exposure=10.0, height=300)
             
-            return video if video else None
+            with gr.Row():
+                download_glb = gr.DownloadButton(label="Download GLB", interactive=False)
+                download_gs = gr.DownloadButton(label="Download Gaussian", interactive=False)  
+    
+    is_multiimage = gr.State(False)
+    output_buf = gr.State()
 
-        def extract_glb(state, simplify, texture_size):
-            if state is None:
-                return None
-            glb_path = converter.extract_glb(state, simplify, texture_size)
-            return glb_path
+    # Handlers
+    demo.load(start_session)
+    demo.unload(end_session)
+    
+    single_image_input_tab.select(
+        lambda: False,
+        outputs=[is_multiimage]
+    )
+    multiimage_input_tab.select(
+        lambda: True,
+        outputs=[is_multiimage]
+    )
+    
+    image_prompt.upload(
+        preprocess_image,
+        inputs=[image_prompt],
+        outputs=[image_prompt],
+    )
+    multiimage_prompt.upload(
+        preprocess_images,
+        inputs=[multiimage_prompt],
+        outputs=[multiimage_prompt],
+    )
 
-        # Connect event handlers
-        generate_btn.click(
-            generate_3d, 
-            inputs=[single_image, multi_images, seed, ss_guidance, ss_steps, 
-                    slat_guidance, slat_steps, multi_algo],
-            outputs=[video_output]
-        )
+    generate_btn.click(
+        get_seed,
+        inputs=[randomize_seed, seed],
+        outputs=[seed],
+    ).then(
+        image_to_3d,
+        inputs=[image_prompt, multiimage_prompt, is_multiimage, seed, ss_guidance_strength, ss_sampling_steps, slat_guidance_strength, slat_sampling_steps, multiimage_algo],
+        outputs=[output_buf, video_output],
+    ).then(
+        lambda: tuple([gr.Button(interactive=True), gr.Button(interactive=True)]),
+        outputs=[extract_glb_btn, extract_gs_btn],
+    )
 
-        extract_glb_btn.click(
-            extract_glb,
-            inputs=[ mesh_simplify, texture_size],
-            outputs=[glb_output]
-        )
+    video_output.clear(
+        lambda: tuple([gr.Button(interactive=False), gr.Button(interactive=False)]),
+        outputs=[extract_glb_btn, extract_gs_btn],
+    )
+
+    extract_glb_btn.click(
+        extract_glb,
+        inputs=[output_buf, mesh_simplify, texture_size],
+        outputs=[model_output, download_glb],
+    ).then(
+        lambda: gr.Button(interactive=True),
+        outputs=[download_glb],
+    )
+    
+    extract_gs_btn.click(
+        extract_gaussian,
+        inputs=[output_buf],
+        outputs=[model_output, download_gs],
+    ).then(
+        lambda: gr.Button(interactive=True),
+        outputs=[download_gs],
+    )
+
+    model_output.clear(
+        lambda: gr.Button(interactive=False),
+        outputs=[download_glb],
+    )
 
 # Launch the Gradio app
 if __name__ == "__main__":