Upload 2 files

app.py

@@ -0,0 +1,212 @@
+import torch
+import gradio as gr
+from diffusers import ShapEPipeline, ShapEImg2ImgPipeline
+from diffusers.utils import export_to_gif
+import os
+from huggingface_hub import HfApi, login
+from PIL import Image
+import numpy as np
+import gc
+
+# Force CPU usage
+device = "cpu"
+torch.set_num_threads(4)
+print(f"Using device: {device}")
+
+def validate_token(token):
+    try:
+        login(token=token)
+        return True
+    except Exception as e:
+        print(f"Token validation error: {str(e)}")
+        return False
+
+def generate_3d_from_text(prompt, token, guidance_scale=7.0, export_format="obj", progress=gr.Progress()):
+    try:
+        if not validate_token(token):
+            return "Invalid Hugging Face token", None, None
+            
+        print(f"Starting generation: {prompt}")
+        progress(0.1, "Loading model...")
+        
+        pipe = ShapEPipeline.from_pretrained(
+            "openai/shap-e",
+            torch_dtype=torch.float32,
+            token=token,
+            revision="main",
+            low_cpu_mem_usage=True
+        )
+        
+        os.makedirs("outputs", exist_ok=True)
+        safe_prompt = "".join(x for x in prompt if x.isalnum() or x in (" ", "-", "_"))
+        base_filename = f"outputs/{safe_prompt}"
+        
+        try:
+            progress(0.3, "Creating 3D model...")
+            with torch.no_grad():
+                output = pipe(
+                    prompt,
+                    guidance_scale=min(guidance_scale, 10.0),
+                    num_inference_steps=16
+                )
+                
+                progress(0.5, "Creating GIF...")
+                gif_path = export_to_gif(output.images, f"{base_filename}.gif")
+                
+                progress(0.7, "Creating 3D mesh...")
+                mesh_output = pipe(
+                    prompt,
+                    guidance_scale=min(guidance_scale, 10.0),
+                    num_inference_steps=16,
+                    output_type="mesh"
+                )
+            
+            progress(0.9, "Saving files...")
+            output_path = f"{base_filename}.{export_format}"
+            mesh_output.meshes[0].export(output_path)
+            
+            del pipe
+            del output
+            del mesh_output
+            gc.collect()
+            
+            print(f"Generation completed: {output_path}")
+            progress(1.0, "Completed!")
+            return "Generation successful!", gif_path, output_path
+            
+        except Exception as model_error:
+            error_msg = f"Model execution error: {str(model_error)}"
+            print(error_msg)
+            return error_msg, None, None
+        
+    except Exception as e:
+        error_msg = f"General error: {str(e)}"
+        print(error_msg)
+        return error_msg, None, None
+
+def generate_3d_from_image(image, token, guidance_scale=7.0, export_format="obj", progress=gr.Progress()):
+    try:
+        if not validate_token(token):
+            return "Invalid Hugging Face token", None, None
+            
+        print("Starting image to 3D generation")
+        progress(0.1, "Loading model...")
+        
+        pipe = ShapEImg2ImgPipeline.from_pretrained(
+            "openai/shap-e-img2img",
+            torch_dtype=torch.float32,
+            token=token,
+            revision="main",
+            low_cpu_mem_usage=True
+        )
+        
+        os.makedirs("outputs", exist_ok=True)
+        
+        import time
+        timestamp = int(time.time())
+        base_filename = f"outputs/image_to_3d_{timestamp}"
+        
+        try:
+            progress(0.3, "Preparing image...")
+            if isinstance(image, str):
+                image = Image.open(image)
+            elif isinstance(image, np.ndarray):
+                image = Image.fromarray(image)
+                
+            image = image.resize((128, 128))
+            
+            progress(0.5, "Creating 3D model...")
+            with torch.no_grad():
+                output = pipe(
+                    image=image,
+                    guidance_scale=min(guidance_scale, 10.0),
+                    num_inference_steps=16
+                )
+                
+                progress(0.7, "Creating GIF...")
+                gif_path = export_to_gif(output.images, f"{base_filename}.gif")
+                
+                progress(0.8, "Creating 3D mesh...")
+                mesh_output = pipe(
+                    image=image,
+                    guidance_scale=min(guidance_scale, 10.0),
+                    num_inference_steps=16,
+                    output_type="mesh"
+                )
+            
+            progress(0.9, "Saving files...")
+            output_path = f"{base_filename}.{export_format}"
+            mesh_output.meshes[0].export(output_path)
+            
+            del pipe
+            del output
+            del mesh_output
+            gc.collect()
+            
+            print(f"Generation completed: {output_path}")
+            progress(1.0, "Completed!")
+            return "Generation successful!", gif_path, output_path
+            
+        except Exception as model_error:
+            error_msg = f"Model execution error: {str(model_error)}"
+            print(error_msg)
+            return error_msg, None, None
+        
+    except Exception as e:
+        error_msg = f"General error: {str(e)}"
+        print(error_msg)
+        return error_msg, None, None
+
+with gr.Blocks() as interface:
+    gr.Markdown("# SORA-3D - Text/Image to 3D Model Generator")
+    gr.Markdown("Create 3D models from text or image input. You need a Hugging Face token to use this app.")
+    gr.Markdown("""
+    > **Important Notes**: 
+    > - Processing time may be longer on CPU
+    > - Keep guidance scale under 10 for faster results
+    > - Number of steps is fixed at 16
+    > - Image size is optimized for quality/speed
+    """)
+    
+    with gr.Tab("Text → 3D"):
+        with gr.Row():
+            with gr.Column():
+                text_input = gr.Textbox(label="Enter description for 3D model")
+                text_token = gr.Textbox(label="Hugging Face Token", type="password")
+                text_guidance = gr.Slider(minimum=1, maximum=10, value=7, label="Guidance Scale")
+                text_format = gr.Radio(["obj", "glb"], label="Export Format", value="obj")
+                text_button = gr.Button("Generate")
+            
+            with gr.Column():
+                text_status = gr.Textbox(label="Status")
+                text_preview = gr.Image(label="3D Preview (GIF)")
+                text_file = gr.File(label="3D Model File")
+    
+    with gr.Tab("Image → 3D"):
+        with gr.Row():
+            with gr.Column():
+                image_input = gr.Image(label="Image to convert to 3D", type="pil")
+                image_token = gr.Textbox(label="Hugging Face Token", type="password")
+                image_guidance = gr.Slider(minimum=1, maximum=10, value=7, label="Guidance Scale")
+                image_format = gr.Radio(["obj", "glb"], label="Export Format", value="obj")
+                image_button = gr.Button("Generate")
+            
+            with gr.Column():
+                image_status = gr.Textbox(label="Status")
+                image_preview = gr.Image(label="3D Preview (GIF)")
+                image_file = gr.File(label="3D Model File")
+    
+    text_button.click(
+        generate_3d_from_text,
+        inputs=[text_input, text_token, text_guidance, text_format],
+        outputs=[text_status, text_preview, text_file]
+    )
+    
+    image_button.click(
+        generate_3d_from_image,
+        inputs=[image_input, image_token, image_guidance, image_format],
+        outputs=[image_status, image_preview, image_file]
+    )
+
+if __name__ == "__main__":
+    interface.launch()

requirements.txt

@@ -0,0 +1,42 @@
+--extra-index-url https://download.pytorch.org/whl/cu121
+--find-links https://nvidia-kaolin.s3.us-east-2.amazonaws.com/torch-2.4.0_cu121.html
+
+torch==2.4.0
+torchvision==0.19.0
+pillow==10.4.0
+imageio==2.36.1
+imageio-ffmpeg==0.5.1
+tqdm==4.67.1
+easydict==1.13
+opencv-python-headless==4.10.0.84
+scipy==1.14.1
+rembg==2.0.60
+onnxruntime==1.20.1
+trimesh==4.5.3
+xatlas==0.0.9
+pyvista==0.44.2
+pymeshfix==0.17.0
+igraph==0.11.8
+git+https://github.com/EasternJournalist/utils3d.git@9a4eb15e4021b67b12c460c7057d642626897ec8
+xformers==0.0.27.post2
+kaolin==0.17.0
+spconv-cu120==2.3.6
+
+gradio_litmodel3d==0.0.1
+https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu12torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl
+https://huggingface.co/spaces/JeffreyXiang/TRELLIS/resolve/main/wheels/diff_gaussian_rasterization-0.0.0-cp310-cp310-linux_x86_64.whl?download=true
+https://huggingface.co/spaces/JeffreyXiang/TRELLIS/resolve/main/wheels/nvdiffrast-0.3.3-cp310-cp310-linux_x86_64.whl?download=true
+transformers>=4.30.0
+accelerate>=0.20.0
+diffusers>=0.24.0
+invisible_watermark
+xformers
+sentencepiece
+peft
+safetensors>=0.4.0
+gradio==4.44.1
+huggingface-hub>=0.19.0
+sacremoses
+numpy>=1.24.0
+uvicorn>=0.14.0
+spaces