---
title: TripoSR + Dust3R
emoji: 🎨
colorFrom: red
colorTo: green
sdk: gradio
sdk_version: 6.3.0
python_version: "3.10"
app_file: app.py
pinned: false
license: mit
tags:
  - image-to-3d
  - 3d-reconstruction
  - triposr
  - dust3r
  - mcp-server
short_description: Image to 3D + Multi-view Point Cloud
---

# TripoSR + Dust3R (CPU)

Two 3D reconstruction tools in one space:

**Tab 1: TripoSR** - Single image to 3D mesh
- Upload one image, get a textured 3D mesh
- Background removal included
- Based on [stabilityai/TripoSR](https://huggingface.co/stabilityai/TripoSR)

**Tab 2: Dust3R** - Multi-view to Point Cloud
- Upload 2-5 images from different angles
- Get a colored 3D point cloud
- Based on [naver/dust3r](https://github.com/naver/dust3r)

## Usage

### TripoSR (Single Image)
1. Upload an image of an object
2. Enable "Remove Background" for photos
3. Click "Generate 3D"
4. Download OBJ or GLB mesh

### Dust3R (Multi-view)
1. Upload 2-5 images from different viewpoints
2. Click "Reconstruct"
3. Download point cloud (GLB)

---

## API

### Python Client
```python
from gradio_client import Client, handle_file

client = Client("Luminia/Dust3R-TripoSR")

# TripoSR: Single image to 3D
result = client.predict(
    input_image=handle_file("photo.png"),
    do_remove_background=True,
    foreground_ratio=0.85,
    mc_resolution=256,
    api_name="/triposr"
)
processed_img, obj_path, glb_path, logs = result

# Dust3R: Multi-view to point cloud
result = client.predict(
    images=[handle_file("view1.png"), handle_file("view2.png"), handle_file("view3.png")],
    image_size=512,
    min_conf_thr=3.0,
    api_name="/dust3r"
)
pointcloud_path, logs = result
```

### MCP (Model Context Protocol)
```json
{
  "mcpServers": {
    "triposr-dust3r": {
      "url": "https://luminia-dust3r-triposr.hf.space/gradio_api/mcp/sse"
    }
  }
}
```

---

## Notes

- Running on CPU - first generation is slow (model loading)
- TripoSR: ~2-5 min per image
- Dust3R: ~5-15 min for 2-5 images

## Credits

- [TripoSR](https://huggingface.co/stabilityai/TripoSR) by Stability AI & Tripo AI
- [Dust3R](https://github.com/naver/dust3r) by NAVER