README.md

metadata

license: mit

TripoSplat: Generative 3D Gaussians with Learned Density Control

TripoSplat converts a single 2D image into high-quality and variable number of 3D Gaussians developed by TripoAI. It can serve as a powerful pipeline tool for asset creation, AR/VR, game development, simulation environments, and beyond.

Read Paper | Technical Blog

Usage

To use with the official inference code, place the files in the following structure:

📂 TripoSplat/
├── 📂 ckpts/
│   ├── 📂 background_removal/
│   │   └── birefnet.safetensors
│   ├── 📂 diffusion_models/
│   │   └── triposplat_fp16.safetensors
│   ├── 📂 clip_vision/
│   │   └── dino_v3_vit_h.safetensors
│   ├── 📂 vae/
│   │   ├── flux2-vae.safetensors
│   │   └── triposplat_vae_decoder_fp16.safetensors

To use with ComfyUI, place the files in the following structure:

📂 ComfyUI/
├── 📂 models/
│   ├── 📂 background_removal/
│   │   └── birefnet.safetensors
│   ├── 📂 diffusion_models/
│   │   └── triposplat_fp16.safetensors
│   ├── 📂 clip_vision/
│   │   └── dino_v3_vit_h.safetensors
│   ├── 📂 vae/
│   │   ├── flux2-vae.safetensors
│   │   └── triposplat_vae_decoder_fp16.safetensors