wshw/texture_pp

Texture++

This repository contains the local super-resolution model used by Texture++: Elevating 3D Asset Texture Resolution with a Region-Aware Diffusion Model (SIGGRAPH Asia 2026).

File structure

unet/
├── config.json
└── diffusion_pytorch_model.safetensors
vae/
├── config.json
└── diffusion_pytorch_model.safetensors
text_encoder/
├── config.json
└── model.safetensors
tokenizer/
├── merges.txt
├── special_tokens_map.json
├── tokenizer.json
├── tokenizer_config.json
└── vocab.json

The U-Net is fine-tuned from Stable Diffusion 2.1-base and takes a 5-channel input (4-channel latent + 1-channel mask). The VAE and text encoder are frozen.