Directional Texture Editing for 3D Models

Paper | Project Page

Introduction

This repository contains the official implementation of Directional Texture Editing for 3D Models(ITEM3D). Our ITEM3D model presents an efficient solution to the challenging task of texture editing for 3D models. By leveraging the knowledge from diffusion models, ITEM3D is capable to optimize the texture and environment map under the guidance of text prompts. More results can be viewed on our Project Page.

Installation

Create the environment by conda.

conda create -n item3d python=3.9
conda activate item3d
pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113 -f https://download.pytorch.org/whl/torch_stable.html
pip install ninja imageio PyOpenGL glfw xatlas gdown
pip install git+https://github.com/NVlabs/nvdiffrast/
pip install --global-option="--no-networks" git+https://github.com/NVlabs/tiny-cuda-nn#subdirectory=bindings/torch
imageio_download_bin freeimage
pip install -r requirements.txt

Demo

In this section, we present an example to edting 3D model. First, generate the mlp texture.

sh run/texture.sh

Then, edit 3D model's texuture.

sh run/direction_edit.sh

Acknowledgement

Thanks to NVdiffrec, Stable-DreamFusion and Fantasia3D, our code is partially borrowing from them.

Citation

If you find our work useful, please consider citing:

@misc{liu2024directional,
      title={Directional Texture Editing for 3D Models}, 
      author={Shengqi Liu and Zhuo Chen and Jingnan Gao and Yichao Yan and Wenhan Zhu and Jiangjing Lyu and Xiaokang Yang},
      year={2024},
      eprint={2309.14872},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}