Awesome

中文版

<div align="center">CraftsMan3D: High-fidelity Mesh Generation <br> with 3D Native Generation and Interactive Geometry Refiner<div>

<p align="center"> Weiyu Li^*1,2, Jiarui Liu^*1,2, Hongyu Yan^*1, Rui Chen^1,2, Yixun Liang^2,3, Xuelin Chen⁴, Ping Tan^1,2, Xiaoxiao Long^1,2</p>

<p align="center"> ¹HKUST, ²LightIllusions, ³HKUST(GZ), ⁴Tencent AI Lab</p>

Usage

from craftsman import CraftsManPipeline
import torch

# load from local ckpt
# pipeline = CraftsManPipeline.from_pretrained("./ckpts/craftsman", device="cuda:0", torch_dtype=torch.float32) 

# load from huggingface model hub
pipeline = CraftsManPipeline.from_pretrained("craftsman3d/craftsman", device="cuda:0", torch_dtype=torch.float32)

# inference
mesh = pipeline("https://pub-f9073a756ec645d692ce3d171c2e1232.r2.dev/data/werewolf.png").meshes[0]
mesh.export("werewolf.obj")

The results should be like this:

TL; DR: <font color="red">CraftsMan3D (aka 匠心)</font> is a two-stage text/image to 3D mesh generation model. By mimicking the modeling workflow of artist/craftsman, we propose to generate a coarse mesh (5s) with smooth geometry using 3D diffusion model and then refine it (20s) using enhanced multi-view normal maps generated by 2D normal diffusion, which is also can be in a interactive manner like Zbrush.

Important: the released ckpt are mainly trained on character, so it would perform better in this category and we plan to release more advanced pretrained models in the future.

✨ Overview

This repo contains source code (training / inference) of 3D diffusion model, pretrained weights and gradio demo code of our 3D mesh generation project, you can find more visualizations on our project page and try our demo. If you have high-quality 3D data or some other ideas, we very much welcome any form of cooperation.

Contents

• Pretrained Models
• Gradio & Huggingface Demo
• Inference
• Training
• Data Prepration
• Video
• Acknowledgement
• Citation

Environment Setup

The mesh refinement part is performed on a GTX 3080 GPU.

:smiley: We also provide a Dockerfile for easy installation, see Setup using Docker.

• Python 3.10.0
• PyTorch 2.1.0
• Cuda Toolkit 11.8.0
• Ubuntu 22.04

Clone this repository.

git clone https://github.com/wyysf-98/CraftsMan.git

Install the required packages.

conda create -n CraftsMan python=3.10 -y
conda activate CraftsMan
conda install cudatoolkit=11.8 -c pytorch -y
pip install torch==2.3.0 torchvision==0.18.0 
pip install -r docker/requirements.txt

🎥 Video

3D Native DiT Model (Latent Set DiT Model)

We provide the training and the inference code here for future research. The latent set VAE model is heavily build on the same structure of Michelangelo. The latent set diffusion model is based on a DiT/Pixart-alpha and with 500M parameters.

Pretrained models

Currently, We provide the models with single view image as condition with DiT. We will consider open source the further models according to the real situation. If you run the inference.py without specifying the model path, it will automatically download the model from the huggingface model hub.

Or you can download the model manually:

## you can just manually get the model using wget:
wget https://huggingface.co/craftsman3d/craftsman/resolve/main/config.yaml
wget https://huggingface.co/craftsman3d/craftsman/resolve/main/model.ckpt

## or you can git clone the repo:
git lfs install
git clone https://huggingface.co/craftsman3d/craftsman

If you download the models using wget, you should manually put them under the ckpts/craftsman directory.

Gradio demo

We provide gradio demos for easy usage.

python gradio_app.py --model_path ./ckpts/craftsman

Inference

To generate 3D meshes from images folders via command line, simply run:

python inference.py --input eval_data --device 0 --model ./ckpts/craftsman

For more configs, please refer to the inference.py.

Train from scratch

We provide our training code to facilitate future research. And we provide a data sample in data. If you want to get the full data, please fill in the form

Due to the costs associated with data deployment. If you could help share our work on social media in any form, we would be grateful. As a token of our appreciation, you will receive download links to exclusive content stored on AWS S3, which can achieve download speeds of 20-100 MB/s.

For more training details and configs, please refer to the configs folder.

### training the shape-autoencoder
python train.py --config ./configs/shape-autoencoder/michelangelo-l768-e64-ne8-nd16.yaml \
                 --train --gpu 0

### training the image-to-shape diffusion model
python train.py --config .configs/image-to-shape-diffusion/clip-dino-rgb-pixart-lr2e4-ddim.yaml \
                 --train --gpu 0

2D Normal Enhancement Diffusion Model (TBA)

We are diligently working on the release of our mesh refinement code. Your patience is appreciated as we put the final touches on this exciting development." 🔧🚀

You can also find the video of mesh refinement part in the video.

❓Common questions

Q: Tips to get better results. 0. Due to limited resources, we will gradually expand the dataset and training scale, and therefore we will release more pre-trained models in the future.

1. Just like the 2D diffusion model, try different seeds, adjust the CFG scale or different scheduler. Good Luck.
2. We will provide a version that conditioned on the text prompt, so you can use some positive and negative prompts.

💪 ToDo List

• Inference code
• Training code
• Gradio & Hugging Face demo
• Model zoo, we will release more ckpt in the future
• Environment setup
• Data sample
• Code for mesh refine

🤗 Acknowledgements

• Thanks to LightIllusion for providing computational resources and Jianxiong Pan for data preprocessing. If you have any idea about high-quality 3D Generation, welcome to contact us!
• Thanks to Hugging Face for sponsoring the nicely demo!
• Thanks to 3DShape2VecSet for their amazing work, the latent set representation provides an efficient way to represent 3D shape!
• Thanks to Michelangelo for their great work, our model structure is heavily build on this repo!
• Thanks to CRM, Wonder3D and LGM for their released model about multi-view images generation. If you have a more advanced version and want to contribute to the community, we are welcome to update.
• Thanks to Objaverse, Objaverse-MIX for their open-sourced data, which help us to do many validation experiments.
• Thanks to ThreeStudio for their great repo, we follow their fantastic and easy-to-use code structure!

📑License

CraftsMan shares the same license with the SD15, which is under creativeml-openrail-m. If you have any questions about the usage of CraftsMan, please contact us first.

📖 BibTeX

@misc{li2024craftsman,
title         = {CraftsMan: High-fidelity Mesh Generation with 3D Native Generation and Interactive Geometry Refiner}, 
author        = {Weiyu Li and Jiarui Liu and Hongyu Yan and Rui Chen and Yixun Liang and Xuelin Chen and Ping Tan and Xiaoxiao Long},
year          = {2024},
archivePrefix = {arXiv preprint arXiv:2405.14979},
primaryClass  = {cs.CG}
}