LAM: Official Pytorch Implementation

LAM: Large Avatar Model for One-shot Animatable Gaussian Head

Yisheng He*, Xiaodong Gu*, Xiaodan Ye, Chao Xu, Zhengyi Zhao, Yuan Dong†, Weihao Yuan†, Zilong Dong, Liefeng Bo

Tongyi Lab, Alibaba Group

"Build 3D Interactive Chatting Avatar with One Image in Seconds!"

如果您熟悉中文，可以阅读我们中文版本的文档

Core Highlights 🔥🔥🔥

• Ultra-realistic 3D Avatar Creation from One Image in Seconds
• Super-fast Cross-platform Animating and Rendering on Any Devices
• Low-latency SDK for Realtime Interactive Chatting Avatar

Chat_Demo.mp4

📢 News

[April 21, 2025] We have released the WebGL Interactive Chatting Avatar SDK on OpenAvatarChat (including LLM, ASR, TTS, Avatar), with which you can freely chat with the 3D Digital Human generated by LAM ! 🔥

[April 19, 2025] We have released the Audio2Expression model, which can animate the generated LAM Avatar with audio input ! 🔥

[April 10, 2025] We have released the demo on ModelScope Space !

To do list

• Release LAM-small trained on VFHQ and Nersemble.
• Release Huggingface space.
• Release Modelscope space.
• Release LAM-large trained on a self-constructed large dataset.
• Release WebGL Render for cross-platform animation and rendering.
• Release audio driven model: Audio2Expression.
• Release Interactive Chatting Avatar SDK with OpenAvatarChat, including LLM, ASR, TTS, Avatar.

🚀 Get Started

Environment Setup

git clone https://github.com/aigc3d/LAM.git
cd LAM
# Install with Cuda 12.1
sh  ./scripts/install/install_cu121.sh
# Or Install with Cuda 11.8
sh ./scripts/install/install_cu118.sh

Model Weights

Model	Training Data	HuggingFace	ModelScope	Reconstruction Time	A100 (A & R)	XiaoMi 14 Phone (A & R)
LAM-20K	VFHQ	TBD	TBD	1.4 s	562.9FPS	110+FPS
LAM-20K	VFHQ + NeRSemble	Link	Link	1.4 s	562.9FPS	110+FPS
LAM-20K	Our large dataset	TBD	TBD	1.4 s	562.9FPS	110+FPS

(A & R: Animating & Rendering )

HuggingFace Download

# Download Assets
huggingface-cli download 3DAIGC/LAM-assets --local-dir ./tmp
tar -xf ./tmp/LAM_assets.tar && rm ./tmp/LAM_assets.tar
tar -xf ./tmp/thirdparty_models.tar && rm -r ./tmp/
# Download Model Weights
huggingface-cli download 3DAIGC/LAM-20K --local-dir ./model_zoo/lam_models/releases/lam/lam-20k/step_045500/

ModelScope Download

pip3 install modelscope
# Download Assets
modelscope download --model "Damo_XR_Lab/LAM-assets" --local_dir "./tmp/"
tar -xf ./tmp/LAM_assets.tar && rm ./tmp/LAM_assets.tar
tar -xf ./tmp/thirdparty_models.tar && rm -r ./tmp/
# Download Model Weights
modelscope download "Damo_XR_Lab/LAM-20K" --local_dir "./model_zoo/lam_models/releases/lam/lam-20k/step_045500/"

Gradio Run

python app_lam.py

Inference

sh ./scripts/inference.sh ${CONFIG} ${MODEL_NAME} ${IMAGE_PATH_OR_FOLDER} ${MOTION_SEQ}

Acknowledgement

This work is built on many amazing research works and open-source projects:

• OpenLRM
• GAGAvatar
• GaussianAvatars
• VHAP

Thanks for their excellent works and great contribution.

More Works

Welcome to follow our other interesting works:

• LHM

Citation

@inproceedings{he2025LAM,
  title={LAM: Large Avatar Model for One-shot Animatable Gaussian Head},
  author={
    Yisheng He and Xiaodong Gu and Xiaodan Ye and Chao Xu and Zhengyi Zhao and Yuan Dong and Weihao Yuan and Zilong Dong and Liefeng Bo
  },
  booktitle={arXiv preprint arXiv:2502.17796},
  year={2025}
}