# LAM **Repository Path**: webos/LAM ## Basic Information - **Project Name**: LAM - **Description**: No description available - **Primary Language**: Unknown - **License**: Apache-2.0 - **Default Branch**: master - **Homepage**: None - **GVP Project**: No ## Statistics - **Stars**: 0 - **Forks**: 0 - **Created**: 2025-05-06 - **Last Updated**: 2025-05-09 ## Categories & Tags **Categories**: Uncategorized **Tags**: None ## README # LAM: 官方Pytorch实现

English | 中文

[![Website](https://raw.githubusercontent.com/prs-eth/Marigold/main/doc/badges/badge-website.svg)](https://aigc3d.github.io/projects/LAM/) [![arXiv Paper](https://img.shields.io/badge/📜-arXiv:2503-10625)](https://arxiv.org/pdf/2502.17796) [![HuggingFace](https://img.shields.io/badge/🤗-HuggingFace_Space-blue)](https://huggingface.co/spaces/3DAIGC/LAM) [![ModelScope](https://img.shields.io/badge/🧱-ModelScope_Space-blue)](https://www.modelscope.cn/studios/Damo_XR_Lab/LAM_Large_Avatar_Model) [![Apache License](https://img.shields.io/badge/📃-Apache--2.0-929292)](https://www.apache.org/licenses/LICENSE-2.0)

###

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

####

SIGGRAPH 2025

#####

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

#####

阿里巴巴通义实验室

####

**"单图秒级打造超写实3D数字人"**

## 核心亮点 🔥🔥🔥 - **单图秒级生成超写实3D数字人化身！** - **WebGL跨平台超实时驱动渲染！手机跑满120FPS！** - **低延迟实时交互对话数字人SDK！**
## 📢 最新动态 **[2025年4月30日]** 我们开源了 [Avatar 导出功能](tools/AVATAR_EXPORT_GUIDE.md)，允许用户在 OpenAvatarChat 平台上接入任何由 LAM 生成的 3D 数字人进行实时对话！🔥
**[2025年4月21日]** 我们开源了 WebGL交互数字人SDK：[OpenAvatarChat](https://github.com/HumanAIGC-Engineering/OpenAvatarChat) (including LLM, ASR, TTS, Avatar), 使用这个SDK可以自由地与我们的LAM-3D数字人进行实时对话 ! 🔥
**[2025年4月19日]** 我们开源了 [Audio2Expression](https://github.com/aigc3d/LAM_Audio2Expression) 模型, 用这个模型可以语音驱动我们的LAM数字人 ! 🔥
**[2025年4月10日]** 我们发布了在 [ModelScope](https://www.modelscope.cn/studios/Damo_XR_Lab/LAM_Large_Avatar_Model) 上的演示程序 !
### 待办清单 - [x] 开源在VFHQ和Nersemble数据集上训练的LAM-small模型. - [x] 部署Huggingface演示程序. - [x] 部署Modelscope演示程序. - [ ] 开源在自有大数据集上训练的LAM-large模型. - [ ] 开源跨平台WebGL驱动渲染引擎. - [x] 开源语音驱动模型: Audio2Expression. - [x] 开源交互对话数字人SDK，包括LLM, ASR, TTS, Avatar. ## 🚀 快速开始 ### 在线试玩 单图生成3D数字人: [![HuggingFace](https://img.shields.io/badge/🤗-HuggingFace_Space-blue)](https://huggingface.co/spaces/3DAIGC/LAM) [![ModelScope](https://img.shields.io/badge/🧱-ModelScope_Space-blue)](https://www.modelscope.cn/studios/Damo_XR_Lab/LAM_Large_Avatar_Model) 交互聊天: [![HuggingFace](https://img.shields.io/badge/🤗-HuggingFace_Space-blue)](https://huggingface.co/spaces/HumanAIGC-Engineering-Team/open-avatar-chat) [![ModelScope](https://img.shields.io/badge/🧱-ModelScope_Space-blue)](https://www.modelscope.cn/studios/HumanAIGC-Engineering/open-avatar-chat) ### 环境设置 我们提供了在Windows系统（Cuda 12.8）上的一键安装包，感谢"十字鱼"的支持.     [视频](https://www.bilibili.com/video/BV13QGizqEey)     [下载链接](https://virutalbuy-public.oss-cn-hangzhou.aliyuncs.com/share/aigc3d/data/LAM/Installation/LAM-windows-one-click-install.zip) #### Linux: ```bash 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 ``` #### Windows: 在Windows系统上的环境安装请参考 [Windows Install Guide](scripts/install/WINDOWS_INSTALL.md). ### 模型权重 | 模型 | 训练数据集 | HuggingFace | ModelScope | 重建时间 | 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](https://huggingface.co/3DAIGC/LAM-20K) | [Link](https://www.modelscope.cn/models/Damo_XR_Lab/LAM-20K/summary) | 1.4 s | 562.9FPS | 110+FPS | | LAM-20K | Our large dataset | TBD | TBD | 1.4 s | 562.9FPS | 110+FPS | (**A & R:** 驱动渲染 ) #### 从HuggingFace下载 ```bash # 从HuggingFace下载 # 下载相关资产 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/ # 下载模型权重 huggingface-cli download 3DAIGC/LAM-20K --local-dir ./model_zoo/lam_models/releases/lam/lam-20k/step_045500/ ``` #### 从ModelScope下载 ```bash # 从ModelScope下载 (如果你无法从HuggingFace下载) pip3 install modelscope # 下载相关资产 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/ # 下载模型权重 modelscope download "Damo_XR_Lab/LAM-20K" --local_dir "./model_zoo/lam_models/releases/lam/lam-20k/step_045500/" ``` ### 运行Gradio ``` python app_lam.py ``` 若需导出ZIP文件以在 OpenAvatarChat 实现实时对话，请参考[指引文档](tools/AVATAR_EXPORT_GUIDE.md)。 ```bash python app_lam.py --blender_path /path/blender ``` ### 推理 ```bash sh ./scripts/inference.sh ${CONFIG} ${MODEL_NAME} ${IMAGE_PATH_OR_FOLDER} ${MOTION_SEQ} ``` ### 致谢 本工作是建立在很多了不起的工作基础之上： - [OpenLRM](https://github.com/3DTopia/OpenLRM) - [GAGAvatar](https://github.com/xg-chu/GAGAvatar) - [GaussianAvatars](https://github.com/ShenhanQian/GaussianAvatars) - [VHAP](https://github.com/ShenhanQian/VHAP) 感谢他们对社区的杰出贡献。 ### 更多工作 欢迎关注我们更多有趣的工作 - [LHM](https://github.com/aigc3d/LHM) ### 引用 ``` @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={SIGGRAPH}, year={2025} } ```