ComfyUI-4DHumans

★ 8

4D-Humans人物模型动作捕捉依赖安装

将 4D-Humans 功能集成到 ComfyUI 的节点集合，便于生成和控制带动作的人体模型，简化依赖安装与使用。

💡 在 ComfyUI 流程中生成并控制带动作的人体模型数据。

🍴 3 Forks💻 Python🔄 2024-08-30

🔗 GitHub 原文

📦

网盘下载

复制链接后前往夸克网盘下载

https://pan.quark.cn/s/8f9eee5e2cdb

📦 requirements.txt

#
git+https://github.com/shubham-goel/4D-Humans
git+https://github.com/facebookresearch/detectron2
pyrender
yacs

📄 README

ComfyUI-4DHumans

ComfyUI nodes to use 4D-Humans

Installation

Install all pip packages via python -m pip install -r requirements.txt.

This has only been tested on Ubuntu and may require some additional packages/libraries to be installed for Windows/Mac.

Ubuntu libraries

apt-get install freeglut3-dev
apt-get install libosmesa6-dev

Models

You need 3 new model directories: smpl, hmr, and detectron.

The directories should look like:

smpl
- basicModel_neutral_lbs_10_207_0_v1.0.0.pkl
- SMPL_to_J19.pkl
- smpl_mean_params.npz

hmr
- hmr35.ckpt (this was renamed from the 35000 checkpoint from HMR)

detectron
- model_final_f05665.pkl

| Model | Link |

|————|———–|

| Detectron | link |

| SMPL (all files in smpl dir) | link |

| HMR | link |

Note:

The SMPL model files are behind a registration wall. Once you download the tar.gz unzip the file and find the 3 files. Then place them in ComfyUI/models/smpl.

Examples

You can find an example workflow in the example_workflows directory.

The node outputs the smpl model, masks, semantic maps, and preview images.

https://github.com/user-attachments/assets/c505ac94-7880-4497-867c-2e66389f9f48