Omnihuman: generate humanoid motion video from text

Understand the project

Practical overview

Consider OmniHuman when a Python-based workflow needs text-conditioned humanoid motion video and can work with the library’s frame-based inputs.

Describe a humanoid motion in text and pair it with an image as input to a video-generation workflow.
Load image or video frames into Python for use with the library.

Before you choose

The README supports image input and shows video input as a commented alternative.
The example’s generation call is commented out, so the README alone does not provide a complete generation run.
The repository specifies the Apache-2.0 license and requires appropriate credit when using, modifying, or incorporating the library and models.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install the Python package and load an image or video into frames using the README’s example.

Install the package with the command below, or clone the repository and install it in editable mode.
Documented command
```
pip install omnihuman
```
Import omnihuman in Python.
Call omnihuman.read_frames with an image path or, as shown in the commented example, a video path.

The example returns image or video data as frames for use in the library workflow.

The README’s model initialization and generate_video call are commented out, so these documented steps stop at preparing frames.
The README does not specify hardware or runtime requirements.

Evidence and transparency

Source record details

Source availability: Active
Sources checked: 17 October 2026
Original website title: mdsrqbl/omnihuman · Official homepage title · retained in its source language