1. Tutorial Introduction

HuMo is a multimodal video generation framework released by Tsinghua University and ByteDance Intelligent Creation Lab in September 2025, focusing on human-centered video generation. It can generate high-quality, detailed and controllable human videos from multiple modal inputs such as text, images and audio. HuMo supports powerful text prompt following capabilities, consistent subject retention and audio-driven motion synchronization. It supports video generation from text-image (VideoGen from Text-Image), text-audio (VideoGen from Text-Audio) and text-image-audio (VideoGen from Text-Image-Audio). The related paper results are "HuMo: Human-Centric Video Generation via Collaborative Multi-Modal Conditioning".

The HuMo project provides model deployment in two specifications: 1.7B and 17B. This tutorial uses a 17B model and a single RTX pro 6000 card as the resource.

→ Click to jump to experienceHuMo 1.7B: A framework for multimodal video generation".

2. Project Examples

VideoGen from Text-Image-Audio,TIA

VideoGen from Text-Audio,TA

3. Operation steps

1. After starting the container, click the API address to enter the Web interface

2. Usage steps

If "Bad Gateway" is displayed, it means the model is initializing. Since the model is large, please wait about 2-3 minutes and refresh the page.  Note: When Sampling Steps is set to 10, it takes approximately 3-5 minutes to generate results.

TIA

TA

Parameter Description

• Height: Set the height of the video.
• Width: Set the width of the video.
• Frames: Set the number of video frames.
• Text Guidance Scale: Text guidance scaling, used to control the impact of text prompts on video generation.
• Image Guidance Scale: Image guidance scaling, used to control the influence of image cues on video generation.
• Audio Guidance Scale: Audio guidance scaling, used to control the influence of audio cues on video generation.
• Sampling Steps: The number of sampling steps used to control the quality and details of the generated video.
• Random Seed: Random seed, used to control the randomness of video generation.

Citation Information

The citation information for this project is as follows:

@misc{chen2025humo,
      title={HuMo: Human-Centric Video Generation via Collaborative Multi-Modal Conditioning}, 
      author={Liyang Chen and Tianxiang Ma and Jiawei Liu and Bingchuan Li and Zhuowei Chen and Lijie Liu and Xu He and Gen Li and Qian He and Zhiyong Wu},
      year={2025},
      eprint={2509.08519},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2509.08519}, 
}