🎬 Omni-Video-Factory
Please give us a ❤️ if you find it helpful. No usage limits beyond your ZeroGPU daily quota.
384 is for learning the workflow only, not recommended for daily use.
I2V aspect ratio follows the uploaded image automatically.
🎬 Omni-Video-Factory
Industrial-grade creator workflow on ZeroGPU
What this project is
Omni-Video-Factory is an OmniAILab experimental integration that combines OmniVideo, ComfyUI workflows, and Hugging Face ZeroGPU to deliver industrial-grade video creator tooling in a lightweight public demo.
About OmniVideo
OmniVideo was first released in December 2025 as a multi-modal generation and editing model for video and image. It has strong compatibility with ComfyUI and is designed for high-fidelity, identity-consistent video generation.
🔓 Open Source & Transparency
Comfy runtime code is based on minimal modifications of
selfitcamera/ComfyUIVideo.
Except for OmniVideo model weights, the end-to-end pipeline code is visible.
Our workflows are also highly compatible with LTX-Video, WanX 2.2, and WanX 2.1. You can use AI coding tools to adapt this stack to corresponding open-source models.
Except for OmniVideo model weights, the end-to-end pipeline code is visible.
Our workflows are also highly compatible with LTX-Video, WanX 2.2, and WanX 2.1. You can use AI coding tools to adapt this stack to corresponding open-source models.
📜 Access & Usage Policy
This demo does not enforce a custom generation count cap. Your available runs are determined by your Hugging Face account quota and ZeroGPU availability.
This demo is for research and experimentation only. Generated production content should not be used for commercial purposes.
This demo is for research and experimentation only. Generated production content should not be used for commercial purposes.