OmniHuman is an end-to-end AI framework developed by researchers at ByteDance. It can generate incredibly realistic human videos from just a single image and a motion signal—like audio or video.

Jan 29, 2025 · OmniHuman significantly outperforms existing methods, generating extremely realistic human videos based on weak signal inputs, especially audio. It supports image inputs of any aspect ratio, whether they are portraits, half-body, or full-body images, delivering more lifelike and high-quality results across various scenarios.

1 day ago · ByteDance trained OmniHuman-1 on an extensive 18,700-hour dataset of human video footage, allowing the model to understand a vast array of motions, facial expressions, and gestures. By exposing the AI to an unparalleled variety of real-life movements, it enhances the natural feel of the generated content.

Feb 5, 2025 · ByteDance has introduced OmniHuman-1, an AI model that generates realistic human videos using a single image and motion signals. OmniHuman-1 uses a Diffusion Transformer (DiT) architecture and an omni-conditions training strategy to fuse multiple input types—audio, video, and pose—to create full-body human animations.

2 days ago · ByteDance’s access to TikTok data could give OmniHuman-1 a competitive edge in realism. OmniHuman-1 competes with established players like OpenAI’s Sora, Runway, and Luma AI in the field of AI video generation. Sora and OmniHuman-1 both create videos using AI, but they’re good at different things. Sora is good at creating realistic scenes.

3 days ago · China is racing fast in the AI game – after DeepSeek and Qwen models, ByteDance has just released an impressive research paper! The OmniHuman-1 paper introduces OmniHuman, a new framework that uses Diffusion Transformer-based architecture to push the boundaries of human animation. This model can create ultra-realistic human videos in any ...

5 days ago · ByteDance’s OmniHuman-1 is a groundbreaking AI model that can transform a single image into a realistic video of a person speaking or performing, synchronized perfectly with a given audio track.

6 days ago · OmniHuman sample videos demonstrate the scope of both human and animal movement. OmniHuman screenshot by Leslie Katz. TikTok owner ByteDance has unveiled OmniHuman, an AI model that can generate ...