TikTok owner ByteDance working on AI tool that generates videos from single image

TikTok’s parent company ByteDance has unveiled its new AI tool called OmniHuman-1 that can generate lifelike videos from a single photo. The model is able to imitate how humans speak, move and gesture accurately. 

A research paper on the tool published shared that it had been trained on more than 18,700 hours of human video. “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,” it stated. 

It also shared sample video clips - one of Einstein delivering a lecture and another of a young girl with a cat. 

TikTok already has an AI video generator called Jimeng integrated on to its platform. 

While OmniHuman is still in its research stage, if released it will compete with OpenAI’s Sora and other AI video generator startups like Runway, Pika and Luma Labs.