Omnihuman: Bytedance AI can clone a person from a single photo
Bytedance Omnihuman-1 AI © Omnihuman
Tiktok's parent company, Bytedance, in turn embarks on the race for artificial intelligence. With its new model, soberly called Omnihuman-1, the firm offers to make more than realistic videos of human avatars. This artificial intelligence aims to be multimodal, it allows you to use a photo as well as an audio sample of a person's voice in order to create a clone.
Clone an individual with AI: various uses
Omnihuman is capable of working from portraits, as from a photo taken on one level. The tower lies here in the fact that the model is able to create a video from a single image of an individual.
One image. One prompt. Infinite possibilities. Omnihuman-1 Transforms Single Photo Into a Lifelike Talking Persona. No Green Screens, No CGI – Just Ai at Work. Watch this! 👇 pic.twitter.com/gs90uwhidl
-Omnihuman-1 ai (@omnihuman1a) FEBRUARY 5, 2025
Bytedance already sees very varied uses of its technology:
- Avatar for face-camura videos : The firm could see here an opportunity to create virtual influencers, whose image gestures would be synchronized with the intonation of discourse;
- Musical videos : The model captures and includes the shades of music. Omnihuman is able to generate good movements and good facial expressions according to the style of music and its rhythm;
- Entertainment : Omnihuman does not confine itself to cloning humans. The company indicates that its model is able to generate videos based on fictitious animation characters, as well as on animals;
More Halfbody Cases With Hands;
Here, we also provides additional examples specificly showcasing gesture movements. Some input images and audio come from Ted, Pexels and Aigc. pic.twitter.com/18nezr7qfa-Omnihuman-1 ai (@omnihuman1a) FEBRUARY 5, 2025
Multimodality is an important characteristic of this new model. In other words, he accepts existing videos at entry to reproduce movements filmed on those of the avatar of an artificially generated video. Omnihuman is similarly able to clone an individual's voice for more realism.
Ethical questions and risks of this AI
If this new AI could open the field to new creative methods for marketing, it is not free from ethical risks. Cloning-virtual-an individual raises many problems, whether they are in the private part or on the public scene.
Omnihuman supports various visual and audio styles. It can generate realistic human videos at any aspect ratio and body proportion (portrait, halt-body, full-body all in one), with Realism Stemming from understanding aspects included, lighting, and texture details. pic.twitter.com/szq1aofqbz
-Omnihuman-1 ai (@omnihuman1a) FEBRUARY 5, 2025
If its use is not properly regulated, this type of technology opens the door to many drifts, including disinformation and harassment on social networks. Beyond the impact on natural persons that these AI models can cause, the questions about the resources and the infrastructures required to make them work are legitimate.
If Omnihuman represents a tangible technological advance, ethical, ecological and financial aspects cannot be overlooked.
