ByteDance's researchers have unveiled OmniHuman-1, a potent AI system capable of generating highly realistic deepfake videos from minimal input—just one image and an audio clip. Unlike existing deepfake technologies, OmniHuman-1 effectively overcomes the 'uncanny valley' problem, producing convincing visuals. It can also edit videos, altering movement and proportions while still facing challenges with low-quality images and angle adjustments. Though not released to the public yet, this breakthrough raises alarm over future misuse, especially following instances of politically charged deepfakes affecting elections around the world.
Researchers from TikTok owner ByteDance have demoed a new AI system, OmniHuman-1, that can generate perhaps the most realistic deepfake videos to date.
OmniHuman-1 only needs a single reference image and audio, like speech or vocals, to generate a video, and it can edit existing videos.
The implications of OmniHuman-1 are worrisome, as evidenced by the spread of political deepfakes influencing elections in Taiwan and Moldova.
Despite its advancements, OmniHuman-1 isn't perfect; low-quality reference images can yield subpar videos, and it struggles with certain poses.
Collection
[
|
...
]