ByteDance, the parent company of TikTok, has introduced a new AI video creator called OmniHuman. This innovative model can transform a single image into a lifelike, full-body video with realistic singing, speaking, and movement.
OmniHuman was trained on over 18,700 hours of video data and can now mimic human-like movements, expressions, and interactions in videos. Unlike previous AI models that only animated faces or upper bodies, OmniHuman can create fully moving characters, allowing users to turn a photo into a video of someone giving a speech, dancing, or playing an instrument.
The potential applications for OmniHuman are vast, including recreating actors from the past, teaching history through simulated historical figures, and creating lifelike digital avatars for social media and gaming. While still a research project, ByteDance’s showcase of OmniHuman’s capabilities suggests that practical uses are on the horizon.
Source: https://www.techradar.com/computing/artificial-intelligence/tiktok-owner-bytedance-has-a-new-ai-video-creator-you-have-to-see-to-believe