Technology Updates : ByteDance’s new AI model goes viral for its deepfake capabilities that animate photos

Source: https://www.scmp.com/tech/big-tech/article/3297500/bytedances-new-ai-model-goes-viral-its-deepfake-capabilities-animate-photos

Apple Intelligence:

• AI Model Capability: ByteDance’s OmniHuman-1 model can create realistic videos of people speaking, singing, and moving.

• Model Performance: The model significantly outperforms existing audio-conditioned human video-generation methods.

• Model Development: ByteDance researchers developed a novel training strategy that combines diverse data sets of text, audio, and movement.

• Video Generation Capabilities: Generates realistic videos with varying aspect ratios, body proportions, and detailed facial expressions.

• Realism and Applications: Features natural head and gesture movements, unlocking potential real-world applications.

• Competition in the Market: Chinese tech firms, including ByteDance, Kuaishou Technology, Zhipu AI, Shengshu Tech, and MiniMax, are competing in video generation.

Opinion:

Really realistic Albert Einstein deepfake.

Creations like these will definitely blur the lines between real and fake information.

I think there will be a desire for more deepfake detection tools.