3 ms·ByteDance's AI model can generate clips based on text, images, audio, and video2 points by speckx 9mo ago