This is crazy... Bytedance just announced DreamActor-M1, a new model that brings full-body, expressive human animation to a whole new level. It outperforms Act-One, Animate Anyone, SkyReels-A1 across all metrics. 10 examples:

fucking nuts

Announcing: Our most advanced speech-to-text model goes beyond accuracy to capture the real-world complexity of human conversation and deliver reliable, source-of-truth audio data. Explore Universal-2 updates 👇

1. Diversity The method is robust to various character and motion styles.

2. Controllability and Robustness It can extend to audio-driven facial animation, delivering lip-sync results in multiple languages.

3. Comparing to SOTA Method

4. Portrait Animation

6. Portrait Animation

7. Portrait Animation

8.

9.

10.

Project page: Paper:

That's a wrap! If you enjoyed this thread: 1. Follow me @AngryTomtweets for more of these 2. RT the tweet below to share this thread with your audience

AI keeps leveling up, full-body animation like this was unthinkable just a year ago

Yeah, was so faaaar away.

did they trained this with the millions of videos ppl uploaded to tiktok? yes.

Data is the new oil

social media is about to get so diabolical we won’t trust anything

the control they have over style and movement is crazy

It's insane, tbh!

What's the CA ? Bullish

@archiexzzz

viado.... @Kapuva2 @Pudimporquesim