Are you choosing between Render, Video, or Seedance?
Actually, it’s about developing robots through training on imperfect motion capture data: the LATENT project — training humanoid sports robots for tennis based on noisy human movements.
The project code can be found here:
https://github.com/GalaxyGeneralRobotics/LATENT
In the comments, people often ask why the robots look so human-like. I have the same question: they seem fragile, unstable, and slow.
One answer is precisely the choice of training data. For example, they use videos with leather suits or mocap suits. Thanks to this, there is material to train such robots.
And also — it’s so cool how it hits the ball that you just want to somehow reward it for such a cute game.
Created with n8n:
https://cutt.ly/n8n
Created with syllaby:
https://cutt.ly/syllaby