New way to encode a profile picture dropped:

https://x.com/Ethan_smith_20/status/1801493585155526675

320 bits is basically a hash. Small enough to go on chain for every user.

so the question I obviously have is...

how do this compare w/ segmentation research, and supposing video data can be compressed, what can be learned about latent representation of pose/motion?

so the question I obviously have is...

how do this compare w/ segmentation research, and supposing video data can be compressed, what can be learned about latent representation of pose/motion?

https://arxiv.org/abs/2406.07550