📢Face Anything: 4D Face Reconstruction from Any Image Sequence
Transformer model for 4D face reconstruction and dense tracking:
- predict canonical facial coordinates per pixel
- tracking as reconstruction in canonical space
- geometry + correspondences in one forward pass
Key idea: a shared canonical space across frames
- correspondences as nearest neighbors
- no motion or deformation estimation
Stable geometry and tracking, even under large expressions and viewpoint changes - check out our results!
🌐 https://t.co/VRF2UFYo6Y
▶️ https://t.co/qMv8IKpy6R
Great work by @UmutKocasa4344, @SGiebenhain, @richard_o_shaw
You wrote a bunch of scalability articles on your blog. You say, meh, I'm going to delete my blog. Then you learn that web archive has your website and a github repo with ~300k favs listed your articles as the first steps to sys design. Respect brother
Lack of context usually leads to a slower progress on any task/goal IMO. I try to ensure my assumptions are correct, make sure my arguments are readable/logical which takes time to check. Any kind of feedback means starting from all over again with the thought process.
Concrete example: I like providing context in the MR descriptions. When I know what I'm doing, I left a screenshot or a couple of links. When I'm not comfortable with the topic, I tend to write a lot, explaining the reasoning behind. It gets outdated really quick.