We're reintroducing and open-sourcing project "See-through". Given a single anime illustration, it automatically decomposes the character into fully-inpainted semantic layers with depth ordering. One image in, layered PSD out. (1/n)
Repo:
https://t.co/w5oB45ljcI