torzdf wrote: ↑Sun Feb 09, 2020 11:33 pm
As to your second, yes, absolutely ![Smile :)](./images/smilies/icon_e_smile.gif)
Cool. I just finished chapter 3 of Deep Learning Illustrated by Jon Krohn, which introduced me to GANs, which blew my mind. In terms of what's relevant to faceswapping, the possibility of being able to autogenerate completely original faces free of legal and ethical encumbarance is very exciting.
Has anyone packaged up any GAN-based face-generation tool? (The Natural Neural Network in my head is guessing "no" based on your previous statements.)
Also, more useful than generating a 2D face would be generating a 3D face model, which could then be posed and lit in any configuration to fit the frame being converted during a faceswap. I'm hoping the CS rocket scientists are already on that...
bryanlyon wrote: ↑Sun Feb 09, 2020 11:35 pm
Best way that I can think of to do tattoo removal in AI would be a Tattoo detection model followed by a patch redrawing (infill) network. But like Torzdf said, I'm unaware of any complete packages that do it.
Oh well. I don't foresee having the knowledge and experience to attempt it myself for a very long time. I'll keep an eye out in the meantime for announcements.