As an AI enthusiast, I love to read different white papers from developers around the world regarding AI tech/development.
Something caught my eye was FSGAN. Face Swapping GAN. No "specific" training required.
What do you guys think?
We're well aware of this, but are waiting on full release to see if others replicate the results. From reading what's going on, it's not terribly different from attempts like GANimation or Samsung's Few Shot which morph an image but actually don't create near the quality of results that a properly trained FaceSwap can.
In general I don't trust published results until I see them replicated by others.