I was curious if anyone has made a vr deepfake before.
Is it as simple as loading the side-by-side 180 vr .mp4 into faceswap and just deep faking it like a normal video,
Or does it take more than that?
When you view VR videos on non 3d video player you get both screens at once s-by-s and they have a fish eye lens effect. Any object (lets say a face) in the very middle of the video will be in proportion but when they start to go towards the boundaries, they start to stretch out like you might expect a fish eye lens to do. Now I'm wondering, do you first have have to edit the replacing(b) faces in just such a way where its all stretched out to similar affect as the replaced-faces in the vr video before running it through extraction? Or is faceswap just smart enough that however wonky and goofy the faces stretch out in the VR video, it will just make the replacing faces similar stretched.