Reface only uses 1 photo - how?
Hi there faceswap-community,
title says it already. I am total noob in regards to AI, deepfakes and faceswap but have been playing around with it a little bit. But only because I had gotten curious about it after trying out an app called Reface.
My question is: How is it possible that Reface achieves really good results with just one photo and 15seconds of processing while I get unsatisfactory results with 1000s of photos after 48h of training with faceswap?
Is reface's code so much better? Is the AI so much smarter? Sorry, I really have no idea about programming and any stuff that is too technical. It took me two days to fight through the faceswap-guide 
But I made it and now I am trying to improve. But yeah, I really want to understand what is the big difference here? Any experts who can explain that to me?
Thank you!
Jamrio
For example, why don't they need 5000 faces? Or why don't they need to train? Do we even know which technique they use?