Do i need FaceTraining when i Can SwapFace with just 1 photo?

[seawithfire]

hi. i like faceswaping and its very fun. but i wonder how much difference there is between "FaceTraining" and "FaceSwap with one photo" result? for example, let say i swap emma watson face with a trained model of her face. Will the result be significantly better than version i create with just one photo? or the difference is not much?

In other words:

1. can we say that face training was for before and now with reactor and other service we dont need that anymore? or still the result with trained face is significantly more natural and good? (in realistic theme)

2. can anyone that already trained a face, show me an example of a SAME PHOTO that swapped with both "trained face" and "just one photo with same person" ? i want to see/feel the difference

3. i used faceswap with reactor and many powerful service like AKOO. in all of them, some faces become significantly similar to person, but for some people result are not similar to them at all. can we say the trained model is for these cases and with trained face i can get a VERY similar result even for them?