face set

[Duhust321]

I started to using this only for few days. It's an amazing tool, especially masking, and that face set!
I have a question for that.
I thought it's better to feed multiple angle of source image and multiple facial expression. So, I made
8 images of same person. (8 is the number I saw in old thread) But, it's on different locations, different time, different light source. The photo taken in same time period. The person looks same for human being, but it has different colors from lighting and color.
Is this good plan to teach AI to show a face structure of the source material? Or AI will confuse with the lighting/color difference?

Is it better to feed two AI reference images of , but way less camera angle, less variation of an facial expression?
thanks in advance.

[LucianoDaluz]

I guess it only takes the face landmarks, so light and expressions are uselesss. It is not like a LoRA

[Shahfahad7866]

Once I trained a faceset with 15 face images, but the result was the same as using one single image or a faceset.