This model is trained on 150 3D modelled faces. This model aims to increase the detail on the face in a hyper realistic way. The images were retrieved from art station so big shoutout for the amazing creators on that platform.

GPT4V was used to caption the dataset, this is my first time using it and it seems to give great results.

I've found if the strength is too high the eyes won't align very well. A strength of 0.8 seems to work effectively (SDXL).

This model is quite experimental so if you have any feedback I'd love to hear it.