Preface:

Im learning how create good LoRA´s at the moment. I think learning to make realistic persons is the first step to see if the LoRA´s are good. It´s hard to make a realistic person that is not overtrained, so i already did around 30 LoRA´s and here is the best so far.

V1:

The face results are awsome!

The problem: its very stiff and not yet perfect with other LoRA´s. I will update as soon as i get better results, i do this 24h/7 at the moment!

Feedback please!

Please if you have some time after testing it give me some feedback how the outcome is for you! Any support is very welcomed :)!

About V2

The new model has much more flexibility! The face looks mostly realistic! I personally don't think V2 is any more realistic than V1 but it has so much more flexibility and still looks very similar. I would upgrade to V2 if you have already downloaded V1.

Reason for new update

I had a bit of a "breakthrough" with my LoRA's and am happy with this model for now. I will now try new characters! If you suggest a person I'll see if I can implement them next, just comment!

Short overview:

Quality: / similar as V1

Flexibility: ++ Way better with any checkpoint!

Combination with other LoRA´s: ++ way better!