Version 2 of my first LORA ever ;)
Pony already offer interesting DAZ3D image but the lora help going a but further, especially to imitate the style of DAZ3D based ?????? games.
What's new ?
- Dataset increasing from 23 to 40
- Repeat lowered from 20 to 15
600 iteration total (epoch:1)
The V2 bring more diversity and does less aberration.
Trigger word is : DAZ3D
I use a custom Checkpoint (unshared) and FaceDetailler for my render.
What's next ? Target is 100 pics dataset /w 15 repeat.
Enjoy :)


