Updated v2 scheduled to release on 01/20/2024

v2 is much more flexible and supports different outfits as well as different poses.

Layla doesn't have much good art so I used a lot of AI art in the v2 dataset, unfortunately this resulted in her bad hands being way worse than usual.

This doesn't always happen when you use AI art for a dataset however the training resolution was set to 768 which means the bad hands were emphasized.

Trained on: 52 images

Training Model: Animefull-final-pruned-fp16

flip_aug: False

Num of Repeats: 10

Unit is Epochs or Steps: Epochs

Number of Epochs or Steps: 10

Training Batch Size: 2

Resolution: 768

Network Dim: Dim=32; Alpha=16

Lora Creation Process took: 1:10:31