Type | |
Stats | 796 1,840 71 |
Reviews | (106) |
Published | Mar 18, 2024 |
Base Model | |
Training | Steps: 16,400 Epochs: 80 |
Usage Tips | Clip Skip: 1 Strength: 0.8 |
Hash | AutoV2 CF9A9875A2 |
This lora was trained for re-create the Ra's art style.
I rlly like see your reviews and generations. They are helpful for fix and train better models. Share some hearts at the previews that you like and feel free to leave your comments or requests, but I can't promise anything xd.
Version 3: LoKr
LoKr version, same dataset, almost 11 hours of training (vs around 4), but the result is better.
Recommended weight: .6 - 1
Version 2: LoRa
I pruned much more the data set and the tags, removing them by co_ocurrence, and I used the base XL bucket resolutions. Now gives an style much more consistent
Recommended weight: .8
Version 1: LoRa
This version recreates the wathercolor style, other epochs do some random styles (The Ra's art style evolved so quickly).
Some characters like Illya, Miku and others could change of clothes because the data set contains several images of them (check the miku example)
Recommended weight: .8