LoRA based on AGM : https://x.com/AGM86997980
Trained using WAI IL V15
Trigger words: agm stylerx
Recommended weight: 0.7-0.9
Note:
This is trained with mixed sfw and nsfw dataset, excels in nsfw AGM style
This style work best with plump body(Style bias)
Lower weight tend to have better background, you might lose the style the lower you go.
Description
Trial run, double the steps, half as light weight as fukuro RXS1
Style and autonomy are great for v1, still need to fix background performance with high weights
Generation in 2:3 may have out of frame issues, should be addressed in RXS2