V0.2 Important Notes:
PSA: Not really good at dark images, so either turn down or load a light dark LoRA.
Trained with Danboru tags and Natural language, real pain in the ass to manually write/edit but, better prompt adherence.
4000 Steps with 120 Images with Tags and 173 with NL
MFW I go through 2000 Images to make a dataset, then have to train 7 times.
A style LoRA for Anima (by extension its finetunes) by CircleStone Labs.
This LoRA was trained on a curated dataset of 120+53 Illustrations from Ramon Nunez. All images were treated to remove text, any censoring and most symbols, then batch tagged with ConvNextV2 Huge Full and manually tagged for fixing/adding tags.
This Lora is best used at 1.0 strength, imo. Trigger word is very optional, but hey, its R3LOAD.
Also this was trained on my chef's kiss AMD 7900XT, (under Linux with Ostris's Ai-Toolkit❤️)
Description
Main Settings: 4000Steps, AdamW8bit, 1Batches, [768,1024], LR 5e-5, WeightDecay 1e-2, Rank32, Repeats 2(TAG) and 2(CAPTION), No X Axis Flip, 120 IMG (TAGS), 173IMG (NL)
[Trained with both Danbooru Tags and Natural Language. Tags weight 1.4, Caption Weight 0.6]