V0.3 Important Notes: Trained with Danboru tags and Natural language, pain in the ass to manually write/edit but holly smokes, WAY better prompt adherence. Upped the dataset size from 46 to a nice 69 (scraped the bottom of the web for valid images) Bumped to 4000 Steps with lower Learning rate, no flipping and less repeats. Fun bonus, example dataset images before fixing (Editing, cropping, upscaling): <img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/fd63ba02-f552-4593-8a89-3b8934edda15/original=true/fd63ba02-f552-4593-8a89-3b8934edda15.jpeg" /> A style LoRA for Anima (by extension its finetunes) by <a target="_blank" rel="ugc" href="https://huggingface.co/circlestone-labs">CircleStone Labs</a>. This LoRA was trained on a curated dataset of 69 Illustrations from <a target="_blank" rel="ugc" href="https://x.com/horikoshiko">Kohei Horikoshi.</a> All images were treated to remove text, any censoring and most symbols, then batch tagged with <a target="_blank" rel="ugc" href="https://huggingface.co/Mooshie/convnextv2_huge.dbv4-full">ConvNextV2 Huge Full</a> 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 H0RIKOSHI. Also this was trained on my chef's kiss AMD 7900XT, (under Linux with <a target="_blank" rel="ugc" href="https://github.com/ostris/ai-toolkit/">Ostris's Ai-Toolkit</a>❤️)

HORIKOSHI-STYLE-ANIMA

Main Settings: 2000Steps, AdamW8bit, 4Batches, [768,1024], LR 5e-5, WeightDecay 1e-2, Rank32, Repeats 3

Files with SHA256: 437fc98c2e1e842ea5fcd472d353c2e320ab3dc29417ba9a76313543a2e355ba