Anima official release!
released BASE1.
This model is based on anima-base-v1.0.
CottonAnima is a merged model based on Anima, combined with a style LoRA (trained on 100 + 60 AI-generated images).
This model adopts a fixed art style, resulting in improved style consistency. Please note that this comes at the cost of some of Anima’s original strength in diversity.
Recommended settings
Sampler: er_sde
CFG: 5
Steps: 24～30
Quality tags:
masterpiece, highres, absurdres, newestNegative prompt:
worst quality, low quality, score_1, score_2, score_3, blurry, jpeg artifacts, sepia, username, watermark, chibi, source_pony,source_furry,source_cartoon, monochrome, greyscale, 3d, realistic, backlighting,Description
FAQ
Comments (5)
cottonいつもありがとうございますillからずっとお世話になってる
Was waiting for this one. Thanks.
CottonAnima is great - It still responds to artisttags, emotions, different quality tags and has great consistency. Thanks for all the work you put into it @uraoto ! 💖💖
base1 is promising, but it may have sacrificed too much flexibility for consistency. It regularly ignores tags or just doesn't understand some pretty simple concepts. As others have noted, it's not very good with male figures and doesn't have hardly any variation in the way it deals with them. It does have a nice, solid style for output, so I hope it'll be developed more.
that true, prompt adherence decrease alot, also its censor my generate most of the time even tho i put on negative