Designed to streamline generation and reduce style stray without breaking consistency (too much).
Recommended settings:
DPM2 + SGM Uniform or Bong Tangent or Simple.
CFG 4 (and up to 6)
30 steps (10 minimum, 40-50 maximum)
Image size around 1 - 1.5 MP
Prompt what exist on image with tags and how it positioned / interacted with natural language
GD folder with testing images with some artist tags that can be used to enhance generation:
@ebifurya, @inoino, @matsunaga kouyou, @tony taka, @a1 \(initial-g\), @hungry clicker, @minaba hideo, @nanashi \(nlo\), @toosaka asagi, @mizuki makoto, @personal ami, @rappa \(rappaya\), @as109, @satou kibi, @imizu \(nitro unknown\), @ririko \(zhuoyandesailaer\), @furaggu \(frag 0416\), @yabuki kentarou, @kanade \(kanade 3344\), @kagami hirotaka, @riichu, @kloah, @misumi \(macaroni\), @sakiyamama, @nironiro, @mashiro yukiya, @unasaka ryou, @sakimichan, @liang xing, @shiratama \(shiratamaco\), @sydus, @shimmer, @xenon \(for achieve\), @makicha \(sasurainopink\), @fugtrup, @shuri \(84k\), @crumbles, @nottytiffy, @goto p, @porurin, @amasora taichi, @siu \(siu0207\), @tama \(tamatamo\), @nogi takayoshi, @kawata hisashi, @seiru \(prairie\), @koutaro, @kobayashi chisato, @karory, @koku, @osashin \(osada\), @ao banana, @morikura en, @maki \(seventh heaven maxion\), @amagi \(amagi626\), @hajika, @nannacy7, @ouma tokiichi, @asakuraf, @guweiz, @yasuyuki, @prywinko, @less, @stanley lau, @miura-n315, @kannko bokujou, @hidulume, @misaka 12003-gou, @sirokuma shake, @monikano, @takeshima eku, @haruri, @himura moritaka, @mitha, @toujou sakana, @tsukasawa takamatsu, @kakage, @oda ken'ichi, @asanagi shion, @jasony, @tomari \(veryberry00\), @ryosios, @ran'ou \(tamago no kimi\), @mochizuki shiina, @takase muh, @janong, @squchan, @flasher \(flasher ura\), @tateha \(marvelous grace\), @nakkar, @medu \(rubish\), @tsurusaki takahiro, @rasukaru, @mimikaki \(men bow\), @yuizaki kazuya, @dantewontdie, @f.s., @softmode, @damda, @neropaso, @aroma sensei, @sankuro \(agoitei\), @reeh \(yukuri130\), @guchico, @himitsu \(hi mi tsu 2\), @saku \(kudrove\), @koretsuki azuma, @yuran, @tsuno no hito, @limgae, @j@ck
Description
FAQ
Comments (9)
Oh I thought you would mean something else by the name of the model lol, gotta try it to see if it can actually work on the artists problems, what did you merge it with?
Anima-preview-3, Cat Tower, Animalka
@eliont dayuum,much appreciated
@Neon_signs
https://drive.google.com/drive/folders/1TBnzwR1w2u_Bp97OtG5faT3h4NNa5a0p?usp=sharing
Tested some artist tags on this models.
@eliont thanks bud
Just for other people to see - this merge works surprisingly well. While it does have slightly more biased style then base preview3, the images are noticibly more stable. Good work. We finally have some chance to go out of SDXL cage :p
https://drive.google.com/drive/folders/1TBnzwR1w2u_Bp97OtG5faT3h4NNa5a0p?usp=sharing
Tested some artist tags on this models.
Why is such great work not appreciated!!?!?!?!?
It does a better job than any other Preview 3. thank you.
Amazing! Best anima model I've seen yet!











