推奨ネガティブプロンプト

bad quality, worst quality, low quality, ps1 screenshot, niji, anime screencap, loli, child, asian, bad hands, bad anatomy, text, watermark

bad quality, worst quality, low quality, ps1 screenshot, niji, anime screencap, loli, child, asian, bad hands, bad anatomy, text, watermark

推奨パラメータ

samplers

DPM++ 2M SDE

steps

30

cfg

5

resolution

1024x1536, 1536x1024

other models

SXZ Illuma 2 Balanced (2566c03c87)

推奨ハイレゾパラメータ

upscaler

4x NMKD Siax, 4xNomos8kSC

upscale

2

steps

15

denoising strength

0.2 - 0.25

ヒント

Use the provided negative prompt as a baseline for best quality.

The model incorporates extensive and varied styles, allowing unique visual combinations.

Support the creator on Boosty to keep the model development motivated.

バージョンのハイライト

First public version of this model.

See description for more info.

クリエイタースポンサー

All the content I make is and will be completely free on CivitAI.
But if you want to support me with a coin, then you can do it here.

SXZ ILLUMA

This model is fine-tuned verison of NoobAI targeted to combine as many great and unique artistic styles as possible.

I am already started to gather some new datasets for next version of this model, so please show some love to current version to keep me motivated. Your generations and feedback will keep me rolling.

All the content I make is and will be completely free on CivitAI.
But if you want to support me with a coin, then
you can do it here.

About v2 Balanced:

To acheve best quality, please use this negative prompt as basic:

bad quality, worst quality, low quality, ps1 screenshot, niji, anime screencap, loli, child, asian, bad hands, bad anatomy, text, watermark

Dataset includes 10.000 manually picked images in almost 200 styles. Most of the images was cleaned of any watermarks and signatures.
Fine-tune process:
10 epochs on 3090TI and MBW merge with original model.

In addition to other styles, the model includes a couple of new concepts. For example:

overdesigned - new tag that helps generate extra details in clothes/armor/ornaments.
altshaped - new tag that helps generate alternative anatomy (something like extra arms or legs, horns and other stuff)

Please note that the model contains both universal styles and those suitable for generating specific scenes, such as characters, landscapes, mechanisms, NSFW content, etc. But nothing prevents you from combining styles in order to obtain a unique visual.

And here is the whole list of new styles tuned in model:

"123wsty"
"39OO27"
"61q83e"
"ahriman, anime screencap"
"akreon"
"albrecht dürer"
"aleriia_v"
"alex ross, retro artstyle"
"alexkonstad"
"aloija"
"amadmir"
"anato finnstark"
"anbudehuang"
"anime screencap"
"animeirl"
"aomrng"
"arcane"
"asahiart, automaton, altshaped"
"avasdmn"
"ayyasap"
"bayardwu, warcraft"
"boris valejo, retro artstyle"
"brault"
"brenda burke, faux traditional media"
"burda, grungetech"
"cagampan, western comics \(style\)"
"caldermoore"
"castlevania, anime screencap"
"cher-ro"
"cherryche, chromatic aberration"
"cherrygig"
"herowars, chibi"
"chuckartt"
"claina"
"clashroyale"
"claude monet, faux traditional media"
"cmg"
"cnlin"
"colehole, warcraft, graphite (medium)"
"conorburke"
"darkest dungeon, high contrast"
"dashiana"
"degenesis"
"demessance"
"deus ex, cyberpunk"
"dimary, graphite"
"dinamuuu3d, 3d"
"disney, flat color, retro artstyle"
"dkeeno"
"dominikmayer, color contrast"
"drakonoart"
"dravacus"
"dreelrayk"
"drkfnts"
"eddie mendoza, photobash"
"edgerunners, anime screencap"
"edjubey"
"elden ring"
"emersontung, tankhead, overdesigned"
"envar"
"esuthio"
"faavesa"
"fanfoxy"
"fangmaomao"
"fortnite, 3d"
"fugtrup, 3d"
"fuhrmann"
"galyosef, 3d"
"ganas"
"gmdv icon"
"greg rutkowski"
"hades"
"hamishfrater, grungetech"
"handpainted"
"hearthstone"
"henry asencio"
"hizipro, high contrast"
"iamdehaf"
"ikenaga, yasunari"
"incase"
"indus"
"ivan kramskoy"
"jakubjagoda"
"jandrew, scenery, environment"
"jebedai"
"jeehyunglee"
"jeremy mann"
"jijiefeng, overdesigned"
"jim lee"
"jinxin"
"jlhwu, 3d, overdesigned"
"john william waterhouse"
"jose parra"
"jparked"
"junejenssen"
"justsomenoob"
"kael ngu"
"khanhphinguyen, warhammer, overdesigned"
"kooka, anime"
"krrat, overdesigned"
"kruglov"
"krysdecker"
"kylepunk"
"laurelin"
"leicai"
"logancure"
"lol"
"loona"
"lovk1y"
"makoto shinkai, anime screencap"
"mangamie"
"marta nael"
"mattsanz"
"mattsellers"
"max kostin"
"mengxuanli"
"metalcore cyberpunk"
"michera"
"mixppl"
"mlwc, 3d, altshaped"
"mohrbacher, divine, altshaped"
"mprizzitano"
"mrvlsnap"
"mtg"
"mxbrdfll"
"nadar, realistic, chromatic aberration"
"nicolasaviori"
"niji"
"nixeu"
"ognjensporin, altshaped"
"othalam"
"roman papsuev"
"persona, anime"
"peterskore, altshaped"
"photo, realistic, cinematic"
"pixar, 3d"
"powl, color contrast, abstract"
"ps1 screenshot"
"qichaowang"
"rakavka"
"syd mead, retrofuturism"
"rinotuna, reference inset"
"ruanzoe, faux traditional media"
"saiya-art"
"sanmingsanri"
"sariya"
"shavrin"
"Shinjuku, chromatic aberration"
"smite"
"sphrthart"
"stumpyfongo"
"sugarpunk, cyberpunk"
"svnsvn"
"sword, swordtember"
"tck, altshaped"
"templarpainter"
"terrycantal"
"tixnen"
"tomas duchek"
"tonysart, altshaped, bestiary creature"
"trufanov, altshaped, overdesigned"
"tshae"
"vablo, altshaped"
"valentin serov"
"valorant"
"vegera"
"viktor vasnetsov, faux traditional media"
"vincent van gogh"
"vntgfnts, movie still, cinematic realistic"
"w3reforged, warcraft"
"warcraft"
"wayfinder"
"weiwang"
"willmurai, warcraft"
"wlop"
"wushenyou"
"xianglongwang, altshaped, overdesigned"
"yumingli, minimalism"
"yuriiiko, flat color"
"zack snyder, movie still, cinematic, realistic"
"zexiguo, overdesigned"
"zhaoziyue"
"znor"
"zooba"

投稿者

前の画像
NightVisionXL - NightVisionXL_v9.0.0
次の画像
Divine Diffusion by Dever (DDD) [SDXL] - v1.1

モデル詳細

モデルタイプ

Checkpoint

ベースモデル

NoobAI

モデルバージョン

2 Balanced

モデルハッシュ

6021750db8

作成者

ディスカッション

コメントを残すには log in してください。