SDXL
Blue-Tear v2.0
- アップロード日時
- 2024/09/02 22:39
- 採用数
- 541
- レビュー
- レビューがありません
- サンプリングステップ
30
- サンプリング方法
Euler a
- CFGスケール
7
- ネガティブ
score_5, score_4, simple background, low quality, lowres,
- 権限
- 画像生成と共有を許可する
- ユーザーによるモデルのダウンロードを許可する
- 生成画像の商用利用を許可する
SDXL
30
Euler a
7
score_5, score_4, simple background, low quality, lowres,
V2 (Crystal-Tear)This Version is the upgraded version of Blue Tear, it has better details and is even more stable compared to v1.Most of the info about this model are already in the apposit section, so here i am gonna go more in dept on what this model can (and cannot) do.What is this model for?Since my last lora, i've received a dozen of request of pony version of them, but i had 0 experience on those, so i've never join that train (till now).This model is the one i plan using as a base for my next works, so it had to have good understanding of the prompt, decent art, and it had to be capable of producing complex images without too many loras.The v1 was a good step in that direction, but was still lacking, so after many reiteration, i've menaged to achieve my objective.This model excells at generating portrait, but it can also used for complex illustration (The Gallery will be slowly updated in the next few days, to give you even more references)i.eThose kind of images can be easily accomplished with few keywords.(P.s i've refrained to use upscale, and adetailer for those images, just to show you the "naked" model, if you use those the quality can truth-fully skyrocket).Model strenghtsThis model has a decent amount of knowledge about anime characters, so it can generate their images without the using of loras (as you can see from the gallery).I find it decent enought, but truth be told, it struggles a bit with side-characters, or with serie that aren't so recent (but worry not - lora's exist for this reasons).Does it support NSFW?Welp, yes, it does. And it does it pretty well.It's flexible, it understand positions and (aslong we don't mess with prompt, like when adding "solo" in a sex scene with 2 people) will produce a pleasing output.Will this model receive more updates?Possibly yes, i don't believe perfection exist, so aslong it can be improved i'll be there trying to improve it.I like the model, how can i help?Just downloading it and using it is enought. But if you truly want to help me, please review it (possibly spotting eventual problems, so i can work to fix them). Another appreciated thing is to upload your generated images here.Please refrain to re-upload this model without permissions.V1Model based on Nova Xl, and created with anime merge.