YutaMix - V9

YutaMix

CHECKPOINT
転載


更新

YutaMix by miaomiao on Tensor.Art
YutaMix by miaomiao on Tensor.Art
YutaMix by miaomiao on Tensor.Art

creator: iselestia

Best anime checkpoint

Fp32 and fp16 avaible on v13 v14

The new version of yutamix is ​​created as clean and beautiful as possible. In order for you to have not to download heaps of other checkpoints just that have other draw style, you just need to connect lora/lycoris to the style of any artist and enjoy completely new and wonderful results. I especially recommend using my styles and other loras because they trained on these models. Yutamix started from a compote with other models long time ago. But now its pure and beautiful

use only Euler a , r-esrgan 4x+ anime6b hires - 10 hires steps

use negative embeddings, atleast - easynegative

if you never used them, time to start, you can set many of them, as default negative prompt and they boost quality of picture,And eliminates the need to write tons of negative prompts.

The model is not intended for other settings and has never even been tested for them. if you want results like mine. do as I say

Also all my loras/lycoris always great work on yutamixes, cause they trained on them.

Works good even on 15 steps

No need use more than 50 steps

Hires steps - 10 denoise 0.35

About all versions YutaMixV14 - upgrade of v13, custom vae built in YutaMixV13 - pure, sharp, and beautiful has built in custom vae.

YutamixV9 - sketch style like

YutamixUR - like v9 but kinda better work with loras, and have some style difference and more sharp than V9

YutamixV11 - sharpest clean anime checkpoint - recommend use for training loras for beginners.

YutamixV12 - sharpest and Colorfull, and works in good quality even with bad quality loras.

バージョンの詳細

SD 1.5
40
7
Sampler: Euler Negative: (pixel:1.0), text, EasyNegative, by (bad-artist:0.6), (low quality:1.3), (worst quality:1.3), (bad anatomy:1.3), (extra arms:1.2), (lowres:1.0), (cropped:1.0) , (worst quality:1.0), (low quality:1.0), (normal quality:0.6), (jpeg artifacts:1.0), (bad hands:1.0), error, (missing fingers:1.0), (missing anatomy:0.6), (missing hands:0.6), (bad fingers:0.6), (extra digit:1.0), (fewer digits:1.0), (bad detailed:1.0), (missing hand:0.6), (bad finger:0.6), (missing finger:1.0), (bad hand:0.6), (bad legs:0.6), (bad reflection:0.6), (missing reflection), (bad background:0.6), (missing background), (bad eyes:1.3),(bad geometry:0.6), (bad light:1.0), (missing light:0.6), (bad shadows:1.0), (missing shadows:0.6), (bad wrist:1.0), (missing wrist:0.6), (blurry:1.3)

プロジェクト権限

このモデルは以下から転載: https://civitai.com/models/123021

転載されたモデルはコミュニケーションと学習目的のみであり、商業利用には適していません。オリジナルの作者は、Discordチャンネル--- #claim-modelsを通じてモデルの移行についてお問い合わせください。

関連する投稿