SD

Void-Break

Void-Break - AI Modelのカバー画像
アップロード日時
2024/01/17 5:34
採用数
1.2k
レビュー
レビューがありません
サンプリングステップ

25

サンプリング方法

DPM++ 2M Karras

CFGスケール

8.5

ネガティブ

(worst quality:1.4), (low quality:1.4), (monochrome:1.1), verybadimagenegative v1.3, bad-picture-chill-75v, zipper, jewelry

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

詳細説明

From https://civitai.com/models/192860?modelVersionId=235293 Created by Vexis (https://civitai.com/user/Vexis)

Void-Break Gamma is currently my personal default model

VoidBreak is kind of a magical mix started from BreakDomain_M2150, a model version that no longer exists, and a near perfect fusion of a few loras. This is an evolution of QBreak, which only had one lora added in, while this has 3-4 more.

It can work with pretty much anything as far as I'm aware. The hands are a bit bigger and less consistent then they had been but I'm not sure if that's from the loras or from dropping a negative imbed to increase uniqueness

These are the base settings I use:

Basic Positive Prompt: 1girl, absurdres, medium soft breasts, curvy, [perfect shadows and lighting], detailed background, incredible high-key lighting,

Negative prompt: (worst quality:1.4), (low quality:1.4), (monochrome:1.1), verybadimagenegative v1.3, bad-picture-chill-75v, zipper, jewelry

Steps: 25

Sampler: DPM++ 2M Karras

CFG scale: 8.5

Size: 512x768

VAE: kl-f8-anime2

Hires Fix + Upscale (Extras) Denoising strength: 0.5,

Hires upscale: 2

Hires steps: 15

Hires upscaler: 4x-UltraSharp,

Clip skip: 2

Upscale by: 2-4 Upscaler(s): 4x-UltraSharp and/or RealESRGAN_x4plus_anime_6B

コメント

Void-Breakで生成した作品