SDXL

Mixstrious

Mixstrious - AI Model封面圖片
上傳時間
2025年12月25日 晚上7:22
使用數
13.1k
評論
好  (4)
取樣步數

30

取樣方式

Euler a

CFG Scale

4

負面

lazyneg, lazyhand, lowres, worst quality, bad quality:1.2, modern, recent, oldest, signature, username, logo, watermark, jpeg artifacts, bad hands, cropped, missing fingers, extra digits, fewer digits, error, bad anatomy, ugly, disfigured, young, long neck

許可權
  • 允許圖片生成與分享
  • 允許使用者下載此模型
  • 允許生成圖片用於商業用途

推薦 LoRA

描述

Thanks for giving this Model a try :)Usage Guide :CFG : 4-6Clip skip : 1 or 2 doesn't matter :)Steps : 30+Sampler: Euler a gives the best results.Use hi-res fix with any good upscalers suitable for anime style.Resolutions :Portrait : 832 x 1216, swap them for LandscapeSquare : 1024 x 1024Hi-res fix resolution : (I use these resolutions you are free to mix up any resolutions you like)15 steps and 0.5 denoise strength.Portrait : base res (512 x 768) upscale this by 2 --> 1024 x 1536base res (683 x 1024) upscale by 1.5 --> 1024 x 1536base res (832 x 1216) upscale by 1.5 --> 1248 x 1824Swap the resolutions for Landscape.Square : base res (640 x 640) upscale this by 2 --> 1280 x 1280base res (854 x 854) upscale by 1.5 --> 1280 x 1280base res (1024 x 1024) upscale by 1.5 --> 1536 x 1536

留言