Membres铆a PixAI

SD

馃毀 (WIP) Everything-ReMix v1

馃毀 (WIP) Everything-ReMix v1 - AI Model cover image
Subido en
15 ene 2024, 12:55
Utiliza
13.1k
Rese帽as
Sin rese帽as
Pasos de muestreo

27

M茅todo de muestreo

Euler a

Escala CFG

7

Excluir de la imagen

(bad-artist-anime, by bad-artist:0.8), (worst quality, low quality:2), bad quality, normal quality, (EasyNegativeV2, easynegative:0.9), (bad eye, unclear eyes, undetailed eyes, missing limbs, fused limbs, broken limbs, fewer digits, extra digits, mutation, deformity:1.4), (interlocked fingers:1.3), (bad anatomy, white background:1.1), monochrome, watermark, ng_deepnegative_v1_75t, verybadimagenegative_v1.3, (NSFW, cleavage:1.2), green skin

Permisos
  • Permitir la generaci贸n y el intercambio de im谩genes
  • Permitir que los usuarios descarguen su modelo
  • Usos comerciales

Descripci贸n

Important: Re-enhance for full body or you won't get decent eyes/hands. If you want to get more quality, try my HD ALL-IN-ONE LoRA. 鈿狅笍 Here's my tutorial on how to get the best quality possible: https://youtu.be/lF5csBbVIX4

Hardly experimental! Everything-Re;Mix is a remake of one my older models named Everything-Mix (that I rename it into Mix 2D), I did this now that I know a little more of block merging. In this model I'll try to include experimental versions, remade in another way to give it different styles and not only 2D. Everything-Re;Mix v1_release is my old 2D version that I want to improve a little with other models. I had to rename my older 2D versions to Everything-Mix 2D, as it was my first 2D model and I don't know as much as now. As it didn't got as I wanted, this version is going to be remade soon from the start. I don't know what do actually do with this model, but probably I'll try a 2.25D style. Once I make this model I'll probably try to create Everything-Mix v1, that is nonexistent due to the name change.

Everything Re;Mix v1_r is actually an slower model than intended, so it suffers on lower steps in PixAI... Don't forget to use face fix if you are going for a non-closeup image. Or you can try to fix it with other models.

Quality for v1: If you want to test different ways to increase quality, try custom resolution 960x610 / 960x530, highres or 1024x resolution. High steps is not something that will work for this model. Don't forget face fix for non-closeups.

路 v1_release has theorically improved hands, face generation and better N-SFW and compared to Everything-Mix v1. The idea was to mix up all my 2D and 2.25D models to create an interesting style, but I decided to remake it again in the following versions. This model seems to give good results with a LoRA with thicker lines, since it lacks some 2D. 路 v1_release.2 has a lot of changes, mainly fixing eyes, giving it better hands, some N-SFW and a better composition overall. This version uses the merges I did in an interesting way. I can try improve it using other models, but right now it works pretty good. Try it with x1024 resolution or use face fix if you are not going for a closeup.

Negatives and samplers not tested enought, but I'd say: Euler a/DDMI: More than 30 steps. DPM++ samplers: More than 20 steps.

Why I decided to create this model? Because I want to keep testing until I make my Everything-Mix v3 model, mixing my best models to create a very good one. Since this one lost some 2D in my opinion as it has more of a 3D body, I'll try to make this one a 2D+ model. For a better 2D style, try Everything-Mix 2D v2.

Issues with v1_r: Lower steps may show wrong eyes, it can be fixed with higher resolution (like enhancing) or using more steps. 27+ Steps recommended.

Comentarios

InicioModelosMio.2Perfil