ANIMA branch of Cyberfalafel

The model is also new to me. But it turned out, as for me, it's not so bad to fix everything.

Recommended settings:

Sampler: Euler A CFG++ (STRONG RECOMMENDATION), ER SDE, Euler, DPM++ 2m.

Schedule: beta, simple, normal, sometimes bong tanget.

CFG: 1.1 for CFG++ sampler and 4-5 for all other.

Steps: 30-50, i prefer 40.

Generic positive: masterpiece, best quality, score_7, newest,

Generic negative: (worst quality, low quality, score_1, score_2, score_3:2.0), artist name, blurry, jpeg artifacts, blurry, blurry background, blurry foreground, blurry edges, blurry vision, bokeh, depth of field, green theme, (red theme:2.0), yellow theme, blue theme, worst aesthetic, bad quality, low quality, (bad hands:2), censored, bar censor, mosaic censoring, artistic error, bad proportions, (bad perspective:2.0), wrong hand, bad reflections, too many fingers, too many hands, too many legs, wrong feet, bad feet, extra digits, bad anatomy, ai-generated, lossy-lossless, jpeg artifacts, scan artifacts, @umanosuke, @roncostar, @ke-ta, @smewed, @melkor mancin, ai-assisted, @dikko, skinny, ribs, fog, outline, white outline, hip bones, missing finger, monochrome, greyscale, (simple background, white background:1.3), ring light reflection, (shiny skin:2), (photo background:2.0), @jasony, @ebifurya, (high contrast:2.0), (long torso:2.0),

V-pred branch of Cyberfalafel

Completely new model type for me, but it's work and work good.

Model have medium cfg sensitivity. CFG 4 or 5 is recommended. Recommended vae xlVAEC_f22 or f231. MaHiRo is also recommended if cfg scale on 4-4.5.

DPM++ 2m. And you can use DPM ++2M SDE by the way.
Euler A.
Euler.
DDIM.

Schedule type:

Beta.
SGM uniform/normal/simple/uniform.
Polyexponential.
Exponential.

For upscale process i recommend DPM++2 m SDE + SGM uniform. For inpainting DPM++ 2m + SGM uniform with 0.25-0.3 denoise and 20-30 steps (if you use forge neo/reforge/forge classic - use formula: steps*denoise = 20-30. 90 steps work fine).

Model can handle large tensor resolution (1536x1536), and i can recommend this on general use if you making composition like portrait, full body standing or something wide shot. Thanks to the new predictor, the model handle with this many times better than the EPS one.

Generic positive: very as2, masterpiece, absurdres, best quality, very awa, very aesthetic, year 2024, newest, ultra-detailed, amazing quality.

Generic negative: (patreon username, patreon logo, artist logo, twitter username, signature, username, logo, watermark, copyright name, copyright symbol, web address, twitter logo, gumroad username, text, artist name, gumroad logo, artist logo, logo, blurry, blurry background, blurry foreground, blurry edges, blurry vision, bokeh, depth of field:1.2), green theme, red theme, yellow theme, blue theme, worst quality, worst aesthetic, bad quality, low quality, (bad hands:1.2), censored, bar censor, mosaic censoring, artistic error, bad proportions, bad perspective, wrong hand, bad reflections, too many fingers, too many hands, too many legs, wrong feet, bad feet, extra digits, missing fingers,bad anatomy,anime screenshot, anime coloring, abstract, ai-generated, lossy-lossless, jpeg artifacts, scan artifacts,

===========================================================

I made cyberfix for falafel mix by ds27sd

This checkpoint has the same reason and goals as the original cyberfix, as well as the creation method. Many thanks to the original author.

Recommended settings: CFG 5, sampler euler A AYS, 28-35 steps.

Original description

Original model: https://civitai.com/models/913998/noobaicyberfix?modelVersionId=1022962

NoobaiCyberFix

Purpose of this merge

The purpose of this model is to fix the anatomy problems of model NoobAI, while keeping 100% of its knowledge and style. NoobAI-Epsilon075 is a great model, it knows almost all characters and styles from Danbooru, but it has some anatomy issues and this merge completely solves this problem. If you use NoobAI-Epsilon075 for SFW purposes, you will hardly notice any difference between this merge and the original model, because the goal was to keep all the knowledge of NoobAI-Epsilon075, and visually this model is almost indistinguishable from NoobAI-Epsilon075. But if you are a man of culture and like Pony models, you will definitely like this one, because this merge gives the model a very good knowledge of anatomy, almost at Pony level

Recommended negative prompt

lowres, (worst quality, low quality, bad anatomy, bad hands:1.3), abstract, signature

Recipe of the merge

1 Create temporary model

TempModel = NoobAI-Epsilon075 + cyberrealisticXL_v4 - stable-diffusion-xl-base-1.0

2 Remove CLIP from TempModel and replace it by CLIP from NoobAI-Epsilon075

I did this step in ComfyUI.

Here is the workflow
https://civitai.com/models/991574?modelVersionId=1110897

Logic behind this merge

cyberrealisticXL_v31 is a general purpose model without any specific style, just like stable-diffusion-xl-base-1.0, and so if you replace the stable-diffusion-xl-base-1.0 weights with cyberrealisticXL_v31 weights in any model, it shouldn't change the model's style and its knowledge much. But at the same time, cyberrealisticXL_v31 has much better anatomy knowledge than sd-xl-1.0, and because of this, it should greatly improve the anatomy knowledge of any model, and in the case of NoobaiCyberFix, it worked exactly as I expected. In theory, you can use this recipe to improve the anatomy knowledge of any model that needs it

Big thanks to the creators of these models!

NoobAI

Illustrious-XL

cyberrealisticXL_v4

stable-diffusion-xl-base-1.0

Search AI Models & Addons

very few results