The superhq model has training and blends of other models, a broad and general purpose model, it will generate images of drawings, cartoons, draw, anime, manga, hentai, porn, it also generates semi-realistic images with some loras models.
I recommend using the Adetailer extension.
Use this extension to fix hand errors:
https://github.com/licyk/advanced_euler_sampler_extension
It is strictly prohibited to distribute, inside and outside civit ai, this template or post it anywhere or on hosting services, on third party image generation services, you can use it for personal use, if you know who distributed it without my consent , report it immediately.
Resolution settings:
Standard resolutions 512 x 768, 768 x 512, 540 x 960, 640 x 960, 960x640 and 1024 x1024
Resolution: 540x540 select the Hires.fix checkbox, result 1080 x 1080 pixels
Resolution: 540x960 select the Hires.fix checkbox, result 1080 x 1920 pixels
Use these recommended settings for generation:
Sampling method: Euler_Max
Sampling steps: 30-50
CFG Scale: 2.0 - 7.0
Skip clip: 1-2
Sampling method: DPM++2M KARRAS
Sampling steps: 30-50
CFG Scale: 2.0 - 7.0
Skip clip: 1-2
Sampling method: DPM++2m
Sampling steps: 30-50
CFG Scale: 2.0 - 7.0
Skip clip: 1-2
Use the recommended negative prompt:
artifacts, low quality, blurry, deformed, distorted, extra limbs, watermark, text, logo, unrealistic, out of frame, low contrast, low resolution, malformed background, malformed fingers, malformed body, malformed arms, malformed face, duplicate arms, blurry face, distorted body, asymmetry, low-quality details, extra limbs, extra fingers, extra eyes, strange eyes, blurry details, bad anatomy, unnatural posture, bad proportions, deformed, low resolution, poor lighting, grainy, unrealistic textures, overly sharp, cartoonish, watermark, text, unrealistic shadows, weird reflections, poorly rendered fingers, unnatural teeth, extra facial features, wrong facial structure
(CyberRealistic_Negative-neg), cartoon, painting, illustration, (grayscale:1.4), (worst quality:2), (low quality:2), (normal quality:2), ugly, fat, 3D rendering, asian;
BadDream, easynegative, UnrealisticDream, bad-hands-5, Asian-Less-Neg, (loli, petite, short, childlike, teen), floating dicks, chubby, fat, BREAK (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, multiple limbs, extra limbs, deformed limbs, normal quality, ((monochrome)), ((grayscale)), FastNegativeV2, fcNeg,
EasyNegative, fat, paintings, sketches, (worst quality:2), (low quality:2), (normal quality:2), lowres, ((monochrome)), ((grayscale)), bad anatomy, text, error, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, bad feet, poorly drawn face, bad proportions, gross proportions, ng_deepnegative_v1_75t, badhandsv5-neg;
Description
v1. Model in implementation environment, this is a preview, still undergoing revisions, I must update the sampling method!
FAQ
Comments (12)
Something went wrong.
When I use sampling dpm++2m karras and when I use sampling dpm++2m, steps: 30, cfg 7, skip 1
2.9GB Pruned model - returns color noise.
6.11GB Pruned model and Full model return images with poor color palette.
I'll check the pruned versions, ok, thanks
Oh, I figured out the color problem. negative prompt "(worst quality:2), (low quality:2), (normal quality:2)" distorts the colors.
@skanbl4 the models have been corrected and are now available
@Dxdesignia, negative prompt "(worst quality:2), (low quality:2), (normal quality:2)" still distorts colors. This is the correct result
@skanbl4 I'll make an update and replace this file I ask you to wait ok
Is this really an SD 1.5 model? It's 6 GB huge vs all other SD 1.5 models which are ~2 GB
This model has an incredible 11.5 gigabytes final version fp32 the others are fp32 pruned and fp16, the data set is very large indeed
@Dxdesignia Yea, but is it really SD 1.5? SDXL/Pony models are usually 6 GB pruned. SD 1.5 pruned models are 2 GB in size. Which models were used in the merge?
@TScitai The base model is anime xl, design_pixar, plus other models that I am developing, answering your question models in fp32 format in sd 1.5 can run the 5.28 gigabytes
Hello, having an error on easy diffusion :
'Attention' object has no attribute 'to_to_k'
This model is working with the stable diffusion forge, latest version, I do not recommend easy diffusion, this model will not be updated for now!
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.



















