IffyMix V6
A true continuation from V 5.0 (not the HD versions or Re:mixed)
Better posing
About the same prompt adherence
Ever so slightly more detailed cartoonish look. But not by much
Lora compatibility high (as no LoRA's that interact with visuals are implemented)Notice your prompt used in the cover images? DM me and ill send you buzz
IffyMix Re:Mixed v2.0
A continuation of polishing the Re:Mixed version.
smoother, cleaner look.
wider style shifting.
artist tags work even better.
prompt adherence is low end of top tier.
natural language works when guided with tags.
Lora compatability is very high.IffyMix Re:Mixed v1.0
A complete remerge of the base models.
updated to to be merged with more modern models
Heavier toon look
Hi-Res Fix - No more orange tint
IffyMix V5.0
Its finally here. Iffymix with a base made of V2,3 and HD1
This has a nicer, smoother pastel like effect.
I feel that this mix gives surprisingly creative results.
Euler A, LCM, and UniPC samplers only!
IffyMix HD Remix V2 - EULER / EULER A ONLY
Various EULER A methods work! ( Automatic / SGM / Beta )
The "HD" like remix using IffyMix v2 + Iffmix v3 as the base.
!!!NEW RULE 34 V2 DATASET INCLUDED!!!
Contrast + Brightness precisely fixed.
Solid prompt adherence
This version continues with the vibrant cartoonish aesthetic.
I would say this version is slightly less 2.5D by just a smidge.
Background detail is improved mildlyIffyMix HD Remix V1 - EULER A IS PREFERED.
An "HD" like remake of iffymix (contains v1-v4)
Produces a clean, vivid, and a wonderfully polished cartoon aesthetic with a mild painterly effect.IffyMix V4 - STRICTLY NON DPM++ 2M KARRAS
Launching iffyMix...
Update found. Apply? [y/n]
V4 Released! (Online starting 6/26/25 @ 19:00)!!!! Currently Online !!!!V4 Update + Preview + Suggestions + unreleased goodies
https://civarchive.com/articles/15470/iffymix-v4-suggestions
Click to help shape and polish off Version 4!
V3 released
Some visual refinements made, added some lora's to increase anatomy, pose and dynamic angles. Increased R34 Lora strength. More predictable results, which sadly means some random creativity has been lost. Background detail has improved. Illustrious v2.0 has been merged in so higher quality gens all the way up to 1536x1536 can be done. Mileage may vary as this is experimental.
V2 Released!
This version adds some visual refinements as well as a huge update in tokens. I've merged in R34 tags/database, which was curated by iamddlta, creator of Pornmaster and it's variants.
This not only has wildly increased how good it gens explicit content, but the accuracy of anatomy and prompt adherence have also skyrocketed.
IffyMix - A Revamp of the best version of YiffyMix.........V51
Whats Different about this version besides its V51 + NoobAI?
Unbaked horrendously noisy lightning step before merging.
Completely prunes noisy, signature filled model, ChromaMix and its variants.
Merged with high quality models.
Restored its natural style's cartoonish look.
Parameters:
CFG - 4.5, seems to be the sweet spot.
Steps: 20 -30
Euler A
Tons of artist tags that sucked on NoobAI, do awesome on here. Totally unintentional, but its something I came across.
Quick Start Prompt
POS+
masterpiece,best quality,amazing quality, absurdres, anthroNEG-
bad quality,worst quality,worst detail,sketch,censor, text, ui, signature, patreon,Description
Illustrious 2.0 base
(still also has Noob AI 1.1 EPS knowledge as well)
Re-balanced composition & creativeness
Injected e621 v2.0 patch for Illustrious
Extra models merged in and rebalanced with R34 lora
SPO added with custom Block wieghted merge. Runs at 1.8 strength.
FAQ
Comments (22)
Speaking of contaminants (like the "removed baked in sdxl lightning")
It seems as though there's an old version of the "NoobAI stabilizer LoRA" baked into this checkpoint, probably always has been. (at least since v2)
https://civitai.com/models/971952?modelVersionId=1230684
Seems to be nbep11 v0.96 (though subtracting v0.114 is also interesting)
strengths around -0.6 to -0.8 (-0.7?) clean things up.
cheers
It must be from one the models merged in? Not sure which one tho.
But how did you figure that out?
@LatteLeopard I just simply read from the author that applying a new version and having it artifact like crazy is likely because there's a different, older version baked into whatever "Mystery Checkpoint" you're attempting to apply it to.
It hadn't even crossed my mind that such things would be stuffed into checkpoints with no disclosure, but so it goes.
So I just took a wild guess that the "NoobAI" parts of this thing's genome likely contained a contaminated model, tried subtracting an old "nbep11" version I had lying around - which helped substantially with applying a fresh new illustrious stabilizer version without crazy artifacting.
then it was simply a matter of dialing in how much negative weight seems to be beneficial, and which seems to distort composition (too high negative) - which was roughly in the -0.7 to -0.78 range
and then just trying all the different versions on a tricky prompt to see which one "did it best"; from which v0.96 and v0.114 had the best outcomes.
this kinda stuff is like being a fish in water to me, good thing I stay safely away from it
btw i can highly recommend trying a Radeon 9060 XT if you wanna install gentoo just to do image generation at 1/4th the speed of a dying rtx 3060, very good user experience (plays videogames good though)
EEYEEAASSSS THAT CREATIVITY I LOVED HAS RETURNED IN V4
I specifically used your prompt to test how well it worked lol
@LatteLeopard les gooo :D
This is an awesome model! I love it! :D
Ty!!! ♥️
What’s the difference between this and Yiff in Hell? Just curious.
Yiff in Hell is a 2.5 model with painterly/impressionist art.
IffyMix is capable of 2d,2.5D and 3D styles for highly cartoonish art with thicklines and vivid colors
Banana Splitz XL is a 2d/2.5D model with a focus on western furry and kemono art
Stormtooh XL is a 2.5 model with extremely high semi realistic detail with vivid and ornate details
@LatteLeopard Do you have anime style furry model ?
@DevillUKU I guess Banana Splitz tbh.
@LatteLeopard Any chance you could somehow pin that, or add that information to the descriptions? Super useful knowing what vision the model creator had when releasing each model version.
@MysticFrostwind I usally describe each model as version come out. Are you asking me to make something that shows EVERY models description by me and then post that model information on each model?
Because I do individually write my visions were for the models, and the describe their style.
The model shows some improvement in character expression. The contrast is higher, which might be preferable depending on individual taste. However, background elements frequently appear cluttered, and lack cohesion, resulting in a distorted visual effect. While some results stand out, many don’t meet the same standard. With continued development, it could become even better.
Youre absolutely right. I didnt expect the background to get like when generating on civitai.
Sometimes I forget that I have to make sure it translates well to A1111, Comfyui and civitai's generator!
Dont worry I am currently working on fixing this. It should never be a problem again in the future.
I really like this model, used it a lot and recognizes most art styles just with the model alone. I do think it should recognize the gremlins from Spiral Knights better as its a hit or miss, but closest to ones you see in game and fan art.
Thanks dude! I appreciate your review!
Unfortunately for the germlins/spiral knight stuff, I have no control over if it shows up or not.
I would either have to bake a lora for it into this model or do a checkpoint training. Which would make this a whole new architecture model, Like NoobAI, illustrious, Pony.
These models like Nova Furry, Indigo, YiffyMix, and others are all model merges.
That means we are finetuning the weights of the U-Net by combining, subtracting, multiplying and adding the difference between multiple models.
So basically Nova furry, Indigo, Mol Keun, WAI-NSFW-illustrious are ALL identical in training data. Whats changing is the style, prompt adherence and composition.
Easiest way i can put it is by using this video game comparison.
Checkpoint Trained = A new video game
Checkpoint Merge = Texture pack for game (think how texture packs for minecraft can be cartoonish or photorealistic.
LoRA/LyCORIS = Mods/Addons/DLC
That could be possible. I've asked since nobody has made a gremlin lora yet, figured that I make one for illustrious. Likewise, I'm new to lora making and I am preparing the dataset for the lora.
Is draft mode glitchy this week?
Im sorry! Im not really sure about that!
Im a bit fimilliar with draft mode but, Im not too sure how this model would work with it.
@LatteLeopard it used to work fine but then civitai did something now it looks glitchy