This checkpoint starts with a major focus on Porn.
Important Usage Tips to get similar results like my images.
Add Realism_Illustrious_Negative_Embedding at the beginning of your negative prompt.
Add Realism_Illustrious_Positive_Embedding at the beginning of your prompt.
Join me on my Patreon for exclusive perks and early access to unique resources.
To discuss custom Lora's or models, feel free to connect on Discord.
👍 Like this model to keep me motivated and inspired to create more!
💬 Drop a comment and let me know what you'd love to see next.
🌟 Review this model to help me improve and make even better creations.
🔔 Hit that notification bell to stay updated with my latest models and updates!
Description
Whats New ?
More Details and some other fixes.
Highly Recommended:
DPM++ SDE
DPM++ 2M SDE
DPM++ 2S a
DPM2 a
DPM++ 3M *
Euler A
Optimal Settings:
Steps: Aim for 27 to 36 for balanced results.
CFG Scale: A range of 4 to 5 works best.
Clip Skip: Set to 1 for ideal output.
Resolution: Generate images at any SDXL resolution
Additional Notes:
For a smooth and cleaner aesthetic, try the Euler A sampler.
To achieve sharper and more detailed outputs, use SDE-based samplers like DPM++ SDE or DPM++ 2M SDE.
Important Usage Tips to get similar results like my images.
Add Realism_Illustrious_Negative_Embedding at the beginning of your negative prompt.
Add Realism_Illustrious_Positive_Embedding at the beginning of your prompt.
Join me on my Patreon for exclusive perks and early access to unique resources.
To discuss custom Lora's or models, feel free to connect on Discord.
FAQ
Comments (15)
Excellent results with v3.0_FP16. Great Work. Thank You.
Thanks for the feedback.
imho it's a little bit shameless to ask for Buzz when you know the average Joe can't buy any.
But, at least I tried the v3 model and it was good.
Totally get where you're coming from 😊
Though honestly, not many “average Joes” hang around here — most folks are deep into AI and curious enough to explore what’s possible 🤖
But hey, if someone feels average while scrolling Civit and dropping comments... well, that’s a different reflection altogether 😉
Anyway, glad you gave v3 a shot. Let’s keep things positive — bitterness doesn’t really train models any better.
@Stable_Yogi @Stable_Yogi Purely positive feedback weakens the gradient signal, precise error reports give a stronger direction for updates.
@DarkSideOfTheForce I think StableYogi takes feedback great, I'm just upset about the nature in which CivitAI finds itself in right now. They are actively encouraging inflated prices for models and that's not any creators fault, they just roll down the tracks that are laid out for them.
But that doesn't mean I can't think it's still a bit shameless to ask for so much Buzz and gate the models for several weeks if you think back to how it used to be, before the Buzz... Everything was free yet everyone shared, nobody had to buy access and that was the golden days, from SD14 to SDXL/Pony when open-source AI was booming.
Now that the model is available, I can review it.
When a model is improved, I somewhat expect the new model to produce something that can be at least a little bit similar to the older model.
This model produces images that's unrecognizable to the images you get in v3.0, and the creative expressions and height differences are all gone. Basically v4 became the boring version of this series by producing less creative results.
Quality and realism went up by a substantial margin, the anime eyes and colors that was bleeding through in the v3 version are nearly gone but it's unfortunate that creativity went down the drain with it.
Side Note:
There used to be a field on the right-side of these model cards where creators could write what's new, I don't see it. Did Civit remove that feature or I'm just blind?
Nevermind, I'm blind.
Thanks for the detailed thoughts — really appreciate the time you took to compare versions 🙏
You're right — v4 is a shift. I’ve intentionally pushed toward more realism, cleaning up things like the anime eyes and color bleed that crept into v3. The trade-off? Some of the spontaneous, quirky “creative” elements naturally took a backseat. It’s not a downgrade — just a different direction for those looking for a cleaner, more lifelike output.
And yep, I get that you'd expect some visual consistency across versions. But each new release brings in fresh data and tuning — which can change the “feel.” Forcing backward compatibility just to match a previous vibe risks overburdening the model. If v3 fits your needs better, totally fine to stick with it — both versions are here for different use cases.
The goal is to hit a realism threshold first. Once that’s stable, future versions will be more consistent in feel, but still evolving.
Oh — and you're definitely not blind 😂
The “What’s New” section is still there on the right — maybe it just didn’t load right away. Happens to the best of us.
Seems like these models all suffer from very heavy over fitting. Like no matter what you do, you keep getting the same female face over and over again.
did you tried to add __lube/allofit__ ?
There are solutions to fight this. Try looking for LoRAs like Real Humans, Ugly faces, Sameface (used at negative weights). They help a lot once you've found your sweet spot using them alltogether 🙌 Keep up
yes that happened to me
I posted a comment on another model-card out of confusion because I couldn't find PornyIllustrious, until GPT explained that this is in fact the model-card but under a different name.
Why is this model cursed?
You went from having a creative model with improving quality all the way up until v3, then with v4 the creativeness disappeared and now neither v3 nor v4 exists, and you renamed the whole model-card. Why couldn't you just make PornCraft its own thing and leave PornyIllustrious alone?
For goodness sake CivitAI is a complete dump as it is without creators making messes of their model-cards.
I loved every version of PornyIllustrious until you released v4, that's when the model became stiff and obnoxious to prompt with. Sure, I'm lazy, I always make short prompts with as low CFG as possible because I want the model to compensate for my lack of creativity, that's why I loved PornyIllustrious v1-3.
Please bring them back and improve upon the v3 model. Just don't bring v4 back cuz that version killed the model.
Hi
Thanks for writing and explaining your concerns.
I will look into this.
Thank you ! 🙏
Fantastic model. Thanks for sharing it.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.