ILLUSTRIOUS GEHENNA
"I want to see you dance on the blood-dimmed tide"
V60 is even further mixed; I hope you find it good and flexible. I actually made it a few months ago but wasn't sure if I wanted to release it until now (interfusor writing this August 2026).
V50 is further improved 3d and realism, but still has a good way to render cartoony pics.
V40 is merged with Illustrious 2.0 and I've further (hopefully) made 3d and realism better if you prompt for it.
V31 is merged with Illustrious 1.1 and I've also tried to make it more "3d-friendly", so do try it out!
V20 is merged with Illustrious 1.0. It's gonna be a very similar look but over all I think everything looks a little better in the new version.
Just a personal mix. Like original Gehenna for Pony, I suggest using tags like realistic -- or 3d, render either in Positive or Negative to change it up well depending on what you fancy.
My advised settings:
Sampling method: Euler a
Sampling steps: 20-30
Width: 832
Height: 1216
CFG Scale: 5-7
For upscaling I like: R-ESRGAN 4x+ Anime6B
These can be placed in Positive before your prompt for some quality enhancing:
masterpiece, best quality, amazing quality, very aesthetic, absurdres, newest
I also find this simple tag to be very useful: lips
And as usual I always recommend Hi-Res Fix, and Adetailer for the faces.
Bonus! How to make Gehenna Girl: 1girl, colored skin, white skin, white hair, side swept hair, long hair, colored sclera, black sclera, yellow eyes, black dress, black lips, black eyeshadow
Or just try the character LoRA here!
Support me on Ko-Fi if you like. My stuff will always be free but if you're interested in requesting a model I'd appreciate a donation. Buzz is fine too!
Description
FAQ
Comments (12)
How just HOW did you make such an amazing checkpoint like this ? I'm using this one since a while and it never failed to impressed me !
Oh thank you! I'm very happy you like it.
The best model! Truly magic! Every picture comes awesome.
Here's version 4 already out and now it's not clear where it's moving, 3d or semi-realism. I tried all versions from 1 to 4 and I didn't like them. For me GEHENNA PONY remains the best.
Fair enough. I'm glad you like the Pony one! I just prefer IL because it's more flexible. If you want, you can try using my new 3d-ish lora for IL: https://civitai.com/models/1549526/fusors-3d-ish-style-illustrious?modelVersionId=1753298
I love your work and Gehenna models, the best thing I found for myself, would like to see your models in versions SDXL, SD 1.5 and maybe Flux
Thank you so much. Is SD 1.5 and regular SDXL still used a lot? FLUX I admit I haven't really touched that much.
@interfusor I recently started working with Stable Diffusion and found that the characters I want to create are only available in specific versions like SD 1.5 and base SDXL. My temporary solution was to train custom LoRAs of these characters myself using Gehenna, which I currently consider the best model for myself.
1. I like the overall (how it handle various) style/s!
2. My only... "drawback" is, it doesnt do the backgrounds i prompt for, still nice ones, but not what I intend for most of my images. (but i often do want "complex" backgrounds, sooo... idk)
3. It even does nice 21:9 images - which i like, alas.. 2. applies.
Ah yeah I think I know what you mean, not sure how I would improve complexity of the backgrounds if I made a v5 but I'll keep it in mind, thanks!
It might lack in detail, but it seems to handle anatomy better than most models.
It doesn't give you that AI crotch/hip crease they all tend to give.
Details
Files
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.