If you like what I do and want to support me you can head over to Ko-fi!
Supporting me there can access links to unreleased and EA models.
Diving into Janku's Unholy Desire to provide a merge which brings us endless mayhem and fun...
JANKU v5 Trained + NoobAI + RouWei Illustrious XL + Diving Illustrious Niji Sketch + Unholy Desire Mix Sinister Aesthetic V5. = AniMayhem!
My second Checkpoint merge is a departure from the semi realistic and further into the anime realm.
I won't try and tell you how to live your life, but I've been very successful with the following
Can go as low as 14 steps and cfg of 2 with decent results using Euler a / normal.
DPM++ 2M and 2M SDE
Karras
30-40 steps
Cfg under 5
Or
Euler A
Normal
24 steps
cfg 4.5 or lower
Can go as low as 14 steps and cfg of 2 with decent results.
As always - adetailer or the like for distance. And use your pos/neg embeddings!!
I prefer Lazy Embeddings for all - however you may have your own preference.
Description
Better low lighting and better dark scenes.
Additional merge of Janku 7.77
Some other little things :)
FAQ
Comments (17)
Thank you for creating such a perfect model! I have just supported you on Ko-fi.😊😊
Thank you!
New model......TAKE ALL MY BUZZZ!!!!
Here we go!! thanks!!
I'm a huge fan of your work! Could you explain the differences between v3.0 and v4.0?
Additional merge of Janku 7.77 into previous bases
- some lighting adjustments for better low light and dark images.
Better character recognition.
@Whistler_ai Thank you so much for the reply!
keep going man. i love your work
The new version feels like a downgrade imo :(
Alot more images come out fuzzy/blurry or oversaturated. Was my fav checkpoint but now I've got to find a new one.
Feel free to share some examples.
Your profile states you make "realistic furry girls" so I'm not sure my checkpoints are what you should be using anyway...
@Whistler_ai Yeah, problem is most furry checkpoints are for some reason, terrible at making fur or at making faces. The old versions worked really well in those regards. I remixed old images of mine (I only use Civitai to make images), and here are some old version vs current version images, using the same loras/configs (all sfw)
https://civitai.red/posts/29797544
Everything is just super washed out now.
@Led123 Well, feel free to keep using V3 then - it's a bit of an edge case - with that many Loras it could be anything that is causing it within the model now.
@Whistler_ai problem is Civitai doesnt allow you to use old versions, only the most current one. But I get I'm not making the type of images your checkpoint was designed for, so it's just disappointing.
I find v4 intend to generate images with faces in shadow. Is there any way to fix it?
@Whistler_ai First of all, thank you so much for creating Pale Rider. It has genuinely become one of my favorite Illustrious checkpoints, and after testing many different models, I still keep coming back to it because of its beautiful style, clean composition, and overall consistency.
I just wanted to share a small idea that I hope you might find interesting for future versions.
If you ever decide to continue improving Pale Rider, I think it could become something truly exceptional by borrowing inspiration from the strengths of a few other checkpoints—without losing its own unique identity.
For example, NoobAI has incredibly strong prompt understanding and follows complex prompts remarkably well, while Ultimate Hentai Anime RX (T-Rex) produces beautiful lighting, rich shading, and very cinematic rendering.
Imagine Pale Rider keeping its own gorgeous style while gaining even stronger prompt fidelity, richer lighting, and slightly improved consistency for hands, feet, and fine anatomy. I honestly believe it could become one of the very best Illustrious checkpoints available.
Of course, this is only a friendly suggestion from someone who truly enjoys your work. Thank you again for all the time and passion you've invested into this project. I'm genuinely excited to see whatever you create next, and I'll definitely continue supporting your future releases.
@Led123 You should be able to choose V3 without issue instead of V4 - it will just default to v4 unless you swap to v3 directly.
@Whistler_ai that's amazing! you're a real one, ty so much










