Updated the Prompting Guide
For business inquires, commercial licensing, custom models, and consultation contact me under [email protected]
Join Juggernaut now on X/Twitter
Juggernaut Ragnarok on RunDiffusion
Juggernaut XI & XII on RunDiffusion
Prompting Guide for Juggernaut Ragnarok by Adam
Prompting Guide by Adam for XI & XII
A big thanks goes to RunDiffusion and Adam, who diligently helped me make it work :) (Leave some love for them ;) )
Hey everyone,
It’s been 8 months since the last version was released here on CivitAI.
Of course, I haven’t been idle during that time . I completed several projects to ensure I’d have the financial means to keep exploring new architectures and possibly do full finetunes on them in the future.
Juggernaut Flux (and its many sub-variants) was a ton of work, but ultimately, I’ve wrapped that chapter up. The training process gave me way too many headaches. To keep my sanity, I spent my spare time working on Juggernaut SDXL with the hope of maybe releasing one final version for you all.
And that day has finally come. :)
Juggernaut Ragnarok has improved in many areas: photorealism, digital painting, poses, hands, feet, and much more.
That said, it’s still an SDXL model, and I don’t recommend comparing it to models like Flux, Reve, or Sora. For example, it still has limitations when it comes to text rendering or faces at a distance.
I recommend using it as part of a pipeline for your projects. Example setup:
FluxDev / Pixelwave / Jug Flux Pro → Juggernaut Ragnarok
A quick personal note about Juggernaut:
Honestly, I don’t know what comes next.
After the release of Sora and similar tools, the open-source image generation space feels a bit dull in comparison.
Nothing has really excited me enough to dive back into training (yes, I’m talking about HiDream too).
So I’m seeing Juggernaut Ragnarok as a kind of farewell, especially since it’s unclear where things are headed with CivitAI in general.
(You can download all Juggernaut versions from HuggingFace, by the way.)
Last but not least:
Have fun with the model, share your creations, and good luck with your projects!
And in case you’re wondering: Yes, you can do anything you want with Juggernaut : merge it, train it, sell the image outputs, etc.
Just a simple shoutout is all I ask. :)
And now, here are the recommended settings:
Recommended Settings(VAE is baked in):
Res: 832*1216 (For Portrait, but any SDXL Res will work fine)
Sampler: DPM++ 2M SDE
Steps: 30-40
CFG: 3-6 (less is a bit more realistic)
Negative: Start with no negative, and add afterwards the Stuff you don´t wanna see in that image.
VAE is already Baked In
HiRes: 4xNMKD-Siax_200k with 15 Steps and 0.3 Denoise + 1.5 Upscale
And now, have fun trying it out. As always, I'm eagerly waiting for your pictures in the Gallery :)
If you liked the model, please leave a Like. In the end, that's what helps me the most as a creator on CivitAI. :)
Last but not least, I'd like to thank a few people without whom Juggernaut XL probably wouldn't have come to fruition:
Dreamlook.AI (Trained 3 Side Sets)
Description
Overall 100k more Steps.
Started to add different body types (Skinny and Chubby/Fat)
Minor Quality Upgrade for the whole Model
FAQ
Comments (21)
Amazing work as always!
Is the VAE the normal SDXL VAE, or have you done anything special with it?
It´s the normal VAE from Stability
Great work! I tested some of my prompts and I got way better images, more details and better imaginary results. Awesome
It seems unfair to leave a review when only v3 is released, however I wonder if anyone else notices photo portraits too often have a cleft chin ("butt chin") in v3? Especially after img2img step. Several SD1.5 checkpoints had this issue and it would be a shame if it got trained into SDXL as well. When you notice it, you can't unsee it! I need to test more though, possible that the aspect ratios I am using is biasing results, or something similar.
Keep up the good work though, as always thank you for your efforts, Juggernaut is my favorite SD1.5 photographic checkpoint.
Chins are not specially tagged in the Dataset. I just took a quick look if there is some kind of bias with the chin. But i couldnt see a bias there, very different chins in the dataset.
So it was prob (or hopefully) just bad luck with Generations.
But i am gonna tag some Chins with the next Update. If their is a bias, it should be fixed afterwards :)
Thanks for letting me know :)
@KandooAI Thank you for checking 😊 I look forward to what you can do with SDXL, I'm sure it will be amazing! I struggled just training character loras, so I appreciate you sharing your expert work here.
If I find a reproducible test case I will put my ComfyUI json on pastebin in case that is helpful.
@halilzeus113 It´s alway helpful when People tell me what doesnt look/work that good, there is no other way for me to improve :D
I know the Struggle to train on SDXL, even now i still train finetunes/LoRA´s that goes straight into the bin :D
You Guys just see the "Endproduct", but i can assure you that i created way more failed Models/LoRA´s then good working ones :D
In Jugger2 and 3, every time my custom face LoRa weights get right about to where they need to be for likeness, the butt chin starts showing up. No clefts in any of the source imagery. I'm about to try with 4 and see how it does, either way, been a huge fan, since the original 1.5 Juggernaut, keep up the great work!
Hi, would it be possible to ask how many images this was trained on and the batch size?
Thats not easy to answer, since their are right now 3 Sets in the Mix that i trained. The BaseModel has roundabout 5k Images, the First Side Model has 500 Images and the Second Side Model has also 500 Images.
So we are overall at 6k images.
The Base Model was trained with Batch Size 1 (I got Bucketing Problems), the first side set i am not sure about (it was trained on dreamlook.ai ) and the current Update Set was on Batch Size 8
@wiiz Just wanted to give u an update about Set 2 from Dreamlook. That was also Batch Size 1
Excellent model. It consistently provides noticeable quality improvements to all of the Loras I created on SDXL 1.0 base.
That´s really good to know since i rarely use LoRA´s . I´ll just tested my own LoRA and the one from Chillpixel, and there also worked pretty fine :)
it's my favorite sdxl model! ...but fingers, they are really bad 99% 😭️
It will get better down the line. Right now i am working on subjects that are more important for me. But Hands will be adressed in some of the later versions. Until then it gets minor improvments with every update
@KandooAI thank you in any case, model is awesome! will be wait for new updates
I don't wanna be that guy but you can try Realities Edge XL model here on Civit, is has really good hands and nice crisp output! <3 Sorry KandooAI if you think it is out of line!
How was SD XL finetuned? Was it using ED2?
Kohya_ss
This is the best XL model so far! Why? It allows me to add my self-trained LoRAs without being overly influenced by the base model. Meanwhile, all other models seem to overwrite my Face-LoRAs with their own facial details or unnecessary skin features (I don't need freckles, red cheeks, or intense facial irregularities all the time). I'm not sure if the author intended this behavior, but I hope it won't change in future updates.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.