All images out of the box, no postprocessing. Model
works best with high sampling levels. Larger resolutions
are also working good. If you have enough vram sometimes
no upscaler needed.
Recommendations (default use):
Optimum CFG-Scale: 7
Optimum Sampling Level: 150
Sampler Best Results:
Euler a
DPM++ 2S a Karras
DPM++ 3M SDE
DPM2 a Karras
Restart
The model is highly optimized on the settings
above, so you should leave out quality prompts
first, if not necessary.
The model is build of the latest trained models
and lora, thanks to all community members
for their work.
I hope you enjoy and share your images.
Ultrium is always a merge of the best and/or latest
avaiable models. So each Ultrium is different than
its predecessors. It's your decision if or which
Ultrium is best for you.
Ultrium 9.333 will be the last Ultrium for SDXL 1.0,
I will see for which AI I continue my work.
Description
new Version
FAQ
Comments (8)
i wanted to get v5 again because 7 is all cartoon based now. And I can't, it doesn't seem to be there now. Why are the older versions photo based and the newer ones cartoon?
Although I really like v7, it feels like a different model than v5 and before. Might it not work better to have 2 versions? A realistic and a bit more towards 2,5D like this one?
seems that v4 was peak Ultrium
The new Versions are always different, because they are created from a new base of trained models. If I refine a model it will become a Subversion of the same Version. Sadly training some aspects more always leads to other aspects getting worse, with the current SDXL it is nearly impossible to create a model that can handle all aspects humans / fantasy / animals / architecture /anime /photorealism etc. with the same quality. V7 is stubborn in some aspects and V4 had other weaknesses. I have not given up on a nearly full balanced model yet. :) Other versions or specialized versions may be a good idea - but you must find the time to build them - if you have a "normal" life as well.
@DSchroeder well like i said there is nothing wrong with v7. Its actually really great. Just that it feels like a completely different model. As in old models were more photorealistic and this one is a bit more cartoonlike. And of course: thanks for the work you put into it.
you sir are a hero amungust men. Be blessed.
to everyone saying it looks cartoony. I'm not getting that at all with this new version. Could it be your prompts or the program you're using? I don't know it looks on Automatic1111 or the others but it looks GREAT on ComfyUI. If you're not using ComfyUI, I recommended trying it out! It looks scary but it's easy to learn. If you are using Comfy and getting cartoon results, I honestly don't know but I'm willing to try and help.
Can your model generate more than one person?
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.