The new version of MBBXL has been trained on >18000 training images in over 18000 steps. It's probably the most significant fine-tune of SDXL so far and the one that will give you noticeably different results from SDXL for every prompt.
The model tends towards a "magical realism" look, not quite photo-realistic but very clean and well defined.
Very flexible model that looks good with most prompts (including simple prompts).
Use of the SDXL Refiner is not recommended with this model. Shouldn't be needed either.
Recommended negative:
blurry, blurred detail, poorly hands, poorly face, enhanced hands, missing fingers, mutated hands, fused fingers, deformed, malformed limbs, disfigured, watermarked, text, extremely grainy, very chromatic aberration, oversaturated
Recommended settings: 1024x1024 or 1336x768.
DPM++ SDE 30 steps or more.
Works very well with hires.fix, try 768x768 upscaled to 2x!
Limitations:
Skin textures are not strong, to come later probably as an add-on lora.
Description
Big update to the model, with an additional 135000 steps on 15000 training images.
The model is much more capable and very different from the base SDXL model (and also very different from all the other currently released SDXL fine tunes).
If you want a model that's actually fun to try out and compare with the SDXL base model, and not mostly the same, this is the model for you!
FAQ
Comments (17)
This model has a sense of humor.
You should see some of my test models 馃ぃ But it is great when the models can make us chuckle 馃榿
Just ad freckles in neg?
I dunno still downloading. Models always appreciated.
Guys... I don't know if you noticed, but there is "standard beauty and sexuality" in my negativ械, which makes such generations completely justified.)
The SDXL associative map is very different from 1.5 and is fraught with many surprises.
@Temp134664333聽Just so! In fact in my testing with newer models now I'm starting out with no Neg at all, and only add them when clearly needed. Opens up another layer of prompt comprehension it turns out, the negatives can confuse the model in odd ways.
@morbuto聽Thanks for the tip. I started remaking my negative from scratch. Many parts of it caused only distortion and nothing more.
Great model! I made a youtube Deforum clip with it (Warning, it turned out a little bit sad!) - Bark Star: https://youtu.be/--Cef97VzMs
Wow that animation is amazing and epic! No exaggeration one of the best ones I've seen made with deforum! Wouldn't say it's too sad... Thanks for sharing!
Best and most versatile model for photorealism
Can you post some examples?
Best Model I have ever used, what is this witch craft?
Thanks! A lot of preparation and work went into this model 馃榿
hey Morbuto, is there anyway to connect with you in social media?
The highest quality model at the moment. Perfectly trained.
Downloaded your model early, was impressed, but then so many other models came out or were updated ... just came back and realized how great yours is for some things... It adds a little more ... Thanks!
In one week ago everything works fine. Now it renders (with review) but at the last picture the results end in a black picture each time. No special errors! Only a black picture. I deactivated all extensions and restart but each time the same issue. Black Pictures. Any ideas?
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.