SDXL
SDXL
I am admittedly kind of a noob at this, but if you enjoy my humble work, any donations are greatly appreciated: https://ko-fi.com/insult_to_ninjary
Your generosity helps me stay motivated to keep improving and to keep trying new things!
At a Glance:
Style: Anime/Illustrated
Works with LoRAs?: "Base" versions - should be fine; "Plus" and "Bonus" versions - mostly yes but maybe not certain style LoRAs
VAE: sdxl_vae is baked (but you can override if desired)
Suggested parameters: Euler a, Automatic, 30 steps, 832x1216/1216x832, CFG 5 (but feel free to experiment!)
Hires. fix: 1.5x, 30 steps, use a quality upscaler (like 4x-UltraSharpV2), denoising strength 0.5 - 0.3
Extra: use ADetailer to fix faces and/or hands if necessary
Description:
All of the model "versions" on this page are variations of roughly the same core merge. Thus, your image outputs across all the models should always be roughly in the same ballpark, and may even be very similar between certain models. But all of the models do have their own particular quirks and nuances that set them apart from each other and make them more suited for different preferences or use cases. I've put brief descriptions of each model type below, but if you want to know more about anything, just leave a comment or send me a message!
General Note: All of the "Base" versions are a little more neutral and generic, and should work well with most LoRAs. All of the "Plus" and "Bonus" versions have some LoRAs already cooked in, so they look a little bit nicer on their own, but may not be as flexible in fringe cases with certain strong style LoRAs or art style prompts (such as pixel art and so on).
Core Models:
Cute: This was the very first merge in this family. It naturally leans more towards a kind of cutesy aesthetic than all of the other models here, but you can prompt away from it if you don't like that.
Default: This is a standard anime model. It leans a bit cute, but it's not too overboard, and you can just prompt away from that aesthetic if you find it popping up too much and you don't like it. And if you do like to have some cuteness, you can also lean into it in your prompt for a stronger effect!
Special: This adds more depth and detail compared to "Default." Unprompted cuteness is drastically reduced, but cutesy styles can still be prompted for if you want them. Colors are a bit more subdued and less cartoony, but still look nice.
Matte: This adds even more depth and detail compared to "Special." That "shiny" effect that a lot of Illustrious models have has been reduced a lot, making everything look a bit more "matte" and textured. Colors can sometimes look a little too washed out for my tastes due to the added realism, but it's not terrible by any means.
"Plus" Versions:
These are versions of the respective core models that are a bit more stylized and add a bit more depth and detail. As a result of the extra stylization however, certain style prompts and style LoRAs may not always work well with these "Plus" versions. Other non-style LoRAs, such as characters, poses, etc. should work fine though!
"Bonus" Versions:
These are the "Plus" versions, but with an extra custom LoRA cooked in. They retain the added detail from the "Plus" versions, but make that detail a bit more clear and coherent. However, these versions have much less of a realistic lean than the "Plus" versions, and instead move back towards the colorful illustrated style of the base core versions, but with more stylization and often more shininess. The same notes about LoRAs in the "Plus" section apply here.
"Alt" Versions:
These are a bit more detailed and stylized than their corresponding main versions, but they also have slightly worse prompt adherence, anatomy, and character knowledge. I was hoping that these could be official second versions of all the models, but in the end I still like the originals a little better, so I decided to release these as "alternate" versions instead. These also have "Plus" and "Bonus" versions as well.
Troubleshooting:
If you're having persistent problems with anatomy (hands, fingers, etc.), fine detail, composition, and/or other jank, just try your best with all the usual fixes - different prompting (including negative prompt), use a different VAE, use adetailer, inpaint, make sure to try the above suggested parameters, etc.
Outro:
If you make anything awesome, please consider posting them to the gallery! That goes for all of my models! I really like seeing what kind of stuff you all are able to make!
If you have any questions, feel free to ask! I will try my best to help! Also, if you have any ideas for future models, please don't hesitate to share them with me!