I wanted to train a LoRa able to render all the characters from the show (like I did for SD1.5) but I failed. Instead it works quite well for giving a cute cartoon look to your characters however.
Copy it to your models/Lora/ folder and add <lora:what_about_mimi_style_illustrious_by_dom83:0.8> to your prompt.
The model I used for all my example pictures is IllustriousModelIllusobXl_v10.
I am sharing the dataset, though not very good quality, if you want to use it to train a version for Pony or whatever, feel free to do so.
Description
FAQ
Comments (7)
Looks cute despite not doing what you planned for.
Did you change your triggers from the 1.5 version? Those were really generic and would explain why they don't call characters properly.
I used the very same training data including the txt files with descriptions, so I expected a similar result. Maybe it has to do with the size of the LoRa, maybe it mixes everything up because it's too small, I don't know.
@Dom83 I doubt it's the size: Both styles I've done will replicate the tagged character (some better than others, but there's clearly an attempt for all but the least tagged characters) and the TD for the Gundam X one was over a GB. It almost certainly is the non-unique trigger words.
@NanashiAnon Thanks for the advice, next time I will pay attention to that.
Can we please get a PonyXL version?
I tried over a dozen times to train various LoRa for Pony or base XL and I always failed for some reason. :( I will share the training data however, someone may use it to create a PonyXL version.
Can Wedding please get a wai-ani illustrious SDXL and illustrious please
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.