Alright, I had this project going on for a long time, and between requests and other similar models being released I hadn't gotten to it. There are no furries. I forgot about Pocahontas, and I realized a bit late, maybe someday with the rest of the decade.
Edit: I completely forgot the credits. @Yoshi__ helped with dataset and a few images on the gallery. @bluvoll helped me a lot to figure it out trainings with CAME/REX and choosing the right algorithm for the job. So, thanks to both!
I have a kofi☕, that is the only way of supporting me and keep my trainings running now. I also take commissions, so if you are interested on a training, contact me or visit my kofi or listings!
About the LoRA:
The LoRA is really a LoCON that includes 4 styles, but the main tag also works as a cartoon one. It cannot do any character by itself, this is a style, not a character pack. I trained separate character LoRAs for this same reason.
The main tag is ict style, the tag flat color can be used as helper along with nose, lips, then the secondary tags are as follows:
The Little Mermaid:
ltme(my favorite and the best flat color).The Beauty and The Beast:
tbatbAladdin:
aladdn. This one is a bit oversaturated; I don't really know why but this can be worked around by lowering the CFG or using weights on the trigger, like(aladdn:0.75).The Hunchback of Notre Dame:
hunchnd. Same as Aladdin but it mutes the colors instead of making them more intense.
Excess of quality tags break the style, as it is very simplistic. For anime girls LoRAs I recommend having nose, lips, narrow waist on the prompt and this LoRA weight at least on 1.2.
That's it! Like the model so it shows for more people! Generate, leave reviews, funny images, sexy images, etc. They really help to stay motivated! Up to X images can get a pin!