#Update

Makes you think, the model I made for SD15 was almost 2 years ago... anyone left here from that time?

Some info about Illustrious version. It's a bit of a pain to train, since most images are less than 500px (both height and width usually).

I tried training two outfits:

The camisole and pants, where the trigger is -- j3n_c@mi
The cheerleader outfit, where the trigger is -- j3n_chr

Obviously, you might need some tags describing the outfits themselves, like purple camisole, white pants for the first one and blue shirt, white skirt for the second (also, I wouldn't expect the P letter print on the shirt to work consistently)

Weight is fine up to 1.0, but it can mess up the fingers sometimes, if that happens, lower it slightly.

I might fix it more in the future, but not right now, since I've already made more versions than I'd like to admit, looking for a good result.

Anybody remember this cartoon?, good times.

Here's Jen from the show 'What's with Andy?'. I wanted to do something from a more dated show, and it really is. The dataset was rather bad, so I'm surprised this actually worked out.

For the prompts:

The trigger word is jen

Useful tags are: parody, purple shirt and spaghetti strip for the camisole she's wearing, black eyes, red hair, and white pants.

If she gets freckles, tag for freckles in negative prompt.

* I didn't shuffle the tags, so moving them around in the prompt can have a somewhat big effect, you can experiment with that. I do recommend putting the character lora and trigger right at the beginning of the prompt

I tested the weights in the range of 0.8 to 1.0, it depends on what you want to do. If you use other loras with it I would recommend dropping a bit.

Search AI Models & Addons

very few results