Lora trained on screencaps, concept art, promotional material and fanart of Brooklynn, Yasmina, Sammy and Roxie from the Jurassic World: Camp Cretaceous Netflix series.
Prompts
Brooklynn Appearance:
brooklynn, pink hair, short hair, single hair bun, green eyes, 1girl, solo
Brooklynn Outfit:
pink jacket, red tank top, jeans, white sneakers
Roxie Appearance:
roxie, brown eyes, brown hair, dark-skinned female, 1girl solo
Roxie Outift:
red shirt, white shorts, socks, brown shoes
Sammy Appearance:
sammy, black hair, short hair, hairband, 1girl, solo
Sammy Outfit:
black jacket, tank top, floral print, jean, boots
Yasmina Appearance:
yaz, yasmina, brown eyes, brown hair, long hair, ponytail, 1girl, solo
Yasmina Outfit:
grey jacket, grey shirt, bike shorts, socks, grey sneakers
This lora was originally just a test to see if I could make a multi conept lora, so it could be improved, specially now that I understand lora creation better, but I've got no current plans to work on it.
The training for this one took so long that my google collab timed out and the process was interrupted while working on the tenth epoch, but the ninth was already giving decent enough results as it turned out.
Quirks
If you don't specify clothes for each character, the lora will give usually dress them in a blend of their various appearances. This works oftentimes, but can also result in really crazy outifts.
The facial features aren't necessarily a perfect match to each character due to bleeding between concepts. This is somewhat more prevalent on Sammy, Roxie and Yasmina since a lot of their tag descriptors overlap.
Yasmina, Brooklynn and Sammy's jackets blend a lot due to tag overlap. Yasmina is the one that's more affected by this since the jacket she wears is skin tight compared to the more baggy jackets worn by Brooklynn and Sammy, so getting her look right might require some clever prompt use.
To reduce the effect of bleeding between character concepts it's best to put the trigger for the characters you are not trying to generate in the negatives.