Download
1 variant available
140 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
00 1 2 3 4 5 6 7 8 9
(6)
Sep 7, 2026
Updated version for SDXL, adding characters from the show and incorporating synthetic data from previous version

2590 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
9040 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
License:
Illustrious LicenseUpdate: New SDXL version, retrained on some of the results of the previous model and adding characters from the "Bad Guys" Netflix series. Should work similarly to previous version otherwise.
Update: New version for SDXL. Mostly trained on Illustrious/Noob but also ran it with Pony, SeaArt, and base SDXL so should work somewhat on any checkpoint related to those models. Using "helper" tags seems to be more important with this version, so include at least tags for species and color for best character consistency. If generating any of the characters in different styles, it may help to negative prompt "bad guys movie style". Conversely, if generating other characters species in that style, put similar-looking characters in the negative to avoid them looking too similar (ex. negative "mr. wolf" if generating another character that is a wolf or canid). Also if anyone cares no images from the second movie were used in training this, only screenshots from the trailers prior to release and fanart.
Another "all-in-one" LoRA for the characters and style of "The Bad Guys"! Includes most of the named characters from the first movie and the new ones introduced in the recent trailer for the sequel. Some of the characters are not quite as refined as I would like, but will work on that more in the future.
Characters included, along with some "helper tags". Most should be reasonably flexible for different styles, male/female:
ash (owl, beak, brown feathers)
She and the other owl characters (chaz, ricki) can generate oddly with separate mouth and beak sometimes, or other non-avian features, use the additional tags and increase LoRA weight if required
baron von tuskington (boar, yellow sclera, tusks)
chaz (owl, beak, brown feathers)
cuddles (human, male, mustache, brown hair, sunglasses)
d. b. cougar (yellow sclera, tan fur)
diane foxington (red fox, female, orange fur, orange body)
doom (crow, female, black feathers, blue sclera)
kitty kat (snow leopard, female, grey fur, green sclera)
misty luggins (human, female, orange hair)
mr. piranha (piranha, male, green skin, orange countershading, yellow sclera)
mr. shark (shark, male, grey body, white countershading)
mr. snake (snake, male, yellow body)
Can have some difficulty separating anthro/feral features, make sure to specify which one you want and negative prompt the other
mr. wolf (wolf, male, grey fur, yellow sclera)
ms. tarantula (tarantula, orange body, brown body, female)
Also has some anthro/feral difficulty, use "taur" or "spider taur" for her body type in the movie, but you may see extra legs when prompting for "anthro". Also try to specify "multi arm/leg" if applicable.
murray (goat, brown fur, broken horn)
broken horn may not always be consistent, helmet is also somewhat baked in
peppy sweet pots (hippopotamus, purple body)
pigtail petrova (warthog, wild boar, female, mohawk, tusks)
Tusks/mouth area often appears unusual, multiple tusks are common, either inpaint the area or roll the dice
professor marmalade (hamster, brown fur, male)
ricki talon (owl, beak, grey feathers)
serpentina (python orange body)
has some issues with anthro/feral separation similar to mr. snake, including generating odd necks, and the pearl necklace/turban are fairly baked in
skulli flame-boni (goat, white fur, yellow sclera)
helmet is difficult to remove
tanya ripper (human, black mohawk)
thalia (yellow sclera, sabertooth, brown fur)
tiffany fluffit (human, female, black hair)
vanna tee (shark, grey body, orange hair)
in some cases have seen it try to generate two separate characters (one shark, one orange-haired human), may be a workaround for that
For the style of the movie, use "bad guys movie style", and depending on the model may help to specify "3d animation" or similar. Models with a heavy style bias will show aspects of the style but may not look "movie-accurate". Best results closest to the movie seem to be using the base model of Noob-VPred with DMD2, use DMD at partial strength and CFG ~2.
