Trained on stills from show and movie.
Works best with AutismMix
The crew is tagged but Zoidberg, Hermes and Bender will not cooperate sometimes.-going to use better source for V2
Strength will vary with some loras
Negative: (source _anime)
Tags to use: circle eyes, overbite, dot pupil, source_cartoon, futuramastyle
old man, Professor Farnsworth, thick lenses glasses
BenderR,robot
Philip J. Fry, orange hair
Leela, purple hair, one-eyed, cyclops
amy wong, pink sweatsuit, black hair
HermesConrad, square glasses, dark skin
Zoidberg, red skin, tentacle mustache, blue sclera
*kif isn't properly tagged but he can appear with: Colored skin, green skin, yellow sclera
Description
FAQ
Comments (15)
Are you using base PonyXL?
no, Autismmix, should show up in recommended resources
@Corruptedcupid Is there meant to be a space in the negative (source _anime)?
@RedAstronaut no type as is, that's how the base pony was tagged.
@Corruptedcupid Thanks. BTW I had a lot of trouble using this LORA and getting even reasonable results took a lot of trial and error. I'm not very good at cartoon style though so it's probably a lot of user error, but wanted to let you know that usage is difficult.
@RedAstronaut did you try to use the image info? Drag into png info and transfer the data.
@Corruptedcupid Yep, examples worked great, but then when I made changes it started to work a lot less. It worked best at just making the regular cast of characters in Futurama world.
I think the training data needs more hair variety and tagged hair (e.g., get more variety for female "short hair"). In my limited experimentation it flattens hair into a few styles.
edit: Female bodies all come out samey too. Amazonians and cast in Teenage Mutant Leela's Hurdles are the only ones that come to mind as possible training data.
I'll keep that in mind for V2, feel free to post your results! Helps others decide on wanting to download the file or not.
@Corruptedcupid I was planning to after the "first image post of day" rolls over (~22 and a half hours).
Hi does this do scene plates (i.e. background, no humans)?
kinda, strength around .8. I'll try to get more usual scenes with less people in it for V2 to make it easier to generate.
Thank you!
very cool. I'm trying to put simpson characaters, family guy characters and futurama characters in the same image. Any ideas on how to do that?
I'm not sure if using multiple LORAS gives best results.. any ideas?
Great lora, shame Zapp Brannigan is apparently missing tho. 😔
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.