This is a model for pony diffusion to generate images based off of Fionna (AKA Fionna the human girl, Fionna Campbell, my wife) as depicted in the various works of the Adventure Time franchise and fanart as well.

This LORA is meant to give pony diffusion the ability to produce detailed (yet faithful) images of Fionna. Fionna will have very good anatomy and the overall rendering will look nicer but certain features of Fionna are meant to never change, the main ones being her black dot eyes and lack of a nose. This LORA, while being SFW-capable, excels greatly at NSFW. If you want Fionna NSFW without much hassle you can literally just type out your scores and “Fionna, nude” and it will give you a really good starting point from just that.

The trigger word being different to what pony diffusion usually uses (being “Fionna the human girl”) means you can rely on PonyDiffusion’s own knowledge to help flesh out Fionna’s appearance if you are having difficulties.

The dataset contains a good few pictures of fionna without her iconic hood, but during LORA creation I did not tag these images accurately, meaning if you want to prompt fionna without her hood, it will be possible but not without some prompting effort and some luck. My advice is to not prompt “Fionna the human girl” specifically and to set the LORA weight to 0.9 or perhaps lower. I will probably update this post with a new version to amend this in the future. This has since been fixed in Version 2!

The LORA also likes to give Fionna’s blue shirt either a cleavage window or a strange yellow-gold symbol. Cleavage window in negative prompt works but i am not sure about how to fix the symbol problem. Again, I will try to fix this should I ever update this lora. Fixed in version 2!

This LORA, in spite of its strict stylization, works amazingly with autismmix. Furthermore, lowering the LORA weight can give you some very interesting results without making Fionna unrecognisable. A more realistic face on fionna can be done on a 0.5 weight. I hope that is good news for those of you that like your adventure time characters to be more realistic-looking.

Dataset contains both SFW and NSFW - Prompt accordingly if you are mindful about that.

Key-Word: Fionna

V1.1 - The symbol issue seems to be completely gone now! Prompting for hair without the hood is still not braindead easy like I hoped it would be but I am okay with this. It is still very simple. Just prompt "no hood", "long hair", "blonde hair" in pos prompt, and "hood" or "animal hood" etc. in neg prompt. Also, do not prompt "Fionna the human girl" as ponydiffusion already associates the hood with that specific prompt. If all that doesn't work then reducing the LORA weight even just by 1 will usually do the trick.

VERSION 2 UPDATE - WORKING HOOD REMOVAL PROMPTS, ACCURATE TEETH, MULTIPLE OUTFITS!

With a dataset that's near-doubled from both version 1's, I have created what I now consider to be the definitive version of this LORA (for ponydiffusion at least). I finally fixed the issue of not being able to remove the hood, and now prompting for it is really consistent and much less of a headache. Furthermore, she now comes with 5 outfits, including her default!

...and I'm unsure if many people care, but I also fixed her teeth, so she should now be as cute as they are in the show! If you want to prompt for these teeth, simply use "buck teeth" in your prompt.

And of course, the model's aesthetics have improved overall thanks to the larger dataset!

I would really reccomend a more powerful model than base PDXL, or have a good detailed style LORA handy, whatever works for you.

Default - blue shirt, bunny hood, thighhighs, long sleeve shirt, blue shorts, backpack
Tour Guide - messy hair, ponytail, brown vest, white shirt, short sleeves, orange bowtie, beige pants

Dream Outfit - dream outfit, very long hair, ponytail, sailor collar, thighhighs, blue top, sleeveless, elbow gloves, pleated skirt, zettai ryouiki, hair ornament, rabbit ears hairclips (This outfit is a bit harder to remove the ears with than the others)
Farmworld - bunny hood, black bra, knee pads, shoulder armor, backpack, shorts, brown boots

The Star - blonde hair, bunny hood, green jacket, open jacket, short sleeves, black bra, shorts, brown boots

(Seems to be missing the chevron pattern on either sleeve.)

You can remove the hood on these outfits by simply not including "bunny hood" in the prompt. If it still doesn't work then put it in your negative prompt.

VERSION 3 UPDATE - ILLUSTRIOUS

It is literally the exact same dataset as v2, but trained for illustrious models. Much of the same information before regarding keywords applies.

Stay tuned for more of my adventure time models to be updated to illustrious too!

If you like the lora, do consider making an image with the LORA and submitting it here! It generates buzz and helps me keep making models.