A character LoRA of Digi, the Renamon girl from the comics by YourDigimonGirl (@YourDigimonGirl on X). In the comic she is a 21-year-old university student. Go check out their work, all credit for the character and the art goes to them.
This is my first character LoRA (until now I've only made style LoRAs), so some issues are expected. Feedback is very welcome!
There are versions for Krea 2, MOSS, Anima and Anima 2.9B. Pick the one for your base model and read its notes (each version also has its own).
Bonus, style use: it was trained for the character, not the style, but at 0.5–0.7 strength and without @digi it can also imitate the YourDigimonGirl art style on other characters. Give it a try! Describe the other character in detail (eyes, fur, body shape) so her traits don't leak in. A few things like her eyelashes, claws or the yin-yang marks can still sneak through, and non-yellow characters work best. I left some examples below in the Gallery.
How to use (all versions)
Trigger: @digi. It already brings her look (fur colors, face markings, eyes and her big round pink glasses), so there's no need to describe them.
Always describe her outfit. Her comic outfits (the navy blazer uniform, the floral bikini) were captioned, so they only show up when you ask for them.
Her purple yin-yang sleeves are optional too: ask for them if you want them.
She is very tall (around 3 m). To show it, put something next to her for scale (a door, a shelf, another character) and say how much taller she is.
Krea 2
Start the prompt with 2D anime style digital illustration, cel shading, and then describe the scene in full sentences, with @digi inside her own description: a full-body illustration of @digi, a very tall adult anthro woman, wearing...
Strength 1.0 (0.8–0.85 with other characters). Epoch 9 of 14, tested on Neutrino v2.
MOSS
No style phrase needed: describe the scene in full sentences with @digi inside her description. It works on its own without a turbo LoRA, but for the best results use it with a turbo LoRA and a detailer.
Some finetunes make the colors quite intense. Ending the prompt with Soft, muted pastel colors, smooth flat cel shading and clean lineart. keeps them closer to the comic.
Strength 1.0. Epoch 10 of 15, tested on OneObsession (no turbo) and mossCatTower (with turbo).
Anima and Anima 2.9B
Start with @digi, then Danbooru tags, and you can add a sentence at the end, all on one line. Use a finetune (the plain base looks washed out). Negative: text, signature, watermark, english text.
Strength 1.0. Epoch 10 of 15, tested on animaElement7. The 2.9B version is the same LoRA converted for Anima 2.9B models.
Known issues
All the training images show her alone, so in scenes with other characters her glasses, long eyelashes or fur can leak onto them. What helps: write the scene as "an illustration of two characters…", put @digi inside her own description instead of at the start of the prompt, describe the other characters in detail (face, eyes, hair, species) and lower the strength to 0.8–0.85.
Background crowds can turn into more Digis. Describe them (for example "human students in casual clothes" or "flat grey silhouettes").
Avoid calling her a fox in the prompt: it gives her a longer, generic fox snout. "@digi, a very tall anthro woman" works better.
Description
Trained locally on Krea 2 Raw: 69 images x3 repeats, epoch 9 of 14. Tested mostly on Neutrino v2.
Start the prompt with 2D anime style digital illustration, cel shading, and then describe the scene in full sentences, with @digi inside her own description, for example: a full-body illustration of @digi, a very tall adult anthro woman, wearing...
Strength 1.0 when she's alone, 0.8–0.85 in scenes with other characters.
Describe her clothes every time; if you don't, she tends to get one of her comic outfits.
With other characters, write the prompt as "an illustration of two characters... On the left is @digi... On the right is..." and describe the others in detail.



