Alicia Pris from the PS1 game "Tail Concerto." The 1st game from the "Little Tail Bronx" series from CyberConnect2. She also appears in the second Little Tail Bronx game "Solatorobo: Red the Hunter"
Triggerwords and prompts between all versions are the same.
Weight: 0.9-1 (Illus/PDXL example images used 1. SD 1.5 examples use 0.9. Experiment with different weights and whatnot.)
Triggerword: alicia
Useful tags: 1girl, short hair, blue eyes, brown fur, animal ears, multicolored hair, dyed bangs, blonde hair, black hair, tail, animal nose
Outfit: eyepatch, midriff, white gloves, black footwear, cleavage cutout, armlet, red shirt, red pants
Interchangeable tags: [blue eyes/grey eyes], [blonde hair/white hair] (because even the official art can't choose just one)
The eyepatch is also optional and easily removable. sPoIlERs!
Illustrious:
Same dataset as Pony. Same prompts. Good results. Not much else to say.
PonyXL:
Trimmed down the images and fixed up some of the old improper captioning for this version.
SD 1.5:
Trained on Indigo V90, 512x768, with 93 images at 2511 steps. Recommend sticking with furry-based models for good results.
Description
v1
FAQ
Comments (11)
An Alicia fan asked me how I do couple characters, particularly with Alica.
Here's :
First, I ask for two characters doing something :
(small anthro cat) sitting on lap of (large anthro wolf), hug, (hands under clothes :1.3), (grabbing breasts:1.3),
I put quality, lighting, etc :
(masterpiece:1.2), (highly detailed:1.2), (intricate:1.2), (best quality:1.2),(8k:1.2),
cinematic light, vivid colors,
detailed background, forest, tree, trail, noon, summer
Then i write BREAK, and ddescribe after that the first character :
BREAK
<lora:AliciaPris:.9>, (small:1.2) Alicia, (eyepatch:1.3), full body, (dieselpunk bikini, dieselpunk bikini), short hair, (cat tail:1.2), cat ears, blush, medium breasts, blue eyes,(detailed fluffy brown fur:1.2), (detailed hair:1.4), (blonde | black | white) hair, dyed bangs, cat snout, looking at another,
(wide open eyes:1.4), (happy, surprised:1.3), open mouth, leaning back, looking back
And the other :
BREAK
tall muscular anthro wolf, wolf snout, adult, (pale yellow fur, light brown hair, front bangs), fluffy wolf tail, (hug:1.4), (dieselpunk clothes:1.3), head tilt, (happy), looking at another
I open Regional Prompter.
Check Active, Check Use common prompt (rarely with Base Prompt), put 1,1 in "Divide Ratio", Click on "Visialize and make template. I don't bother for Width and Size : Look like SD don't use it.
I try a few Generate. Work more of less, and the two zone are not exactly airtight. Avoid to use too many Loras.
I just discovered that, in the prompt, "common" come BEFORE "base", not after.
Yeah, that explains a lot of problems...
we need more ltb models. though pdxl or sdxl are more widely used from what ive seen
Agree so much that it's the reason why I started making loRAs
@Oxidize do you think you could make a pdxl version of this one?
@ChaosFoxOf96 Definitely will. Actually already did without modifying the dataset used for the 1.5 version. Wasn't terrible, but it wasn't amazing either. Gonna need to touch up all the images and remove a bunch of them to get it to where I want. Think I'll do it once I get my next LoRA up tonight or tomorrow.
Also while I'm here posting, wanted to let you know I've been working on Merveille/Opera. Have the image collection step done for them. Gonna make Merveille first.
@Oxidize omg, that was gonna be my next question lol
PDXL Version is up. Hope you like it.
@Oxidize awesome, thanks!
@Oxidize which character are you gonna do next?
@ChaosFoxOf96 My next LoRA is finished, but it's unrelated to LTB and likely something you won't care about.
After that's posted, Merveille will be next. Should be able to finish that once I have a full day or night to work on it.