About V2.

Now it is 190 images with mostly manual tags. Seam and stuffed fixed. A lot of new possibilities for creativity. Main tags - taxidermy, on display, seam, stuffed, specimen. Head only, head display for head. More tags can be found in image prompts. But not all of them.

Key features:

• Can change styles. Even if you never asked for it.

• Undocumented properties may appear if tags are combined irresponsibly.

• Unpredictably predictable outputs.

• Two miles high. Two miles wide.

• Keter-class lora.

• Maybe best thing i ever did. Maybe worst.

About V1.

Experimental concept, results can be not safe, 70 images, auto and manual tags.
This lora intended to make anime girls looks like disassemblable sex doll (or zombie, or taxidermy, if you want more horror and darker themes).
Sometimes it have problems with styles and characters lora, but works fine with styles and characters in checkpoints (genshin characters in pony for example).
Main tag - taxidermy. Important tag - on display. For head - head only, head display. For more stuffed appearance - neck cut and seam (but seam sometimes have problems, my bad).
For more hardcore - no arms, no legs, neck stump, behead or headless.
There is more tags, upside, leg hook, arm display, leg display, impaled, but this works strange and can produce disturbing images.