Download
1 variant available
1300 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
240 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
60 1 2 3 4 5 6 7 8 9
(7)
Feb 23, 2024
Retrain of an old dataset taking into account new experience.
Use tag - mlzs, long sleeves, animal ears, short hair, bangs, blue hair, gloves, white gloves and other at lora info
Sampling steps - 20-36
CFG Scale - 7
Lora Weight - 0.6-1.2
Upscaler - LDSR
Hires steps 20-44
Denoising strength - 0.34-0.44
Show more


480 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
530 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
8230 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9

This is not the lora of one character, these are all melusines in one heap, in a common train.
Well yes, that's.
It Lora "works" in most models. No, well, of course, I didn’t run it directly on everyone, but on those whom the test package ran, the result suits me.
Options:
Token - mels
Steps - 10-48
CFG - 3.5-15
Clip Skip 1-3
Lora weight - 0.7-1.3
P.S. - I don’t know how to train lore, I’m just “learning”

