Sign In

Kisaragi Reona TI

170
728
13
Updated: Oct 5, 2024
character
Verified:
PickleTensor
Type
Embedding
Stats
728
Reviews
Published
Jun 10, 2023
Base Model
Other
Training
Steps: 3,500
Epochs: 1
Trigger Words
kisaragiReona
Training Images
Download
Hash
AutoV2
115535E3B3

The character from the not so popular Full Dive RPG anime. Very open ended model. Keeps the character, but will follow the prompts much closer with little need for weight adjustments. Works great at almost any step and scale config, compatible with most LoRAs and other embeddings.

To get the in-game outfit consistently, just type "fairy" into the prompt. Might also want to try adding "wings, thigh highs, gauntlets" for a closest match.

All training images were taken from r34, squared in paint3d, rescaled to 512x512 in birme, and then upscaled x4 in SD using R-ESRGAN. Trained at a rate of .0005 for 5k steps, but best result was around 3500-4200.