Sign In

Steven Universe Style (LORA)

68

472

9

Verified:

SafeTensor

Type

LoRA

Stats

472

0

Reviews

Published

May 15, 2023

Base Model

SD 1.5

Training

Steps: 18,000
Epochs: 3

Trigger Words

steven universe style

Hash

AutoV2
D4A113C65E

This style model is based off of the beloved Cartoon Network show Steven Universe. I used a dataset of 60 images and 3 epochs in order to make the art style as accurate to the show as possible.

Recommended LoRA weight: ~0.5-1

CFG Scale: ~5-7

I trained this on SD 1.5, but you can create this with other models such as AOM3, Deliberate, DreamShaper, etc. The possiblities are endless. And this isn't a character LORA, but you can try and make characters (like Garnet, Steven, Peridot, etc.) as accurate to the show depictions as possible if you'd like.

Make sure you add the trigger word "steven universe style" every time or else SD won't detect it.

Feedback is always welcome! Please enjoy!