Sign In

Pixel Art (Pony, SDXL)

151
1.1k
3.6k
70
Verified:
SafeTensor
Type
LoRA
Stats
850
2,605
Reviews
Published
Jun 30, 2024
Base Model
Pony
Training
Epochs: 6
Usage Tips
Clip Skip: 2
Strength: 1
Trigger Words
ArsMJStylePony
Pixel Art
Hash
AutoV2
FAB96382A6
Created on Civitai

The Lora is trained on MidJourney Images that use my "Personalised Style" + Pixel Art

Works without the trigger words ArsMJStylePony, Pixel Art, manyP, fewP
You can find v2 of this model here:
https://civitai.com/models/629354?modelVersionId=703625
v2 have different color structure and the tagging described below works significantly better.
IMPORTANT:
All 8bit samples in the dataset are tagged with fewP . You can add it to the negative/positive depending on whether you prefer that type of Pixel Art .
All "higher quality" 16bit+ samples are tagged with manyP

Try lowering CFG if everything is too "melty"

Will start adding visible effects at 0.35+
Works great in the 0.4 - 0.8 range, depending on your preference and Lora Mix

I recommend 1+ for best "Pixel Art" results, if you are not going to mix it with other Loras.

Few of the images in the gallery are with the final epoch that was overbaked, but still produced interesting results. If anyone is interested I will upload it as well as a separate version.