Type | |
Stats | 686 831 |
Reviews | (132) |
Published | Apr 29, 2024 |
Base Model | |
Usage Tips | Clip Skip: 2 Strength: 0.9 |
Hash | AutoV2 6B556B1675 |
Makes chibis similar in style to the artist Ikiyouz.
How to use
Start with "1girl", "chibi" (semi-optional) after that prompt as normal. For the classic look add "smile" and "open mouth". Other tags to try are "portrait", "upper body" and "looking to the side".
Landscape seems to work slightly better than portrait.
This style is somewhat basic and best at lower resolutions which Pony XL somewhat work against so you probably want to scale down your outputs before publishing. Personally, I'd probably use the NAI-based version since output quality is about the same and it requires less resources.
MISC
Dataset of 85 images trained at 1024px resolution on PonyXL.
All sample images are raw txt2img, no upscaling, no helper embeddings/negatives used. E.g. - things should look even better if you do.
The training was done on a small subset of images, as such the output of this LoRA is not an accurate representation neither in quality or style of the source material. Go support the artist if you want the good stuff.