V2.0 is Baked VAE

prompt: score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, just describe what you want, tag1, tag2

negative prompt: score_6, score_5, score_4, ,...

Important Information: Ensure you use clip skip 2 when loading this model, otherwise you will get low-quality image chunks. The model supports a wide range of styles and aesthetics, but provides an insightful default prompt template that allows for high-quality samples to be generated without the need for negative prompting and other default settings. score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, simply describe what you want, tag1, tag2 (the previous Pony Diffusion model used the simpler score_9 quality modifier, the longer version for the V6 XL version was unable to be corrected during training due to training issues, you can still use score_9, but it is much weaker in effect compared to the full string. You can learn more about these tags here). The model is designed to not require negative prompting in most cases, and does not need additional quality modifiers such as "hd", "masterpiece", etc. Other special data selection tags include 'source_pony', 'source_furry', 'source_cartoon' and 'source_anime', as well as ratings of 'rating_safe', 'rating_questionable', and 'rating_explicit'.

source_pony　ponyスタイルの画像
source_furry　furryスタイルの画像
source_cartoon　cartoonスタイルの画像
source_anime　animeスタイルの画像

rating_safe　安全なコンテンツ
rating_questionable　ちょっと怪しいコンテンツ
rating_explicit　露骨なコンテンツ

data from: https://note.com/asuri/n/n65610e991405