I redownloaded my venv folder and realized the lora was made wrong the first time, so I remade it.


















All images shown are raw generations.
FOR ANYONE USING THE ON-SITE GENERATOR AND NOT RUNNING SD LOCALLY: AbyssOrangeMix3 does not seem to be available on the list of usable models, but you can use "AnyOrangeMix - Anything + AbyssOrangeMix" which is a hybrid of AOM3 and AnythingElse v4.5. This is your next best option if you want good results.
Recommended strengths are from 0.7-0.9, and I used Euler a as my sampler. Its able to do most color combinations of hair and hoodies, and it can even do a few different hair styles (so long as they work within a hoodie, so a ponytail wouldn't work for example). It can also do some simple facial expressions, such as narrowed eyes and smiling, but you need to emphasize it quite a bit. Trained on AOM3, and that is highly recommended for generations. I did however realize that AOM3 is not very good with ???????, but you can always use other loras/inpaint to patch up any weird ?????. If you have any questions, feel free to ask.
RECCOMENDED EMBEDDINGS:
EasyNegative, bad_prompt Negative Embedding, Bad artist Negative embedding, negative_hand Negative Embedding, Deep Negative V1.x, veryBadImageNegative, badhandv4
(note: I am not sure how or if negative embeddings work with the on-site generator, if they don't work, then sorry)
RECCOMENDED TAGS:
mask, eyelashes, sleeves past wrists, (round eyes, wide eyes), (faceless), black eyes, no nose, [[belt]], (<INSERT COLOR> hair, (<INSERT COLOR> hood), <INSERT COLOR> skirt, hood, <INSERT COLOR, WHITE WORKS BEST> thighhighs, <OPTIONAL> garter belt, garter straps)
Negative prompt: (realistic, lip, nose, tooth, rouge, lipstick, eyeshadow:1.0), eyebrow, eyebrows
Remove tags when applicable. For example, if you're making a butt-focused image, I would remove the skirt part of the tags, as it might not generate correctly. Some poses will not work with a garter belt, such as sitting. If you want narrowed eyes, replace (round eyes, wide eyes) with ((narrowed eyes)).
I redownloaded my venv folder and realized the lora was made wrong the first time, so I remade it.
1. Quyền đối với các mô hình được đăng lại thuộc về người sáng tạo ban đầu.
2. Người sáng tạo gốc muốn xác nhận mô hình vui lòng liên hệ nhân viên SeaArt AI qua kênh chính thức. Nhấp để xác nhận
