Style embedding trained on Missymuscle's art. Generates very muscular and busty women. Can generate both SFW and NSFW. Tested on Anything V3 model
Description
FAQ
Comments (8)
can i ask you your basic settings for textual inversion embeddings ? i have a really hard time making one from game screen capture (3D anime look)
any help is welcome, i also try to make some muscle models too
Usually I tried 8000-8500 steps regards of the dataset size, but I was told that I was training wrong this whole time. You need about 200 steps per image to make better style embeddings (for a style embedding try 150 images or more)
thanks for the quick answer ! very much appreciated :D
btw do you know if the dataset in TI must have a similar style ? (can differents artist bundled together work ?)
what do you use to get advice and help on your side when youtube and github don't provide much ?
I must be doing something wrong. Even when using the same prompts as in the examples, I never get any anime-style images, only photorealistic images of dark-skinned bodybuilder. Can anyone please explain what I'm doing wrong?
Maybe you are using photorealistic model, you need to use models for anime/cartoons
@agntperseus I thought this missymuscle-8100 was the model. Is it a different checkpoint (from sd-v1-4) that I need?
Is this a Lora, hypernetwork, or something else?
这肌肉lora对于女生来说有点太大了，不够有美感，不过用在男生应该效果不错
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.