The image sample was only 8 images with 100 steps, so it will be further refined in the future.
I don't use trigger words to teach style, they are more needed for certain objects.
Trained on abyss orange mix 2 NSFW, but works on other models as well.
It is better to set the Lora force 0.5-0.9
Description
Test
FAQ
Comments (4)
trigger word ?
I don't use trigger words to teach style, they are more needed for certain objects.
Oh, you too into Calm style, great ha-ha, and looks delicious, definitely gonna try it! Maybe I'll learn something.
But is 8 pictures enough? 🫥
Yes this is one of the first attempts to make a lore that should reflect only the style of the artist, not the shapes, not the poses, not the faces. According to my latest tests, the ideal ratio is about 10-15 paintings with an accurate rendering of style, a comprehensive description, and as much different content as possible. Soon I will update this lora, because it was mashing during the training very noisy image, which gave a parasitic texture in the generation of the image. Translated with www.DeepL.com/Translator (free version)