LIKE TO SEE YOUR RESULTS!
Any words would be welcome...
Attention! Do not use quality and score tags!!! They greatly spoil the artist's style
If you use unnecessary tags on quality, the artist's style will greatly deteriorate. The skin will be especially distorted
Streng 1, model autismix
v1 Promt: x_style
v2 Promt: NO!
Use Hiresfix and ADetailer
Description
improved style. No more need for a trigger word
FAQ
Comments (8)
Is the v2.0 without the trigger or using the same trigger as v1.0. I can not confirm it from pictures. (no prompts there)
Hi! In the "About this version" column I indicated that the trigger word is no longer needed. a little later I will redo other sought-after models along the same lines.
If it bothers you that I don't show my data, the reason is simple. I don't use anything special :) . If a person can't see the data, they are trying it the way they know how to do it themselves. That's how his own and unique result is obtained. So that there are no problems, I describe the details. In this case I mentioned that SCORE can mess up the style. But I am not forbidding you to use something.
As a result, the description is short and clear :)
@Simisir I just needed confirm a trigger, that's all. No need to assume more. :) If you think it might bother someone, you can put it in the model description. :) Thank you.
@HugMeIntoFace Got it, good luck!
@HugMeIntoFace and please share your result if you like the style...
>Attention! Do not use quality and score tags!!! They greatly spoil the artist's style
How true is this? also, are you talking of your loras? or in general
It happens in different ways. It may be applicable to other LORAs. Training is always different, for different datasets. Just try it. If you are satisfied with it, use it for good. It's just that I often see people writing a lot of things in Promt, after which the style deteriorates.