Home Artists Posts Import Register

Content

Hi everyone, I'm going to rant for a bit because I hear people saying things like "ai generation is so easyyyyy". Welp, yeah, it's way easier than drawing it yourself. I'm not delusional to say that this is anything more than having a good PC or knowing how to use resources, and being a master of text prompts, embeddings, loras, vaes, models, and other things that come with generating pretty AI art. 

HOWEVER, some folks clearly misunderstand the process of AI generation. To get finely looking AI art, you spend hours generating it because many times, even with the right negative embeddings and good positive prompts, you can still get ugly/deformed limbs (the weakest point of AI art). This happens so often that I have to sometimes generate the same prompt over and over for hours to get at least 5-10 nice pictures. And even within those, you can tell that the limbs are usually not perfect! Yes, they are not completely deformed, but they are still FAR from perfect. 

I have two solutions for this.

Firstly, I usually try to hide their hands. I put them behind their back or above their heads. This is usually working but the AI does not always "listen to you" even if you want it to do it. Weird, right?

My second method is that I give up on generating perfect limbs. I'm usually satisfied if the limbs/fingers look kinda okay, that's why the hands on my AI art are sometimes "meh". I'm in no way a perfectionist because then I would generate content for hours and could use maybe 5 pictures from 5 hours of generation. 😂

I hope this clears some misunderstandings that come with AI art generation. 

Files

Comments

No comments found for this post.