Having fun with Stable Diffusion 2.1 // Chat GPT3

Ive been playing with AI for the last few weeks. Trying to see what its good at and where its flawed.
I basically focused on Stable Diffusion and ChatGPT. Id say that at this point text generation is a bit more advanced than the the art AIs.

Some of the issues I found with ChatGPT is that it kind of sucks when it comes to following very specific commands. Even when you explicitly tell it not to do something, it still does it. Many times you have to repeat the command for the AI to get it right.

Ive played a bit with the prompts to see what it reacts to the best. A good way to go about prompts if you cant think of something specific enough is just telling the AI to write a prompt for you. It will add much more creativity to the prompt than you can think of in a short amount of time.

When it comes to Stable Diffusion ive seen some amazing stuff and some not so amazing. But overall, most of the things the model makes, if I had to describe the emotion you get looking at it, it feels kind of cold. PlaygroundAI.com is something Ive been using the most.

Something you will notice scrolling the gallery is that in 99% of these images, there are no hands in the frame.
Ive played with the remix button, using the same prompts being used in the "best of" gallery but with hands it distorts them almost 100% of the time. If you try and add an instrument to the image and have your character hold the instrument or some other object you get a whole mess of things.
Maybe I just suck at making these images but its something Ive noticed others mention as well.

Some examples:

To get it to look right its a lot of trial and error. When it comes to instruments Ive rarely seen it work well.

The practice of making AI art is fun, sometimes silly, but I definitely see why some are saying AI will eventually replace artists. It will replace some for sure. I see no reason why AI couldnt make the generic Bauhaus, Wallmart prints that are sold on mass.

What AI wont replace is artists with a unique vision of their own, that create brands of themselves.
Id say that art will become more about the person behind the piece rather than the piece itself which is something the AI cannot replace or replicate.