Raw LLM Responses

Great video and good examples. Just want to commenting as a developer in the AI field. Everyone imagines an AI artist being some1 that goes to ChatGPT and types "An image of a cartoon scene" which is possible but leads to boring results no one wants. You also have no control over the result. The real part of an AI artist is creating datasets for Loras, use those Loras to generate images and trust me, you spend at least 1h+ of parameter tweaking to get a decent result. Everything in AI art is try'n'error. No one knows the "best" parameters because there are no. Steps are usually between 20 and 30, doesnt mean that your image wont be good at 50. Then, after creating the art the real hard part comes into place. Using either Affinity or Inpainting to remove the errors or artifacts. Use outpainting to extend your image,... Thats what real AI artists do. And thats the AI art that looks impressive and this art is impossible to do with ChatGPT or Dall-E3. Also why do people (including me) even use AI to create art? The same reason why a painter draws an image. For fun, for passion. Another reason is financial state. I myself can't afford to pay an artist to create textures for games that I'm developing as a hobby. Anything AI generates is better than what I can do using my own drawing skills. And to conclusion, I still respect the art of real artists. I'd never release or ship a game full of crappy, ChatGPT generated AI art. I either want real AI art I put effort into or art from a real artist that got paid for its job. (Little Edit): ComfyUI also is a great example of putting effort into art. I got pipelines which I created like weeks ago and I am still finetuning.