3 ms·

I feel like everyone is collectively pranking me with these generative AIs. Everyone posts wonderful images and then every single time I try to get the damn th

• kadoban 4y ago
  You're just not seeing the hours they spent learning prompt engineering and/or the random results they picked through to get the good one(s).
  • pmoriarty 4y ago
    As someone who's generated many thousands of images on Midjourney, I agree. People think they can waltz in and immediately get great results from using AI's to generate images... and they can, if they're lucky or if they copy somebody else's prompt. It's a lot harder to do so consistently, or if you want your images to look both good and original, and not like mere copies of what everyone else is doing.
    • kadoban 4y ago
      Yeah I thought my copy of stable diffusion was broken at first because all my results were awful. Then I copied someone's prompt and got really great ones. I suspect eventually there will be tools you can just fire up with no knowledge, but all of them I've seen so far still do require a bit of expertise and time.
• diabolo96 4y ago
  You can : 1.ask chatgpt to generate a prompt of what you want by giving it a few exemples from a random SD prompt sharing website.(this alone gave me stunning results) 2.(optional) Use Controlnet for the pose you want,from the posture of the body down to each finger individually. 2.5 use multi Controlnet for multiple characters. 3. correct any errors with img2img. 4. Enjoy It takes 10 to 20 minutes (mostly in getting a good pose) but the results are always good and you can later reuse the pose again.