No announcement yet.

AI Text to image generation

  • Filter
  • Time
  • Show
Clear All
new posts

  • AI Text to image generation

    None of these exist. "Most beautiful speakers". Text to image AI from

    The above were generated by Stable Diffusion 2.1. Pretty interesting. Here's a Most Beautiful Integrated Amplifier....

    When I chose to render the same text in Vintedois Diffusion the following resulted..

    This is interesting. And a little hilarious. It highlights the different approaches, and source "learning" material that various IE image generation models employ. Vintedois seems to be aiming for the sci fi fantasy anime user. That's would seem to be much of what its model has digested from the web and learned from. Hence the more garish and fanciful. And the vaguely facial featured components and unmistakably perky "protuberances" on that last speaker?


  • #2
    I've played with it too, i always get a little weirded out by the way it renders grilles as some kind of lizard skin texture