I tried sachin/vit2distilgpt2 to caption an abstract image I have, and it seemed to output a blank string for the caption. I'm new to this, but my goal is to caption some of my abstract artwork - for instance describe the colors, sharpness, vividness, what it may be representing or abstractly looks like.

What could I use to approach setting up and/or training an AI model to characterize and caption abstract artwork?

  • 3
    $\begingroup$ To get a sense of the possibilities you can use the "describe" command in midjourney $\endgroup$ Jun 26 at 13:27
  • 1
    $\begingroup$ Check out the "Image to Text" section: pharmapsychotic.com/tools.html for clip interrogation techniques. $\endgroup$
    – Rob
    Jun 27 at 16:57


Browse other questions tagged .