Art in Art in Art (DALL-E, MidJourney, Stable Diffusion)

Unsupervised AI creativity

A Japanese artist with her works (MidJourney)

Prompt Design / Prompt Engineering / Prompt Poetry — there are many old and new terms for this novel way to create unique visual works using text input.

We’ve already seen, how precisely Transformer-based Diffusion models like DALL·E keep attention to every detail. It follows your photo metadata like Lens and Aperture. It also uses StyleTransfer, and (in the case of DALL·E) even the artist’s essence if you refer to some specific artist (“in the style of”).

But that’s not enough for me. As you know from my previous explorations, I am looking for unsupervised creativity of a machine.

DALL·E

That’s why I created the series “Photography of an Art Gallery with different artworks displayed on the walls.”, which I presented in my virtual exhibition merzDALLEum:

“Photography of an Art Gallery with different artworks displayed on the walls” (DALL-E, merzDALLEum)

Instead of some “art in the style of”, DALL·E created a photography of a gallery space for me — with canvas on the walls. I left it up to AI to imagine what these canvases should contain.