Fun with DALL-E 2

If you follow me on Twitter, you might have gathered that I recently signed up for the AI image processor, DALL-E. Warning: This thing is addictive.

DALL-E uses artificial intelligence to produce images based on descriptions you provide it. For instance, to make the image above, I asked DALL-E for something like “3D horses standing on a checkerboard in front of the Grand Tetons.” I thought the result was pretty good (unlike a lot of them!). Could be a good album cover for a prog rock band.

My earliest efforts yielded disappointing or just plain bizarre images, such as this one, where I asked for “baseball players with ham-like arms.”

I’ll be having nightmares about her for the rest of my life.

I typed in simply “literary novelist” and got this guy:

And “cat with afro hairdo riding a Stingray bike” got me this one (and I’m using it as my Twitter ID photo now):

“Several giant 3D peaches resting on a proscenium stage with blue velvet curtains on either side and a cherub in the middle of the arch” yielded:

Very Monty Python. This taught me that you need to be super-specific if you want an interesting pic to come out of it. I mean specific.

Sometimes, though, specificity is no guarantee of future results. I typed in “photographic representation of the roman coliseum filled with spaghetti and meatballs like a pasta bowl” and got this uber-strange pasta holder:

Give DALL-E a try sometime yourself. It’s pretty much free, unless you go hog wild and use up your credits fast. If nothing else it’s a fun time-waster, and it might give you a blog post or two.