On Halloween night, while waiting for kids to come begging for candy, I sat on the couch and started throwing strange but simple prompts at DALL-E 3, and basically just having fun. In case you haven’t spent much time with DALL-E 3, it’s different in that, on the back end, ChatGPT will take your simple prompt and rewrite it, adding various enhancements. Generally speaking, the output is far more interesting than my wimpy prompt should generate. For instance, I asked for a “Landscape picture of a horse wearing a space helmet. His legs are made from piano keys” which got me this really cool picture, even if my request for a space helmet was ignored.
I asked for “Landscape image of a soldier made from baseballs, riding on a mechanical robotic spider,” and got the following. Note that the spider, not the soldier, is made entirely from baseballs, but still cool.
Then, since it was Halloween, I tried the following; “Young girl in a superhero costume standing triumphant on top of a mountain of Halloween candy.” The AI happily created two images. One was a cartoon, like the one below, and the other was a photorealistic image of a girl in a mishmash superhero costume, with a cape, and a familiar “S” logo on the front.
