Sometime back in April 2023, I started playing around with generating AI (artificial intelligence) art. In a nutshell, you provide a written prompt describing what you want the picture to look like, then the program creates four different takes on the prompt. The initial platform that I played around with was Microsoft’s Bing Create. At the time, Davison and Brock were in a musical theater production of “Joseph and the Amazing Technicolor Dreamcoat”. So, I started by creating some variations of pictures with a prompt something like “boastful young man wearing a long multicolored coat, standing on a hill with wheat bowing down to him, evening, ” and then some basic art styles (impressionism, abstract, anime, etc).

Bing Create has some limitations, the primary two of which are the image size and the ability to iterate on existing images. The image size is 1024 pixels by 1024 pixels. If you are printing on a laser printer, you are probably printing at 300 dpi. So these images would work out about 3.25 inches x 3.25 inches. The second limitation is that you cannot ask Bing to create a variation on one of the existing pictures; you just re-run the prompt or change it slightly. Bing also gives you a limited number of tokens to speed things up. On initial sign-up, this was 25 tokens. But it arose that Bing bumped your account up to 100 overnight each night, so that didn’t end up being a limitation for a dabbler like myself.

I ended up creating a large poster for the Joseph production to have out in the lobby, basically to provide something of interest for folks walking through. It was 6 rows of 4 pictures each, resulting in 23 different takes on Joseph in his multi-colored coat (plus one slide for the show title). Here are a number of the “Joseph” pictures.

_a56255b8-33fe-4df6-a212-28e45d8852ab
_18090694-0525-4596-b849-9f3fa4be78fb
_3262cdf5-f9b3-46bc-b339-c64c6dd428f0
_6abf4756-9aab-43e7-87fa-803ba49923ae
previous arrow
next arrow

0 comments

Leave a Reply