Photographed in Philadelphia by my old friend, the illustrator and children’s book author (and many more things), Brian Biggs. The beauty of such homemade fixes is in how they change over time. Why the instruction to “ring bell” needed to be added is beyond me. Perhaps there was a new tenant, and they wanted to eradicate the “JBS” by writing over it. On the one hand, you’d think that they might have painted over the full plank of wood. On the other, if you’ve regularly observed such handmade fixes, you know full well that this sort of accrual of temporary fixes is the very foundation of that visual vernacular.
I’m working on doing a better job of added alt text to images to aid the visually impaired. I’m not sure how much is too much or too little detail in such text. This is what I wrote for the above photo:
This is a color photograph of a handmade doorbell. It’s wood, painted red, and it’s been rubbed raw over time. Long ago it was labeled with three letters — J, B, and S — with a stencil. Sometime after, someone stenciled two words on top: “ring bell.”
Naturally, after having done so, I figured I’d see what the DALL·E 2 software would divine from that mundane, descriptive spell through the modern magic of artificial intelligence. Here’s some of what I got. For a tool built to interpret text, DALL·E 2 doesn’t produce text all that well.
Brian being awesome, he proceeded to then take my alt-text description of his original photo and feed it into Craiyon (craiyon.com), another image-to-text AI tool. As Brian has noted, Craiyon is particularly good at archaic vibes. He got the following, which are both quite visually compelling — they feel grounded in some other reality, more China Miéville than The Repair Shop — and even less literate than what DALL·E 2 served up.