💾 Archived View for idiomdrottning.org › image-descriptions captured on 2024-06-19 at 22:50:22. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2024-06-16)
-=-=-=-=-=-=-
I used to be really bad at writing text descriptions for images. Here’s what I’ve learned so far.
I find this is a little easier than describing my own.
Focus on the meaning behind the image. What was the image’s creator trying to say? Sometimes this means you can get really terse. “Meme guy likes flowers (but they’re wilting) but hates weeds (but they’re growing)”. Done.
I sometimes see text descriptions that are like “OK, so in the top left corner there’s a red shape and then in the center top, it gradually turns indigo”, trying to describe the image in a way that would be great for a printer or a painter or an ML image generator but not so good for someone who is looking for an alternate decription of the image! We have the PNG file, if you’re gonna scan it through line by line we can run scripts on the pixel data to do that for us.
Spatial relationships are often super important in real life but often not relevant to the meaning behind the image.
[Colors] are rarely something you need to bring up but feel free to do so when there’s semantic relevance to them, like you might write “The area is marked off with warning colors (yellow and black stripes)”.
One of the worst descriptions I ever saw was “It’s [name of famous painting] but [a detail they had changed]”. While not always possible, it’s good if your description is self-contained and doesn’t require looking a bunch of stuff up.
I do like putting references in but I try to remember to put them last or close to the end.
Alt descriptions are also good for dyslexic people or low-bandwidth people or color-blind people, or people using auto-translation apps, or just people like me who are easily confused about what’s supposed to be the purpose behind an image. And there are many low-vision people who have some vision but can get more clarity if they have the alt text.
So on Fedi people often like to put in all of the text that’s in the image, if they can, since dyslexic people who can see the image can see that there is text there and might want to use a screenreader or similar to find out what it says.
The colorblind crowd is also a case for sometimes mentioning color, when that’s relevant.
This is not as easy and I’m still trying to strike the right balance. When I see someone else’s image, I can notice what my own brain’s first impressions are and that helps me write the description. “Oh, it’s a dude in a cowboy hat.” But that’s not how I draw. I draw from the side of my head that’s beyond words. I’ll just keep trying my best here. Lately I’ve been posting a series of images that I painted last fall, and that time distance makes it way easier to describe them.