@Danie van der Merwe
Given that, I'm not sure how long a screen reader takes to read 40,000 character alt-text per image?
Don't make the old mistake of taking "alt-text" and "image description" for being mutually 100% synonymous. Alt-text can be more than image description, and an image description does not always go into the alt-text.

The 40,000 characters don't go into the alt-text. If they did, Mastodon would chop 38,500 characters off and throw them away, as would Misskey and all forks of both.

The 40,000 characters go into the post text body. The toot, if you want. Where you have a limit of 500 characters, I have none at all. On Hubzilla, which supports in-line embedding of images, they go directly below the images.

Into the alt-text goes a second, different, separate, independent, much shorter image description.

Each one of my images usually has two descriptions. One full, long, detailed description with all explanations and transcripts in the post text body. One significantly shorter description in the alt-text.

The long description in the post text body is there to deliver all necessary information without Mastodon, Glitch, Ecko, Hometown, Misskey, Firefish, Iceshrimp, Sharkey, Catodon etc. etc. putting the axe to it at the 1,500-character mark.

The short description in the alt-text is there to satisfy those Mastodon users who absolutely demand there be an actually sufficient image description in the alt-text, no matter how. Even if there already is a much longer image description in the post.

After all, they can't see the image description in the post. The post is hidden behind a content warning because it is longer than 500 characters. I always put all posts that exceed 500 characters behind a Mastodon-style content warning that starts with a StatusNet/Friendica/Hubzilla-style summary of the post, followed at least by "CW: long post (<number of characters in the post, raw count> characters)".

Sometimes I take the extra effort upon me to specify how many of these characters are actual post, and how many are image description, which makes it clear right away that there is an image description behind that CW.

What I always do is to the alt-text the extra information that a longer, more detailed image description with explanations and text transcripts can be found in the post.

Depending on how much room the short image description leaves me in the alt-text, I also add where exactly the long image description can be found. And that's two different locations, depending on whether you're on something that supports in-line images (e.g. Pleroma, Akkoma, Friendica, Hubzilla, (streams)) or not (e.g. Mastodon, Misskey and their forks).

The rest of us can just look at the image so we won't use alt-text as we read the post that accompanies the image for context.
On Mastodon, you'll only see the image and a content warning at first. The content warning starts with a summary, unusual for Mastodon, but standard on Hubzilla. Then it warns you about the post being long, including how long the post is. That'll most likely be a five-digit number.

When you open the content warning, you'll see a few hundred or a very few thousands characters of actual post text. Then, after two blank lines, comes the long image description, still in the post. And that's usually several tens of thousands of characters. If the post is recent enough, the image description is announced by a HTML headline so even those who don't have images in the post know where the actual post text ends and the long image description begins.

If you want to know from personal experience what an image post of mine looks like on Mastodon, search for the hashtag #UniversalCampus. The second or third post from the top is a good example. It has a long image description in the post text body of not 40,000, but over 60,000 characters, my longest to date. The alt-text is precisely 1,500 characters long, 1,402 of which are image description.

By the way, this is what the self-same post looks like on Hubzilla.

An example with multiple images can be found under the hashtag #OSFest2023. You have to scroll down until you find a post whose content warning starts with the summary, "Classic creations by Arcadia Asylum a.k.a. Lora Lemon/Aley at OpenSimFest 2023". This one is hopelessly outdated now, by the way.

This is the Hubzilla original.

#Long #LongPost #CWLong #CWLongPost #AltText #AltTextMeta #CWAltTextMeta #ImageDescription #ImageDescriptions #ImageDescriptionMeta #CWImageDescriptionMeta

(Apologies for not protecting Mastodon users from this long post and from the contained eye contact, but I lack the technical capabilities for either. I'm not on Mastodon. I'm on Hubzilla which is not Mastodon, which is not based on Mastodon, which is not influenced by Mastodon, which is older than Mastodon and which works differently from Mastodon. On Hubzilla, this is not a stand-alone post. It's a comment like a blog comment. Comments don't have a field for summaries which are content warnings on Mastodon. Also, on Hubzilla, I have no means at all to blank sensitive images out on Mastodon.)

@Alt Text Hall of Fame
@Jupiter Rowland I don't think it's necessary to describe everything in a picture.
That depends on the picture.

Real-life photographs, the most common kind of image in the Fediverse, are something people are familiar with, and they usually show something that's at least halfway familiar, so they don't necessarily need everything described.

Pictures from obscure 3-D virtual worlds, and that's what I post, are pictures taken in places that nobody knows and full of stuff that nobody knows. If I do what one usually does with real-life photographs, if I only the describe the bare minimum without explaining anything, the people simply won't get my pictures.

For demonstrational purposes, I have attached three pictures with fairly minimal alt-texts. They don't describe what anything looks like, they don't explain anything, not even where the pictures were made, and they don't transcribe any of the text in the pictures. They're basically useless even though they're still longer than most alt-texts for real-life photographs.

Full descriptions can be found in the post linked below which is almost 77,000 characters long due to the image descriptions (the same content warnings apply here again). Due to technical difficulties, I can't link to a Mastodon copy of it, only to the Hubzilla original. If you want to see it on Mastodon proper, search the hashtag #OSFest2023.

https://hub.netzgemeinde.eu/item/8c2b4728-dda5-498b-9f84-2f11e163a4a5

This poll is a bit hard to read because of the included "view results" option, but it seems that blind people do generally prefer short, to the point descriptions over overly detailed ones.

#^https://stefanbohacek.online/@stefan/110848657796445471
Yes, but about two of them for each one who prefers detailed descriptions.

How about using a shorter description that focuses on why the picture is important to you, and then elaborating in a reply for those who do want more detail?

Just a thought.
From what I've read so far, putting the image description someplace else than the image isn't really accessible.

#AltText #ImageDescription #ImageDescriptions #Accessibility #A11y #Long #LongPost #CWLong #CWLongPost #EyeContact #CWEyeContact