@jupiter_rowland ¡Ouch! Okay, so working within those limits, you could:
Toot the first toot with the image and brief alt text.
Toot a second toot with just the full image description.
Edit the first toot to provide a link to the second toot.
The issue about link text is not, to my knowledge, solvable within current Mastodon. I need to file an issue if one has not already been filed. Obviously, you can't do anything which is not supported by the Mastodon code.
tl;dr: Clear and concise alt-text goes in the alt attribute. Full description of absolutely everything in the picture plus explanation if necessary goes on the linked page with the link being placed immediately following the text.
Added: Be sure that link is meaningful, for example, "Alt text for image of the lighthouse control room" and not "Alt text for above image" since blind people will sometimes listen to a list of link text with no other context.
@jupiter_rowland @fediverse And sorry if this is topic drift. Alt text is both easy and hard.
Also, just so you know, I had to do this as separate posts because my server has a much smaller character limit than yours.
@jupiter_rowland @fediverse And a reminder that an image of a table will be unfriendly to people on mobile devices. An HTML table can be reformatted for mobile by a website using Tablesaw, but an image of a table can't be reformatted.
@jupiter_rowland @fediverse Also, my understanding of WCAG 2 is that it requires a text equivalent be provided, not that the text equivalent always be in an alt attribute.
In particular, if you are going to provide an image of a table (something I'm appalled by, but it happens), that definitely belongs in the linked page, not in alt text, so that blind people can navigate the table to their preference rather than experience it as a data dump.
As for stuff that would be beneficial to non-technical viewers, that belongs in the separate alt-text blog post, since most folks not using a screen reader won't get to see the alt text attribute contents.
This is all my take as a sighted reader who needs to prepare and audit website content and who rarely uses a screen reader.
@jupiter_rowland @fediverse I've wrestled with this over the years. Pictures of text is clear: it must contain all of the text. But in the situations you're describing, I'd say do enough to let the listener know if they want to learn more, then add a link to an alt-text blog post which gives the whole shebang. (I've heard varying reports that longdesc is inconsistently supported, so a separate link immediately below the image is best.)