I guess I'll throw my hat in.
Here's how I'll be doing it in my answers (here's an example):
- The hover text. This part is what ordinary readers will be interested in. Note that this doesn't benefit the visually-impaired users (hat tip to @doppelgreener), so if the picture is self-explanatory, it is better to leave this field empty. In fact, the hovertext is a function that is better used sparingly - if you want to clarify something about the image, the place to do this is the question body.
The hovertext is edited in the end-text links by putting a comment in quotes directly after the link, like this:
`[foo]: https://i.stackimgur.com/pic.jpg "your comment"`
Which will look like this (hover your cursor over it!):
[![Saitama in his natural state][Saitama]]
Transcript. In my understanding, this is the most important part, given the specific need @Stokhet has pointed out.
After posting a pretty picture with fancy hover text, it comes down to explaining what exactly is going on the picture. Putting this in the hovertext is absurd, as sometimes a transcript may be a couple of paragraphs (damn you, Alan Moore!).
The one thing I am not sure about is the content of the transcript. Two alternatives I see are:
Full transcript. This would require one to transcribe everything on particular slide, meaning all the speech bubbles, description bubbles, and even the graphical content. This is a lot of work, but I think this option is what we should be aiming for.
Short, synopsis-like description. In this case, you outline, in general terms, what is going on on the picture. This requires a bit less work (but more creativity), and could be less informative than the previous option. Note that this option is pointless if there are important quotes on the picture.
In any case, it would be nice to put a transcript in a quote, directly after the picture. As Tsundoku and Laurel point out in the comment below, per the HTML standard the
<sub> tags shouldn't be used for this purpose.
1: Needless to say, abuse of this feature, including intentional misdirection of unsuspecting users to TVTropes, is punishable by death.
This answer uses the information from this answer over at Mi Yodeya, and the editing help page, especially the parts on images and links.