Reposting this amazing guide on how to write helpful alt text from https://www.perkins.org/resource/how-write-alt-text-and-image-descriptions-visually-impaired/, originally posted by @karengregory (including her original image description) so that I can pin it.
@karengregory Thank you for sharing this!
As some pointed out, the example alt text could be improved for folks not too familiar with capybaras, or yuzu fruits.
Something like this might work:
"A capybara, a large rodent animal with brown fur, sitting in water, with its head poking out, looking relaxed.
Yellow yuzu fruits, a citrus fruit similar to lemons, are floating in the water, and one is balanced on top of the capybara's head."
Still, very much following this guide. Good reminder to always keep your audience in mind!
@alttexthalloffame @karengregory
ooo I like that. I always do alt text but I'm never sure if I get it right or express myself properly. This is good.
@MostlyTato @alttexthalloffame @karengregory
Isn't it great, I hope to keep trying to improve my #alttext
Type of image
I usually write something like "digital rendering from within a 3-D virtual world, generated using shaders, but without ray-tracing" in the full description which goes into the post.
Last time, however, I only added the part before the first comma to the alt-text and forgot the details after the first comma. I hope I'll remember to do better next time, especially seeing as I still had lots of room in the alt-text.
Text
Most of the time, I transcribe all text anywhere in the image absolutely verbatim down to every last spelling mistake. This ought to be a good example. I even include text that's
[list]
But I just remembered that I got sloppy last time (see Type of image for a link). I didn't mention that the room numbers and the "Office" label are written on the doors of the motel on the advertising poster, let alone in which typeface and colour, and I neither described the license plate hanging above the office door, nor did I transcribe anything on it. I guess I wanted to get this over with. For the same reason, I omitted to describe what can be seen through the open blinds inside some of the motel rooms.
Who is in the image
Since my pictures always come from places that probably nobody who reads this knows exist, namely very obscure virtual worlds, I have to mention who the avatars in an image are, if any, explain who they are and describe in detail what they look like and what they wear.
Also because of this, it's always only myself and maybe also my in-world sister @Juno Rowland who is in the pictures. It'd be too much of an effort to document the identities and looks of two dozen avatars in a party picture and research and explain who they are and what they do. Besides, chances are that some of them teleport out between me taking the main picture and me getting to identify them and take detail pictures for the description.
Setting of the image
This really takes up a whole lot of characters when I write an image description. It definitely only ever goes into the detailed description in the post, never into the alt-text, also because there's no room for it in alt-text anymore. The last time (again, see Type of image for a link), it took me almost 2,300 characters to explain where the image was made, 800 characters or over 50% more than what fits into alt-text on Mastodon, Misskey and their mutual forks.
But I have to write that much because, again, nobody knows the place where the image is from. So I have to write at which location the image was taken, on which sim, in which grid, in OpenSim, what OpenSim is, what Second Life is, what a grid is, why it's called a grid, what a region is, what a sim is, what a varsim is if necessary, what this specific grid is, what this specific sim is and maybe even what this specific place is. Last time I stopped just short of explaining who Edgar Wallace was because that would have been important for non-German and/or too young readers to understand the name of the location.
Color(s)
I try to remember to mention the colour of everything I describe, usually down to the hue. Okay, it was easy last time because it was a picture of a monochrome avatar in a fully monochrome location, save for some very slightly blu-ish or teal-ish hues on the OpenSimWorld beacon which are barely visible in the image, if at all.
Expression and emotions
Pass. That's because I quit showing human or otherwise variable faces in my images. And that, in turn, is because there's absolutely no way to hide them from Mastodon users, and some autistic Mastodon users must have such images concealed from them due to the danger of being triggered by eye contact.
But even if I showed faces: Avatar faces are mostly expressionless because it takes special animations triggered either by hand or by script for them to show an expression for a few seconds. So there wouldn't be anything specific to describe.
Item locations and descriptions
I usually go from the foreground to the background, unless the actual background dominates everything (see my second link) in which case I'll start with that. From top to bottom doesn't always make sense; the last time, I would have started with mostly feature-less and completely unimportant off-sim mountains and the sky and then continued with the trees from left to right, jumping back and forth between trees farther in the back and trees farther up front.
Otherwise, if at all possible, I describe everything I can that's at least partly in the image and not fully concealed by something in full detail. I mean, my images consist entirely of stuff that next to nobody has ever come across. And people, including blind or visually-impaired people, might be curious. Besides, regardless whether someone is sighted or not, they might still need extensive explanations to understand certain elements in the image.
Still, it often happens to me that I realise after posting a described image that I've skimped on certain details too much. I think the only time I was halfway satisfied was the last time I broke my own image description length record with almost precisely 40,000 characters for one individual image plus over 1,300 characters for the preamble for both images. It would have been much longer, however, if I had transcribed repeating bits of text and described repeating fabric textures whenever they occurred.
#Long #LongPost #CWLong #CWLongPost #FediMeta #FediverseMeta #CWFediMeta #CWFediverseMeta #AltText #ImageDescription #ImageDescriptions #ImageDescriptionMeta #CWImageDescriptionMeta
@alttexthalloffame @karengregory I'm probably coming across as incredibly stupid but how do the focal point correctly. Just don't get what it means in relation to post. Apologies in advance if that made no sense.
@fifi65 @karengregory Not at all!
I'd say it's just the most important part of the image you want to bring the attention to?
Or maybe you can elaborate on the question, if I'm missing something.
@alttexthalloffame @karengregory Yeah I thought as much. I'm guessing paw on book should be okay. Also I was just curious as to what it's purpose was, if that makes sense & thank you for replying.
@fifi65 @karengregory No problem!
I'm guessing that it helps decide how to crop the image on platforms/in apps that don't show the full image by default.
@alttexthalloffame makes sense! got it, thank you.
@fifi65 My pleasure!
@alttexthalloffame @karengregory I'm just happy that such photo exists
@alttexthalloffame @karengregory
You had me at "capybara". I loves me some capybaras.
@FosseLeo @karengregory They're awesome!
@alttexthalloffame @karengregory So anschaulich erklärt, verstehe sogar ich das. 
Im Ernst: erklärt einige Punkte, die ich schon immer fragen wollte. 
@alttexthalloffame @karengregory @tbroyer Hah I used this very image in my argument for how we should do content modelling with assets and that alt-text is nothing to do with captions but is the textual asset for the media item. It's a great example.
@alttexthalloffame @karengregory Thanks this is amazingly amazing useful as a reference to share.
I do wish more apps provided direct access to alt-text for copy/pasting (beyond ferreting from browser inspector), e.g. here in mastodon web, I can only see alt text as hover, why cant I easily get to it? I can only see as hover text.
@alttexthalloffame @karengregory Can't get bored of this capybara picture tho.
@alttexthalloffame @karengregory YES! 
This is so good!! Thank you for making this 
Is there some way to nominate users or posts that have exceptionally well-written ALT text? I'd love to shine a brighter light on the people who go out of their way to make their posts accessible!
@GoodNewsGreyShoes @karengregory Feel free to tag me, happy to re-share!
Some people also use the #AltTextAwards hashtag.
@alttexthalloffame @karengregory Brilliant alt text writing guide and accompanying illustration.
@alttexthalloffame @karengregory
Needs a good acronym, though. Almost works out to IDIC, which would be a good mnemonic, but that "e" gets in the way.
@alttexthalloffame @karengregory I've seen people say not to describe the medium? ie don't say "photo of..." or "drawing of...".
Is that useful information to include? I don't want to assume what I think would be useful would actually be useful to include.
Thanks
@ScotttSee @karengregory There are differing opinions. I generally include it if it's relevant, if it's something I'd like to know about the image if I couldn't see it myself.
In all other cases (analogue photograph, digital painting, oil painting, pencil sketch, newspaper clipping, digital rendering etc. etc. pp.), do.
#ImageDescripton #ImageDescriptions #ImageDescriptionMeta #CWImageDescriptionMeta