-

2025-12-30 21:09:40

micr0@wetdry.world

i want to hear from people who use screen readers: what makes alt text useful to you? what doesn't help?

#accessibility #a11y #blind #alttext

2025-12-31 00:56:17

Amber :neodog_box:

puppygirlhornypost2@transfem.social

@micr0@wetdry.world I don't use a screenreader but I think I can speak on behalf of the visually impaired given that 1. I am visually impaired 2. I have a visually impaired partner w/o central vision and 3. I talk to a lot of the visually impaired.

The most useful things in alt text is not the literal description. Say that there's an image consisting of a black cat, a fireplace, a family gathering around the fireplace and so on. Describing the exact state of the cat is not useful. Saying "People positioned in front of a fireplace" (aka what altbot says the most) is not that helpful either. It feels like it's missing something. It doesn't convey the warmth/comfort of being near one's loved ones in an environment like that. There's also some nuance when it comes to alt text because there's a difference in people blind from birth and people who were later blinded in life. People blind from birth have trouble with visual language. They may understand basic direction like up/down right/left, but when it gets into 2 or 3 dimensions, it's not really that easy. This is especially true with diagrams hence why sometimes you'll see me just straight up skip alt-texting data visualization because I know that for my blind friends that it's utterly useless, if anything it's even more confusing than just being handed the raw data. There are some exceptions to this where it's entirely possible. An example of that:

Maslow’s Hierarchy of Needs, a triangle where each "layer" represents its own category of need, where the topmost layers require the bottom layers to be fulfilled. Within this diagram the first layer is physiological like food, clothing, shelter.

This can be conveyed fairly effectively, but again, sometimes it's literally better to dump the organized data. One might use the visually descriptive alt text I just wrote or someone might go "Maslow's Hierarchy of Needs grouped by dependencies" and then start off with "The first group is [...]". There's not really a simple answer but once you play around with it you can adjust/adapt.

2025-12-31 00:59:44

Amber :neodog_box:

puppygirlhornypost2@transfem.social

@micr0@wetdry.world My alt text is getting to the point where sometimes, instead of uploading template memes, I will just write the alt text for it because everyone can visualize it. Sometimes it's templated I like to do specific things like screenshot of a tweet by $display_name $handle "$content" it's not all done the same way. When alt-texting images that are just like a PDF scan of text, I like to just transcribe the raw text. Sometimes when I see characters I recognize from media that isn't widely popular I will add the context (an example of this on popular media would be saying "Spongebob Squarepants from the hit show Spongebob Squarepants". Idk how to like, explain that my alt text is a bit more than just describing what's in the image. I try to give some context to the image if I can. I learn from some of my mistakes. For instance, I uploaded a gif that was popular on tumblr because it looks lewd but my alt text did not actually emphasize that cultural background. This meant people who relied on the alt text did not get the same sort of context a sighted individual would have. It's about trying to break down that sort of information loss.

2025-12-31 01:07:11

Amber :neodog_box:

puppygirlhornypost2@transfem.social

@micr0@wetdry.world I used to have a good example. It was a women's suffrage image. There was a line of topless ladies holding up signs protesting something (I honestly don't remember, but I know it vaguely enough). I was particularly disappointed to see alt bot describe the scene as a bunch of naked women at the beach while skipping out the fact that they were in the act of protesting something (probably sexualization of women's nipples and the fact that they can't be topless without it being indecent exposure). It's not that the alt bot fails to describe the image; it's that it fails to convey meaning. It might see two people holding hands, there might be a wedding ring on both of them but it's not going to say "A married couple" it'll say something like "two people holding hands with visible wedding rings". When it's broken down like that it seems pretty stupid right? Like. Okay? Are they married? Why are you telling me that?

2025-12-31 15:58:43

Jupiter Rowland

jupiter_rowland@hub.netzgemeinde.eu

@Amber :neodog_box: @Micr0byte Actually, I strongly recommend not to explain images in the alt-text, and I strongly recommend not to add any other exclusive information in the alt-text that is neither in the image not in the post. That's because not everyone can access alt-text. For example, depending on the frontend, reading alt-text may require at least one working hand, so a physical disability may make alt-text inaccessible to certain people.

So explanations and the like must always go into the post text. Granted, that's often easier said than done. Mastodon has a hard-coded character limit of 500, changing which technically speaking requires a soft fork. Sharkey has a configurable limit that defaults to 3,000. I'm on Hubzilla whose character limit is imposed by the database, and it's over 16.7 million.

That said, I myself always have a lot to write when I post images which is why image posts from me are extremely rare. I've only posted new images twice in 2025; before anyone goes looking for them, they aren't on this channel.

That's because when I post an image, it's always about a very obscure topic. It's either a meme about the Fediverse where Fediverse does not mean only Mastodon. Or it's a meme about usually super-obscure 3-D virtual worlds. Or it's an in-world rendering from one of these super-obscure 3-D virtual worlds, and these image posts need the most work.

Whenever I post images, I follow three premises.

One, next to nobody knows anything about the images in and the topic of my post. After all, just about anyone anywhere in the Fediverse could be my audience. For my virtual world pictures, this means that those who can't see them don't even have a remote idea what anything in them looks like.

Two, on the other hand, my posts about virtual worlds demonstrate something highly unexpected: 3-D virtual worlds, something that was considered the hottest thing ever in the early 2020s and that's now widely considered dead, are not dead after all. So while people don't know anything about them, they might pretty well be super-curious about them.

Three, especially some Mastodon users don't like to look up for themselves what they don't know. And they don't like to ask either. They prefer all information necessary to understand a posed served to them on a silver platter right away in the same post. This goes even more for image descriptions: Having to ask for a detail to be described is almost as bad as having to ask for an image description in the first place.

This means that my meme posts come with appropriate alt-texts including transcripts of all relevant text for all images. But they also come with explanations for a) the meme templates and b) the topic in the post text. I simply cannot assume that absolutely everyone who comes across the post will be familiar with both and able to understand my post with no explanations.

That is, if I can link to useful explanations, I do so. This means that I always link to KnowYourMeme for meme template explanations. And if I can link to someplace that explains or helps explain the topic, I do so as well. But if I can't, I write my own explanations, and I often have to do that. This leads to posts with several thousand characters worth of explanations.

When I post original images from within virtual worlds, it gets even worse. Not only does next to nobody know anything about these worlds, but nobody who can't see the image knows what anything in the image looks like. But they may be curious about what everything in the image looks like because they may be curious about these virtual worlds in general.

In addition, when I post an image from a virtual world, I often don't want to show just one specific element of the image. I rather want to show the entire scenery with everything in it. So there isn't that one element in the image that's important within the context of the post because everything in the image is important within the context of the post all the same. Even more so to those who are super-curious.

So I end up having to describe these images twice.

I start with a fully detailed description that includes everything within the borders of the image. It also includes transcripts of any and all text that's at least partially within the borders of the image, regardless of whether or not it can be read in the image itself. If I can read it in-world, I will transcribe it. And it includes all necessary explanations.

This description alone usually ends up exceeding all known character limits in the Fediverse by magnitudes. For example, it always takes me well over 1,000 characters alone to explain where the image was made. I've once had to describe an in-world object that doesn't exist in real life, and I ended up describing what it looks like in over 1,000 characters and explaining what it is and what it does in another over 4,000 characters. All in all, the long description regularly ends up tens of thousands of characters long. It goes into the post text, of course. I guess it inflates my posts to such sizes that Misskey and its forks reject them due to being too long.

But I also need an appropriate image description in the alt-text. So I start with an explanation that there is a longer, more detailed image description with text transcripts and explanations in the post text itself and where and how to find it; this is a relic from times when Mastodon only hid the post text behind a content warning, but not the images. Then I take the long description and distill a shorter one from it, short enough that this description plus the note about the long description aren't longer than 1,500 characters.

Some people say it takes 20 seconds to describe an image. Well, it tends to take me literal days to describe one virtual world rendering. I haven't posted any virtual world images since 2024 because I have had the image descriptions for a series of avatar portraits in the making since late 2024.

(By the way: Don't go looking for image posts on this channel. I no longer use Hubzilla for posting images because Hubzilla can't make Mastodon blank out sensitive images. (streams) can do that, and so I use two separate (streams) channels for posting images now: @Jupiter Rowland's (streams) outlet for virtual world images and virtual world memes and @Jupiter's Fedi-Memes on (streams) for Fediverse memes.)

#Long #LongPost #CWLong #CWLongPost #FediMeta #FediverseMeta #CWFediMeta #CWFediverseMeta #CharacterLimit #CharacterLimits #CharacterLimitMeta #CWCharacterLimitMeta #AltText #AltTextMeta #CWAltTextMeta #ImageDescription #ImageDescriptions #ImageDescriptionMeta #CWImageDescriptionMeta