@Halla Rempt Since you wanted criticism and advice, here it is. I'll add links to the corresponding pages in my wiki if there are any. (In case you're unaware: Parts of the Fediverse can do embedded links without a URL in plain sight. So if it has a different colour from the rest of the comment here, it's a link even if it isn't a URL.)

First of all: Don't start alt-text with "Photo of". Do mention any medium that isn't a digital photograph. But do not say if something is a digital photograph. It's generally considered the default medium on the Web, so mentioning it is redundant and needlessly inflates your alt-text.

Next: Don't use line breaks in alt-text. Yes, they make your alt-texts look prettier. But those who rely on alt-text can't see them anyway.

Besides, most screen readers expect alt-texts to be only one paragraph. They generally start reading out alt-text with something like, "Graphic." If there are multiple paragraphs, they'll take each paragraph for a separate alt-text and start reading out each one with, "Graphic."

Don't use the quotation marks on your keyboard in alt-text. Again, yes, they make your alt-texts look prettier. But, again, those who rely on alt-text can't see them anyway.

Besides, these quotes are not generally accepted as standard elements in alt-text. Hence, many frontends don't support them, not even in the Fediverse. Mastodon does.

But Hubzilla, for example, which is actually older than Mastodon (and which I'm commenting from right now), doesn't. Hubzilla keeps these quotes in alt-text as their HTML entity: &⁠quot;.

So, for example, you have this alt-text:
Photo of three books. These are manuals with grey covers, entitled "Owners Manual", "BASIC Users Manual" and "DOS Users Manual" The background is once again my painting table.

Hubzilla renders it as:
Photo of three books. These are manuals with grey covers, entitled &⁠quot;Owners Manual&⁠quot;, &⁠quot;BASIC Users Manual&⁠quot; and &⁠quot;DOS Users Manual&⁠quot; The background is once again my painting table.


Now, let's assume someone blind uses Hubzilla with a screen reader. The screen reader will read your alt-text out loud like this:
Photo of three books. These are manuals with grey covers, entitled and quot, Owners Manual and quot, and quot, BASIC Users Manual and quot, and and quot, DOS Users Manual and quot, the background is once again my painting table.

And then there are (streams) and Forte. The former is a fork of a fork of three forks of a fork (of a fork?) of Hubzilla by Hubzilla's own creator, the latter is a fork of (streams) by the same guy. These two internally use this very same quotation mark as an alt-text delimiter. This means that once they hit a quotation mark in an alt-text, they assume it marks the end of the alt-text.

Hence, they render your above alt-text like this:
Photo of three books. These are manuals with grey covers, entitled
And they continue right after the end of your alt-text while completely ignoring the rest of your alt-text.

While we're at it: If you really want to make sure that screen readers pronounce your acronyms correctly, write them in a way that ensures just that.

Let's take "BASIC" as an example. Some screen readers may pronounce it, "basic" because they recognise it as a word. Some screen readers may pronounce it, "bee ay ess eye see" because they spell everything in all caps out.

If you want all of them to read it out as a word, don't write it in all caps.

In contrast, if you wanted all of them to spell it out, you'd have to insert full stops like so: "B.A.S.I.C."

These are two exceptions of the rule that text must always be transcribed 100% verbatim.

Don't explain things in alt-text. Explanations must always go into the post text where everyone can access them.

Why? Because there are people who cannot access alt-text. Often due to physical disabilities. In order to access alt-text, at least one sufficiently working hand is required. And there are more than enough cases in which people do not have sufficiently working hands at all.

Also, GNU/Linux users who run graphical browsers on minimalist window managers such as i3wm that are entirely controlled by keyboard can't access alt-text either. They'd have to move a mouse cursor either to the little "ALT" button or until it hovers above the image. But they don't have mice or other pointing devices. They control their machines entirely by keyboard.

But if all these people can't access your alt-text, they can't read your explanations. At all. They're lost to these people.

Lastly: Keep your alt-texts and image descriptions strictly neutral. Alt-text is no place for personal opinions.

CC: @Alt Text Hall of Fame

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

@Halla Rempt I'm working on a very extensive wiki about how to describe images and write alt-texts in and for the Fediverse. It's still very incomplete, though. If you're interested anyway, here is the link.

I could still take a look at your alt-texts.

CC: @Alt Text Hall of Fame

#FediMeta #FediverseMeta #CWFediMeta #CWFediverseMeta #AltText #AltTextMeta #CWAltTextMeta #ImageDescription #ImageDescriptions #ImageDescriptionMeta #CWImageDescriptionMeta

@abovearth Makes me wonder where I'd end up:

• 1,400 characters in alt-text describe the image, but with no explanations and with no text transcripts
• Another 100 characters notify the reader of a long, detailed description in the post itself
• 60,000+ characters (you've read that right, over sixty thousand) in the post text describe the image at full detail, even more than full detail because they cover details that aren't even visible at the image's resolution, complete with extensive explanations and 20+ individual text transcripts
• Especially the latter description tries hard to adhere to as many image description rules and guidelines that I've read in the past as possible
• When I learn about another image description rule/guideline, I promptly declare both the alt-text and the long description obsolete

Guess that'd be lawful evil or something.

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

@Author-ized L.J. That's the problem: Whatever I do, I'll lose either way.

On the one hand, I feel a great pressure to describe and explain everything in advance. That way, nobody would ever have to ask me to describe a detail or explain something. And nobody, not even the most die-hard Mastodon alt-text activists, could say that I'm careless and that I only do the very bare minimum or not even that. There are people out there who are eager to block everyone who doesn't describe their images enough or lecture them or attack them for being lazy.

The last time I've described an image for Hubzilla, I refused to write detailed descriptions for the images within that image. That would have escalated and cost me weeks to describe them all because I'd also have had to describe dozens of images within these images and even more images within these images. Mind you, someone who travels to the place I've described couldn't actually see what I'd have described because the images in my image themselves have a limited resolution. But I genuinely felt bad for not describing these images.

Besides, if I only described my original images once, namely in the alt-text, and then briefly and concisely, and if someone came and asked me to describe certain elements at greater detail, I couldn't always do that. Sometimes I couldn't go back to the place shown in the image and take a closer look and write a more detailed description because that place simply doesn't exist anymore, or it has been modified, and it doesn't look like the image anymore. The details that I'd have to take a closer look at are gone.

On the other hand, my experience is also that posting more than 500 characters at once reduces my reach on Mastodon tremendously. I think I must have over 700 or 800 followers, but my reach on Mastodon is similar to that of someone with not even a dozen followers. And I don't think that's because what I post is so uninteresting or because of my rather controversial thoughts about the Fediverse, accessibility in the Fediverse, image descriptions etc.

Basically, I can't possibly post images without risking being sanctioned by anyone.

I've briefly considered putting my long descriptions into separate HTML documents and linking to them. In theory, that would reduce the length of my image posts greatly. However, this is entirely untested. I don't know if it'd work at all, i.e. open the HTML document in someone's browser rather than downloading it to their device as a file. I don't know either if a plain HTML document with no style sheet would be accessible to screen reader users.

What I do know, though, is that Mastodon hates external links with a flaming passion. That's also because the vast majority of Mastodon users is always on phones, using dedicated Mastodon apps. They hate their browser popping open when they tap a link all the same. Also, they tend to distrust external links because the linked documents or pages may not be sufficiently accessible.

Everything would be a whole lot easier if there were Fediverse-wide standards for image descriptions that take the requirements of blind or visually-impaired people into consideration as well as Mastodon's unique culture. If these standards were known to everyone both on Mastodon and in the non-Mastodon Fediverse. If everyone from blind or visually-impaired users to neurodivergent users to fully sighted alt-text activists agreed upon these standards all the same. And if these standards covered extreme edge-cases like mine as well. If there was a generally agreed-upon consensus on a whole lot of questions like:

• Is it okay to have to ask for detailed descriptions of certain details in an image that don't matter within the context of the post?
  Or do they have to be described right away if there's a chance that someone might be curious about them? What if nothing specific in the image matters more within the context than everything else?
• Is it okay to have to ask for explanations if you don't understand the topic of an image?
  Or do images about very obscure niche topics have to come with enough explanations for everyone to understand them right away (not counting technical or jargon terms which always have to be either avoided or explained)?
• So there's the rule that all text within an image must be transcribed verbatim. How far does this rule go?
  Let's suppose I have a few dozen individual bits of text within an image. Most or all of them are so small that they're unreadable. Some are so tiny that they're actually invisible at the image's resolution. Still, technically speaking, they're there. And: I can read them. Instead of reading them in the image, I can read them at the source. So I can transcribe them all.
  What is the rule then?
  Do I have to transcribe them although they're unreadable because the rule says all text has to be transcribed?
  Do I have to transcribe them although they're unreadable because not doing so and writing that they're unreadable with no transcript is or may be considered lazy?
  Do I have to transcribe them because they're unreadable, and even fully sighted people need a transcript to know what's written there?
  Mustn't I transcribe them because they don't show themselves as text in the image at the image's resolution (if they actually don't)?
  Mustn't I transcribe them because I must only describe what's visible in the image at the image's resolution to the naked eye?
  Do I have to transcribe them in my special edge-case in spite of the two above lines because this might be my last and only chance to transcribe them, for they may be gone tomorrow, and I would no longer be able to transcribe them if someone asked for a transcript? Or must I remember to keep personal transcripts of all the texts I come across in my images, just in case someone asks for a transcript of a bit of text that no longer exists?
• Must all text transcripts always be in the alt-text as opposed to an extra long image description in the post? Even if I have 20+ individual text transcripts to squeeze into Mastodon's limit of 1,500 characters of Misskey's limit of 512 characters?
  Or is it okay to
  
  • transcribe them in a separate long description in the post text
  • not put these transcripts into the alt-text
  • mention in the alt-text that there is a long image description in the post, that all the texts in the image are transcribed there, and how exactly to find that long image description?
• If any of the above requires a separate long image description because the image description won't fit within the alt-text character limits, is it preferred for the long description to be in a linked document that will open in the browser (given one has the means to write and host such a document, and users on Hubzilla, (streams) and Forte do have these means)?
  Or must the long description be where the image is at all costs? Must it be in the post itself for the convenience of app users even if it inflates the post to a hyper-massive length to the inconvenience of Mastodon users?

Unfortunately, this would require some very extensive discussions on Mastodon, involving mostly Mastodon users. But Mastodon isn't fit for this kind of discussion or debate at all.

Worse yet: I've recently found out that none of the things above must be discussed on Mastodon. Ever. You must not discuss that stuff. You must do it. But you must do it right off the bat. For whichever individual definition of "right".

#Long #LongPost #CWLong #CWLongPost #FediMeta #FediverseMeta #CWFediMeta #CWFediverseMeta #CharacterLimit #CharacterLimits #CharacterLimitMeta #CWCharacterLimitMeta #AltText #AltTextMeta #CWAltTextMeta #ImageDescription #ImageDescriptions #ImageDescriptionMeta #CWImageDescriptionMeta #Transcript #Transcripts #A11y #Accessibility

@Richard Lewis That's the trouble: The only way to actually speak with screen reader users is to find them, single them out and mention them personally. And even then they will have to want to discuss these matters with you.

This is also because next to everyone in the Fediverse who isn't sighted is on Mastodon and only on Mastodon. And Mastodon has no support for groups whatsoever. Discussing things would be much easier if Mastodon had had full-blown group support, either simply compatible with existing Fediverse group solutions or with its own solution that's fully compatible with what else supports groups, already before Musk announced he'd take over Twitter. Then Mastodon's culture would include groups rather than being completely oblivious of groups.

What I know, though, is that blind and visually-impaired Mastodon users are happy to have some alt-texts. On the commercial social platforms, they got nothing. So they generally don't have sky-high demands. In fact, unless it's a matter of life and death, they don't really care how accurate a description is because they can't verify the accuracy anyway. Also, some like a bit of whimsy with their alt-texts.

But I'm rather safe than sorry. Besides, it isn't the blind or visually-impaired people who police alt-texts and image descriptions. Mastodon's alt-text police are fully sighted. And it's them who sanction you and who decide whether you're allowed to have any reach in the Fediverse, based on how you describe your images.

However, due to Mastodon's limitations, the alt-text police don't talk to each other either, nor do they ever talk to anyone who isn't sighted. So everyone enforces different quality standards while believing their standards are the official gold standards.

#Long #LongPost #CWLong #CWLongPost #FediMeta #FediverseMeta #CWFediMeta #CWFediverseMeta #AltText #AltTextMeta #CWAltTextMeta #ImageDescription #ImageDescriptions #ImageDescriptionMeta #CWImageDescriptionMeta #AltTextPolice

Quote-post of an actual image post of mine, complete with image alt-text and a long image description in the post itself; CW: long (over 64,000 characters), quote-post, alt-text meta, image description meta, AI mentioned Artikel ansehen Zusammenfassung ansehen

@David Mitchell :CApride: If you really believe I never describe any of my images, here's a counter-proof: my last post with an actual image here on this Hubzilla channel before I moved my image-posting to (streams). It's from May 16th, 2024. By the way, the image should be embedded within the post, right above the "Image description" headline.

The image has an alt-text of exactly 1,500 characters with as detailed an image description as I could possibly fit into it, and in addition, it has a long image description in the post text itself that measures over 60,000 characters. It has to be the longest description for a single image in the history of the Fediverse. It took me two whole days, morning to evening, to research for and write this image description, and I wrote the alt-text in the morning of the following day. All without using any AI.

Fair warning: The image description is outdated in the ways that dimensions and colours are described, and parts of the explanations may be factually wrong. Besides, I didn't try hard enough to either avoid or explain technical and jargon terms. But I didn't know better back then, and I don't go around and edit all my image descriptions whenever I learn something new.

I could quote-post more image posts with alt-texts and either explanations or full descriptions in the post if this one post doesn't convince you. But this is just about my only image post that has nothing potentially triggering in the image. All the others have potentially triggering eye contact which would end up on older Mastodon versions and probably in many Mastodon apps in plain sight.

RE: https://hub.netzgemeinde.eu/item/f8ac991d-b64b-4290-be69-28feb51ba2a7

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

I've described all my images since I've learned about alt-texts, and I put more effort and knowledge into them than anyone on Mastodon; CW: long (almost 8,700 characters), Fediverse meta, Fediverse-beyond-Mastodon meta, alt-text meta, image description meta, character limit meta Artikel ansehen Zusammenfassung ansehen

@David Mitchell :CApride: Now listen here.

Ever since I've learned about alt-texts and image descriptions, I've described all my images. And unlike most Mastodon users, I've improved my image-describing further and further.

Whenever I learned something new about image descriptions, be it a rule, a guideline, a good practice or a Mastodon preference, I used this new knowledge in new image descriptions and declared all my previous image descriptions obsolete. And I've learned a lot over the years.

I've learned from Mastodon that if explanations are necessary to understand an image, they must be delivered immediately with the image post. Ever since, I've explained everything in my images that needs explaining. And since all my image posts are about extremely obscure niche topics, they need a whole lot of explanations.

I've learned from a physically disabled Mastodon user that not everyone can access alt-texts. She, for example, can't. Thus, explanations in the alt-text are lost to her. I've learned from her that explanations go into the post text. I've put all my explanations into the post text ever since.

I've learned from Mastodon that Mastodon tends to love long, detailed image descriptions. Considering how obscure the contents of my original images are and how nobody knows what anything in them looks like if they don't see it, I came to the conclusion that someone somewhere out there might need full, detailed descriptions. I've given my original images full, detailed descriptions ever since.

I've learned from various sources that alt-text must only describe what's important within the context of a post. But judging from my observations of Mastodon, its culture and its love for long alt-texts override this rule. If someone wants to know about all the small details in your images, the context doesn't matter. Thus, how detailed my image descriptions are depends on whether or not I have to expect someone being curious about the details.

I've learned by experimentation that Mastodon truncates long external alt-texts from outside at the 1,500-character mark. Also, Hubzilla (where I am) can only display so many characters of alt-text, and alt-text cannot be scrolled. Since the audience of my alt-texts is pretty much exclusively on Mastodon, I've put the full, long, detailed image descriptions into the post text.

I've learned from a blog post that alt-texts must never contain line breaks. Line breaks in alt-texts have a nasty side-effect for screen reader users: After each line break, screen readers assume that they're reading a new alt-text for a new image. And they start whatever they consider an individual image alt-text with something like, "Graphic." Thus, I write all my alt-texts as one single paragraph.

I've learned from another blog post, as well as personal experience with various Fediverse server applications, that alt-texts must never contain the double quotes commonly found on keyboards. Different frontends may misbehave in different ways, some fail very ungracefully. Thus, I no longer use these quotes in my alt-texts.

I've learned from Mastodon that even if there is an image description in the post text, there must always be an accurate and sufficiently detailed image description in the alt-text regardless. Otherwise you risk being sanctioned. I have described all my original images twice ever since: with a long and fully detailed description in the post text and a shorter description in the alt-text.

I've learned from blog posts and websites about alt-texts that text in images must be transcribed verbatim. However, nowhere that I've seen this rule written down, I've seen it mention text that's unreadable in the image while the author knows what's written there. My conclusion is that there is no exception for these texts. I tend to have many such texts. Thus, I transcribe all bits and pieces of text within the borders of my images if I have a way to read them. And I usually have.

I've learned from other blog posts about alt-texts that colours must not only be mentioned in image descriptions, but they must also be described. After all, blind people cannot be expected to know what e.g. Burgundy red is. Also, dimensions must be given not simply in absolute measures, but relatively to what else is in the image or to something that everyone is familiar with, namely the human body. Unfortunately, I've learned that so recently that I only have one original image post in which I make use of these techniques; hence, all my older original image posts count as obsolete.

I've learned from yet elsewhere that races must not be mentioned, and genders must not be assumed. I abide by both when describing meme images. My original images, on the other hand, never contain actual human beings. Whenever I show an avatar, it's always one of my avatars whose gender I have personally defined, and these avatars can't really emulate real-life human phenotypes.

Most of the above has never been taken into consideration by anyone on Mastodon. I'm literally the only one in the Fediverse who takes describing images to such levels.

But I go beyond alt-texts and image descriptions.

I've learned from Mastodon that if there's something, anything in a post of yours that might disturb anyone in some way, the post requires a Mastodon-style content warning that mentions in which way the post is disturbing. Here on Hubzilla, that's a summary. It's the same thing, and Hubzilla had summaries before Mastodon had CWs.

From observing both Mastodon and the Web outside the Fediverse, I've compiled a list of potentially triggering topics. Even excluding national/state/provincial/regional politics, I've gathered 111 of these so far. I do my best to include each one whenever necessary. On top of that, I add CWs for many things I post about because I guess I go onto people's nerves when I post about them (the Fediverse, alt-text, image descriptions, hashtags, character limits, quote-posts, actual quote-posting etc.).

However, Hubzilla is not a Twitter wannabe. It's more like Facebook or blogging software. It only offers a summary (Mastodon: CW) field for posts and DMs, but not for comments (it has two different editors for when you reply and when you don't). I could try to add a summary (Mastodon: CW) using a pair of BBcode tags, and I've done so here, but I know from personal experience that the summary tags do not translate to a Mastodon CW in comments. I'd add an individual CW to each one of my comments, but Mastodon users will neither get an actual CW nor understand that I've tried.

So I double almost all my CW'd topics up with an appropriate set of hashtags. This is in line with the culture where I am: Here on Hubzilla and in its whole software family, we don't force poster-side CWs upon each other. Instead, we have them automatically generated for ourselves, reader-side, tailored to our individual needs. But this requires keywords to trigger the automated hiding of content behind CWs.

Also, I know just what may disturb people. The best example is eye contact. You think that eye contact can only be triggering in full-face portraits of a person looking directly at the viewer? Wrong! It's triggering if there's at least one eye in the image. I've been told that some people in the autistic spectrum can detect an eye in an image if it's only a tiny fraction of a pixel. I have to expect this to extend to other potentially triggering things as well.

Thus, if it's potentially triggering and somewhere within the borders of one of my images, even if it's hardly discernible or completely invisible to the neurotypical, I still consider the whole image potentially triggering, and I treat the image and the whole post as such.

In fact, I've stopped posting potentially triggering images here on Hubzilla altogether. That's because Hubzilla has no way of making Mastodon blank an image out. And not long ago, Mastodon's CWs only hid the post text, but not the images belonging to a post. I can't rule out that certain Mastodon apps still behave this way. So I can't even use CWs to hide a triggering image. This is why I only ever post images on (streams) now: (streams) makes Mastodon blank images out when a post contains one or two certain hashtags.

Again, nobody on Mastodon goes even only nearly that far.

Please tell me in which ways exactly this is still insufficient.

#Long #LongPost #CWLong #CWLongPost #FediMeta #FediverseMeta #CWFediMeta #CWFediverseMeta #AltText #AltTextMeta #CWAltTextMeta #ImageDescription #ImageDescriptions #ImageDescriptionMeta #CWImageDescriptionMeta #CW #CWs #CWMeta #ContentWarning #ContentWarnings #ContentWarningMeta #CharacterLimit #CharacterLimits #CharacterLimitMeta #CWCharacterLimitMeta #Hashtag #Hashtags #HashtagMeta #CWHashtagMeta