When uploading an image, you can provide alt text for screen readers. When sending a sticker, it should have a canned alt text, probably taken from the filename? Stickers are usually associated with an emoji in most apps, but that’s not necessarily the case here, so this does present a small issue.
Actual behavior
Currently, when you send an image or a sticker, a screen reader will simply read it off as an “unlabeled image”. There is no option to provide alt text.
mastodon is social media, delta chat is private chatting, what is the point of sending a sticker to your blind friend??? stickers can come from keyboard etc so there is no filename etc
being ask for a text description for every sticker is also quite unpractical
if you are in a chat with some blind or visually impaired person, the more logical common sense sensible thing to do is to avoid sending stickers
I used Masodon as an interface example, but other messaging apps also support alt text, namely Discord, from which a lot of communities and groups with folks who are visually impaired are coming from. Group chats exist with people of mixed vision, and alt text is also helpful for those who are not fully blind but might not always be able to make out all the details.
Obviously, it doesn’t make sense to send images to someone who can’t see them, but if you have someone who is blind or visually impaired in a group chat, then having the image already have a description in the correct place is very helpful and much more user focused than having to call it out every time in a separate message.
Stickers are often reused. Might it be possible to include the altext in the file, in the same way that metadata is included in many file formats? Is there a standard for this?
If the provider of a sticker pack supplied the stickers with alttexts, that would be minimally burdensome.
a cool thing would be, if sticker had an “alt text” in their format. that way, receiver could benefit without the sender doing action. and no UI is needed. and there is no need to describe the same sticker over and over, which conflicts with the way how stickers are usually used: tap and forget.
not sure, if there is a standard for that, maybe you know smth about that, that would definitely be more actionable.
otherwise, i’m also not sure the extra UI clutter and maintenance is worth the effort for a 'friends & family" app. esp. as workarounds exist, and the issue is for sticker only, which are would need a better UI at all first - there is not even a sticker selector
but other messaging apps also support alt text, namely Discord
Discord is often used for large groups and is kind of social media as well.
so for comparison, usually, messengers as Whatsapp or Signal fit better.
Not only the visual impaired use screen readers - one could be doing a job where they aren’t actively looking at the screen of the chat app, but would prefer to be read aloud updates when their hands are full.
We should not force the visual impaired to give up their privacy by announcing this in every single group and private chat they are in. Such health information is also considered highly sensitive according to the GDPR.
If one would keep requesting others to be more mindful about their attachments due to using a screen reader in each group they are present, they will be kicked out or at least bullied pretty soon. I know, I’ve tried at quite a few places and it did any good in exactly 0 of them.
The issue is for both uploads and stickers, with stickers being the trickier aspect of it
I would also argue that while maybe not most, many Discord groups I’ve been in have been smaller, more friends and family oriented, and thus are prime for moving to apps like Delta. But your point is fair, though I haven’t used Whatsapp or Signal personally so I don’t know what they do for this. I’ve mostly stuck around the FOSS space, with Discord, and historically Skype, for keeping in touch with friends outside of those groups.
The libexiv2-27 software library seems to be capable of manipulating the IPTC alt text and extended desc properties. ‘apt rdepends libexiv2-27’ suggests that the Gimp indeed does not have this functionality, which is bad, but at least other packages like Krita and a number of image viewers might. There are also AI tools claiming they can semi-automate adding descriptions.
I haven’t tested specific workflows, but it seems there are adequate tools and standards: it would be feasible for stickerpack author to add the accessibility metadata, if DC surfaced it. And while DC can’t currently render SVGs inline, it could render their alttext instead.
DC’s family-and-friends format also means that people can edit their own messages. Images with no alttext could have an icon indicating this, which any group member could click on, to add an alt text in-chat. Perhaps one could click an acceptance icon (or set to “always accept“, for some groups) to edit the alttext into the image file, so that it would be present in re-posts.
I don’t know how stickers are stored and thus whether the would be any exfiltration risks there.
nowadays with AI accessibility tools and assistants that can describe what you see in your screen I am afraid that is what most visual-impaired people would use instead of relaying on the good will and effort of people, because even if you add a feature to set description, quite a lot of people will just not use it, that is the case already in mastodon and that is why there are bots that can describe images there
@r10s in the case of Telegram and Signal stickers that are part of sticker packs have an emoji associated, because stickers are basically fancy versions of emojis to express some emotion, that doesn’t apply to stickers shared via keyboard tho, only stickers that are part of a sticker pack in that platforms, if we supported importing sticker packs at some point, the base filename of the sticker could be the related emoji this also allows sticker suggestions in telegram when you type an emoji, and that is kind of how the StickerBot we have works, it takes Signal sticker packs and sets the emoji in the filename, and that is also why you can send to it an emoji and it will send a related emotion sticker because Signal also has an emoji tied to sticker, this also makes it nicer when showing message summary in notifications and chatlist because you already now the emotion of the sticker ex “ sticker” in the notification
and a note for the creative users: letting your imagination run wild on ideas of “you could do this and that” is easy to say, not easy to implement&maintain, nor provides good UX for the end user, if you don’t even realize the problems for UX overcomplicating the UI with such options like “describing stickers” which already a bizarre thing to have to do, then you missed the point
we want something that is simple to use and understand, not over-bloated with features and options like Discord…
Firstly, I will say that this thread has been very focused on the stickers side of things, which is by far the more difficult problem here. Image uploads are a much easier prospect and would likely get more use as those are done on a case-by-case basis anyways.
While I understand the need for simplicity, we also should provide clear and contextual understanding for our low vision users, who are just as important as our sighted users. For uploads, one optional string tucked into the image upload is not going to over bloat the UI.
For stickers however, I do think that it could easily become “refactor the entire sticker system” which should be avoided if it can be. Something as simple as taking the filename as the default alt text would be a start, though, and would give us something to work with.
Just to restate though, alt text for uploads is the most doable part of this, and would probably do the most good for the most people by my guess.
Does DC currently allow a screenreader to read alttexts and/or extended descriptions that are already part of the file? Sorry, I can’t test this myself as my sound hardware is pooched.
If I understand, currently, a person sending an image in DC is never prompted to add an alttext; ditto a person sending a sticker. When you say “upload”, do you mean the same as “attach” or “send” an image, or something else?
Currently, in my client, I get a preview of the image when attaching is as an image, but not when attaching it as a file (that is, uncompressed). If the image preview contained a preview of the alt text (from the metadata or filename), that would encourage awareness.
A WebXDC app could be used to manually add an alttext to a file when sending it to a chat. The user would need to manually attach the app to the chat, though, then attach the image in the app, a slower and kludgier UI.
Having a UI option to pin a minimized draft-message-editing app to the bottom of a chat might make app functionality much more useful for this sort of purpose. It could be used for custom emoji-pickers, other special characters, stickerpacks, shared-secret encryption, file-format conversion, unit conversion, and other output-centric uses, for which sharing the WebXDC zip file is not really needed.
There have been people on the forum who assemble and distribute groups of files and designate these groups as stickerpacks, as I recall. An ordinary WebXDC app could be used to collaboratively collect and caption images for this purpose, and would encourage alttext use.
Running a machine-learning image-description model on my mobile, or most of the world’s mobiles, would be unusably slow. You’d have to add a trusted bot to the chat, or use some other cloud tool. I don’t know if a describe-all-alttextless-images-in-chat bot exists.
No, DC does not expose any alt text from the file. The images are read as “unlabeled image” in all cases for me.
And yes, by “upload” I mean the same thing as attach or send, specifically sending image files as image attachments where they are previewed in the chat.
Okay, there is zero point in discussing if and how to prompt users to create alt texts if the interface does not surface alt texts already in the file metadata.
How hard would it be to surface alt texts to screenreaders, and to the visual interface if the image cannot be rendered? With the filename as a fallback?
For the desktop app at least, the alt text comes from the image element in the UI (HTML based UIs have a lot of bult-in context thankfully, so it’s less work to make them accessible). Screenreaders know it’s an image, and adding the alt text is just a matter of putting it there AFAIK.
Most platforms don’t read file metadata for alt text though, and as far as I’m aware it’s not well supported overall. I wouldn’t personally stress making that a major issue for this.
for normal image messages you can provide the description in the text part of the message
there are two mindsets:
I have a problem that I want to solve
There is a feature in Discord and I want it here
if the objective is 1, you can do with what you already have, you can provide text descriptions together with the attached images, you can avoid stickers in the group of friends where there is a blind person, etc.
sticker” in the notification