• tal@lemmy.today
    link
    fedilink
    English
    arrow-up
    1
    arrow-down
    2
    ·
    17 hours ago

    I mean, inline images are kind of trouble brewing for accessibility, since you’re going to create problems for blind users. Yeah, okay, alt text can help, but not all users are going to write alt text (and I…don’t actually think that Threadiverse Markdown has a convention for adding inline image alt text at all, though I guess you could stick a manual caption beneath each image). AI image recognition and generation of descriptions is neat, but it’s subject to the limitations of AI models, and someone has to be spending compute resources running it.

    If I were going to try to structure an environment to be accessibility-friendly, I’d probably just omit inline images. I happily used Reddit for a long time when there were no inline images, and have used other forums that don’t support inline images. That’s doable. But…I think that a lot of users do want inline images. I was just reading some discussion on the Threadiverse in the past week or so with someone saying that the reason that they didn’t like gopher/gemini was because they don’t support inline images.

    • lmmarsano@fedinsfw.app
      link
      fedilink
      English
      arrow-up
      2
      ·
      edit-2
      8 hours ago

      inline images are kind of trouble brewing for accessibility They’re not: the other image intensive platforms mentioned before handle them fine. The fediverse builds on the web, and web accessibility is quite mature.

      In particular, the alt attribute has existed since HTML 1.2 draft in 1993 (same year as initial release of HTML) when text-based web browsers (which still exist) were common. The web was accessible since the beginning.

      This reinforces the earlier point that web accessibility is not a limitation of capacity, but of will. The threadiverse is lagging.

      Yeah, okay, alt text can help, but not all users are going to write alt text

      That can be enforced & promoted in the environment. Platforms that do this better even in the fediverse were mentioned.

      (and I…don’t actually think that Threadiverse Markdown has a convention for adding inline image alt text at all, though I guess you could stick a manual caption beneath each image).

      It’s basic markdown.

      To add an image, add an exclamation mark (!), followed by alt text in brackets, and the path or URL to the image asset in parentheses. You can optionally add a title in quotation marks after the path or URL.

      AI image recognition AI recognition is unnecessary and faulty. The rest of the web doesn’t need it, and the other platforms mentioned before don’t rely on it.

      As WebAIM states

      Although technology is getting better at recognizing what an image depicts, algorithms alone cannot understand what an image means within the context of the overall page. A maple leaf might represent Canada, or it might just illustrate the leaf of a tree. Web page authors must provide alternative text that represents the content and function of their images.

      a text alternative is a substitute that conveys equivalent purpose in context. Unless AI reads an author’s mind, it will miss contextual meaning & intent. It will also garble text, misinterpret non-text content, sequence complex content wrong.

      The alt attribute in particular is typically accurate & equivalent, succinct, not redundant. Complex images get a succinct alt attribute and an adjacent text alternative or link to text alternative.

      Of course, an image of text doesn’t need to be an image at all. Culture & enforcement could discourage that anti-pattern.

      For the threadiverse, the top issue is images lacking text alternative. However, text alternatives apply to any non-text content[1] and web accessibility encompasses more than that, which the platform mostly addresses.


      1. the formal guideline:

        Provide text alternatives for any non-text content so that it can be changed into other forms people need, such as large print, braille, speech, symbols or simpler language.

        ↩︎
    • early_riser@lemmy.world
      link
      fedilink
      arrow-up
      3
      ·
      14 hours ago

      As a blind person I want to like Gemini specifically because it doesn’t support inline images. Or more to the point because it’s a line-oriented text-first protocol. But somehow… SOMEHOW Gemini is less accessible than the web despite being nothing but text. No windows Gemini client I have tried works with a screen reader, and the one reasonably accessible iOS client, Elaho, is no longer in the App Store.

      • tal@lemmy.today
        link
        fedilink
        English
        arrow-up
        2
        ·
        edit-2
        12 hours ago

        Aren’t there any terminal clients that also run on Windows? I don’t know whether those will be optimized for screenreaders, but they should be at least usable with them. I had some discussion a few days ago about using emacspeak — which probably has a steeper learning curve than most would want — and elpher, but even outside of that, there have got to be console clients out there.

        searches

        What about amfora? It sounds like it runs on Windows.

        Amfora aims to be the best looking Gemini client with the most features… all in the terminal. It does not support Gopher or other non-Web protocols - check out Bombadillo for that.

        It also aims to be completely cross platform, with full Windows support. If you’re on Windows, I would not recommend using the default terminal software. Use Windows Terminal instead, and make sure it works with UTF-8. Note that some of the application colors might not display correctly on Windows, but all functionality will still work.

        EDIT: Actually, I guess that if someone’s using a screenreader, that they’d be better-off with…I guess it’s called “line-oriented” software, not…I dunno, whatever you call menu/ncurses/visual terminal software. Whatever the term is for software that works like line editors like ed, where you don’t have a constantly-updated display filling the screen, but type a command and then get back a finite quantity of text. That is basically designed around something that I’d expect to work well with text-to-speech systems.

        searches

        It looks like there are a couple of gemini clients that work like this: all of the items on this page marked “Terminal(CLI)”:

        https://en.wikipedia.org/wiki/Gemini_(protocol)

        I don’t see any explicit Windows packages, but two of these apparently are in Python and in PyPI.

        • offpunk. Not explicitly packaged for Windows, though it looks like it’s Python, so it might be practical for someone to package it for Windows; it looks like the dependencies are Python packages, and it’s packaged in PyPI, which can be used in Windows (the pip/pipx command). Even without a package…I’d probably try making a Python venv and using pip in it to install it out of PyPI. On Linux and presumably Windows, this is something like first installing Python and the Python venv and pip packages, then $ python -m venv venv in a terminal in a directory to create a venv there, then activating the venv for that terminal, which on Windows is apparently done via running a .bat or .ps1 file. Then finally $ pip install offpunk in that terminal to install offpunk. Then whenever one wants to run it, one first activates the venv and then runs offpunk.

        • AV-98. Ditto; Python, not explicitly packaged for Windows, but…in PyPI.

        • early_riser@lemmy.world
          link
          fedilink
          arrow-up
          2
          ·
          13 hours ago

          I’d like to take advantage of structural navigation. On the web I can jump from link to link with tab, headings with h, lists with l etc. A terminal client is a wall of unstructured text.

          I could use a Gemini web proxy but then I lose client certificates.