I've been thinking through approaches though I haven't yet attempted incorporating eInk Mode yet.

My basic document model is:

  <header>
  <main (flex layout)>
    <nav>
      <.headlines-nav (column layout)>
    <.sections>
      <section>
        <article>
        <article> ...
      <section>
        <article>
        <article> ...
      ...
 </main>
 <footer>

I've been talking to someone who'd prefer a printed-out version of the page (motor-neural limitations), and this might be one path toward that.

FWIW, I've considered adding images to the output, though in general I find these don't add much, and in fact frequently detract from, actual content. I've done some manual edits to add in images on a few instances. Automating that process of selection and incorporation is another challenge of course.