EPUB 3 Working Group FXL A11y Task Force Telco — Minutes

Date: 2021-06-08

See also the Agenda and the IRC Log


Present: Rachel Osolen, Gregorio Pellegrino, Matthew Chan, Wendy Reid, Ken Jones, Charles LaPierre, Hadrien Gardeur



Chair: Wendy Reid

Scribe(s): Matthew Chan


Wendy Reid: https://w3c.github.io/epub-specs/epub33/fxl-a11y/

Wendy Reid: update: the outline document has now been updated with content
… there are some issues with markdown (esp. with images), but otherwise it seems to be working

Ken Jones: https://demo.colibrio.com/?publication_url=https://www.circularsoftware.com/samples/DonKomarechka/macro-book-48pp-section_IDexport.epub

Ken Jones: i’ve got some sample content that i mentioned last week from Don Komarechka
… should be viewable in your browser
… its got double columns, technical illustrations (e.g. comparative images, table as an image)
… pg 380 has a full-bleed two-page image
… might be interesting to work on together, to put into good reading order, with descriptions
… right now it has come straight out of INDD
… this is just a section, but we are allowed to use as much of it as we want

Wendy Reid: yes, some of these will be good examples for demonstrating techniques

Charles LaPierre: on the tables, our recommendation has always been to take out the image and create the equivalent table with CSS styling
… for more complicated tables a solution is to have a link out to a separate page with a full-blown table, and then with link back to the main book content

Rachel Osolen: agree, and I can help with examples of alt-text
… and there is also already good documentation about how to describe images in a series, and how to describe images in a spread

Wendy Reid: yes, the tables documentation somewhat bled over into alt-text
… so the next section we should write is a comprehensive guide on how to write alt-text

Rachel Osolen: i can take that on, and anyone who wants to help is welcome

Charles LaPierre: I can help. Diagram Center has lots of examples. We have an image description sample book that we did with help from BrailleNet
… we also have POET, which is a training guide for writers of alt-text

Ken Jones: should we choose this document to make as fully accessible as we can?
… so how about we draft some alt-text for this document?

Wendy Reid: both. We want to have the guidance document explain how to write, and then refer to our examples

Gregorio Pellegrino: I would focus on making it accessible from HTML and CSS point of view, but not focus so much on the content
… the actual text descriptions might be a little out of scope. We’re not advising on how to create tech to write alt-text

Rachel Osolen: We should focus on FXL. And there are resources out there for cases where there are full page images across spread, images of text, images with text embedded alongside complex diagrams

Charles LaPierre: we can have best practices reflected with examples drawn from multiple different types of sources (children’s books, cookbook, etc.)
… and then we should also have an actual epub containing all the examples

Hadrien Gardeur: at DeMarque, we see a lot of comics/manga, then children’s, then non-fiction/cookbooks, etc.
… as a group are we strictly trying to address the best practice sorts of things, or if we are trying to tackle some of the more difficult genres?
… e.g. for manga, just alt-text wouldn’t work
… in some of our past meetings we talked about alternate style sheets
… this wouldn’t work for the most complex cases either, and it is also problematic from RS perspective, as we don’t have the ability to trigger these alternate CSS
… one possible solution is to have spine alternates
… this seems easiest to implement from RS perspective
… where does the group stand on that?

Wendy Reid: we’re talking about alternate CSS, but it doesn’t differ too much from spine alternates, but because its still so early in our design, we haven’t been very specific yet
… this is still very open to discussion
… re. comics, agreed that it is very complex in terms of formatting, and there’s much more to discuss there

Charles LaPierre: on the manga and comics, do we have any leads on getting example content?
… and does the flow change (e.g. can it go from left to right to vertical, are there other types of reading orders?)

Wendy Reid: yes, reading directionality of the page can change from page to page, subject to a general flow (either left to right, or right to left)

Ken Jones: i’ve started writing the section on navigation, and there’s region-based navigation that is in the standard, but little supported
… we could talk about using that to guide reading?

Ken Jones: https://en.wikipedia.org/wiki/Pepper%26Carrot

Ken Jones: would help with small screen reading, and visually impaired reading
… link is to free comic that we can draw examples from

Hadrien Gardeur: region-based is part of spec, but its also mostly implemented by Barnes and Noble, maybe one other
… it relied a lot on renditions, separate XML documents, and was never widely adopted for authoring
… region-based navigation is widely relied upon, but everyone does it in their own separate way
… the spec method is complicated, and maybe relies too much on multiple renditions
… also, maybe not much desire to standardize out in industry today

Wendy Reid: agree, spec is complicated, especially for authoring
… tools are not widely available

Charles LaPierre: a group at Benetech India is augmenting children’s picture books by augment each page with audio on each page
… the human narration also includes the alt-text
… we could mention that as a solution

Ken Jones: in the same way that film can also be audio described, maybe that would be a solution for comics

Gregorio Pellegrino: one issue is that we’d be imposing a reading order. It works for video, but in a comic book this is not so. Readers find it important that they can choose the order

Hadrien Gardeur: not sure I agree. We’re talking about adapting content, which involves interpretation.
… so for comic, each page could have HTML with semantic structure to indicate where you are
… it could also be used to provide information to RS, which could provide context
… but it would be a lot of work for the author

Ken Jones: comics are story-based, which usually do have a logical sequence

Wendy Reid: has NNELS ever had comics requested?

Rachel Osolen: I wrote a paper on this topic, and we have staff that are avid comics readers, so inside NNELS we’ve been discussing this topic
… I can link to resources
… there are people out there outside of the WGs that are working on innovative ways to make comics more accessible
… the presentation I’m referring to also has first hand accounts of difficulties that readers are actually having with comics
… at the time we were working within the confines of creating a DAISY book
… so our research was about how to narrate visually-driven content
… and there’s been a lot of work done on web comics
… mostly by way of how to draft text alternatives
… and then there’s a company working on making that text alternative navigable

Wendy Reid: link above is to a comic that is accessible via a number of different modalities (closed captioning, high contrast, translations, change text size)
… author felt that it was a cool experiment, but that it was a lot of work
… unsure that it was workable for a whole book
… we might not be able to do complete what this example implements, but we could take some ideas

Hadrien Gardeur: most of the time this content is produced in bitmaps, and there is not a lot of work done in terms of structure beyond reading order
… even pagelists are sometimes not very useful in these types of books
… but to have something that is widely adopted, it must be compatible with this sort of workflow
… and it would have to be able to be added to this workflow at the end
… there is a spec called Media Fragments at W3C, which allows you to refer to a rectangle in an image
… you can image left and right pages, but with TTS to read the HTML content and with media fragments to locate the HTML within the image
… advantage of this is that it relies on existing web-technology without creating anything new
… and it doesn’t require authors to change the way they author

Wendy Reid: our documentation is primarily aimed at guiding the primary authoring stage, since we want people to start thinking about accessibility, but it can also guide processing the results of authoring
… I also want to experiment with web technology like grid, flexbox, or even media fragments to see what can be done

Hadrien Gardeur: with my RS hat on, there’s still a lot of inconsistency in terms of how to implement FXL
… I see that being an issue
… anything that doesn’t change the way FXL is handled, but is more of an addition, has a better chance of uptake

Wendy Reid: so Rachel_Osolen and CharlesL will work on alt-text best practices section
… we’re going to use Ken_Jones’s sample from the photography book, we probably want to pull out some representative pages (e.g. one with the table, page 37 with the group of images)
… Ken_Jones to work on navigation section

Hadrien Gardeur: “Page Blanche” could be used as a sample as well: https://github.com/IDPF/epub3-samples/tree/main/30/page-blanche

Wendy Reid: I will try to get my hand on more sample comics content (maybe with more complex formatting)
… and we also still have the experiments

Ken Jones: looking at the media fragments, they seem similar to region-based navigation
… is there something about sync navigation that also integrates with region-based navigation
… they’re both existing web technology

Hadrien Gardeur: region-based navigation was informed by media fragments, and extended

Matthew Chan: media-fragments can only be rectangles, but region-based navigation can have different shapes

Hadrien Gardeur: but region-based navigation isn’t widely implemented, where as media-fragments is used (e.g. by IIIF)
… in that way media-fragments aligns more closely with the web

Wendy Reid: AOB?

Ken Jones: would alt-text be enough to describe each fragment? Or would there also be a description of the overall scene?

Charles LaPierre: I think that would be the bare minimum for WCAG A, but we could raise the bar by adding a scene description, but it really does depend on the page
… there are some pages where just describing the fragments could lose some of the meaning in the context

Rachel Osolen: for comics, one thing i’ve found is that you do need that extra level of the scene description
… this is the way that NNELS has approached it: 1) whole page, 2) each panel, 3) and then try to break apart each panel
… there’s a lot of additional information in the scene that is relevant to comics, such as the size of the panel, shape of panel, border, positioning, etc.

Charles LaPierre: the author’s notes to the illustrator are super important for describing comics. Would be great to have these, but difficult to get

Ken Jones: do we have samples of any of these?

Rachel Osolen: these are more experiments, I could share those
… I can also share the paper I wrote related to the DAISY comic book
… and I’ll also reach out to that company working on navigation by headings for comics

Ken Jones: is ARIA describedby something we can use, potentially for scene descriptions

Charles LaPierre: there’s ARIA details that is starting to get some support, describedby is not what we want here because that strips out all the HTML semantics, i think

Wendy Reid: describedby stores an ID to some other HTML element
… you do lose some semantics in doing that

Charles LaPierre: same idea with ARIA labelledby

Wendy Reid: okay, you have some tasks, we’ll talk again in 2 weeks
… thank you all for coming!