EPUB 3 Working Group FXL A11y Task Force Telco — Minutes

Date: 2021-05-25

See also the Agenda and the IRC Log


Present: Ken Jones, Wendy Reid, Gregorio Pellegrino, Laura Brady, Matt Garrish, Vijaya Gwori Perumal, Charles LaPierre, Will Awad



Chair: Wendy Reid

Scribe(s): Matthew Chan


Wendy Reid: let’s get into updates on what everyone is working on

Ken Jones: https://paper.dropbox.com/doc/Visual-to-Textual-Explainer–BLb5F5hgHAn~4G_Vi9LYjG09Ag-K3nAwKn2vlVqRpFyZ9KHN

Ken Jones: i’ve had a go at writing an explainer for visual-to-textual solution
… would love to get feedback from this group, if you also think it’s a good idea
… we can’t get around fxl ebooks
… so one solution is a way to present the FXL content, in a different way
… a textual presentation would maintain reading order, semantic hierarchy etc. but disregard some of the visual stuff
… regarding disregarding text styling, this was meant to be analogous to a plain “reading view”, like you would get on a phone
… also disregarding synthetic spreads
… that pages would automatically reflow, except where pagebreaks are explicitly inserted

Gregorio Pellegrino: would this be an RS feature, or would this be something embedded in the epub?
… if embedded, maybe this could be done using some media attributes to present different CSS to different RS

Gregorio Pellegrino: https://developer.mozilla.org/en-US/docs/Web/CSS/@media

Gregorio Pellegrino: but then it would be incumbent on authors to provide these
… conversely, the idea about the “reading view” is more something that is controlled by the RS

Ken Jones: open to your suggestions for how to define this
… the whole idea was not to change anything in the epub standard, but to layer on this solution that would address FXL problems with zooming, fonts, etc.
… turning back to the doc, discussing images, if this is purely for screen-readers, then surely no need for image itself if a description is available
… few different options are presented, including disregarding all images, just presenting the description

Charles LaPierre: for low vision users, it would be important to present the image itself in context with the description (so option 3)

Ken Jones: perhaps for a textual view you’d want images presented in a different way? e.g. a cropped version?

Charles LaPierre: that might work

Ken Jones: wasn’t sure how to treat media query content. Should I add a note that where this exists, it will also be converted to textual equivalent?

Wendy Reid: sure

Ken Jones: add something to the accessibility metadata that the ebook has been prepared in this way
… do we have any first impressions?

Gregorio Pellegrino: my concern is that I think we should specify who is responsible for preparing this textual version of the ebook
… i was also looking at the app Pocket(?). They published the algorithm that they use to get the main content out of a webpage
… this was quite complex due to the variety of content on a webpage
… i was thinking this could also been an approach

Ken Jones: i would say that this would have to be done by the content creators
… we should say that reading order etc. is something that’d have to be done for FXL anyway, but that this textual view is another reason for FXLs to be prepared properly

Wendy Reid: yes, a lot of this falls on the content creator, but there will have to be an RS component
… and we can’t force any RS to do anything with this, but can provide guidance on what they can do if they encounter it
… i’m going to share this with dauwhe and get his opinion as well

Gregorio Pellegrino: i was thinking that since this feature should not break anything, we can have an epub with polyfill with this feature, as a proof of concept
… i also think this can be done with a single HTML5 file base, where some content is FXL and some is reflow, right?

Wendy Reid: i think you can do that already, using the package level metadata in the spine, or something as simple as not including the necessary CSS for the textual version where we don’t want it to apply

Ken Jones: and i’m open to this being redrafted

Wendy Reid: we want to try to use existing web tech as much as possible, e.g. media queries level 5 (e.g. prefers reduced motion, inverted colors etc.)
… we might be able to take advantage of those
… we have a lot we can work with without reinventing anything
… and adding a metadata flag to package is something we could do, since we control that
… otherwise, do we have any other updates?
… i still have a couple sections of text that I will be adding to the best practices document

Gregorio Pellegrino: a publisher sent me some sample files that we looked through together, they are now annonymizing the sections that we decided we might share

Wendy Reid: i talked to ralph about a fair use document we can distribute to publishers who are submitting samples

Ken Jones: i worked with a photographer who is doing a book about nature photography, and he’s said that we can use a section as long as we credit it

Wendy Reid: now that we will soon have examples, we have to decide how we want to use the examples in our document
… should probably use some of them in the best practices document
… maybe display the image, and then how we’d transform it to make it more accessible?

Charles LaPierre: for Diagram Center, we had sample images broken down by category (e.g. complex spread, etc.)
… we could do something similar

Gregorio Pellegrino: I wouldn’t be too strict. Those examples should be a starting point. Authors can be creative in how they achieve the goal of accessibility
… so maybe we explain what the problems are, and what technologies are available, with some examples

Ken Jones: does it only have to be documentation, or can we have video walkthroughs?

Wendy Reid: i think we can use videos in our documentations

Charles LaPierre: sure +1 for video walkthroughs etc. captions :) yup

Wendy Reid: we have to provide transcript if we do, but its possible

Charles LaPierre: +1 to code examples ;)

Wendy Reid: we should probably have code examples beside epub pages, demo ebooks
… what else do you think production people at publishing houses would find helpful?

Ken Jones: biggest for me is an example document that people can dissect and replicate

Wendy Reid: so we still have some documentation to write, sample files to create
… we also have the F2F this week, where we might do an update on our work here (incl. visual-to-textual)
… if you would like to contribute and you’re not sure how, let me know

Charles LaPierre: regarding doing visual-to-textual with CSS, maybe you could have a button in the document itself, using JS, that toggles between the visual and textual views

Wendy Reid: for an experiment we can do that (since we don’t have compatible RS), but in practice that wouldn’t work
… a lot of RSes have separate renderers for FXL and reflow content, that a JS button wouldn’t be able to toggle between

Gregorio Pellegrino: maybe for this proof of concept we can do it in a webpage where it is more simple to toggle between FXL and reflow

Charles LaPierre: great idea Gregorio

Ken Jones: we could also have the same content, but in two epubs (one visual and one textual)
… someone could then open the two books side-by-side

Wendy Reid: okay, reach out to me if you need a task assigned to you
… otherwise, you know what you need to do
… AOB?
… okay, thanks everyone!