Render images from the document in a reading view

[cary-rowen]

Is your feature request related to a problem? Please describe.

Bookworm offers an accessible reading experience, but unfortunately, it currently doesn't display images within documents unless you view them in the web viewer. However, this method has some significant drawbacks, such as unnecessary extra steps. More importantly, it disrupts the reading context—like losing your place in the document—and forces users to waste time searching for where they left off or trying to find the images they wanted to see.

Describe the solution you'd like

Find a way to present images within documents seamlessly, ideally in a way that feels as natural as reading web content.

1. This would allow users to use screen reader OCR or image descriptions to interpret images in the document.
2. It would also ensure that Bookworm is usable by individuals without visual impairments.

Describe alternatives you've considered

As a temporary solution, it would be acceptable to display images when the user clicks on specific areas where images are located. However, this should not be the final approach.

Additional context

n/a

[pauliyobo]

Hello.
Unfortunately, the widget we use to display text, wx.TextCtrl, does not natively support rendering of images.
I believe that if we wanted to go this route at best we could display a panel that renders the image, but I'm a bit on the fence on this mainly because of user experience.
The other alternative, as you also pointed out would be to open a dialog with the whole image rendered. Currently, we do something similar with tables, as in, pressing ctrl + enter on a table will render it in a HTML view.

[cary-rowen]

Hi @pauliyobo

One improvement I think we can implement soon is to do something similar to the table and display the images in a separate html view.
Perhaps we can do more in HTML views, such as introducing AI capabilities to make images more accessible.
for example:

1. Convert tables rendered as images into accessible text.
2. Convert mathematical content presented as images to mathml
3. Perform image description on any image.

Thanks,
Cary

[pauliyobo]

The HTML view sounds ok to me.
I'm confused though, why should we convert tables to rendered images?
And as far as AI goes, we could, but I'm not totally convinced. Perhaps it should be a different issue entirely, as the scope of this one is to understand how to display images.
If nothing else at a minimum we could solve issue #148

[cary-rowen]

Hi @pauliyobo

You wrote:

I'm confused though, why should we convert tables to rendered images?

Sorry for the confusion.
What I mean is that an image containing a table can be extracted using AI into a formatted html table that can be accessed by a screen reader.

I often encounter tables in books that are inserted using images and do not contain any altText