Inclusion of Non-Tactile images

[clapierre]

Since tactile images can be in multiple formats (PDF, JPG, PNG, or SVG) there is no way based on file type alone to know if these images are visual images or tactile images. Reading systems will therefore treat any image the same and try to display it which could have various degree of success. With the advancement of AI image descriptions, a Reading System could instead of trying to display a visual image get an AI description and display that text instead of that image itself. Tactile images would just be displayed outright.

So if we don't ban the inclusion of visual images which could have some use-case in the future, we need a way to identify a tactile image verses a visual image.

Proposed Solution:
In the manifest, all images must be identified already, we could add an extra property indicating it is visual in nature for any visual image file. The default being if the property doesn't exist would imply it is tactile in nature. This way a Reading System would know ahead of time if the image should be rendered, ignored, or do some other advance AI processing.

[ManfredMuchenberger]

I would not ban visual images. In some cases we would probably seek ways to include both the original visual image in the "plain text rendition" and the representing tactile image in the braille rendition.

Proposed Solution:
I think it would make more sense to add the extra property to the tactile images. I don't remember the results from the images working group. But it could maybe even make sense to have a set of properties for tactile images.
e.g.

• Image optimised for tactile use - e.g. with reduced details: property = tactile
• Image optimised for swell paper: property = tactile:swell paper
• more....
• visual image: no property