Hey all,
What if we managed the alt text on the node level rather than the media level (as it might apply to multiple expressions of the same object) and grabbed it when displaying image media as images?
I'm only talking about normal images right now - for large images, it seems like HOCR being readable, selectable text would solve the accessibility issues.