> Treating image as character (with unusual width and height) means > indefinite number of potential characters and if a machine (not human) > does not able to differentiate between "text characters" and "image > characters" it renders character sets unusable.
Sure, but the idea is that the actual description of the image lies in a different layer (no, I don't have any idea how these will be linked) and only a descriptive placeholder will appear at the relevant coordinates (recall that I'm advocating a two-dimensional representation to replace the current linear simplification). But your point is certainly relevant. And I'm only tossing ideas around, no deep theories involved. ++L