I've looked up some examples of OpenStax images on ddg image search,
just for some idea of what is and isn't possible.
In theory all the below should be possible on a Graphiti, although I
couldn't comment on in which cases it would be practical. For example
how well fine lines are handled, or how print text inside bitmap images
is handled.
So I will mostly cover what the Canute can in theory do and not do,
automatically or manually (via a transcriber). Automatic conversion
would only be possible assuming there is a source file, ideally as a
CSV, but possibly as SVG or similar.
--------
Search term:
https://duckduckgo.com/?q=openstax+diagram&iax=images&ia=images
- Graph, time against signal, 'Time plot of a guitar's string':
https://external-content.duckduckgo.com/iu/?u=https%3A%2F%2Fi.pinimg.com%2F736x%2Fad%2Fb4%2F09%2Fadb409230619200c1de9be583551e73c--cnx.jpg&f=1&nofb=1
Generate automatically for Caqnute, would work quite nicely, probably
from one page to 40-ish pages, depending on level of detail wanted
(overview at one extreme to lossless at the other).
- Diagram of cross section of bone and ligament.
Manual creation necessary, probably would create overview page followed
by close-up of each element over four more pages.
- Diagram outline of 2x arachnids.
Impossible to do any justice to on a Canute. Would work nicely on a
Graphiti I think.
- Elaborate 3D cross section of a cell, with 18 labelled parts.
I doubt it is possible to meaningfully represent this sort of cluttered
image in any tactile medium, digital or paper, now or in the future,
without going right back to the literal drawing board.
- Block diagram of the body and thermal dynamics.
If you had the original source this could be automatically generated
really nicely on a Canute. However working from the bitmap, as is far
more likely with textbooks, it would require manual creation.
---------
One thing that interests me as an approach to SVG diagrams is to scrap
the actual object outlines and simply note their position and scale and
their labels. Then to position those labels around a 2D space
appropriately, with a key off to one side that elaborates on the meaning
of the label.
This would only work in some circumstances, but I think it would work in
a lot of the most *practical and productive* circumstances. Often the
most useful information from images is (A) the labels of objects and (B)
the relation in space between the objects.
In many cases, such as understanding a knee joint, it could be better
(in digital 2D at least, where a superior medium like analogue 2.5D
embossed plate isn't available) to 'illustrate' the different parts in
accompanying text whilst keeping the diagram to something simple that
shows each part relative to every other, as a label.
Even if this approach loses a lot of information it could be useful
because the information it does deliver can be read so much faster,
enabling a reader to decide quickly whether they need to dive into an
exhaustively detailed bitmap that has every contour, or whether they
already have enough information and can move on.
--------
Final thought: If it is subtle gradient of colour you are looking to
note then, so far as I know, the Graphiti is the only game in town, and
likely to stay that way, because of its multi-height pins.
--------
Just a few thoughts to give people an idea of the various challenges
when we're designing tactile graphics machines.
Ed