If you change the size of the dataType I’d say you need to define a customCap, but doesn’t this depend on your interpetation of Icap_ZoomFactor == 1, and on your presicion requirements?
When I initially read your question I think this sounds like a good place for a customCap, but if you could be happy with the range and presicion of a 16bit floating point value then maybe you don’t need a custom cap.
It doesn’t seem to me that the spec defines what setting icap_zoomfactor to 1 means (or 65535 for that matter).
Depending on the precision you’re targeting you *might* be able to defend the argument that you choose to interpret tw_int16 as (internally) Fix16.
It may make your value(s) look odd and may not get you invited to next twain develops luau, but as long as you limit yourself to just using the 16 bits of tw_int16, your interpretation of the value is pretty much up to you.
That said, I don’t think implementing icap_zoomFactor’s tw_int16 as a Fix16 is very intuitive but then again this is twain so …
.