I don't know if this is due to some kind of psychoacoustic thing or other, but do interpolated libraries sound, well, not quite as good as stepped libraries - obviously this is VERY subtle, but still? I mention this because I noticed that the UA 610 EQ from Gemini - at least to my ears - sounded better when using the stepped programs as opposed to the interpolated instances (the combo programs). I looked at the manual and it does indeed mention a difference - with the single programs being more 'authentic to the hardware' than the combo programs. As I say, this is very subtle, but still noticeable to my ears. Is this a common thing with all interpolated libraries? I suppose ideally each 0.25 db or whatever stage would be dynamically sampled but that would be a pain to do!
Chris, interpolation to put simply is like a guess between samples. It is a way to surpass the hardware represented in the library by of course allowing for "in-between" values not necessarily present in the hardware. Some folks like to request this feature in libraries and some for the reasons of accuracy and authenticity, do not want it. But because of the nature of it, yea you may get inaccuracy. That is too contingent on how close values are to each other or how much sampled info there is to work with that determines the quality of the interpolation.