dianoda and i have been working on this krog. the problem with scanners is that they are quite difficult to color calibrate (although i don't know, maybe you have a drum or flatbed that _is_ calibrated with an appropriate measurement device?
however, recall that flatbed scanners are just upside down copy stands with a motor. that is, they're cameras. but, for some dumb reason, i have yet to find a flatbed scanner that will just give you its raw sensor output.
thankfully, actual cameras are much better about this, and at the very least. actual light source color temperature can be measured quite easily and corrected for. additionally, if you have a fully calibrated chain (camera, monitor) and high quality chips of known emissivity (eg, a pantone swatch), you can build a corrective profile to apply to that sensor output.
anyway, to this end, i sent dianoda a set of abs chips last week and he's going to start on them shortly, if he hasn't already. two sources of imaging would actually be nice for comparison purposes, and process refinement but krog, you starting on the PBT chips first would be more expedient in terms of getting _some_ relatively accurate values to work with. in particular, matt3o wants to pull averaged RGB data for his key picking front-end.