Mauro Boscarol wrote: > In this way seems to me that the RGB->PCS (i.e primary model or forward > model) is not involved. > So I do not understand what it means "against the forward model". > >> You can check the inverse model (PCS -> Device) against the forward model > > Can you comment on this? It's perfectly reasonable to check the overall inverse model by printing a chart of (in gamut) PCS target colors and checking the measurements against the targets (although you need to allow for the effects of intent of course, making absolute colorimetric intent the easiest intent to test). That's a separate test though, and include test instance repeatability errors in both the device and instrument. The profcheck and invprofcheck are intended to act as a verification on the steps used to create the profile, based on the measurement data that it has been created from. Since Argyll doesn't attempt to directly construct an inverse profile from the raw data, there is less point in directly verifying this, since it doesn't point to any particular step as the cause of errors. Graeme Gill.