Am 06.04.2010 01:14, schrieb Graeme Gill: > Currently the instruments are assumed to provide correct XYZ values. > Generally you get what you pay for in this regard. Some colorimeters > don't work so well on some displays, and this is a good reason for > buying a spectrometer. That's what I thought... unfortunatelly though, there is wide-held believe that colorimeters – in some regards, surpass the accurary of spectrophotometers when measuring emissive devices such as displays. I just stumbled upon an example of that, where an i1 pro returned the black level of a display to be around 0.8 cd/m^2, while a colorimeter more accurately would measure 0.12cd/m^2. Seems to be that the accurary of the spectrophotometer suffers greatly when luminosity approaches the black level of most displays. I do not know about how the i1 pro would behave using Argyll, as measuring over a longer period of time might compensate for this disadvantage (I think iColor Display was used in that particular case). Also, some people argue 10nm resolution in the spectrum is not enough, and 5nm resolution would improve on the accurary of results. Then again, other people say there would not be much of a difference. But everybody has an opinion, which is fine. There is just one problem for the "regular user": there is currently no satisfactory option to just "go and buy a device" as some have, or seem to have, disadvantages in one area, whilst others perform "mediocre" elsewhere. > While it's perfectly possible to create correction matrices for particular > display/colorimeter combinations, in practice there are stumbling blocks. > > A major one is that it's a combinatorial explosion. I don't even have > access to all the displays that support in-display Luts, never mind the > thousands > of other LCD displays out there, so I can't supply such correction > matrices. (I suspect it would be a full time job keeping up with them, > even if one were suppled monitors by all the manufacturers). The other > problem is that one particular instance of a colorimeter doesn't > necessarily > represent the average for that type of instrument. One needs access to a > statistically significant number of instruments to overcome this problem. I guess half of my brain was offline while composing that message. One can be lucky if the colorimeter one buys does not stray to far from the "reference" device for which the calibration was done. The DTP94 seems to be very good in this regard, but still. I do not expect you or anybody to come up with a solution (which just might be impossible), but as there was no information available on this topic regarding ArgyllCMS, I was simply curious. Thanks for the information, Steffen