As Andrew notes, the i1Pro does a decent job. It covers the full range of any wide gamut monitor I have measured, and I have not seen much weirdness arising from spectral resolution limitations. Where the i1Pro comes up short is in shadows. If you have a monitor that offers exceptionally good shadow detail -- a magically preserved Sony Artisan would top the list, although Eizo's best come close -- calibrations made with the i1Pro will show less shadow detail, more posterization, and sometimes color crossovers in darker shadows that are absent if the DTP-94 is used instead. A Spyder3 also outperforms the i1Pro in shadows, but not by as much as the DTP-94. This is particularly apparent if your display is of sufficiently good quality to use the Absolute Black calibration rather than Relative Black. Eizo's CG series are the only ones I have had success with here.
We have an entire fleet of colorimeters and spectrophotometers available for in-house measurement. For wide gamut displays such as the Eizo CG221, HP LP2480, and Samsung XL24 we find the Spyder3 outperforms the i1Pro when used with ColorEyes Display. This is based on both qualitative visual evaluations and quantitative measurements using an RPS 380 spectroradiometer. On both the Samsung and HP displays, the most saturated midtone and highlight greens, yellows, and oranges are more accurate and have smoother tonality when the i1Pro is used for measurements. This is offset by poorer shadow detail and artifacts that are not present in the Spyder3-generated profiles and calibrations.
The color filters in the DTp-94 are evidently tunable, and this would probably give the best results of any instrument. Unfortunately, X-Rite has not seen fit to provide a wide-gamut optimized model. They are instead pushing the i1 Display 2, with models specifically tuned for NEC displays among others. Unfortunately the tuning does not compensate for the overall poor performance both in terms of accuracy and repeatability of the i1 Display compared to the DTP-94 or Spyder3.
A tip for using any colorimeter in monitor calibration is to plug the puck in and leave it sitting on your desk for five minutes before starting the profiling process. Powering up the device for a while eliminates most of the internal heating drift that would otherwise take place.
Unfortunately their support forums with the information are down right now, but the people at
SpectraCal (you could say that their CalMAN software is the equivalent to ColorEyes for home theatre calibration) have done extensive testing on all the meters they support, which includes things like testing for thermal drift. For all the (inexpensive) contact colorimeters they support, the only one where thermal drift is not a concern is the
X-Rite Chroma 5. This is NIST certified, uses glass filters unlike most of the cheaper colorimeters out there, and has far better thermal drift compensation.
I own one of these, but unfortunately ColorEyes doesn't support it. SpectraCal reckon it can measure down to 0.05cd/m² accurately, at least when using their algorithms—they have sophisticated noise reduction, rather than simply averaging in their software package, which is an order of magnitude better than the i1Pro.
Anyway, their recommendation for the majority of sensors is to have it connected to the computer and left on the display (since it's generally going to be warmer than ambient temperature) for 20–30 minutes before doing any calibration.
Furthermore, their software actually warns you when it is time to recalibrate the meter. (as HT calibration is done by hand and you often have 9-point gamma controls, colour management systems to tweak etc. it can be quite time consuming) The DTP-94 needs a black calibration every 10 minutes.The i1 Display 2 can go 20 minutes between black calibration, however it is far more sensitive to thermal drift, and as such many people consider it to be unsuitable for Plasma displays (or others that produce a lot of heat)
It's unfortunate that the i1Pro is really the only spectro that is available (and supported by profiling packages) at a reasonable cost. The i1Pro is getting pretty old at this point. Sure there have been revisions to improve its reading speed, but it would be nice to see a more accurate, affordable spectro. (rather than a cost-reduced one like the ColorMunki)
It would be great if ColorEyes or some other calibration package supported some form of meter profiling. What CalMAN lets you do is measure the display primaries/white-point with the i1Pro, measure them again with your colorimeter and create a profile. As long as you do not move the position of the RGB primaries (note: greyscale adjustment is not moving the primaries' position) you then get readings with the colorimeter that have the accuracy of the spectro with the speed/low-light capabilities of the colorimeter. (this is why I own both the Chroma 5 as well as the i1Pro)
SpectraCal are soon to be releasing a hardware calibrated Spyder 3 colorimeter for HT calibration which is supposed to be closer in accuracy to the Chroma 5 than any of the others. I wonder if this would work with the standard calibration packages like ColorEyes. If it does, that might be the best compromise for now. It should be more accurate than a standard Spyder 3 for colour, but also have the benefits of better low-light handling. As they are only just supporting the Spyder, they haven't released information on how well it copes with thermal drift etc.
I'll send them an email and ask if it would be suitable for use in other calibration packages, and how well it handles thermal drift.
I'm surprised you found the Spyder 3 to be more accurate than the i1Pro on some displays though. SpectraCal did a test a while back, and even the $4000 X-Rite Hubble colorimeter was (in my opinion, significantly) less accurate than the i1Pro for gamut measurements, being roughly the same as all the other colorimeters (they were generally equally inaccurate) when compared to a newly calibrated Minolta CS-200. It would be great if they could do another test including the Spyder 3 and one of the reference spectros they sell. (Photo Research PR-655 or Orb SP-100) Their recommendation is that if you are going to use a colorimeter, it needs to be profiled off a spectro for each display you use it with for the best accuracy. Even if the colorimeters generally agree, it's often an equal inaccuracy rather than necessarily meaning that the i1Pro is wrong.