Monitor Calibration

I recently purchased a BenQSW2700PT monitor and have some questions about calibration of this monitor as well as my older HPLP2465.  

After calibrating both monitors (w/ Xrite Color Munki) when viewing the same pictures on each monitor the same picture will look significantly different.  It seems to be a difference in the white balance with the BenQ a bit cooler and the HP a bit warmer.  I have both set to the D65 target, luminance at 80.

Due to the difference I'm wondering if I am doing something wrong in the calibration or if the difference is simply coming from the fact that the BenQ is a superior monitor?  If it's not this are there other settings I should be looking at?

I know the best way to calibrate the BenQ is with the pallet master elements software however it is not compatible with the color munki so until the new calibrator arrives I was just trying to get a decent calibration with the color munki.  

 

What operating system and video card are you using?

Windows 7 Professional and Nvidia GE Force GT 640.

I think the problem is the HP 2465 - it is a VERY old monitor and was never designed for a luminance level of 80.  It's color rendering starts to fall apart below about 120 and there's nothing the color profile can do for it below that.  I had two of these in the distant past, more than 10 years ago, and I set them to 120, it's the lowest I could get away with and still get accurate colors.

Another issue, even if the HP was capable of 80 lumens, is that it's an older technology and only covers the sRGB gamut while the BenQ covers the Adobe RGB gamut so the same photo will look different because they render the colors in a different color space.

Thanks, E.J. I had a feeling that was the case.

Okay, lots of potential issues with this situation. We used to discuss it frequently quite some time back, but it hasn't come up for awhile.

First, E.J. is correct that your aged LP2465 may not be reacting well to a target luminance of 80 cd/m2. It's a very old monitor. Also, its backlight is a CCFL (fluorescent) light. These age and become troublesome to calibrate; the lower the attempted target luminance, the more troublesome an old CCFL becomes. Even brand new, as I recall it was a bit challenging to accurately calibrate the LP2465 below 100 - 120 cd/m2.

Second, your two monitors have quite different colour gamuts. Even if both calibrated perfectly, as E.J. notes the same image may look different because of the gamut limitations on the HP. Further, since the colour gamuts differ from each other non-uniformly in the separate Red, Green and Blue channels, you may get what appear to be colour shifts that in reality are just one monitor having more or less ability to render certain saturated hues compared to the other monitor.

Third, your new BenQ has an LED backlight. Even if both monitors were brand new, calibrated perfectly, and had the same colour gamut, the difference in backlights between them will most likely create a noticeable difference. LED-based monitors calibrated to D65 in my experience still seem slightly cooler in appearance, compared to CCFL-based monitors also calibrated to D65. There's a bunch of colour stuff that creates this difference, effectively through a combination of metamerism, inability of the filters found in a colorimeter to highly accurately read white points, and the challenges of creating curves to calibrate the white point of a CCFL vs the various types of LED's (which can be white, full RGB, GB, or RB as used in the BenQ).

Finally, there's the age-old issue on Windows that you need either multiple video cards, or a single video card with multiple LUT's, in order to properly calibrate more than one monitor at a time when using standard profiling software. That's because the calibration data (white point, black point and the curves that adjust them plus grey linearity) are all stored in 8-bit LUT's found within the video card. If the video card only has a single LUT, even though you can create multiple monitor profiles, one of them will override the other when its calibration data is loaded into the video LUT. Since your two monitors are quite different in their native gamuts and white points, a single video card LUT will not allow both to look accurate since their calibration data will be quite different from each other.

Fortunately your BenQ monitor has its own high resolution (14-bit) LUT built into it. Once you're able to run the PaletteMaster software, you can calibrate the BenQ directly within its own hardware LUT, bypassing the 8-bit video card LUT. However you still may not be able to accurately calibrate both monitors if the video card has only a single LUT, because loading that LUT with calibration data for the HP monitor would impact the colour data being output to the BenQ as well, which you wouldn't want.

Unfortunately, I can't tell if the Nvidia GT 640 supports dual LUT's or not. Manufacturers rarely quote this in the specs for their video cards. One way to test is to create a new custom monitor profile for only one of your monitors, targeting a gamma curve of 1.0 for example, instead of the normal 2.2. This will produce a display on that monitor that is whacked. If the video card has a single LUT, both displays will be whacked as soon as the gamma 1.0 profile is loaded for the one monitor. But if there are two LUT's, loading the gamma 1.0 profile will whack the monitor it was made for, while the other monitor should still look normal.

The only ways to run two accurately calibrated monitors on Windows are to run a video card with multiple LUT's; or run multiple video cards, one per monitor, to assure separate LUT's; or run multiple monitors like the BenQ, NEC, Eizo, etc. that each have their own internal LUT and then calibrate them using the proprietary software that can drive the monitor LUT's. The generalist software that comes with over-the-counter calibrators like the Spyder, X-Rite ColorMunki, i1Display Pro, etc., can't drive the proprietary monitor LUT's.

Even using one of these 3 ways to create multiple profiles, some of the other above issues may still limit your ability to get a very close match between multiple monitors. The more different they are from each other, the more the limitations of calibration may allow noticeable visual differences to creep in.

Royce I have an old NEC SpectraView 2690WU which is the same model you use to have I believe. I have been calibrating it to a luminance of 85. It really needs to be set at 80 but I have been reluctant to set it so low for fear of messing up the color. In your opinion do you think the monitor can be safely set to 80? My current work around requires me boosting the exposure by 0.25 so that the print looks correct. When my Epson 4800 was working (died 3 years ago) I set the luminance to 95 and prints matched the monitor well and the print was well exposed. I got an Epson P800 SC as my new printer and this is when I had to set the luminance to 85 and then increase the exposure by 0.25 to get the prints to look correctly exposed. When I go back and print TIFF's that printed perfectly on the 4800 I now have to increase the exposure by 0.25 to have the print look correctly exposed when the printed on the P800. Any thoughts?

Hi Royce,

Thank you for the detailed information.  It sounds like things will become simplified a bit once I can use the Pallet Master software and once I upgrade to having to have 2 of the BenQ monitors.

A few questions though;

1.  Is there an Nvidia card that you know has 2 LUT's.  I will hopefully be upgrading my computer in the near future and was looking at the Nvidia 1070.
2.  Once I can use the Pallet Master software is there a way to confirm that I have calibrated things correctly?  I'm concerned that since i have calibrated both monitors now with the color munki those setting might interfere.
3.  If I understand correctly with the Pallet Master software the calibration is done with this software but internally in the monitor's hardware (LUT) instead of through Windows like a 3rd party (color munki) would do?  If so once I have 2 - BenQ monitors there may be a slight difference in calibration but they should be very very close?

Thanks, Aaron

signgrap wrote:Royce I have an old NEC SpectraView 2690WU which is the same model you use to have I believe. I have been calibrating it to a luminance of 85. It really needs to be set at 80 but I have been reluctant to set it so low for fear of messing up the color. In your opinion do you think the monitor can be safely set to 80? My current work around requires me boosting the exposure by 0.25 so that the print looks correct. When my Epson 4800 was working (died 3 years ago) I set the luminance to 95 and prints matched the monitor well and the print was well exposed. I got an Epson P800 SC as my new printer and this is when I had to set the luminance to 85 and then increase the exposure by 0.25 to get the prints to look correctly exposed. When I go back and print TIFF's that printed perfectly on the 4800 I now have to increase the exposure by 0.25 to have the print look correctly exposed when the printed on the P800. Any thoughts?

The NEC 2690W was a tank. I finally upgraded even though my 2690 was still trucking along, because I needed more screen real estate and couldn't wait for it to die. Plus I wanted to switch over to DisplayPort connections with newer video cards, etc.

 I wouldn't have hesitated to calibrate my 2690 to 80 cd/m2. I'm confident it would do that luminance quite well while still within its useful operating life. I'm speaking here of using the SpectraView II software; if using some other general monitor calibration software that can't drive the monitor's internals, then results may not be the best. This would be due to falling back on 8-bit curves loaded in the video card rather than using the higher-bit curves in the monitor itself.

Having said that... the 2690 is a pretty long in the tooth now. Whether yours will still accurately calibrate at 80 cd/m2, I can't say. But if you run the SpectraView II software at that target luminance, it likely will tell you through various results afterwards (curves and delta-E stats) whether any significant problems may have been encountered. So give it a try.

It's not necessarily surprising that you have to adjust files that used to print bang-on on your 4800, to get good results on the new P800. The reformulated inks in the new generation of Epsons are quite different; it's the biggest change in inks for Epson in quite some time. One notable change is the blacks are now much darker. So it makes some sense that you need to up the exposure of old files slightly to get them to print not quite as dark as they otherwise would.

Aaron Jors wrote:Hi Royce,

Thank you for the detailed information.  It sounds like things will become simplified a bit once I can use the Pallet Master software and once I upgrade to having to have 2 of the BenQ monitors.

A few questions though;

1.  Is there an Nvidia card that you know has 2 LUT's.  I will hopefully be upgrading my computer in the near future and was looking at the Nvidia 1070.
2.  Once I can use the Pallet Master software is there a way to confirm that I have calibrated things correctly?  I'm concerned that since i have calibrated both monitors now with the color munki those setting might interfere.
3.  If I understand correctly with the Pallet Master software the calibration is done with this software but internally in the monitor's hardware (LUT) instead of through Windows like a 3rd party (color munki) would do?  If so once I have 2 - BenQ monitors there may be a slight difference in calibration but they should be very very close?

Thanks, Aaron

1. I have tried researching this a few times, and have not found a source of info that I would trust. So I don't know for sure which Nvidia cards support multiple LUT's and which don't. I gave up looking all that hard at some point, because my monitors all have internal LUT's which made the video card a moot point for me.

2. PaletteMaster's interface is more rudimentary than most other higher-end tools like NEC's SpectraView system. So it's a bit short on details that can tell you stuff about what was actually done. Running both PaletteMaster and the Munki software definitely would not be recommended unless you know for sure that your video card has 2 LUT's, because the Munki software will load something into the video card LUT, interfering with the signal going to the BenQ monitor. If you really want to confirm the status of your monitor calibration, there is such a function called Display-Check, within the PatchTool software made by BabelColor. This function will validate a calibrated monitor against a bunch of values (ranging from simple to comprehensive), to verify that output flowing to the display through the monitor profile is accurate. I use Display-Check whenever I think something weird may be going on, to prove for sure whether the display calibration is accurate and properly engaged.
http://www.babelcolor.com/patchtool_display-check.htm

3. Yes, you've got it pretty much. If you run 2 BenQ monitors (or 2 NEC or 2 Eizo, etc.), then both monitors will be calibrated uniquely with their own internal LUT's, and neither will interfere with the other. The video card doesn't matter at this point since it's 8-bit LUT is not used. Two paired high-bit LUT monitors should calibrated closely enough that most people won't see a difference, unless it's caused by something else like inconsistent viewing angles or uneven ambient light falling on the monitor surface.