Looking at Monitors

We stare at them day in and day out, yet for all of our assessment of capture processes and equipment, we mostly take displays at their word. I didn’t start this comparison with the intention of a write-up. I had a PA242 that I was having calibration consistency issues with and so asked NEC if they would mind sending me a PA243 to test so that I can more confidently keep them in my suggestions for institutional workflows. Ran into other issues with that (see below) so I directed the same question to EIZO for their CS2420 in roughly the same price range. Having had sat with them all for some months, I thought I’d share my findings.

NEC PA243 and EIZO CS240 are the current sub-$1000 options for wide-gamut, LUT accessible monitors from the two brands. I haven’t had a chance to work with Benq and others so I cannot offer a comparison with alternatives.

Specs

Let’s put some specification differences aside first.

  • Similar gamut coverage, wled backlight on IPS panels of the same size and resolution.
  • Most sources list the NEC as a true 10bit panel while EIZO is listed as 8bit+FRC. EIZO said “we are not able to discuss the panel types and technologies used to achieve a 10-bit a color-accurate monitor”
  • They both have USB3.1 Hubs with the NEC containing a KVM switch that allows you to switch between two computers with independent USB connections.
  • NEC has one USB port on the side while EIZO has all three ports easily accessible on the side
  • The NEC has VGA and a 3.5mm in/out if which can be useful in installing old video art
  • CS2420 uses a 1d 16bit LUT while the PA243 uses a 14bit 3D LUT. When the software utilizes the 3D LUT is a bit obfuscated, but it can be addressed by Lightspace.

Build

For the most part, they are built similarly, being slightly trimmer versions of our panels of yore. They both have handles in the back and feel put together. The EIZO has a touch-sensitive interface that is a lot more minimal and frankly more intuitive to use, I sort of brush my hand roughly around where the power button is and it’s on and ready to go, labels appearing on-screen above the buttons.

Panel flex and frame pressure

Where build matters is how the panels are assembled and affixed to the frame. As displays fight to have slimmer bezels, panel assemblies get thinner, leaving little space for load distribution.

This shows on the PA243 where touching the bezel of the panel causes visible shifts of brightness in the screen as the polarizing films bend from the pressure. Touching the panel directly you can see and hear it move. Even without pressure, you can see how gravity is enough to cause some asymmetry in the image heightened by the lack of retardation films(these reduce the ‘glow’ in IPS panels). With gentle pressure, you can permanently realign pressure points, affecting appearance measurably.

After pressing/pushing on panel
After pressing/pushing on panel
After pressing/pushing on panel
Before pressing on the panel
Before pressing on the panel
[thumb][thumb]

To make sure that this was not a one-off issue, I had them send me another PA243 and sure enough that had the same issue, so does my PA242, and for the record, an older CX271 from EIZO. The CS2420, on the other hand, is well mounted and decoupled from the bezel and does not exhibit any pressure points.

Stands

The stands are similar column-on-turntable affairs, except EIZO has moved its tilt hinge to the column rather than the VESA mount which is a clever little change that rises the midpoint of the display as you tilt it back. At my height (6’2″), this allows it to reach a more ergonomic position from keyboard level. Neither has substantial cable management, NEC has a more significant clip on the back of the column.

Panel Quality / Out of box appearance

They use similar IPS panels: Same resolution, same subpixel array. the NEC has a slightly slimmer black matrix on the color filter layer.

They are both fully matte, anti-glare displays. CS2420 is virtually the same as the PA243W, perhaps a hair less diffusing. For some reason, the EIZO(at least the copy I have) is very difficult to clean without leaving streaks. In comparison, the PA242W had more of a semi-matte finish with a less grainy appearance, though it did not diffuse reflections as well.

Moire in capture is messing with it but you can see the difference between the green which doesn't show stripes and the other colors. The right/left end of the blue bars is a good approximation of it's appearance.
Moiré in capture makes it difficult but you can see the difference between the green which doesn’t show stripes and the other colors. The right/left end of the blue bars is a good approximation of it’s appearance.

NEC PA243W suffers from some sort of dithering / moire effect that shows as a pattern of vertical lines with certain colors. I don’t know why it would be dithering if it’s a 10bit panel, and I’ve not seen it on 8bit panels like the pa242w or the EIZO CS2420. The default blue accent color used in OSX which Spectraview II utilizes in its UI is where it happens to be most prominent . When I first received the monitor, I thought it was a malfunctioning unit and was sent a replacement, alas, it appears to be the ‘nature'(NEC’s choice of word) of this panel.

To make sure it wasn’t just my eyes trained to spot such aberrations, I asked a couple of others (of non-visual professions) to see if they spot anything strange with the display and they were able to immediately point to the stripes. I am a bit surprised to have not seen it mentioned anywhere else but this is a dealbreaker for me and would drive me batty working on images (It shows up on photographic images as well, like warmer skin tones and skies).

Both come with calibration reports from the factory with several modes that are tagged to that calibration (AdobeRGB, sRGB etc), and a custom uniformity correction matrix/map. If you disable the uniformity correction, you can see that the EIZO has a symmetrical vignette that is fixed by darkening the middle of the screen (which affects contrast and makes it appear a little duller). While NEC has some left-to-right variation in color, the symmetrical vignetting is not as pronounced though certainly present. NEC allows you to set the correction in degrees (0-5) while it is either on or off on the EIZO. Make sure that you do your characterizations at the intended uniformity correction level.

NEC
EIZO
[thumb][thumb]

Pulling up a black screen (RGB000 fill in Photoshop), EIZO appears quite a bit darker and more even despite measuring similarly in the 0.12cd range. See the build section for how the ‘IPS Glow’ of the NEC can shift with pressure.

NEC
NEC
NEC
EIZO
EIZO
Black levels and uniformity

Calibration Software

Spectraview II UI & Options

Nec – Spectraview II 1.1.40
EIZO – ColorNavigator 7.0.8.3

Color Navigator is more of an always-present app that lives on your menu bar and allows you to change between characterization presets on the fly. Spectraview hasn’t changed much over the years and is more of a launch, characterize and disappear sort of affair. It will check whether the monitor preset matches the mode selected in the software, but does not offer preset switching in the same way.

There used to be a Spectraview Profiler application from Basiccolor with more characterization options but that doesn’t seem to be getting updates. On Windows, you can also use Lightspace for either display with a lot more control over the calibration/characterization options.

Spectraview is a little more basic when it comes to presets and mode management, and rather lacking in validation settings. While they offer access to similar diagnostic tools and characterization parameters, Color Navigator allows you to install custom validation charts in csv form, defaulting to ISO 12646. NEC only does a basic primary+grayscale validation, reporting only on the grayscale delta E(CIE94) along primary readouts.

CN7 UI

CN7 allows you to see and modify all the presets stored on the monitor. You can change some parameters such as luminance on the fixed [STD] modes, and store unlimited independent presets on the Custom [ADV] slots. You can switch between the five stored calibration presets on the fly, as well as between the saved preset parameters under each advanced slot. CN6 was a lot more intuitive and clear in this process with CN7 feeling a bit strapped together by engineers.

You can see the vast array of option-settings (checkboxes that toggle other checkboxes) under CN7 settings. It looks a bit like the complete option exposures handed to UX designers who would then design an end-user friendly UI.

option settings

At the point of writing, there are some bugs in CN7: The test patterns show 8-bit graphics which might lead you to worry about banding, the calibration window makes it seems like it is skipping lighter gray values (they likely forgot to put the gamma calculation in the interface script that puts a dot on a neutral axis to approximate measurement location), and there are some translation mistakes.

I have not used Color Navigator Network designed to administer multiple monitors over a network, but they do offer such a solution for environments where operators are not supposed to fiddle with their display. The use would be a little more limited in the case of the CS2420 as it doesn’t have a built-in colorimeter.

Performance and General Notes on Reference Displays

Reference Equipment

Unfortunately, I don’t have access to a proper spectrophotometer at the moment, having sold mine when I moved from London. The tests would have been much better with a Klein. I had an Xrite i1Studio from a while ago and so I sent that, along with my i1d3 units to x-rite for an assessment, and they all came back within specs, so here is the lineup:

  • x-rite i1 display pro Rev B (i1d3)
  • x-rite i1 display pro NEC rebadge Rev A (i1d3 NEC)
  • x-rite i1studio spectrophotometer (i1studio)
  • Eizo ex3 aka. spyder5 (ex3)

The displays were connected one at a time via displayport (see HDMI 8bit note issue in the Other Notes section) and all measurements were done in virtual darkness.

X-rite tested the two i1d3 units and their technicians cannot see a difference between characterizations using i1profiler. I can see and measure it, so I don’t know what to make of that response.

The photographs are taken using a D850 using a 50mm prime with the monitors placed (to the best of my ability) at the same height and distance, leveled. I shot in RAW to avoid local adjustments and set them all to the same white balance. I have intentionally defocused the images at capture to minimize moire and false color issues affecting comparisons.

Appearance

NEC differs visibly from calibration to calibration, generally measuring warmer than the EIZO when using the same device (id3).

Comparing Grainger Rainbows made in ProphotoRGB, there is a distinct step in the horizontal gradations of greens to reds and in the vertical gradation of blues towards pinks in the NEC. EIZO renders it a little more smoothly. This is not necessarily indicative of anything but I thought I would note it.

See the Panel Quality section above for a black level comparison.

Consistency

Calibration to Calibration

With the NEC, I had visible differences from calibration to calibration apparent in validations that were ran right afterward. Perhaps this has to do with shifts over time that go beyond the suggested warm-up period, either way, it was unsatisfactory. I wasn’t able to get consistently low values in validations immediately following calibration except for once (0.46W/0.91AVG after 96min warmup). I think some of the shifting might be due to the panel flex caused by the weight of the calibration device. I switched to a near-vertical angle to minimize this potentiality.

Note here that spectraview does not include the dark values in the little validation results popup, you have to switch to the tracking tab and click the checkbox to see validation results that include dark values.

EIZO was pretty consistent from calibration to calibration using the same device, and validations immediately after calibration read quite well. (0.01W/0.45AVG after 93min warmup). The Delta E 0.5 difference between the displays is not significant, the comparison is provided for the context of getting consistent results.

Over Time / Warm-up

Regardless of manufacturer suggestions, let your monitor stabilize for at least an hour before calibration

I was very surprised to find the NEC to have a wild color cast after long periods of sleep, neutrals appear entirely moss green (starting at Delta E Max 9.37 Avg 5.15) for more than 10 minutes at times. I checked in with them and they confirmed that this was to specification(!).

After Warmup
After Warmup
After Warmup
Before Warmup
Before Warmup

With this knowledge, I decided to measure stabilization performance and was surprised to find that both monitors take a lot longer than advertised to stabilize.They vary quite a bit from one wake-up to another so I took a few measurements, but did not have the privilege of a controlled experiment with precise sleep/wake times. I tried to keep the sleeping period long enough with the hope that beyond several hours it would not matter.

The NEC appears to continue the shift in the white point past the one-hour marker moving away from the validation reference. I don’t have enough data to see whether it simply fluctuates or whether it has to do with the warmup time at the point of original calibration(2hr+ for all measurements)

Once stabilized, both yield similar results with the EIZO doing ever slightly better. The black point is consistently off from the ideal target of absolute black by about 1.5-1.7 Delta E, White point is off by 0.2-0.35 with the EIZO dropping to 0.1s in one measurement cycle, and the average is about 0.5-0.7 off.

EX3 reads a much higher Max. Delta E as it just cannot cope with low luminosity measurements.

Turning DUE off can drop the black point delta E a little further to the 1.35s and the white point a bit as well, I haven’t graphed it since both require uniformity correction for our purposes.

For reference, the older PA242 can consistently hit 0.45 Average 1.05 Max despite having a visually warm white point that can measure in the low 0.0x region. (i1studio measurements would dispute that white point).

Across Calibration Devices

Choose and stick to a single calibration unit in your studio, don’t expect calibration devices to yield comparable results, even if they are the same model

This has been a bit of a pet peeve of mine that I periodically bother x-rite and display manufacturers about. There is a large and visible difference in characterizations made not only with different devices but also with different revisions of the same device (i1d3). As they are collectively marketed to be ‘calibration’ devices that bring your display to a known and common state, most people assume that they can be mixed and matched with some consistency. Observation and measurements suggest otherwise.

On the NEC I was able to bring the two i1d3s close enough to each other by virtue of calibration errors, and the neutrals compression of the RAW processor is not doing it justice in the photos, but they remain visibly different.

EIZO
EIZO, in focus with moire
NEC via NEC
NEC via NEC
NEC via NEC
NEC via X-rite
NEC via X-rite
NEC
[thumb][thumb][thumb]

White point of EIZO measured with CM after it was characterized using an i1d3

CM: CCT = 6356K (Duv -0.0004) / XYZ: 102.827439 105.871928 116.803909 / D50 Lab: 102.227434 1.237393 -20.744766

White point of EIZO measured with CM after it was characterized using an EX3

CCT = 6836K (Duv 0.0019) / XYZ: 106.702574 111.450896 128.239556 / D50 Lab: 104.268671 -1.222888 -24.327730

White point of NEC measured with CM after it was characterized using the Xrite i1d3

CCT = 6459K (Duv 0.0011) /  XYZ: 104.056893 108.122416 119.304509, D50 Lab: 103.059276 -0.320252 -20.901932

White point of NEC measured with CM after it was characterized using the NEC i1d3

CCT = 6502K (Duv 0.0016) / 108.484070 113.080948 124.907870, D50 Lab: 104.852175 -0.875619 -21.297631

White point of EIZO measured with i1d3 after it was characterized using an i1d3

CCT = 6628K (Duv 0.0044) / XYZ: 99.666623 105.624309 115.767630 / D50 Lab: 102.135190 -3.653796 -20.237664 

White point of PA 242W, the older model, measured with CM after it was characterized using an Xrite i1d3

CCT = 6334K (Duv 0.0000) / Result is XYZ: 103.406250 106.707764 116.933123, D50 Lab: 102.537746 0.856544 -20.292531          

Characterizing the NEC with the i1d3 Rev B and Rev A yields visibly different results, i1studio yields yet another. Since the i1studio is a spectrophotometer that can make absolute readings, I used that to measure the white points achieved with the two i1displaypro units.

Spectraview has a custom EDR file (spectral reference for the colorimeter and the specific panel+backlight combination), but I don’t know what revision of i1d3 it was made for. NEC wasn’t able to clarify which revision or device the .edr file pertained to and simply noted that I shouldn’t expect different devices to match.

EIZO does not support the NEC-OEM i1d3 so I asked them for an EIZO EX3 to add some variation to the mix. Their engineering department told me that they did not use spectral reference files for the X-rite i1d3 but going through their CN7 App resources I was able to locate a few generic .edr files from X-rite. One could potentially inject custom edr files for the panel. I haven’t tried it as I don’t know whether they are being used, and I don’t trust the i1studio as a reference spectrophotometer to base all other measurements from. See https://github.com/fhoech/ccss2edr for making your own .EDR references.

On the EIZO, EX3, i1d3, and i1studio yielded visibly different results. EX3 was very similar to the factory calibration with the i1d3 being significantly red, and the i1studio significantly yellow in comparison. Unfortunately, EX3 cannot measure low luminance values well, so it consistently read low contrast results and yielded a slightly shifted black point.

I asked EIZO about why they don’t have reference files made specifically for their panels since all it takes is a couple of measurements with a spectro, and they said they don’t provide support for third-party meters. Of course, if you try to correlate displays, or profile those without internal sensors, you have to use a meter. It is also a bit baffling that they have chosen to source Spyder5( Rather than the newer Spyderx ) as their colorimeter of choice.

Sidenote on the Spyders, both the EX3(Spyder5) and the newer SpyderX use dye-based filters with questionable long term accuracy.

NEC PA242W vs PA243W

Calibrating with the same devices, the two monitors look nothing alike. I tried to photograph it but the raw processors compress near grays too much, essentially, the PA242W appears totally green/yellow next to the PA243W to the point that if they were prints I would consider it a toned print. Something to keep in mind if you are considering one as a second display. If both are set to factory presets, the PA242W is still quite different but not as green while the PA243W looks similar between factory and calibrated presets.

Other Notes / HDMI warning

With my OSX(Mojave) Desktop, HDMI uses 8bits rather than 10 despite reading RGBA101010 in system profiler. Throwing up a 16bit gradient in photoshop, you can see stepping on one and not the other. Would be happy to hear if you have similar results. This is with a RX560 GPU that has a displayport, HDMI 2.0b and DVI outputs. the Displayport works fine. Technically HDMI 2.0 supports 10bit RGB, perhaps it is a card limitation.

In Sum

Stick to a single calibration device across the studio unless you can confirm that they can deliver similar readings.

Let your screens warm up for 90 minutes before a calibration, and 45 minutes before starting color critical work.

Don’t expect two displays to match, especially if they use different software.

As far as the two monitors go, the NEC PA243W has significant issues with the stripe patterns, panel flex, poor dark uniformity, and lackluster calibration-to-calibration consistency. While CS2420 has some issues, its general performance and consistency makes it the better of two displays.

If you can afford it, the CG2420 step up might be worth it for the internal calibrator, a better polarizer for black uniformity / increased contrast, a hood, and (advertised) faster warmup. You have to step up another block if you need Rec709/video specific calibration. On the NEC side, PA271Q is an option that is supposed to be certified for quick warm-up, I reached out to see if they would be interested in sending a test unit.