NEC 2490WUXi display profile tests- deepest tones

[AnthonyAdachi]

A few days ago I purchased a NEC 2490WUXi. Sean Reid's review and this forum were of a considerable help in this decision.

I was wondering how others have found the deep shadow detail/dark tonal separations with their NEC 2490WUXi or NEC wide gamut xx90 series?

After calibrating my NEC using the SpectraView II package I tried the first test described in Andrew Rodney's "Testing your display profile" article which reveals the accuracy of black and the ability to distinguish the subtle differences deepest tones.

My results were, with target settings of 120 cd/m2 for monitor intensity & a contrast ratio of 300:1, I'm getting around to a value of 5 before I just begin to see separation.

With the default Photo Edit settings (140 cd/m2) I get around to level 4 before I can see separation.

These tests were done at night in a dark room with no lights on. i.e.- the only illumination was from the display so as to reduce the variable of ambient light affecting the results.

Interestingly, according to Andrew's article, with his Sony Artisan, he could go from level 0 to 1 and see separation. I've heard other comments that the Sony Artisan is significantly better than even the best LCD's in showing deep shadow detail.

If any NEC xx90 owner have done this test, what where your results?

Thanks,

-Anthony

[digitaldog]

You may wish to start at 150 cd/m2 as a "minimum" lower level for luminance and see how the test pans out.

[WillH]

There are a couple of different issues here:

1. The Artisans were typically using a gamma of 1.8 instead of the much higher 2.2 value being used nowadays. This higher gamma means that the first few levels of the response curve are practically zero.

If you do a spreadsheet of a gamma 2.2 curve, and calculate the expected luminance for each step for all levels from 0 to 255, you can see just why you can't see the first few steps:

Digital Step : Calculated Luminance (with 120 cd/m^2 maximum)
0 : 0.0000
1 : 0.0006
2 : 0.0028
3 : 0.0068
4 : 0.0128
5 : 0.0210
6 : 0.0313
7 : 0.0440
..
..
254 : 118.97
255 : 120.00

Hopefully that makes it clear why the first few levels are all but invisible.

2. The way the gamma curve is typically encoded in the ICC profile typically means that digital levels 0 and 1 are indistinguishable with gamma 2.2. This is due to the number of bits used to store the curve data.

3. The Artisan software kinda-sorta "enhanced" the display response in the blacks to make sure this particular "test" passed each time and that each individual step was distinguishable.

A good way to look at the true response of the calibrated display (bypassing all of the color management / ICC stuff) is to use the test pattern built into SpectraView. Look at the black "Quality Control" pattern and you should be able to distinguish all of the letters in the "Quality" and most in "Control".

Some would argue that using L* instead of gamma 2.2 is a better alternative since it provides more steps in the black area.

[AnthonyAdachi]

Hopefully that makes it clear why the first few levels are all but invisible.

That does clear things up a bit.

From my understanding one's eyes have a harder time distinguishing detail in the darkest tones compared to the lightest tones.

What would be a visibly distinguishable difference in luminance in dark tones in a dark room with no other light sources other than the monitor? A difference of .0100?

A good way to look at the true response of the calibrated display (bypassing all of the color management / ICC stuff) is to use the test pattern built into SpectraView. Look at the black "Quality Control" pattern and you should be able to distinguish all of the letters in the "Quality" and most in "Control".

I'll take a look at the test pattern.

Thanks,

-Anthony

[Nill Toulme]

...Some would argue that using L* instead of gamma 2.2 is a better alternative since it provides more steps in the black area.
[a href=\"index.php?act=findpost&pid=200613\"][{POST_SNAPBACK}][/a]

This does indeed make a very noticeable difference.  I'm calibrating my 2090uxi way down at 85 cd/m² because my workroom is dim, and anything higher results in my prints coming out looking too dark in comparison.  At gamma 2.2 I was typically getting a black point of about .38 and a contrast ratio between 200 and 250:1.  Looking at that test target (of which I was previously blissfully unaware), I wouldn't have known there were any letters at all in the black square if you hadn't said so, and on peering at it carefully I could just barely discern the Q.

Using L* (by which I assume you mean monitor native gamma) I have a black point of .19 and a contrast ratio of about 450:1, and I can make out (again just barely, if I really try) through perhaps the T in CONTROL.

Is there any downside to using L* instead of 2.2?

Nill
~~
www.toulme.net

[digitaldog]

Using L* (by which I assume you mean monitor native gamma).....

Actually that's not clear as there IS an L* option in the software (click on custom curve) and there's the option you selected (Native Gamma). They should be quite different.

LStar is not really a gamma but a gray scale constructed so that each step visually appears to be an even increase in brightness from the step before. It is sometimes referred to as the "Equal brightness" scale but if you plot it against a gamma curve in the 2.3 - 2.4 range, the differences should be pretty small.

[Nill Toulme]

Ah... OK, I tried again at L* and this time got numerical results very similar to 2.2:  black point of .35 and contrast ratio of 244:1.  BUT... the low level differentiation is even better than native.  I can see the R in CONTROL pretty clearly and can even make out the last O fairly well if I squint.

Nill
~~
www.toulme.net

[Czornyj]

Ah... OK, I tried again at L* and this time got numerical results very similar to 2.2:  black point of .35 and contrast ratio of 244:1.  BUT... the low level differentiation is even better than native.  I can see the R in CONTROL pretty clearly and can even make out the last O fairly well if I squint.

Nill
~~
www.toulme.net
[a href=\"index.php?act=findpost&pid=200646\"][{POST_SNAPBACK}][/a]

I think L* is not a good idea. If you'll use sRGB, Adobe RGB or ProPhoto as your editing space, you'll lost the better differrentation as soon as you'll start working in a color managed application. The Tonal Response Curve (TRC) of sRGB is something similar to gamma 2,2, Adobe RGB's TRC is gamma 2,2, and ProPhoto's TRC is gamma 1,8. If you'll calibrate your display at L*, the Photoshop's CMM will display an 2,2 or 1,8 gamma image in L* using 8 bit palette for conversion, and as a result you'll lost a few tones - so the image gradation will have worse precision and smoothness.

I'd rather recommend to stick to a TRC of your editing space, and if you want to get better precision in the shadows, try a calibration at a higher black point, or try to fine tune the TRC of the display using gammacomp software.

[digitaldog]

I think L* is not a good idea. If you'll use sRGB, Adobe RGB or ProPhoto as your editing space, you'll lost the better differrentation as soon as you'll start working in a color managed application. The Tonal Response Curve (TRC) of sRGB is something similar to gamma 2,2, Adobe RGB's TRC is gamma 2,2, and ProPhoto's TRC is gamma 1,8. If you'll calibrate your display at L*, the Photoshop's CMM will display an 2,2 or 1,8 gamma image in L* using 8 bit palette for conversion, and as a result you'll lost a few tones - so the image gradation will have worse precision and smoothness.

I'd rather recommend to stick to a TRC of your editing space, and if you want to get better precision in the shadows, try a calibration at a higher black point, or try to fine tune the TRC of the display using gammacomp software.
[a href=\"index.php?act=findpost&pid=200649\"][{POST_SNAPBACK}][/a]

I quite agree! Well said.

[Nill Toulme]

Back to 2.2 then?  And what about native gamma?

Nill
~~
www.toulme.net

[Czornyj]

Back to 2.2 then?  And what about native gamma?

Nill
~~
www.toulme.net
[a href=\"index.php?act=findpost&pid=200656\"][{POST_SNAPBACK}][/a]

If you work in Adobe RGB/sRGB, then gamma 2,2 is the best solution (or, in case of sRGB, it might possibly be better to use a real sRGB TRC) . Native TRC of the display will cause just the same problem like L*.

If you really want to work with a L* calibrated display, then you should render your images to ECI RGB v2 editing color space, that is based on L*. That may be problematic - for example ACR doesn't render images to ECI. And  the conversion of images that were alredy rendered to AdobeRGB/sRGB may cause colorimetric errors.

[digitaldog]

Back to 2.2 then?  And what about native gamma?

Not really necessary or useful since this display handles the adjustments in high bit internally (where native is useful in displays that don't to avoid banding). Also, with Native Gamma, you can't adjust contrast ratio.

[Nill Toulme]

OK thanks.  I don't see a place to adjust contrast ratio with my version of Spectraview II (1.0.30).  There is a choice in Preferences for either "Maximize Contrast Ratio" or "Best greyscale color tracking."  I've been selecting the latter.

Nill
~~
www.toulme.net

[AnthonyAdachi]

Look at the black "Quality Control" pattern and you should be able to distinguish all of the letters in the "Quality" and most in "Control".

•In the black "Quality Control" pattern, with 120 cd/m^2 and a target contrast ratio of 300:1 if I cover up the white pattern I can just barely make out the "Y" in "Quality" but cannot make out any of "Control".

•With 150 cd/m^2 and a target contrast ratio of 300:1 and covering up most of the display except for the black QC pattern I can see the Y and just begin to see the "C" and "O" in "Control".

By the way, I made my target settings by copying the default Photo Editing Target Settings. I didn't see an option to Target a Black Level but the info window says it was calibrated at .49.

If it's relevant, ColorComp was set to On via the NEC monitor's buttons.

Also, in NEC SpectraView prefs what should the "Calibration Priority" setting be? "Maximum Contrast Ratio" or "Best greyscale color tracking"?

Thanks for your help,

-Anthony

[digitaldog]

Also, in NEC SpectraView prefs what should the "Calibration Priority" setting be? "Maximum Contrast Ratio" or "Best greyscale color tracking"?

For most users, Best Grayscale (since you're unlikely to be using max contrast anyway).

[AnthonyAdachi]

OK thanks.  I don't see a place to adjust contrast ratio with my version of Spectraview II (1.0.30).

They must have changed the interface in the later versions. As, I don't see a place to adjust the black point in version 1.0.42.

According to NEC's support site 1.0.42 is the latest version of Spectraview II.

The  Spectraview II pdf manual which came with my NEC seems to indicate that there isn't a way to change the black point with LCD's and only offers a way to change the ratio between the black point & intensity. i.e.- contrast ratio.

-Anthony

[Nill Toulme]

What are the considerations in choosing the contrast ratio — "monitor default" vs. something else?

EDIT:  Well, to answer my own question at least in part... Monitor Default gives me a black point of .34 and contrast ratio of 245:1 at 85 cd/m² and gamma 2.2, while asking for a contrast ratio of 400:1 produces a black point of .45 and contrast ratio of 185:1!  Guess I'll stick with the default... the non-default choices must only be effective for decreasing CR at higher luminance values.

Nill
~~
www.toulme.net

[digitaldog]

An ideal contrast ratio matches that of the print. Since you can build multiple targets (if you're using multiple papers), its easy to target based on that output.