OSX El Capitan + Eizo black clipping

[larkis]

When I export jpg images from Lightroom and open them using apple's preview, the black levels seem to be clipped. I have noticed that this happens only on my Eizo screen, not my secondary apple monitor. This behaviour was not happening with previous versions of the operating system. Does anyone have a similar problem ?

[digitaldog]

When I export jpg images from Lightroom and open them using apple's preview, the black levels seem to be clipped. I have noticed that this happens only on my Eizo screen, not my secondary apple monitor. This behaviour was not happening with previous versions of the operating system. Does anyone have a similar problem ?

Check to see if the Eizo software is building a Version 4 (V4) ICC profile and if so, set it to V2 and test again.

[kirkt]

This issue has been bugging me for a while.  I think I have found a solution, but you should try it for yourself to see if it works for you.  I have noticed that opening a JPEG, for example, in Photoshop versus Preview is an easy way to see if the problem is affecting your system.  It sounds like you have some candidate images that appear blocked up in the deep shadows/blacks - this is what we are trying to cure.  In my case, the affected image is rendered correctly in Photoshop but appears blocked in Preview.  Same image, with an sRGB or any other color profile embedded, etc.

In the Eizo ColorNavigator software, when you are creating a new profile for your display and manually setting up the various parameters, you get to the last screen (see attached screenshot) where you can name your new profile before you proceed with the calibration and profiling.  At that dialog, select the "Customize profile" button.  It will take you to a dialog where you can adjust things like the version (as Andrew pointed out above - maybe select v2 if you want here).

What I found was the culprit was the checkbox that is labeled "Reflect Black Level in Tone Curve."  UNCHECK THIS BOX and run your calibration and profiling operation again.  Ideally, you want to make a profile with the box checked and then repeat the process with the exact same parameters, but without this box checked.  Then open an image that shows the blocked shadows and, with the image displayed in Preview, switch between the two profiles (using the System Prefs > Display list).  You should see the problem go away with the profile where the black level in tone curve box has been unchecked.

Hope this solves your problem - it has been plaguing me since Mavericks, but it only rears its ugly head every once in a while. 

Attached is an example from my machine on an image that displayed the blacks issue and a comparison I made to demonstrate the effect of the above solution.

Kirk

[tho_mas]

Does anyone have a similar problem ?

same here. Everyone using a monitor with a black point above 0 cd/m2 will see this issue as the OS ignores the black point of the monitor profile (the system assumes black is RGB 0|0|0, L*0 respectively). So everyone with a decently calibrated monitor...
The native "white point" of almost all monitors shows a pretty cold (blueish) black point. Now, when you calibrate your monitor for a neutral grey axis at, say, 5400 Kelvin from black to white the RGB gain of the monitor has to be adjusted. To get a "white balanced" black point the reds and greens will be lifted (in the monitor profile)... and this is why the absolute luminance of the black point raises. But since the OS assumes pure black the increased luminance of the blacks get clipped.
Me personally I do boost the black point of my monitor to 0.3 cd/m2 intentionally (for several reasons) ... and I only have 2 softwares that handle the black point of the monitor profile (and black point compensation) as supposed to: Photoshop and Capture One (well, and Adobe Acrobat...). All other imaging softwares fail with regard to displaying the blacks correctly.
The workaround Kirk described works very well.. but of course the profile version that ignores the actual black point is pretty inaccurate ... so don't forget to switch back to the main monitor profile for serious editing (in Lightroom + Photoshop).

[digitaldog]

same here. Everyone using a monitor with a black point above 0 cd/m2 will see this issue as the OS ignores the black point of the monitor profile (the system assumes black is RGB 0|0|0, L*0 respectively).

Well an Lstar 0 document appears identical to me in Photoshop and Preview. Black level calibrated (SpectraView) to 0.50 cd/m^2.

[tho_mas]

Well an Lstar 0 document appears identical to me in Photoshop and Preview. Black level calibrated (SpectraView) to 0.50 cd/m^2.

well... monitor-black is monitor-black :-)
what I am talking about is levels short above pure black (that may get clipped to pure black... depneding on the black point of your monitor and how softwares apply BPC).

The attachment (best to be opened in Photoshop) shows screenshots of the same TIF file opened in Photoshop, Capture One, Affinty Photo, Iridient Developer ... and the undefinable soup on the right is Apples "Preview".
Photoshop and C1 match 100% on my machine (at least when working with RAW files ... but the image below is a TIF file and for some reason there is a minimal mismatch between PS and C1). Affinity and ID also do apply BPC... but different to PS and C1 (and different to each other). PS, C1 and AP do show all numbers from 1 to 8 ... though in AP the low levels are a bit too dark. ID clips the numbers "1" and "2". Preview clearly clips all values close to pure black...

[digitaldog]

well... monitor-black is monitor-black :-)
what I am talking about is levels short above pure black (that may get clipped to pure black... depneding on the black point of your monitor and how softwares apply BPC).

Ah, got it, thanks. Yes, I do see a visual difference with Lstar2 between Preview and Photoshop. Preview is clearly wrong.

[tho_mas]

Preview is clearly wrong.

Color Management on Mac was much better (more consistent) before Apple "converted" to Gamma 2.2 ... and redesigned the whole systemwide color managment. To be honest I can't figure out how it works exactly. Then again, who cares?! :-)

[Tim Lookingbill]

Thomas, your posted jpeg test image shows up identical in both Preview and Photoshop CS5 on Mac OS 10.6.8 and using a Colormunki Display non-LUT profile.

Firefox browser shows it slightly lighter where the "2" in Mac's Preview app barely shows up and measures 666RGB in Apples ColorMeter and measures 111RGB in Photoshop and appears black. All absolute blacks are at 000RGB both in Apple Colormeter and Photoshop.

It appears their may be quantization errors in calculating the arc shape of the (LUT?) curve that attempts to scale tonality smoothly out of black across several apps as you've shown but I wonder if a LUT display profile is compounding the issue. Or is LUT shown in the Eizo software the vctg tag Macs use? I'm never sure how LUT is defined in this regard.

I also never know if the curve shape is customized mathematically on the fly or is a standard curve that just happens to be slightly differently shaped in the shadow regions across different color managed image viewers.

[tho_mas]

Mac OS 10.6.8

with regard to CM Snow Leopard was a dream to work with compared to later OS X versions ... IMO...

Or is LUT shown in the Eizo software the vctg tag Macs use? I'm never sure how LUT is defined in this regard.

"LUT" in the Eizo software refers to the correction table that gets loaded into monitor (hardware), AFAIK. The monitor profile available on the system (iColor Sync) is a matrix based profile in any case (4 KB size, Chromix Color Think reads a 256 point TRC out of it...).

I also never know if the curve shape is customized mathematically on the fly or is a standard curve that just happens to be slightly differently shaped in the shadow regions across different color managed image viewers.

not quite sure here but I think the main difference is how softwares apply black point compensation (starting with how the actual black point of the monitor is being taking into account).

[Doug Gray]

with regard to CM Snow Leopard was a dream to work with compared to later OS X versions ... IMO...
"LUT" in the Eizo software refers to the correction table that gets loaded into monitor (hardware), AFAIK. The monitor profile available on the system (iColor Sync) is a matrix based profile in any case (4 KB size, Chromix Color Think reads a 256 point TRC out of it...).
not quite sure here but I think the main difference is how softwares apply black point compensation (starting with how the actual black point of the monitor is being taking into account).

Yep. LUTs refer to the EIZO monitor's internal LUTs. Also, the vctg tags are just unity gain and reset the video cards to unity. That's a good thing. No jumps like you get in some profiling software w/o smart monitors.

TRCs vary. They default to do BPC to the monitor's black point, which is settable. You can also get TRCs that implement the sRGB ramp with a linear ramp attached to a gamma (2.4 I think) which is the standard sRGB tone curve.

I normally default to sRGB and switch over to high gamut in Photoshop if working with more saturated colors.

[Tim Lookingbill]

Yep. LUTs refer to the EIZO monitor's internal LUTs. Also, the vctg tags are just unity gain and reset the video cards to unity. That's a good thing. No jumps like you get in some profiling software w/o smart monitors.

TRCs vary. They default to do BPC to the monitor's black point, which is settable. You can also get TRCs that implement the sRGB ramp with a linear ramp attached to a gamma (2.4 I think) which is the standard sRGB tone curve.

I normally default to sRGB and switch over to high gamut in Photoshop if working with more saturated colors.

Where is this BPC settable, Doug? In Eizo's software during the creation of the profile? If so, with all the variations to BPC slope curve lift out of black shown in thomas's screengrab, what's the point of setting it?

[Doug Gray]

Where is this BPC settable, Doug? In Eizo's software during the creation of the profile? If so, with all the variations to BPC slope curve lift out of black shown in thomas's screengrab, what's the point of setting it?

EIZO's CN software lets you set the black point at either the minimum (highest dynamic range) or a specific value. I use the minimum, which is recommended, even though the ambient light off the monitor is about 300:1. I sometimes turn the lights out

You can also either hard clip at the BP or have it smoothly approach the BP. You can even load your own tone curve. I normally run with the minimum set and don't hard clip. When I soft proof all my papers have a much higher BP than the monitor and this gives the best setting for softproofing using paper black PB or paper white. There are other workflows people use. Many prefer setting the monitor BP to .3 or so which can better preserve shadows in a bright environment and is better for producing images for web viewing but then if you select paper black or white soft proofing it can over lighten shadows.

[Tim Lookingbill]

Doug, on the Eizo set to minimum black point how much lighter is 000RGB to the black plastic bezel surround. Is the difference close to the example shown below?

[Tim Lookingbill]

Good grief, this is why black point calibration can be a bitch when editing shadow detail. That first sample I posted of my 27" LG LED and processed in CS3 found was too light and applied a tweak to the Medium Curve in ACR to make it darker.

On second inspection it's actually too dark and switched to the unedited Medium Curve in ACR CS5 and now looks as it should. Apple Colormeter readings indicates I barely made a dent but visually there is a difference.

[Doug Gray]

Tim,

I'm using Win 10 and a CG318 and ColorNavigator. I don't have a colorimeter and my spectro is not very accurate at very low luminance. The monitor does have a built in colorimeter so I assume the minimum brightness RGB(0,0,0)from that is accurate. It's about .05 Nits. with white at 80 nits.

[Tim Lookingbill]

So you don't have a black bezel surround like I have shown in the posted image of my LG?

[Doug Gray]

So you don't have a black bezel surround like I have shown in the posted image of my LG?

I do but the question is complicated because the brightness of the bezel is proportional to room illumination while the brightness of a monitor depends on the dynamic range, what the black point is set to, and the room light reflected from the monitor surface. In normal environments my bezel is brighter than the screen. In a darkened room at night this can change.

The bezel is roughly the same blackness as the black patch on a colorchecker. It reflects about 4% of the light hitting it. The light shining on the monitor's surface is approx. 20 lux which would produce a reflected brightness of 20/Pi, or a bit over 6 Nits on a perfectly white bezel. Since the bezel is fairly dark and only reflects 4% of the light hitting it, that is about .25 Nits equivalent from the Bezel. Reflected light from the screen is about .2 Nits with only about .05 Nits coming from the LCD backlight, so the bezel is slightly less dark.

But that's with the lights on. With the lights off the background light drops to about 2 Lux (from a curtained window) so the bezel reflected light is now the equivalent of a .025 Nits. The reflected light from the screen drops from .15 Nits to .015 Nits. The .05 Nits from the LCD backlight adds to that yielding .065 Nits effective from the monitor screen displaying RGB(0,0,0). So the monitor is now slightly brighter than the Bezel with the lights off.

[tho_mas]

There are other workflows people use. Many prefer setting the monitor BP to .3 or so which can better preserve shadows in a bright environment and is better for producing images for web viewing but then if you select paper black or white soft proofing it can over lighten shadows.

actually raising the black point a bit (in the range of 0.2 - 0.5cd/m2) is not aimed at viewing web images but at calibrating the monitor to a contrast ratio that is a bit more print-like. As a rule of thumb the monitor contrast shouldn't be higher than 1:400 ... also because at a higher contrast ratio it's stressful for our human eyes to differentiate very dark tonal values. When softproofing shadows may over lighten a bit (colormetrically) but in the range of the values we are talking about here the differences are beyond perceptibility. Much more important is how good shadow differentiation of a print is represented on screen (with black ink simulation or paper white simulation enabled)... and in this regard raising the black point of the monitor a bit is rather helpful than misleading.
In addition all "consumer" colorimeters are pretty weak in measuring very dark tonal values accurately. So lifting the black point of the monitor a bit will improve the calibration result since the grey axis will be more accurate (I am talking about hardware calibration, of course).

[Doug Gray]

actually raising the black point a bit (in the range of 0.2 - 0.5cd/m2) is not aimed at viewing web images but at calibrating the monitor to a contrast ratio that is a bit more print-like. As a rule of thumb the monitor contrast shouldn't be higher than 1:400 ... also because at a higher contrast ratio it's stressful for our human eyes to differentiate very dark tonal values.

I mostly agree with this but note that EIZO, who markets primarily to this demanding segment, recommends the default BP of "Minimum."  I see advantages and disadvantages of both. Note that most web pages are not viewed in subdued lighting where one needs high dynamic range. Reflected light is the biggest problem. A high BP used in creating these will provide a better viewing experience when in typical web viewing environments.

When softproofing shadows may over lighten a bit (colormetrically) but in the range of the values we are talking about here the differences are beyond perceptibility. Much more important is how good shadow differentiation of a print is represented on screen (with black ink simulation or paper white simulation enabled)... and in this regard raising the black point of the monitor a bit is rather helpful than misleading.

Indeed, when room illuminance is such that the ambient reflected light from the monitor's surface is much more than the monitor's intrinsic black point the darker areas are subsumed by the reflected light. That's the best argument for setting the BP higher than minimum.

In addition all "consumer" colorimeters are pretty weak in measuring very dark tonal values accurately. So lifting the black point of the monitor a bit will improve the calibration result since the grey axis will be more accurate (I am talking about hardware calibration, of course).

Spectros are much worse than colorimeters at measuring low luminance levels. In any case the noise floor creates some uncertainty about a monitor's true black point but, in general, reflected light, not the monitor's actual "black" levels dominate in most room environments.

So it really depends on how much reflected ambient light as to where to set the BP. In a brightly lit office environment one might even want to set it at 200:1. In more subdued lighting 400:1 could be best. In a darkened room the minimum BP would work well.

It's also a tradeoff between accurate soft proofing and normal workspace editing. For matte type prints with a Dmax of 1.6 or so the tradeoff is minimal and setting a higher monitor BP is quite helpful and softproofing is hardly affected. For higher Dmax prints, the softproof image dark areas can appear lightened too much.

Ideally, color management should, in addition to the monitor's physical BP capability, include luminance estimates of reflected room lighting.