- This topic has 83 replies, 5 voices, and was last updated 9 months, 2 weeks ago by
.
-
Posts
-
On a printing press, by altering the amount of “100% Cyan”, “100% Magenta” and “100% Yellow”, it’s possible to affect how the “300% CMY” prints, more “brown”, more “green”, more “blue” and so on. (Vital for IDEAlliance “G7” method)
Ideally, we would want the 300% CMY to print “neutral” but that’s not always possible. Historically, on a sheetfed press, on good coated paper, the 300 CMY used to print “Brown” but not anymore. Today, it can be “anything” – really.The idea is that, to some extent, the appearance of that 300% can’t be controlled.
Now what about on a monitor? Can the appearance of RGB = 0,0,0 be “controlled”, in some way?
I am thinking through the OSD, for instance? If I change the “Gain”, is Black affected? Or is it “independent”?Something tells me it is NOT idenpendent. Maybe on the “better” displays, it is?
I have yet to try this on my Dell…
You can try to alter dark grey color by “offset” RGB setting in OSD. It may modify white cofigured by RGB gains (the ones you use to tweak white), so you may need to go from gains to offset a few times to get what you want.
DisplayCAL popup to tweak whitepoint should have an icon (RGB=black) to change to this dark grey setting an use offsets.
Do not push offset way off its 50 default value in your dell.Full black point correction will be destructive in contrast so usually it it not recommended. Fixing dark ger with offset RGB controls in OSD should be enough.
Additional comments on LCD response and setting blacks.
Per the DisplayCal documentation, calibrating an LCD leads to a tradeoff to fit a power function response (gamma) to the LCD limited contrast ratio.
BACKGROUND:
CRTs had a natural power response where the contrast ratio was a function of an industry convention for peak white and a black level set to slightly above veiling glare in a studio (very dim) or office (dim) environment. According to the BT.1886 Reference EOTF (baseline decode for Rec.709 content) the contrast ratio for legacy studio monitors was about 50,000:1.
Today’s commodities IPS LCDs have a CR of about 1200:1 max and 700:1 typical. This response is a window that slides up and down with peak white (the LCD is an aperture for the backlight and so defines the CR). The out-of-box response curve of an LCD is a nominal gamma 2.2 or 2.4.
Because of the limited contrast ratio, a calibration choice must be made between tracking an actual power law response from peak white. Your choice is to incur a roll-up of response near black (with some attendant crush) or offset the entire response curve to maintain proportionality across the response range while incurring a brighter than spec response. The former is truer across the majority of the response curse and the slight crush is very tolerable for mass media (ignoring engineering and medical applications). The latter is less true to 1886 spec across the range but favors detail in the darkest range.
DisplayCal also gives you two controls for black calibration with its “Black output offset” and “Black point correction” controls.
Black output offset selects whether the darkest range should roll-up towards display black level, or offset the entire response range as just described.
Black point correction trades off neutral tracking near black. The best neutral tracking near black incurs loss of contrast ratio because the only way to color correct black on an LCD is by raising black level, due to LCD fixed lower limit for aperture based on polarization. As vision sensitivity to the neutrality of black is much lower than for white (except in unusual conditions), the default is to seek best contrast and let color of black fall where it may. But you can choose the degree of this tradeoff.
See the DisplayCal online docs for “Calibration.”
NATIVE RESPONSE FOR A DELL UP2516D WCG MEDIA MONITOR
Based on the above observations, a question that may be in the back of your mind is what trade-off might Dell choose for mapping power response to the limited contrast ratio?
I’ve attached an image that shows some reference responses and measurements for a Dell UP2516D (this was a semi-professional WCG IPS display from 2016).
You will find that the UP2516D native response tracks an inverted 709 OETF for the top 1/3 of its response, then dips with a knee towards the BT 1886 reference CRT EOTF response and rolls up near black to fit its input range into the limited contrast ratio of about 1200:1. Dell’s alignment preserves highlights, enhances mid-tone contrast, and sacrifices some shadow detail in a way that fairly matches overall response expectations for photo-realistic media.
CHOOSING SETTINGS FOR YOUR DELL
Based on my experience with 4 generations of Dell monitors, I believe you’ll find the smoothest response and widest dynamic range by starting with factory default settings then using manual color controls, custom RGB gains to set white with the aid of your i1Display.
On the UP2516D the “brightness” control (a little sun) sets the backlight intensity, while the “contrast” (a split circle) limits the LCD as an aperture. Default brightness (=75) maximizes dynamic range without clipping whites. Lowering brightness (<75) closes the aperture linearly across the response range to compress the contrast ratio (black remains constant for a given backlight level). Raising brightness (>75) clips whites and expands contrast of mid-tones. To set white manually, there is a “custom color” setting with “gain” and “offset” controls. The gain affects all levels equally across the response range. The offset affects darker levels more than lighter levels. Lowering offset cannot expand max contrast, but raising offset reduces contrast, so leave offsets at default and set white with gain.
Wire,
Two questions / clarifications, if I may be so bold to ask you.
First, you wrote :
Black output offset selects whether the darkest range should roll-up towards display black level, or offset the entire response range as just described.
By default, the value is set to 100%. I noticed that, as soon as I change the value down, the selected Tone curve entry changes to “Custom”. My question is how does interpret this 100%? It’s not longer a “pure gamma value” from white to black?
Would you whether that setting is actually the same as Argyll’s dispcal “-f” flag :
-f [degree] Amount of black level accounted for with output offset (default all output offset)Second question. You wrote :
Black point correction trades off neutral tracking near black. The best neutral tracking near black incurs loss of contrast ratio because the only way to color correct black on an LCD is by raising black level, due to LCD fixed lower limit for aperture based on polarization. As vision sensitivity to the neutrality of black is much lower than for white (except in unusual conditions), the default is to seek best contrast and let color of black fall where it may. But you can choose the degree of this tradeoff.
I confess this language would leave many users scratching their head. I won’t criticize you for employing technical terms (geek speak) as “tracking” but… Good thing I experimented with this particular setting on my own and found out exactly what that does to “Black calibration”. I’m not as familiar as you with regards to the precise hardware components involved in a typical LCD monitor but I get it that this is a trade off because of the fixed nature of the beasts.
Needless to say, better monitors like the NEC PA271W I had on my desk for years don’t suffer from the poor black point discrepancy that this “Office” Dell clearly exhibits.
“50,000:1” contrast ratio? Say the white Luminance was set to 100 cd/m2, what does that mean for the microscopic black luminance?
Thank you for your comments. Always well-regarded.
Needless to say, better monitors like the NEC PA271W I had on my desk for years don’t suffer from the poor black point discrepancy that this “Office” Dell clearly exhibits.
Although different Quality Control may play a role in results… after reading ArgyllCMS mail list it seems that you have been using all your LCD displays in a wrong way and this is the main source of your problems.
And if PA271W is not exhibiting these issues are likely because another undisclosed software (likely Basiccolor) is doing it right for you by DDC/CI + HW cal.Contrast OSD = default, unless some HW paper contrast ratio is needed
Brightness = white level
RGB gains = fix White point (try to keep one RGB at 100%)
RGB offset = try to correct dark grey color, do not push them too far away from start
So any further attempt to request assitance should start by using your LCD displays this way, because otherwise it will be pointless
“50,000:1” contrast ratio? Say the white Luminance was set to 100 cd/m2, what does that mean for the microscopic black luminance?
CCFL LCD have no actual dynamic contras ratio, marketing gibberish. Same for LED LCD … unless full led array local dimming fo HDR ans such.
So 800-1200:1 typical static conbtrast ratio.
Thank you for your comments. Always well-regarded.
Also it is meaningless to compare white point and black point by CCT, it should be compared by actual color distance (visual color distance percieved by humans).
So a 0.1 nit black 8000K to 100nit D65 dE00 should be lower than 1 nit black 8000K to 100nit D65 dE00So I would start over and it’s very likely that with this reciepe most of your issues wil go away as with any other DisplayCAL/i1Profiler user
Contrast OSD = default, unless some HW paper contrast ratio is needed
Brightness = white level
RGB gains = fix White point (try to keep one RGB at 100%)
RGB offset = try to correct dark grey color
Roger wrote:
Black output offset selects whether the darkest range should roll-up towards display black level, or offset the entire response range as just described.
By default, the value is set to 100%. I noticed that, as soon as I change the value down, the selected Tone curve entry changes to “Custom”. My question is how does interpret this 100%?
“Custom” simply demarks that a non-conventional response is chosen, which can be important when we examine profile metadata, reports/logs.
I haven’t investigated the precise effects of these controls. Maybe Vincent can speak to this?
It’s no longer a “pure gamma value” from white to black?
In order to discuss this, we’ve got to be clear about the term “gamma”:.
So here’s what I think about the term.
First and foremost, the term gamma is a holdover from engineering of CRT display devices. You might find the Greek letter on display device schematic next to an icon for a light-emitting vacuum tube to remind the reader of a nominal power response characteristic of the component.
(As to the nominal response, according to my studies, Sony produced a paper from around 1998 breaking down the traits that define gamma to 5 degrees of approximation for their professional CRTs displays.. Sony’s conclusion is that their CRTs, when carefully aligned for their target environments yield an effective power response of 2.25. This paper was produced in the context of the sRGB standard being finalized. I can find it for you if you’re interested).
A PURE GAMMA DISPLAY RESPONSE IS GENERALLY IMPRACTICAL
(Pls excuse my strained language and any typos in the following I didn’t proof read it)
There can be no “pure gamma” response in a practical display device because a true power function has infinite range. Actual displays present a contrast window.
What do I mean by contrast window?
There are at least two important factors of theory to contend with when examining contrast in display application:
One: In a practical visual sense, every conventional panel-type display is contrast limited by a combination of its peak brightness and by ambient light scattered on the display surface (glare) .
Two: The instantaneous adaptation of vision has a relatively low static contrast of approximately 100:1, but sensory adaptation, this instantaneous responses slides and is integrated dynamically over time to discern many orders of magnitude more range. How the factor of sensory adaptation comes into play depends on the relative brightness of the environment to the face of the display, and by the relative brightness of portions of the image being displayed.
So It is within these constraints I say there’s a practical “contrast window”.
Ultimately, any careful analysis must entertain both of these factors of display application to make sense of performance, so let’s keep in the back of our minds our application assumptions of a medium with a rectangular face with a neutral surround situated in dim viewing environment, presenting a peak luminance limited to about 200 nits, and a effective contrast ratio of around 1000:1.
Notice that image data we intend to present may be of any range (ignoring format assumptions) but in order to display it, the data range must be fit to the medium of presentation. This implies the presentation of the data may need to be compressed or clipped.
Also notice that since the first daguerrotype we haven’t generally been troubled by these limits as compared to the reward we enjoy by the of recording and presentation of photographic images.
If you have ever been interested in high-fidelity audio reproduction you have come across people with a fascination with details and nuances of orchestral music reproduction, yet no matter how obsessive the aficionado to accuracy, his system will reliably intentionally include a device for introducing enormous distortions called a volume control.
An idea similar to that of the hi-fi volume control pertains visually with the display’s “gamma” response.
On one hand, gamma is a fortuitous condition for digital imaging. To paraphrase Charles Poynton from his extensive study of the subject: the existence of CRT gamma as a nearly perfect inversion of human visual response was a happy coincidence for the engineering of digital imaging, because if the CRT gamma physics were not present, it surely would need to be invented to permit numerically efficient coding of image data.
But on the other hand, imaging system gamma is like a volume control which lets us fit arbitrary data into practical displays, where contrary to any instance upon accuracy, it is precisely the nature of a control for distortion that makes the system tenable.
So the term “pure gamma” is a misnomer. Gamma is a term for a practical display tonal response that adheres to a convention (spec) for converting data into luminosity while fitting the response into the available contrast window. In analog days, the technician adjusted the CRT monitor black level against a PLUGE. Today a combination of alignment features ranging from display physics, to monitor controls, to GPU VCGT, to video driver features, to color-management modules and profiles, to SW libraries, to OS control panels and applets, on and on, all work in concert to tailor image presentation and create the nightmare we suffer today.
Would you whether that setting is actually the same as Argyll’s dispcal “-f” flag :
-f [degree] Amount of black level accounted for with output offset (default all output offset)DisplayCal is a GUI wrapper for Argyllcms and operates by issuing Argyll command.
Second question. You wrote :
Black point correction trades off neutral tracking near black. Needless to say, better monitors like the NEC PA271W I had on my desk for years don’t suffer from the poor black point discrepancy that this “Office” Dell clearly exhibits.
Vincent suggested a good explanation: your setup includes various features that work behind the scenes, e.g., via DDC/CI, profiles, vendor specific display firmware. The black behavior is intrinsic to LCD panel. However NEC corrects this out of box, the underlying physics are at still at play. The cool thing about Argyll/DisplayCal is that these open the door to understanding, at the cost of time spent learning.
“50,000:1” contrast ratio? Say the white Luminance was set to 100 cd/m2, what does that mean for the microscopic black luminance?
Yes, you are pointing a finger directly at the physics. Pls recall that vision power response is logarithmic, so the magnitude of the numbers is misleading: in photographic terms that’s a dynamic range of 15 stops, which is high side of practical. Note that 8-bit (256 level) quantization could not handle it without gamma.
My reference was to the BT.1886 spec, which retrospectively codified HDTV CRT monitor traits for posterity. The 1886 “reference EOTF” has a contrast window of about 50,000:1. But the “CRT measured data” supplied in the spec shows a contrast ratio of about 10,000:1. LCDs are more limited, but some professional designs such as Sony reference use panels with two layers, allows CR of greater than 1,000,000:1.
Thank you for your comments. Always well-regarded.
My pleasure. Best to you.
My NEC PA271W like the NEC PA301W I used at Transcon, was always handled through NEC Multiprofiler. The Eizo CG301 was handled by ColorNavigator. Both were handled by Kodak MatchPrint Virtual on press, though.
I am quite happy with the performance of my Dell U4326W as far as the ‘best’ calibration it can handle. I don’t think I can get better for my needs. I’m leaving Brightness at 100 and Contrast at 50, for the moment. The Red gain is at 100%, Green is 42% and Blue 65%. Granted there are other RGB gain combinations that will also get me close to my ideal D50 chromaticities but I don’t know any ‘better’ for the moment. I have 85 Cd/m2 and the Black point Luminance is a little high at 1.79 for my taste. I know I’m taking a beating in Contrast Ratio but the shadows appear neutral to my eyes. Currently have 219 Lux (5400K CCT) falling on my screen in the middle of the afternoon.
I’m not touching Black Output Offset. Leaving it at 100%.
Black point correction is at 100%. I’ll experiment with lower levels at some point.
Incidentally, I wish there was a way with either Argyll or DisplayCAL to experiment with Black point settings without having to incur the whole calibraiton?I’m leaving Brightness at 100 and Contrast at 50, for the moment. The Red gain is at 100%, Green is 42% and Blue 65%.
That’s bc you are doing it all wrong. NEC SV2, NEC MP, Eizo CG, or even i1Profiler DDC/CI controls won’t do it as bad as you are doing.
You are the only one limiting your U4326W performance.
I have 85 Cd/m2 and the Black point Luminance is a little high at 1.79 for my taste.
And that’s your fault, not monitors. Your are not using a CRT anymore!!!
I used NEC Multiprofiler with my PA271W at about the same Luminance settings for years. I wish you would use a less “judgemental / criticising” language than “You are doing it all wrong” and “You are not using a CRT anymore”. And don’t feel obligated to comment on my posts, Vincent. Nothing compels you to. I’m only aiming to share my experience, for those who are interested.
It’s not just an explanantion of your issues but also a warning to other users, an example of all things that can be done wrong, you did them wrong.
Your are using an backlit LCD (1), and your are driving white level through RGB gains (2). All these things point that you got stuck on CRT era and that you are unwilling (to learn how)to use an LCD/TFT monitor in a proper way.
Then do not blame that Dell about not matching your desired performance because your are the first culprit.“white level for LED backlit = brightness OSD level” and until you understand that, there is little any calibration app can do to help you, other than “hide” the proper way to handle white level from you on a fancy interface and DDC/CI access to display
Then explain to me why I have OSD Brightness at 100%? And the only way I can raise White Luminance is through the Contrast setting? Currently at 50%?
See attached screen capture.
I cannot touch RGB gains without affecting white point chromaticities.
Could you please make an effort to use less offensive language?
Could you please make an effort to use less offensive language?
I’ve done it through all these threads…
Then explain to me why I have OSD Brightness at 100%?
Bc you set it to 100% or you are using a locked OSD mode at 100% brightness
And the only way I can raise White Luminance is through the Contrast setting? Currently at 50%?
See attached screen capture.
I cannot touch RGB gains without affecting white point chromaticities.
If brightness is low at 100% brightness is because at ~350nit max ? (100% OSD brighteness) you are controling white level with RGB gains, hence the high black level.
Screen capture does not show at least one channel at max level. Is Red about 80% in the screenshot?Also keep in mind that G will drive most of brightness. Didn’t you push down it too much chasing blue? If G/B +-1 step moves too much WP in one direction or another then maybe you want to control that last 3-4dE step in white in GPU VCGT calibration, rather than chase a GB gain combination that lowers a small dE in a faulty, or at least limited RGB gain OSD control. If that is what’s happening, bc information about your case is extremely limited.
Now let’s assume all what you say, that monitor is so bad that cannot achieve D65 unless you push 2 RGB gains near 50%, like in very bad TN display from $300 office laptops. Hard to believe for a non faulty $1000 display!!!
But let’s assume that on your behalf, that 90% of Dell U4326W from factory have a yet “undisclosed” native whitepoint far from what you aim to (d65?) so you have to push green near 50% be get less than 4-5dE from D65.
Then maybe you want to aim to cooler than D65 daylight whites. They’ll look a little cooler “but white”. This is the easy solution for those $300 laptops, get rid of the green cyan cast on screen and make them look white although a little cool. But I doubt that such dusplsy is behaving like a 500:1 TN laptop.Also wrong configuration in measurement devices, like wrong colorimeter correction so your are pushing xy coords beyond they should (AFAIK undisclosed mesurement device and settings), or input range mismatch (legal vs full) on computer graphics card or display may be adding extra black level (RGB 16) and limiting brightness. IDNK what you did in those steps.
So let’s make another assumption, that at 30% brightness / factory contrast 50% / RGB gains 100% all other settings reset to default, that $1000 350nit display is about 140nit and has a bluish-cyan white 7×00 K CCT… it’s even expected.
Now if keeping red gain at 100% and lowering little by little G and B cannot get “close to” 4dE D65 unless you push G to 50%… i’ll say that it is faulty unit (return it), or measurement was done in a wrong way (maybe that’s happening)Also keep in mind that G will drive most of brightness. Didn’t you push down it too much chasing blue? If G/B +-1 step moves too much WP in one direction or another then maybe you want to control that last 3-4dE step in white in GPU VCGT calibration, rather than chase a GB gain combination that lowers a small dE in a faulty, or at least limited RGB gain OSD control.
Let’s assume no user misconfiguration for legal-full range in GPU and no measurement device misconfiguration. Then:
Factory reset, RGB gains to 100%. Measure whitepoint xyY at 50% brightness OSD then 30 and 20% to get a hint about brightness ramp with OSD in Y (nits) and xy (whitepoint color).
Let’s say that native white (RGB gain 100%) is cool a cyan white, it’s even expected.
Let’s say that near 30% brightness you get about 130 nit, maybe not so far from your target, whatever it is. Lower brightness near to your target + a generous headroom for G correction downwards (+10-20nit?)
Then lower G and B little by little, first a little B then a little G, then again and if you cannot achieve <2dE WP to D65 unless you push green near 50%, ***maybe you should have stopped earlier in a significatively higher G gain*** because up to 6dE whitepoint correction can be done in GPU sacrificing less than 10% unique grey levels at GPU output.
That would imply a very limited OSD control or weird behavior of gains… in those cases, let GPU VCGT handle that last jump to <2 dE, like in a laptop display without RGB gains.Brightness 100% to aim 100nit D65 lowering RGB gains near half factory value, (your last configuration) is an abnormal configuration that should be avoided in almost every case.
Hey Roger, I can clarify a couple points
The backlight spectra is fixed so the only way “gains” can work is by the LCD (closing the shutter) . As an LCD has a fixed native contrast, lowering gains necessarily reduces contrast. So you want to set color temp by keeping at least one gain at max and reducing the others. To get warmer than native, red is left at max. (You might note that roughly red & blue set CCT and green finds the daylight offset from CCT.
An LCD peak output is controlled both by backlight intensity and by the panel. For consistent response, you want to raise / lower the backlight, but leave the LCD panel at factory default, which has been optimized for contrast and tonal response according to manufacturer’s design.
Almost every LCD offers a “brightness” / “contrast” control. The definitions of these terms are historically ambiguous, dating from television (CRTs), but you will find that almost every design uses language appropriated from television, where ” contrast” sets peak intensity and “brightness” sets the black level. The terms gain and offset are analogous to contrast and brightness. You should notice that the manes of the controls are opposite of their typical function. Moreover, due to sloppy technical writing, the names of the controls are sometimes reversed!
So the question is which one controls the backlight?
And to make matters more confusing, Display designers might make these controls affect both backlight and LCD. Generally, Dell does not do this.
You need to figure this out, then leave the control that affects the LCD at default, and set white point using backlight intensity and RGB gains, with red at max and reducing blue and green to set desired temperature. Most LCDs these days hold while temperature very well over the range of backlight intensities so you can adjust white intensity on the fly safer a calibration as the contrast ratio which is defined by the physics of the LCD will remain consistent as it’s independent of the backlight (ignoring dynamic contrast)
Looking at your attached readout, we see you’ve reduced all 3 gains, and this limited contrast. Use the backlight control to set your desired intensity.
You asked : “Is Red about 80% in the screenshot”‘?
Red is bang on 100%, dear Vincent.
Now, do keep in mind my main “application” is PrePress. By definition, I *don’t* live in a D65 world : if I can’t get decent D50 White point out of this Dell monitor, whatever the price, $1,000, then I will quickly get it off my desk.
The other point to keep in mind is that I don’t have to live by any standard. While I use the internet for research or put together a magazine page in InDesign or create a logo in Adobe Illustrator, I don’t need to blind my eyes with 160 Cd/m2 (I thought I long read that “nits” is a CIE deprecated unit, See Vigiano RIT?). 120 and even less, iike 100 and even 80, as the sRGB Luminance standard is defined is plenty for me.
Third, keep in mind that I use “Black point correction = 100%” under Calibration settings. That is why I get such a relatively “poor” black Luminance. For the work I need to do now, Vincent, that is *fine*. It’s a trade off, right? Remember, my “native” (aka ‘As meaured’) Black chromaticity is blue, like 10,000K blue.
I just noticed I uploaded the WRONG screen capture? SO SORRY!!!
Attached is the correct one.Here is a lof of Gains / Chromaticities / CCT / Luminance :
1 Raised Red from 227 to 255 ; Y = 98 / CCT = 4028 K
x 0.3747
y 0.35912 Red 255 Green 173 Blue 153
Y = 100 / CCT = 4686 K
x 0.3494
y 0.32523 Decide to raise Green from 173 to 181
x 0.3476
y 0.335
CCT 4822 K4 Decide to raise Green again, from 181 to 191
x 0.3461
y 0.3473
CCT 49445 Raised the Green again from 191 to 195
x 0.3456
y 0.3501
CCT 4975
Y 1156 Raised Green from 195 to 200
x 0.3449
y 0.3546
CCT 5017
Y 1187 Increased Blue to 150
x 0.3462
y 0.3750
CCT 4979
Y 1208 Decrease Green from 200 to 197
x 0.3471
y 0.3549
CCT 4937
Y 1159 increase Blue from 150 to 152
x 0.3452
y 0.3521
CCT 4997
Y 11710 increase Green from 197 to 199
x 0.3447
y 0.3546
CCT 5028
Y 11810 Decrease Blue from 152 to 151
x 0.3463
y 0.3573
CCT 4977
Y 11611 Decrease Blue from 151 to 150
x 0.3462
y 0.3572
CCT 4981
Y 11912 Increase Green from 199 to 200
x 0.3462
y 0.3571
CCT 4979
Y 11913 Increase Green from 200 to 201
x 0.3451
y 0.3613
CCT 5034
Y 12114 Decrease Blue from 150 to 149
x 0.3491
y 0.3678
CCT 4918
Y 12015 Decrease Green from 201 to 200
x 0.3499
y 0.3630
CCT 4870
Y 11816 Increase Blue from 149 to 150
x 0.3460
y 0.3574
CCT 4988
Y 117I stopped at step 16.
