Custom gamut vs "Native"?

[Mais78]

What are the benefits of using a user defined gamut (via xy values for RGB) vs using "Native" vs using AdobeRGB in a wide gamut monitor with internal calibration?

[digitaldog]

What are the benefits of using a user defined gamut (via xy values for RGB) vs using "Native" vs using AdobeRGB in a wide gamut monitor with internal calibration?

What kind of system? Native (gamma, white point etc) is usually a good start for displays that don't have high-bit control internally. You end up with less banding since you profile the native behavior rather than try to adjust it. Not sure what you mean by defined gamut. I suspect you either want the widest gamut possible out of the unit and/or (if you can produce multiple calibrations) one for sRGB and one for as wide a gamut possible.

[Mais78]

Hi digitaldog, I am glad you replied as I read with a lot of interest your old posts.
Let me give you a bit of background so you understand where I am and we can frame the discussion.

First of all I want to say that I am not an expert in color management (even though I am familiar with the concept of monitor calibration) so I struggle with the jargon most of the times. My previous monitor was a Dell U2414M, cheap and chearful and I always calibrated (GPU) via Colormunki and DispCal with great results. Now I bought a Dell UP2414Q which is a 4k wide gamut and has internal calibration that works with the supplied software (Dell Color Calibration Solution - I will call it DCCS) and 1i Display Pro (that I also bought). The new monitor has a range of options that I did not have before and I am a bit confused.
I have discussed the matter also here http://en.community.dell.com/support-forums/peripherals/f/3529/t/19616624?pi23185=1#comments and here http://en.community.dell.com/support-forums/peripherals/f/3529/t/19616853#comments there is a guy called yumichan who seems to know color management inside out but he is a bit arrogant and uses a jargon i don't understand.

The issues/doubts i have are the following:

1) Blackbody vs Daylight. When I calibrate with DCCS and check the profile in Dispcal, in the measurement report I see that I always fail the whitepoint delta test. DCCS, contrary to the option selected, seems to calibrate to black body instead of daylight D65 (I say this because if I check the "assumed target whitepoint is blackbody" option top left of the measurement report I pass the test). How "bad" is that? Can you do photo editing with a monitor calibrated to black body 6500 vs D65? I understand from a post of yours from many many years ago (2006 - http://forum.luminous-landscape.com/index.php?topic=9814.0) that the difference is small, is that still correct with 2015 technology?

2) Native vs Custom Gamut. The guy on the Dell forum categoracally said that calibrating a wide gamut to Native is a non starter ("DCCS' "Native" target is a WRONG choice for 99% of users. It's true that it aims for native gamut but also to native white (GB-LED white is a green-white @7000K, out of daylight locus). My advice is to set manually R,G,B xy coordinates, and your desired gamma in "Custom"  DCCS preset". I wanted to understand in simple terms what is wrong with "Native" and the supposed benefits of Custom, still not clear. He suggested to calibrate for R 0.685 0.31 while keeping G and B xy coordinates the same as in AdobeRGB space. I did this and I got results that are visually quite different (shadows more open in Custom vs Native), while the gamut coverage seems similar (I posted some charts in page 4 of that Dell forum discussion). Which one is "right"? Visually I prefer Native, but don't have a printer to test the results (I either post to my website or print books with Blurb, rarely print on my own).

3) DCCS vs DispCal profile. The last doubt I have is whether I should keep the icc profile created by DCCS or profile in DispCal after the internal calibration with DCCS? I tried both and the output on screen seems identical (even though the DispCal measurement report tells me that its profile has lower Average and Max color delta). Still in DispCal if I pull up the gamut chart for both, the one with the DispCal profile has a much wider coverage (I posted the charts in that same page). In case a DispCal profile is advisable I was wondering what the best options are: so far I always used Curves + Matrix, No blackpoint conversation, but I am being told to use XYZ LUT with blackpoint compensation for wide gamut?   

Sorry for the long post, many thanks in advance!

[digitaldog]

I have no idea how or what the Dell software is doing, sorry. You certainly don't want to pin a White Point just for wider gamut. And the values are kind of meaningless until you define what you are trying to calibrate for.

[digitaldog]

I have discussed the matter also here http://en.community.dell.com/support-forums/peripherals/f/3529/t/19616624?pi23185=1#comments and here http://en.community.dell.com/support-forums/peripherals/f/3529/t/19616853#comments there is a guy called yumichan who seems to know color management inside out but he is a bit arrogant and uses a jargon i don't understand.

I didn't read every post but what I did read from yumichan, it seems like he knows what he's talking about. I found this comment interesting too:

BTW it is IMPOSSIBLE that a widegamut (U2711) in its full gamut configuration (Standard or custom) does not show banding in black to white gradient (I mean in profile colorspace) AFTER you modify GPU LUT in order to calibrate with a gamer nvidia or intel GPU. IT IS NOT POSSIBLE since that hardware has 8bit LUT.

[Mais78]

Thanks Andrew, understanding that you have no direct experience of the Dell software, would you mind sharing your view on those questions in more general terms, i.e.:

- Is it bad to have a monitor calibrated for blackbody withepoint instead of D65 for photo editing?

- His advice not to use the Native gamut and define a Custom one instead is a general reccommendation or has to do with The Dell software? I mean if you had a NEC or Eizo widegamut monitor, would you calibrate to a custom space yourself or you would pick Native? (The ultimate goal is still accurate photo editing).

- Also the last question, when you have a monitor with internal calibration capabilities, in general, are you better off sticking to the profile generated by the proprietary software or Dispcal in general does a better job at profiling?

The guy in that forum definitely knows what he is talking about (!) but the issue is that he assumes that everyone is as knowledgeable as him!

Thanks!

PS If you want to play with it, the software is this http://www.dell.com/support/home/uk/en/ukdhs1/Drivers/DriversDetails?driverId=VJVPH&fileId=3371733744&osCode=W764&productCode=dell-up2414q&languageCode=EN&categoryId=DD

[digitaldog]

Is it bad to have a monitor calibrated for blackbody withepoint instead of D65 for photo editing?

I don't think it matters. The numbers are all kind of meaningless in the end. The black body radiator is theoretical.

- His advice not to use the Native gamut and define a Custom one instead is a general reccommendation or has to do with The Dell software? I mean if you had a NEC or Eizo widegamut monitor, would you calibrate to a custom space yourself or you would pick Native? (The ultimate goal is still accurate photo editing).

My NEC is widest gamut.

- Also the last question, when you have a monitor with internal calibration capabilities, in general, are you better off sticking to the profile generated by the proprietary software or Dispcal in general does a better job at profiling?

Too many variables, most software for this task sucks! But I'd try proprietary first if I knew it communicated with the panel or had more options over the calibration targets.

[Mais78]

 - Thanks then I guess I will live with black body whitepoint, it is my only option if i want to do internal calibration

- "My NEC is widest gamut" meaning Native? Maybe on your monitor is not called "Native" but what I mean is that you choose the widest preset as opposed to dialing in your own xy coordinates, correct?

- I know most software sucks, e.g. at the time of my U2412M with DispCal I was getting much better results than the Colormunki own software. All I know is that with DispCal profile I get lower deltas vs the Dell proprietary software, is that alone a good reason to profile (and profile only - not calibrate) in Dispcal or there might be other things interfering that are not captured by the deltas? 

[digitaldog]

SpectraView calls it "Native (Full)".

[Simon Garrett]

Just as a general comment, I seem to recall that the tftcentral reviews I've read generally reported better accuracy when calibrating/profiling in native gamut. 

[GWGill]

I didn't read every post but what I did read from yumichan, it seems like he knows what he's talking about. I found this comment interesting too:

BTW it is IMPOSSIBLE that a widegamut (U2711) in its full gamut configuration (Standard or custom) does not show banding in black to white gradient (I mean in profile colorspace) AFTER you modify GPU LUT in order to calibrate with a gamer nvidia or intel GPU. IT IS NOT POSSIBLE since that hardware has 8bit LUT.

That's not actually true with a lot of modern graphics cards - it's typically the connection to the display that's the problem.
My VGA connected display gets 10 bits from the NVIDIA 8600GT VideoLUTs, but if you are using typical DVI then yes, this limits it to 8 bits/component, but something like Display Port should handle more than 8 bits. So if your graphics card has 10 bit VideoLUTs and you are using Display Port or some other connection that hands it to the display, then using VideoLUTs based calibration curves can give you satisfactory results with the right calibration software.
Now an 8 bit/component (ie. 24 bit total) frame buffer is a different question though, and getting 10 bit/component frame buffer (30 bit total) working with operating systems and application is something of a trick.

[Mais78]

Thanks All, we got some color management heavyweights in this thread! 

@Simon I don't think TFTCentral ras reviewed this monitor yet

@GGill what DispCal settings would you reccomend to profile this wide gamut monitor?

[Simon Garrett]

@Simon I don't think TFTCentral ras reviewed this monitor yet

When I said that "the tftcentral reviews I've read generally reported better accuracy when calibrating/profiling in native gamut" I meant "tftcentral reviews of monitors with factory presets".  

Your original question was a general one about wide gamut monitors, and I gave a general comment about wide gamut monitors!

To expand on that: any user-defined gamut or factory preset depends on the monitor emulating a gamut that is not the native gamut, and emulating RGB primaries that are not the natural RGB primaries.  It seems possible that these emulated primaries may not be quite as stable at all brightness levels as the native primaries.  For that reason, it does not surprise me that for monitors with this capability, the most accurate calibration and profiling might be using the native gamut.  

However, I've no specific knowledge of the UP2414Q, and I can't find any technical review that assesses accuracy calibration (e.g. the sort of review that tftcentral or prad.de do) other than a rather light-weight one by trustedreviews.com that doesn't say what they tested and is next to meaningless.  

[GWGill]

The issues/doubts i have are the following:

1) Blackbody vs Daylight. When I calibrate with DCCS and check the profile in Dispcal, in the measurement report I see that I always fail the whitepoint delta test. DCCS, contrary to the option selected, seems to calibrate to black body instead of daylight D65 (I say this because if I check the "assumed target whitepoint is blackbody" option top left of the measurement report I pass the test). How "bad" is that? Can you do photo editing with a monitor calibrated to black body 6500 vs D65? I understand from a post of yours from many many years ago (2006 - http://forum.luminous-landscape.com/index.php?topic=9814.0) that the difference is small, is that still correct with 2015 technology?

That's an error of about 3.3 Delta E. Maybe you can fix it by setting your own white point target as a custome space as D65, i.e. x,y of 0.3127, 0.3290

2) Native vs Custom Gamut. The guy on the Dell forum categoracally said that calibrating a wide gamut to Native is a non starter ("DCCS' "Native" target is a WRONG choice for 99% of users. It's true that it aims for native gamut but also to native white (GB-LED white is a green-white @7000K, out of daylight locus).
......
Which one is "right"? Visually I prefer Native, but don't have a printer to test the results (I either post to my website or print books with Blurb, rarely print on my own).

Native giving you some random white point is a different issue to not getting the full range of colors. You're better off with native and then calibrating to your desired white point if you want wide gamut. If you want to use the in-built calibration, then do as suggested and measure the native R,G & B and plug them in as the R,G,B target chromaticities + your target white point.

3) DCCS vs DispCal profile. The last doubt I have is whether I should keep the icc profile created by DCCS or profile in DispCal after the internal calibration with DCCS? I tried both and the output on screen seems identical (even though the DispCal measurement report tells me that its profile has lower Average and Max color delta). Still in DispCal if I pull up the gamut chart for both, the one with the DispCal profile has a much wider coverage (I posted the charts in that same page). In case a DispCal profile is advisable I was wondering what the best options are: so far I always used Curves + Matrix, No blackpoint conversation, but I am being told to use XYZ LUT with blackpoint compensation for wide gamut? 

Sounds odd if the two profiles have distinctly different gamuts. A verify should indicate which one is the more accurate though.

What DispCal settings would you reccomend to profile this wide gamut monitor?

If your concern is mainly color managed applications, then there is nothing special needed - choose a white point appropriate for your work - e.g. D65 if the display is used in isolation, or something closer to D50 if you are going to be comparing to prints in a viewing booth, and a perceptual gamma curve, something like an effective gamma of 2.3 or L* curve. At the end of the day you want it to have the right white point, contrast ratio and gamut and for it to be well profiled.

[Mais78]

Thanks all

That's an error of about 3.3 Delta E. Maybe you can fix it by setting your own white point target as a custome space as D65, i.e. x,y of 0.3127, 0.3290

Dispcal says around 2.2-2.5 delta. I tried also the custome space but i always get blackbody, so black body is my only option if I want to do internal calibration. Otherwise I have to do GPU LUT calibration, but I guess it is still better to do internal cal even with this limitation?

Native giving you some random white point is a different issue to not getting the full range of colors. You're better off with native and then calibrating to your desired white point if you want wide gamut. If you want to use the in-built calibration, then do as suggested and measure the native R,G & B and plug them in as the R,G,B target chromaticities + your target white point.

Yes, I know it is a different issue. Not sure I understand the part in red, I do want white gamut but also want to use internal cal. What is the advantage of plugging RGB vs using Native?

Sounds odd if the two profiles have distinctly different gamuts. At the end of the day you want it to have the right white point, contrast ratio and gamut and for it to be well profiled.

Indeed, very odd. I get a contrast of only 585 if uniformity compensation is on, is that bad?

Thanks

[D Fosse]

Going back a bit:

Is it bad to have a monitor calibrated for blackbody withepoint instead of D65 for photo editing?

A black body/Kelvin defined white point only takes into account the blue/yellow axis. The native white point could be miles off on the green/magenta axis and still be 6500K. So yes, assuming black body could be bad and that's the point everybody's trying to make.

D65 (or any of the D series standard illuminants) puts the white point on the daylight locus, defining it on both axes, blue/yellow and green/magenta. That said, don't assume the calibration will necessarily get you there without some manual tweaking, for a number of reasons. What you want to see on screen is paper white.

I don't see how gamut has anything to do with this. Calibrate to native gamut and be done with it - perhaps with a secondary sRGB emulation for use with non-color managed software. No need to overcomplicate this.

[Mais78]

Going back a bit:

A black body/Kelvin defined white point only takes into account the blue/yellow axis. The native white point could be miles off on the green/magenta axis and still be 6500K. So yes, assuming black body could be bad and that's the point everybody's trying to make.

Got you. DispCal tells me the delta E vs D65 is ~2.4, so probably I am not miles off, right?

I don't see how gamut has anything to do with this.

Correct, it has nothing to do with this. Above I said that whatever combination of gamut and whitepoint settings I choose I always get black body whitepoint, meaning that it is not a bug of a particular setting, I tried all combinations.

Calibrate to native gamut and be done with it

Ok, it does not cost me more time to calibrate to Custom vs Native. I think also Graham is suggesting to choose a Custom gamut dialling in the measured RGB. Probably Native is good enough for me but I was puzzled by that comment made in the other forum, by a very competent person, that "Native is wrong for 99% of users" and that Custom should be used instead.

[GWGill]

I tried also the custome space but i always get blackbody, so black body is my only option if I want to do internal calibration. Otherwise I have to do GPU LUT calibration, but I guess it is still better to do internal cal even with this limitation?

Sounds like the internal calibration is rather innacurate if it can't hit the target x,y within 3 delta E. You could try counteracting the error with the x,y target value, and see if that drags it to where you want - unless they are silently editing your target by mapping the x,y to the black body locus, in which case nothing will help.

Not sure I understand the part in red, I do want white gamut but also want to use internal cal. What is the advantage of plugging RGB vs using Native?

Well the supposed advantage is to get the neutral and white point calibration implemented using the displays own tables, which (hopefully) are better than 8 bits.

Indeed, very odd. I get a contrast of only 585 if uniformity compensation is on, is that bad?

That's up to you to decide, although in my view 500:1 is OK. Real life prints don't have a huge contrast range, so if soft proofing is the aim, any display contrast range better than the print is fine. If, on the other hand, you are watching Video in a dark room, then purists might want a better contrast ratio - which is why they love Plasma displays.
[/quote]

[GWGill]

A black body/Kelvin defined white point only takes into account the blue/yellow axis.

If you define a particular temperature on a particular locus, then this defines the x,y exactly. The same temperature on the Black body and Daylight locus are different x,y values, a little way apart.

D65 (or any of the D series standard illuminants) puts the white point on the daylight locus, defining it on both axes,

A temperature of 6500K on the black body locus also defines the chromaticity on both axes.

[GWGill]

Probably Native is good enough for me but I was puzzled by that comment made in the other forum, by a very competent person, that "Native is wrong for 99% of users" and that Custom should be used instead.

Because if you use native and don't calibrate the white, you end up with a non-standard white point. So you want to use the native RGB chromaticities, but balance their maximum Y value to arrive at your target D65 white point when they are all 100%.