Monitor calibration - 3D LUTs for colour space emulation

[richardj21724418]

I came across this article.  I hope some of you will find it interesting and educational.  (I didn't write it!)

https://photographylife.com/the-basics-of-monitor-calibration

[D Fosse]

Nothing new here. All of this was explained in your other thread - which, BTW, I don't think should be locked (even though it did go round in circles).

I'm getting used to explaining the difference between calibration and monitor profile, a distinction very few people get at first. It usually takes some time and perseverance, but I always get there in the end

[NB, colourmanagement · Accepted solution]

richard

2C from me

I agree with this statement you wrote

"I 'program' my BenQ SW2700PT monitor to AdobeRGB, the profile I use is the standard AdobeRGB profile.  If it's been done properly there should be no need to have to characterise the monitor and produce a profile unique to that monitor. 

If I understand correctly what you want [accurate calibration to a standard], its a situation not unheard of in our industry - that’s what's been largely achieved in the world of offset and flexo. print.

The work of the print experts at BVDM, FOGRA , PSO and GRACoL has enabled the standardisation of CMYK file separation for most sectors of the print industry.

Yes, ICC profiles are used in the separations to CMYK, but all the data from any user (for the same target sector) is the same - it's a standard space not a custom ICC for each press. Kind of what you'd like with sRGB and displays.

Also, what you apparently seek is pretty close to what the video industry does in edit suites largely using LUT boxes. Calibrate with a hardware LUT, then chuck the unaltered data through that [no ICC profile - there is no mechanism for a profile] and it looks "right".

but I don't agree with this statement:

"That's 'old school' thinking.  If I do attempt to characterise and profile the monitor, the profile I get should be the same as that of the standard AdobeRGB profile, within the bounds of accuracy of the process.  It's a pointless procedure."

It's not old school to use ICC profiles in display situations where accuracy is needed. It's actually a great way of getting good (or at least acceptable) visual performance out of non ideal fairly low cost displays.

As you've seen its hard to achieve what you desire at low cost, maybe that’s why BenQ can't do it.

What you are describing [calibration (or programming) only to a standard space] is what Barco used to do (at massive cost) - and Colormatch too - long in the past. That's old school.

When HP and Microsoft came up with sRGB it was with the idea that it could be used to characterise the "average" display - [sRGB is based on the display characteristics of consumer-grade CRT monitors manufactured in the early 1990s].

But IMO sRGB's not good enough for

A: accuracy (displays are generally not consistent and most do not have the mechanism for accurate calibration)

B: those who sensibly like to use larger and various document color spaces* with print in mind.

     [*yeah a TRUE Adobe RGB capable and calibrated display would largely solve that but at what cost for a display that can 100% match Adobe RGB!]

c: those who want to make the best of far more capable [increasingly wider gamut] display systems

And, this is a good point:

"It could be argued that sRGB is the most important profile to emulate accurately, because it is the profile assumed by applications that don't colour manage."

Right, agreed.

But we need to get rid of applications that don't color manage, not dumb everything down to sRGB.

But what this conversation is all, about is, it seems,  proving that your BenQ display is incapable of what you want, a true emulation of sRGB, and asking us to agree - end of story.

We agree. It can't, but we don't care because an ICC profile used with a savvy application solves the issue.

The wide gamut display of course introduces another conundrum. Non color managed applications are useless on these displays - yes you could limit the gamut to sRGB but would you? Why pay for 98% of Adobe RGNB and do that.

ICC color management solves the problem, at least for intelligent applications.

Good hardware calibration displays can do it - basICColor SW claim to properly match sRGB with emulation in their display SW on both Eizo and NEC.

In your ideal world where ICC profiles are not needed, it seems -  everyone with a computer who is viewing images requires a high quality hardware calibration display [constrained to sRGB). That's not going to happen anytime soon and sRGB . is not the way the industry is going, in the interim we have the well tested and successful solution of ICC profiling.

If what you seek is a world where "calibrated (or programmed)" displays come off the production. line that may be some time in the future. And even if they do they likely will not be sRGB.

It's what HP seem to have achieved with may of their simple photo printers, sRGB files print pretty well and making custom profiles often may not improve upon that.

But it's unlikely to happen for displays

I hope this is of interest

thanks

neil barstow, colourmanagement

[richardj21724418]

Thanks Neil for taking the time to reply and for your excellent reasoned comments.  I appreciate it.

It seems you're agreeing with the principle that if a display which has been perfectly programmed to meet a standard colour space then it can be used the ICC profile for that colour space and obtain correct results.  It's something I seem to have struggled to get some people to accept, but perhaps that's a failure on my part to express it well.  It's just you're sceptical that dedicated hardware in consumer level products can perform to the accuracy required.

Once calibrated, my BenQ actually seems to do rather a good job of emulating AdobeRGB space.  I've confirmed the accuracy by subsequently profiling using the i1Profiler, but I'm sure a professional like yourself would look at it more critically than me.  I tried programming to L* too and that also appeared to work, but it was only a brief inspection and I didn't confirm the accuracy.  So it appears the monitor should be perfectly capable of doing a decent job of emulating sRGB - the problem is just the software doesn't provide the option to program it with the correct sRGB TRC curves.

I think as hardware 3D LUT technology has become cheaper it's become feasible to incorporate it into consumer level displays.  It's no longer 'rocket science'.  I'm not saying the BenQ implements it this way but supposing you had a big (4GB?) LUT to translate 30-bit input space to 30-bit native space.  How much better do you need than that?  Is the software 3D translation technology in PC side profile conversion really superior?  I can't see why you'd go to the trouble of designing a monitor that can be programmed to emulate a standard colour space, then expect to have to characterise it and use it with a non standard profile.

Effectively we're talking about differences in the profile connection space between the computer and the monitor.  This can be some arbitrary space, determined by the native characteristic of the monitor.  Or it can be a standard space, provided the monitor has been setup to respond correctly.  Much as your CMYK analogy.

Regarding sRGB.  I'm certainly not intending to promote it as a working space but I guess some could argue there are merits in using sRGB for display, and not just because of compatibility with non colour managed applications.  That's because of the bit depth bottleneck of the link.  I imagine most of us have 10-bit component connections between computer and display, some may be 8-bit.  This limits the number of values you can communicate to the device, so you don't want to waste valuable tonal resource on describing a larger gamut than you need.  For instance, in 8-bit mode in Photoshop I doubt you'd want to use a wide gamut working space like AdobeRGB or ProPhoto RGB, if you know your colours are all going to be contained in sRGB gamut.  Why opt for a larger tonal granularity than necessary for the job?

No, it isn't a colour space I'd normally display in.  I just came across this sRGB mode 'misconfiguration' when I was investigating something else and thought I'd mention it and ask if it seemed acceptable in a hardware calibrated monitor.  It was never my intention for it to evolve into some deep philosophical discussion!  I'd mistakenly assumed the concept of a hardware calibrated monitor was already a given.

[TheDigitalDog]

richardj21724418  wrote

Thanks Neil for taking the time to reply and for your excellent reasoned comments.  I appreciate it.

It seems you're agreeing with the principle that if a display which has been perfectly programmed to meet a standard colour space then it can be used the ICC profile for that colour space and obtain correct results.  It's something I seem to have struggled to get some people to accept, but perhaps that's a failure on my part to express it well.

Nope, those posting in the locked thread got that clearly; pretty much unanimous acceptance other than from you. What you apparently didn't understand, maybe still do not, is that even if a display like a PressView produces ColorMatch RGB perfectly, a display profile is still absolutely necessary and always used in ICC aware applications; always! They cannot not have such a profile in the chain. It's used for Display Using Monitor Compensation EVEN if there is zero compensation. And had you understood and accepted that fact, the rabbit hole you dug about TRC/Gamma would have been clear as just that, a rabbit hole.

No, we're not talking about the profile connection space and the monitor. That's simply one small and insignificant part of the process where a display is calibrated (calibrated, that's the correct term used for decades, there's no reason to state it's 'programed') and the profile that desirables that state for the CMS. As other's correctly stated, you're making this far, far more complex than it has to be for some reason. No, none of this is LUT specific; the way display calibration and the associated profile operate in a color managed system doesn't have to and isn't related to LUTs any more or less than if we use matrix based profiles. Any more than if you change the display resolution! This is a basic and well established architecture that's been around for 20+ years.

And yes, in terms of your article which was probably posted to start this all over again and risks a locking down, NOTHING NEW.

[richardj21724418]

thedigitaldog  wrote

What you apparently didn't understand, maybe still do not, is that even if a display like a PressView produces ColorMatch RGB perfectly, a display profile is still absolutely necessary and always used in ICC aware applications; always! They cannot not have such a profile in the chain.

I don't know why I can't express this.  I've never said an ICC display profile isn't necessary for a color managed environment.  Just if you have a display that emulates a standard colour space, you use the ICC profile for that space.  You shouldn't need to characterise the display and produce another profile.  If you attempt it, the result will simply be a profile for the standard space.  Why bother?

If you find the need to characterise the display and produce another profile, then the display isn't emulating the standard colour space properly.

[D Fosse]

OK, I'm giving this one more chance.

if you have a display that emulates a standard colour space, you use the ICC profile for that space.

To which I ask: why?

This whole exercise that you're proposing is only meaningful if your intention is to get rid of icc profiles altogether. If your monitor emulates sRGB perfectly, you don't need icc profiles at all. You have just defined an alternative model for color management that doesn't use icc profiles. You create your file in sRGB, save it out without any profile, and display it on your sRGB-compliant monitor. Done.

This was, BTW, the original intended use for sRGB IEC61966-2.1. That was before modern color management was invented.

The whole point we're trying to make here, is that standard color management is very much simpler and much more efficient (not to mention vastly more flexible).

Getting a monitor to exactly match sRGB is a tall order. It takes a lot of adjustments. The profile does all that with much less cost and overhead. You don't have to physically change the behavior of a monitor - you just change some numbers, and it's all good to go.

In theory your approach is perfectly possible, no one ever denied that. The question is - why? What is it you're trying to accomplish that standard color management can't?

[TheDigitalDog]

https://forums.adobe.com/people/D+Fosse  wrote

OK, I'm giving this one more chance.

if you have a display that emulates a standard colour space, you use the ICC profile for that space.

To which I ask: why?

No need to ask. He's trolling. He asked a question in the other post, didn't accept the answers and now this. Time to move on. This is clearly CWOBaT (colossal waste of bandwidth and time) for those of us who post in an attempt to aid Adobe users.

[TheDigitalDog]

richardj21724418  wrote

thedigitaldog   wrote

What you apparently didn't understand, maybe still do not, is that even if a display like a PressView produces ColorMatch RGB perfectly, a display profile is still absolutely necessary and always used in ICC aware applications; always! They cannot not have such a profile in the chain.

I don't know why I can't express this. 

Because it's wrong!

richardj21724418  wrote I've never said an ICC display profile isn't necessary for a color managed environment. 

Sure you did, in the locked post, unless you're a different richardj21724418

https://forums.adobe.com/thread/2559451

"After calibrating to sRGB PME does output an ICC profile, but it's useless.  Regardless of the calibration colour space selected, the profile is always written for the native gamut.  This means colours look way undersaturated, as you'd expect for a profile that's been designed for a wide gamut."

And that's wrong and it was explained why it's wrong.

You seem to be under at least two misconceptions here about these forums. The first is that this is a forum for arguing after asking a question, getting multiple and consistent answers from people who know more about the subject than you do. That you're asking a question and will argue until you hear the answer you want to hear, not the one that's correct. I suggest you go to this video as an assistance in that agenda: Monty Python's Argument Clinic - YouTube

Your 2nd misunderstanding about the forums is you're unaware this is a user to user forum for Adobe customers. You state oddly you've never stated an ICC Profile isn't necessary for a color managed environment failing to grasp that outside such environments the profile isn't understand and that all Adobe applications that deal with color use them and are color managed. So you're ranting, you're not accepting correct answers, you're off topic, this new topic isn't anything new but an excuse to continue going off topic after the original post was locked. As this one will be.