Display Panel Uniformity Acceptable Standard? Dell U2413

Hello,

I've recently bought a monitor, Dell U2413 (yes, it's old :) and am currently trying to evaluate if it's worth keeping, based on the panel uniformity.

I've profiled and calibrated the monitor with i1Display Pro and DisplayCAL (unable to hardware calibrate the monitor using DUCCS, dCAL has proven to be better than DUCCS at software-only calibration), the gamut measurments of 98% AdobeRGB, 99.9% sRGB are good enough for me.

* What i don't really understand is the panel uniformity-report: I am wondering, is it "good enough" to work with, or am i looking at a sub-standard panel, unsuitable for photography-editing?

* I'm using the monitor to produce my own images for print: pigment inkjet, c-41.

Below is the uniformity-report expressed as ISO 14861:2015:

This is the uniformity-report expressed as Avg Luminance & deltaC:

I have a vague understanding of what the report means. Obviously it's not fantastic. I'm unsure whether to return it or not because it seems to be on the border-line of what i understand to be acceptable...

...is the centre (9 grids) of the screen good enough to work with?

...i can see in ISO14861 the screen completely fails in the bottom-row, however the rest of the panel acheives at least a nominal tolerance -is this nominal tolerance good enough to work with if producing images destined for print?

...is over 10% luminosity difference simply not good enough?

So, i'd like to know, from anyone who edits images for print, in your technical or experience-based opinion:

(i) Does the screen uniformity pass a minimum standard for photo-editing work?

(ii) Would you keep the screen if you had no other AdobeRGB option?

(iii) Is this an expected /accepted uniformity report from *this (price-range of) monitor? If you had bought this *new*, would you deem the panel acceptable, or would you send it back?

Where i'm coming from is, i'd like to work in 98%+ AdobeRGB and this is the only monitor i can get for the forseeable future which acheives that. It's age, lack of hardware calibration (DUCCS won't work on this Mac OS) is not an issue for me. I'd like to keep it, but not if the panel uniformity is going to be counter-productive to me learning how to produce prints from my images. I'd be using -and learning- on the screen for probably a year.

Perhaps there's only one question here, not three. Yes /No reply(ies) to these questions is fine, longer replies welcomed -all appreciated.

With thanks,

Asim :)

First a question: Is this report from measurements you made by moving the Colorimeter over multiple areas of the display?

In terms of deltaE, this is a metric of difference (distance), lower values are better. Less than one, invisible. But the question becomes, is this ONE sample or multiple and if multiple, hopefully the average dE. When you don't get a max dE of one area, the idea of dE of the uniformity is kind of suspect.

The easier approach is to forget this and simply create a document in say Photoshop that's the pixel dimensions of the display. Fill with middle gray (or Lstar 50) and the image should fill the screen at 100%. How does the uniformity appear to you?

There's this too:

http://digitaldog.net/files/20Testing%20your%20display.pdf

Hi,

Yes, the report is a 5x5-grid sample of the screen, 25 seperate measurments.

Regarding deltaE, the max and average numbers are given for each grid. A recommended tolerance 'pass' is given to a grid with less than 2 max dE; a nominal tolerance 'pass' for less 3.9 max dE and a 'fail' for 4 or over max dE (i think).

So for example, for ISO 14861:2015, the centre 9 grids have nominal or recommended passes, with the greatest deviation being -10% luminance, max dE 2.8, avg dE 2.4.

So i wonder if these centre 9 grids, based on the numbers, are effectively a good-enough standard for colour accuracy...? -If so, i can adapt my work-practice and view the image within this centre-space when fine-tuning. I guess i'm hoping that someone can give me their opinion based on what these numbers are saying.

I filled the screen with L*50 grey as suggested... obvious bleed-light from the edges, a magenta-ish tint in some corners... i feel i could work with it -however, i don't know any better, this is my first screen which hasn't been a laptop-screen.

Regarding the link -assessing the screen by raising luminosity 1% at a time- what result should i expect /accept for my needs? -a change at 5% (but not at 1-4%) is OK?

Thanks for getting back to me :)

Regards.

Hi,

Yes, the report is a 5x5-grid sample of the screen, 25 seperate measurments.

Regards.

Measurements you made, right?

Max dE of 2.8 isn't too bad, it could be better with display that has better control over purity.

This isn't about accuracy per se. Well it kind of is as the quadrants are 'less accurate' to the reference but what it's telling you is about purity of output over those 25 quadrants.

Too much magenta tint in corners not too good, does it 'jive' with the report? Seems not to but maybe it doesn't bleed into the measurement areas.

I don't think raising or lowering the luminance of the calibration is the answer unless it's really low AND you can raise it plus produce the calibration you desire. IOW, most calibrate for a goal, like a print matching a display. If you had a really low cd/m^2 for that result, raising it would make the match worse unless you can also raise the viewing conditions of the print which is an OK fix IF doable. So if you raise cd/m^2 and get better purity, can you also raise the print viewing conditions IF that's the goal of the calibration? It's not worth taking one step forward (better report) and two steps backwards (display and print no longer match).

Yes, i made the measurments after first calibrating the monitor to my working conditions. (D65, g2.2, 80cd/m).

The tint in the lower-corner, and the light-bleed from the edges seems to correlate with the report... max and avg dE is much higher in these grids, even when they get a nominal pass, compared to the dE's from the centre-9 grids.

A misunderstanding on luminosity -i have no intention of changing luminosity for the screen, i was referring to the measured difference in luminosity from one grid to another in the report, and saying that the centre-9 grids are all within 10% of the optimum (in my case, 80/cdm).

I mentioned this because i read a (well written, well argued) article on panel uniformity, where the author gave 10-15% luminance difference as an acceptable number for what is essentially an entry-level, not quite 'pro', wide-gamut display. I guess the idea being that if the uniformity was within these numbers, the display, while not 100% colour-accurate, would be a tool you could trust enough to work with, creatively, if not so technically.

Be aware that a new Dell U2412m, a very good monitor, is going for a little over $200 at B&H (link) They claim it will be back in stock soon.

Hi Hector. The U2142m isn't a 98% AdobeRGB gamut display; the reason i'm persevering with the U2413 is because it is -otherwise i would've already sent it back :)

Yes, i made the measurments after first calibrating the monitor to my working conditions. (D65, g2.2, 80cd/m).

A misunderstanding on luminosity -i have no intention of changing luminosity for the screen, i was referring to the measured difference in luminosity from one grid to another in the report, and saying that the centre-9 grids are all within 10% of the optimum (in my case, 80/cdm).

Ah thanks, I was confused. So yes, you're seeing the 'error' it seems less in color than luminosity which I would believe is the lesser of two evils.

So yes, you're seeing the 'error' it seems less in color than luminosity which I would believe is the lesser of two evils.

That is something i hadn't thought about, although now it seems obvious :)

Do you know, in this respect, if that is essentially what the 'Avg Lum & deltaC' measurement (the 2nd of the two reports) relates to? -Colour.

With thanks for your input.

That is something i hadn't thought about, although now it seems obvious :)

Do you know, in this respect, if that is essentially what the 'Avg Lum & deltaC' measurement (the 2nd of the two reports) relates to? -Colour..

From the BableColor CT&A manual:

∆C*: The chroma difference between the two sides.

In addition:

• ∆L*: The lightness difference between the two sides.

• ∆H*: The hue difference between the two sides.

• ∆h: The hue angle difference between the two sides.

The ∆L*, ∆C*, and ∆H* parameters are defined in the CIELAB & CIELUV section; they represent the individual contributions of lightness, chroma and hue to the global ∆E color difference. ∆h, defined in the L*C*h section, is

also shown because this angular difference is readily associated with a 360 degrees hue circle whereas ∆H* is an indirectly derived value.

Since ∆L*, ∆C*, and ∆h can be positive OR negative, it is important to identify a reference and a sample. InCompare mode, the reference is always on the LEFT side and the sample on the RIGHT side. In Convert mode, the reference is always on the side being converted "FROM" while the sample is the side being converted "TO".

* ∆C*: The chroma difference between the two sides.

• ∆L*: The lightness difference between the two sides.

• ∆H*: The hue difference between the two sides.

• ∆h: The hue angle difference between the two sides.

 

OK. The deltaC report nominally-passes every grid.

 

-But actually, i think something's clicked!

 

The numbers are saying that where i've got:

dE <1 = technically colour-accurate display;

dE 1-2 = display which can be regarded as colour-accurate for the majority of eyes;

dE 2-3 = there is gonna be a perceptible difference between actual colour and displayed colour -a deal-breaker, or workable?

dE 3-4= not suitable for colour-accurate evaluation.

 

Is that fair enough to say?

 

If this is so, perhaps the question should be:

 

Is deltaE 2-3 accuracy an acceptable minimum for a photography oriented display?

 

Would you, for example, be comfortable working on a display which was accurate to dE 2-3?

(bit of a mis-post, not sure what happened there...)

OK. The deltaC report nominally-passes every grid.

-But actually, i think something's clicked!

The numbers are saying that where i've got:

dE <1 = technically colour-accurate display;

dE 1-2 = display which can be regarded as colour-accurate for the majority of eyes;

dE 2-3 = there is gonna be a perceptible difference between actual colour and displayed colour -a deal-breaker, or workable?

dE 3-4= not suitable for colour-accurate evaluation.

Is that fair enough to say?

If this is so, perhaps the question should be:

Is deltaE 2-3 accuracy an acceptable minimum for a photography oriented display?

Would you, for example, be comfortable working on a display which was accurate to dE 2-3?

I think that's fair enough. Now the question is, can you run this in a month, then two months etc, and compare? Is the display stable in that behavior?

Then the question is, dE 2-3 for the display, great. Now what? You're soft proofing and making prints? Is the report pertinent for all colors? If you're only dealing with images on the web, what you see and what other's see is quite different so dE 2-3 doesn't really matter.

I wouldn't stress over the report much. But it would be useful to know how stable that display remains over a longer period of time.

In all fairness, the Dell U2413 is a reasonably affordable screen; apart from the consideration whether its panel uniformity would be acceptable or not, I'd also consider what it would cost to improve upon it, and whether that still fits the budget. Little doubt there are better screens available, but most of those command a pretty serious amount of money more than this Dell.

Personally I have the Dell U2412M mentioned before, and sure it's not 98% Adobe RGB and not perfect by any stretch of imagination, it does the job for me because I know a serious step up would cost significantly more (screens as the NEC SpectraView 24"), which at present I couldn't justify. My prints are consistent enough with the screen on most papers, again it's not perfect but it's not so far off that I need to regularly throw away prints. So, while things certainly could be better, it is working now - maybe not a perfect screen, but a good screen nonetheless, and good enough for my needs. And that's the point: most of us have to draw a line somewhere when it comes to budget, and then it's about getting something that does the job good enough since perfection is too expensive.

Of course, if your budget stretches well beyond this Dell, I would look around for a better screen. But at a budget, you will have to factor in compromises, and you have to consider just how vital the "colour correctness" is between your screen and printer - it's fair to say somebody doing prepress work or delivering high-end art prints have different requirements than amateurs who appreciate a well-done print.

@ digitaldog

Thank you for your time on this. Discussing this on the forum has helped a few pennies to drop in my understanding of calibration and the numbers, i feel i know what i need to know now, and can direct my attention there.

For example, i can identify exactly which colours are innacurate (dE 3+) in different parts of the screen by doing full measurment-reports from different parts (rather than the minimal uniformity-report which i first posted). At the moment i'm noticing that only specific colours in the shadows return a dE of 3-4, the others are all within recommended-tolerance -this tolerance being what i need. Will be repeating the tests again tomorrow to confirm repeatability.

So yes, i feel i've progressed, and can make a decision (after further measurments and calibrations) that i'll be comfortable with.

Thanks again.

Regards

:)

I

And that's the point: most of us have to draw a line somewhere when it comes to budget, and then it's about getting something that does the job good enough since perfection is too expensive.

Hi Wouter,

Yes, this is it, what everyone who can't go out and buy an Eizo has to weigh-up and balance-out :)

Like most people, i think i've read about it and thought about it too much. However, i feel like i've come through the worse of it and have a level of clarity about my 'acceptable-minimum' needs.

Yes, i bought the U2413 because of what it cost (about £130) and because of its wide-gamut and potential to be colour-accurate and uniform. The latter i confirmed through reviews of the model from sources whose approach i respect. So, it's not that i feel i need an Eizo (CS2420, btw, if ur wondering) for my work. I bought it with the expectation that if it was a good copy then it would satisfy my needs.

When i originally posted, the place where i was coming from was: Is this U2413 copy a good one, or is it a dud? -Homogenity across copies has always been a reasonable concern for this model, according to the web. I thought i might hear from someone who had a U2413 they were happily working with, who would indicate the quality of my measurment-report compared to their own.

So, i'd be more than happy with this monitor if (i knew) the panel was performing to expected standard. If i decide to return it, it will only be because i think it's a sub-standard copy.

If i do return it, then god, yes, i'll be back in that hole of weighing-up and balancing out. AdobeRGB gamut is the way i'd choose to go... but also, i have no issue working in sRGB (100% gamut, colour-accurate across the display to <=2 dE) if that's the way it has to be -better making photos than evaluating colour-spaces, right? Thanks for the monitor recommendation, i've put it on the Plan (sRG) B list.

Regards,

:)

@ digitaldog

 

Thank you for your time on this. Discussing this on the forum has helped a few pennies to drop in my understanding of calibration and the numbers, i feel i know what i need to know now, and can direct my attention there.

:)

But wait, there's more....

Delta-E and color accuracy

In this 7 minute video I'll cover: What is Delta-E and how we use it to evaluate color differences. Color Accuracy: what it really means, how we measure it using ColorThink Pro and BableColor CT&A. This is an edited subset of a video covering RGB working spaces from raw data (sRGB urban legend Part 1).

Low Rez:

High Rez: http://digitaldog.net/files/Delta-E%20and%20Color%20Accuracy%20Video.mp4

But wait, there's more....

Helpful information. I've listened to your piece on AdobeRGB /sRGB as well.. which encourages my thoughts that AdobeRGB is the colour-space i need to be working in.

I've finished doing my test-measurments from multiple points in the centre of the display (roughly the area of the 9 centre grids from my original, 25-grid report). -Because this centre-area is large enough for me to view photos in. Would you confirm that my interpretation of the measurments is correct?

In-depth measurments from the centre are generally dE max <=2, passing all recommended-tolerances (according to the measurment standard). In measurments taken from the centre-edges, which give the poorest results, it is only the White-point which fails even a nominal tolerance. -Below are 2 of the reports which are typical of the best and worst results; the first sampled from off-centre, the second from edge-centre:

2:

* Are these results saying that i can trust the centre-area of my screen to be colour-accurate, except in respect to luminance?

* Grey-balance and Shadows are going to spot-on?

(I guess this confirms what you noted in the original uniformity-report, that colour-values were less off than luminosity?)

I have one more question relating to measurments and consistency: I've taken measurments from the the *same spot* on the screen 3 times, with 5 minute intervals between each measurment. Each time i get slightly different results; the greatest difference being in RGB grey balance.

Is this to OK? -Or does it indicate that either the display or colorimeter is not consistent?

The 3 measurments:

I have one more question relating to measurments and consistency: I've taken measurments from the the *same spot* on the screen 3 times, with 5 minute intervals between each measurment. Each time i get slightly different results; the greatest difference being in RGB grey balance.

You cannot nor should you expect a dE of 0.00 or even close with multiple measurements from any device. It should be a small fraction of 1.0 however. So I don't see this as an issue. The other question (and there's a lot of type/text in the reports and I haven't scanned it that closely) is what dE formula is used. Some like dE 2000 are 'better' for reporting small differences than others. This is simply to point out that the reports can vary and differing formulas are better for some tasks than others.

You cannot nor should you expect a dE of 0.00 or even close with multiple measurements from any device. It should be a small fraction of 1.0 however. So I don't see this as an issue.

(Sorry, a little lost in language there -) You're saying if i run a patch-test multiple times from the same point on my display, a variation of up to dE_0.25 is OK...?

Formula-wise, yes, dE 2000. The ISO 14861:2015 reports from displayCAL seem to be quite deep (seems a better tool for verification than i1Profiler, to me anyway).

(Sorry, a little lost in language there -) You're saying if i run a patch-test multiple times from the same point on my display, a variation of up to dE_0.25 is OK...?

 

Formula-wise, yes, dE 2000. The ISO 14861:2015 reports from displayCAL seem to be quite deep (seems a better tool for verification than i1Profiler, to me anyway).

Yeah, that's not out of the realm of 'noise' seen from such an instrument making multiple measurements. Nothing to worry about.

Here's a report of an i1 iSis Spectrophotometer where I've measured 567 colors two days apart. Notice too how nice ColorThink Pro reports this data showing average dE (so of the 567 patches), max (the worst single difference), then best and worst 10% and so on. And in this case, there was one case where two patches indeed produced a dE of 0.00.

--------------------------------------------------

dE Report

Number of Samples: 567

Delta-E Formula dE2000

Overall - (567 colors)

--------------------------------------------------

Average dE: 0.09

Max dE: 0.78

Min dE: 0.00

StdDev dE: 0.09

95th %ile dE: 0.11

Best 90% - (509 colors)

--------------------------------------------------

Average dE: 0.07

Max dE: 0.17

Min dE: 0.00

StdDev dE: 0.03

Worst 10% - (58 colors)

--------------------------------------------------

Average dE: 0.31

Max dE: 0.78

Min dE: 0.18

StdDev dE: 0.14

--------------------------------------------------

--------------------------------------------------

Well those are rather excellent figures :)

As i said before, something has (fortunately) clicked and i've got the perspective i needed on this subject. I decided to send the monitor back based on the overall innaccuracy of the panel... i'd rather work in 100% sRGB on a uniform, colour-correct display, i think (the limitation of sRGB colour-space is not a limitation on technique, as long as the display is accurate). Anyway, a good introduction to getting the right photo-dedicated display -i feel confident i can assess the next one, whatever it may be.

With thanks for your help and advice.

Regards.

How about working on a display that has excellent purity through electronics, can produce both wide gamut and sRGB?

NEC SpectraView.

At a cost.... ;)