Dell U2711 Brightness/Contrast Spyder3

[demianov]

<p>Hello.<br /> I've had this panel for 2 years now and have been color-managed much longer than that. This is not your garden-variety I-need-a-quick-fix amateurish gibberish. If you do decide to take me seriously, I ask that you try to understand *everything* in this post before offering advice. Either way: thank you for your time.<br /> I'm still beating myself up for getting this panel. However, until it gets replaced, I have to make the best of it. The question I'm about to pose is more of a fundamental nature anyway.<br /> I've read quite a few reviews online about the U2711 panel in search of the 'ultimate setting'. Also on the forums guys are posting how *they* managed to achieve *the perfect picture*. The settings vary *greatly*, which only underlines that there is no real practical standard of any kind out there. Still, what *I* want to achieve is a picture that looks RIGHT. By that I mean: accurate colors, brightness and contrast. After 8 years of dealing with color managing, that definition still eludes me, namely because display systems vary greatly in brightness of the projected image and *that* affects the picture pretty much in every way.<br /> With all the adjustments on the hardware and software levels how can I know for sure that what I'm seeing is really the way it is supposed to look? The majority of the panels out there aren't calibrated *at all*. The very few people I know who *do* work in a color-managed environment, all have their preferences. During the last year or so I did get the chance to see my photographs rendered by a handful of other display systems and what I saw differed at least to some extent from what I thought those photos should look like. I haven't done any printing in years, so to me this is not about display-to-print matching. This is about doing all of my post-processing *right*. I send out my photos to publishers and agencies. A number of times I got to see my work in printed material and every time the colors and contrast looked way off. I'm 99% sure that they applied at least some adjustments, and that's the norm, really, but then how can I know that what they initially saw was exactly (or even close to) what I had seen on *my* display? I haven't received any complaints yet but can you blame me for simply wanting to do my job right? Enough of this. Let us get down to business.<br /> The environment I work in is always pitch black. Just trying to be consistent. The Brightness level of the panel that I am comfortable with is from 4 to about 8. With anything above that I get killer headaches the following morning. I realize that that is WAY below what other folks shoot for. However, the picture doesn't appear dark to me, plus I couldn't make it brighter even if I wanted to. Now for the Contrast.. In my CRT days I always had the contrast at 100% and brightness at 0%, which looked great and made a lot of sense too. Something tells me that is not the way to go with *this* display. I ran the Spyder calibration at various settings and reverted to 50% Contrast, the way it's always been. Based on my experience as well as the online reviews the 50% seems to be the sweet spot. Am I right at least on this one? Even though the brightness is way low?<br /> Before going any further here are my system details:<br /> Display: Dell U2711 set to maximum resolution and bit-depth. It's connected via DisplayPort to: ATI Radeon HD 5670 display adapter. All drivers are up to date and I've also made sure that the video card isn't affecting the picture in ways I don't want it to. The OS is Win 7 Pro 64-bit. The spectro and software: Spyder3 Elite.<br /> The color settings on the panel are set to 'Custom Color' and I'll tell you why in a minute. The Contrast is set to 50 and the Brightness: let's say to 6. In the Spyder utility I select that my display has RGB sliders. I set my target Gamma to 2.2, White Point to 6500K and brightness to: Native. <--- Now is this the right thing to do in my case? I'm pretty sure that that's what it needs to be even though it says that the recommended value is 120. I set the Gray balance calibration to: Iterative (which my panel *really* needs). Then I choose to do a "FullCAL". I place the spectro in place, it takes a few preliminary readings and then the fun part begins. The calibration process halts and the software lets me adjust the RGB sliders on the display to bring them within range. This is supposed to be a very good thing because this panel (at least in theory) has a 12-bit LUT. If I adjust the picture on the display itself, the adjustments made to the video card LUT are minimal thus any banding is minimized. I'm not really sure and the information wasn't readily available, but can anyone confirm that my display adapter has a 10-bit LUT? Would be nice to know for sure.<br /> That's the way I've been calibrating for 2 years and I need you to tell me what you think of my method. I work strictly in the Adobe RGB colorspace. Naturally, I want to have a wide gamut but the colors have to be spot-on and even more so - the brightness (and the contrast). The last thing I need is a couple of years down the road, to find out that I've been processing my photos all wrong and I need to redo them all.<br /> With Spyder3 Elite software there's probably even a way to take some Lab measurements to make sure that I'm at least in the ballpark. Any ideas on that?<br /> Hope I didn't put anyone to sleep.</p>

[sheldonnalos]

<p>Everything you are doing sounds fine, except I wouldn't use "Native" as the brightness setting. I would find the cd/m2 setting that suits your eyes and gets the prints to match properly. If you use the native setting, it may drift over time with the monitor and the calibrator won't correct for that. Perhaps choose a brightness of 80 or 90 cd/m2 and see if that works for you. Those seem like reasonable values given your totally dark working environment.</p>

<p>FWIW I calibrate my U2711 with an i1 Display Pro, gamma set to 2.2, I use the custom color mode to adjust the native whitepoint to 6500k before profiling, and I calibrate to 120 cd/m2. The software has me set the contrast at 77 and the brightness at 18 to get an optimal result.</p>

<p>As far as getting a true 10 bits displayed on screen, you'll need to do some extensive research on that. From what I understand every component in the chain has to be compatible from video card to drivers to OS and then software. I think you also have to use a Displayport cable. Your component list sounds pretty modern so you may be able to get it to work, but you'd have to look into that. I also don't think that 10 bit display is that dramatically different from the standard 8 bit display that almost everyone works with currently. You might see it in a grayscale image with fine gradations, but that's about it. Even if you are only getting 8 bits on screen it's not really an issue for the output side on prints.</p>

[digitaldog]

<blockquote>

<p>I would find the cd/m2 setting that suits your eyes and gets the prints to match properly</p>

 

</blockquote>

<p>Exactly. The right setting is that which produces a visual match to the prints, properly viewed. I’ve never heard of a ‘native brightness” but there are native white point options in some software that can be useful, again, if the WP produces a visual match. If not, you have to tweak. </p>

<p>http://www.luminous-landscape.com/tutorials/why_are_my_prints_too_dark.shtml</p>

[demianov]

<p>Thank you, guys, but like I said, I don't care about the print-matching (at least for now). What I need is to work out the optimal Contrast setting suitable for my Brightness level. With this display, the minute I touch the contrast dial - *everything* starts to change. And I mean: the apparent brightness begins to go up, then the colors start to do a funny thing. I guess I should want maximum Contrast, because that's what gets you the maximal Dynamic Range. At the same time I don't want to max anything out, I still need all of my highlights. This is the only non-CRT panel that I have ever gotten to calibrate, so I don't know how it goes with other displays out there. All I know is that *this* panel behaves in an awkward manner. I guess I need to cancel any adjustment that the Spyder3 software is doing, then take some Lab readings while adjusting the brightness and contrast to my desired level and use *those* setting as a constant for any further calibration. But what kind of image should I measure? What method should I use? Obviously I should be measuring a patch of total white but then how do i know I'm not maxing out? There's a million stupid mistakes I could make here so I need a sound strategy, a starting point at least, a way to check that I'm on the right track. The dynamic range that I need to achieve is the maximal dynamic range for my brightness level and that probably means that I need to take a cd/m2 reading and keep it constant throughout the calibration. If I begin increasing the Contrast on the panel I also have to reduce the Brightness because the image becomes significantly brighter. A 100% Contrast value doesn't look right at all but then what should it be? Mr. Nalos here suggested a value of 77, which makes sense but then, it can only be right for *his* panel. I need a proven method to figure this out. And yes, choosing a fixed target brightness makes a lot of sense too, so thank you, so I need to figure out the value here.<br>

As per the Neutrality of the tones, with this panel, at least, the minute you touch the contrast dial, the colors start acting crazy. Call it split-toning, call it whatever you want, but they look terrible, even after running the Iterative calibration. No matter what the setting are or what I do, I can always see some hue shifts in my BW images and that's a damn shame because I'm about to process a whole bunch of them. The darker values look neutral enough but then as the values start to go up there's a greenish tint to them, as they go higher yet they become magenta, and then back to neutral again. It's not that bad and I notice those things only because I know how to look at an image. Something tells me though that that's as good as I'm gonna get it with this panel and *this* spectro. Achieving that true *grayscale* is actually secondary to me right now, as like I already mentioned, I first I need to figure out the best Brightness/Contrast combination that my eyes can also handle. That combination has to make sense too, it cannot have any adverse effects on the way my photos are displayed AND it has to be something that my Spectro/software can handle well.<br>

Where should I begin?</p>

[digitaldog]

<p>If you are not trying for any kind of match, it does not matter, set the display to look as you wish. All the display is for is showing you a big pile of numbers. Calibrating from that point forward ensure the behavior remains consistent. </p>

[demianov]

<p>I'd have to disagree on that. If I only cared about how *I* saw my photos I wouldn't bother with this kind of display, much less - color management. I'd just get me some crappy LCD for $139.99 and *tweak* it to my liking. Apparently that's what most people do, even those who actually care a little about what they pictures look like. I'm no weekend photo-snapper whose only goal is to impress his grandmother with a *cute* picture of her aging cat Bootsy once in a while. My work is seen by thousands of people daily and it's only gonna pick up from there. Some of them actually care about photography and it's only fair they see quality photographs, especially if they're paying money for it. I want to give *them*, the publishers or whoever's involved, the best starting point. Obviously, all my files are tagged with an appropriate profile, but the actual RGB values have to be correct. Dark is supposed to have very low values and *appear* close to d-max rendered by a proper imaging system. Bright is supposed to be close to white and not 'light gray', the gradation - smooth where it needs to be and all the colors have to be spot on. Each of my photo spends a great deal of time in post-production to achieve just that. I know I'm wasting my time if what I *see* is not the way it really is.</p>

[digitaldog]

<p>I can’t answer why you purchased this display for what task (nor can I say it is anything special in terms of a great color managed display system ala Eizo or NEC and the Spyder isn’t all that state of the art either). The fact is, <strong>you have to calibrate a display</strong>. That means you have to specifically select target calibration aim points which at the very least is used to define White Point, back light luminance and TRC Gamma (and if you are fortunate, contrast ratio, something Eizo and SpectraView provide). The targets are based on <strong>something</strong>. 9 times out of 10, the goal of a screen to print match. You don’t care about this. So set the targets to whatever you wish. At least subsequent calibration will bring this unstable output device back into calibration so the RGB values you see today and in a year will look the same. </p>

<p>Once you figure out what behavior you want the display to be, setup the calibration targets (hint, I’d suggest you do this to match a print, even if you have no print today to match. That way, if the time comes you do, you will have the same display behavior for all your legacy images).</p>

[demianov]

<p>Thank you for your response. I completely agree with you on the fact that the hardware in question is not of the highest grade. I pretty much made a mistake, because I was led to believe it was enough for my needs. Apparently it is not sufficient, as I have realized later on. For the meantime I have to make the best of it. Like I said in my initial post: this is more about that illusive concept of how an image is or should be rendered by any given imaging system. What are the factors that make it look *right* and once it does look that - what can be done to make it look its best. Please read the following and express your opinion whether my reasoning is sound.<br>

My understanding of the fundamentals is such. The 'viewer' is always in some sort of environment. Be it a screening room with black walls or an outdoor location with no roof during midday. Let's assume that the light that hits the projected image surface and the viewer's eyes has no color casts, the light intensity itself would still vary greatly from case to case. Provided the device is capable of displaying an image in a way that suits the environment the *Brightness* (cd/m2) would be very different in each of the two examples above.<br>

The way to calibrate a system that isn't affected by any ambient light would have to be thus (and I'm really talking about my case here):<br>

Set the Brightness level to 0. Project an image that shows a proper 'grayscale'. Adjust the Contrast so that the 'pure white' is close to 'maxing out'.  By that I mean: the 'pure white' strip has to be *almost* as bright as the display will render it (without adding Brightness). Next: Load on screen a patch of pure white alongside another patch that holds a value below that of the 'pure white'. What does that value have to be? The spectro has to be able to tell the difference between the two areas, but just barely so. Start increasing the Contrast until the spectro can no longer tell the difference between the 2 areas. Decrease the Contrast to the last value the spectro was able to take 2 different readings.<br>

In theory, the Contrast dial on a display, actually increases the Brightness of the images. A more contrasty image is just that: brighter 'whites' and darker 'blacks', except if the darkest values that the panel can project are already at d-max, they won't be made any darker, and just the Brightness would go up.<br>

What to do with the Brightness? The Brightness setting on any given display is for the sole purpose of adapting the device to the environment. The ambient light can have very different and ever-changing intensity. How to *set* the optimal Brightness? First, the spectro has to be able to differentiate between the hardware d-max of the panel and the next value above that. Increase the value of Brightness until the spectro can do just that. If at 0 Brightness the spectro can tell the difference between RGB 0,0,0 and 1,1,1 you shouldn't be adding any Brightness at all, since you'd be only hurting the d-max. In theory, the brightest values will be increased too in a linear fashion in relation to the darkest values but that's not the case in the real world. Once you've got your Brightness where it needs to be, you should check that 254, 254, 254 isn't maxing out (and make appropriate adjustments if it does).<br>

If all of the above gives you a picture that's too dark for your taste, you can increase the Brightness but know that you're doing that at the expense of your d-max. Add Brightness until you're satisfied but make sure your 254,254,254 isn't maxing out.<br>

Take a reading of 255,255,255 to determine the cd/m2. And have that value as your target brightness.<br>

Once you've reached this point, you're ready to let the calibration software do its thing and guide it in the right direction if needed. The gamma, tones, gray neutrality are all to be handled by the software + spectro combination (at least in my case).<br>

I realize there are different approaches to this but can someone confirm that there are no mistakes in my logic? Remember, it is specific to my case, which is: no ambient light at any time, AND a relatively low cd/m2 output. (and obviously, my set of hardware)<br>

Thanks.</p>