Major update of my article on digital darkroom lighting

<p>I've updated my article on digital darkroom lighting. You can find it here:</p>

<p>http://solux.net/ies_files/Digital%20Darkroom%20Lighting%202013.pdf</p>

<p>The major points/updates are:<br>

- monitor gamma has nothing to do with the non-linearity of human vision<br>

- new list of sRGB and Adobe RGB monitors with IPS panels<br>

- why 6500K is so prevalent<br>

- why prints often come out too dark<br>

- working in total darkness is a surefire way to develop bad habits and get bad results</p>

<p>I have no ties to SoLux other than liking their products.</p>

<p>Your list of available IPS monitors I found helpful and reassuring and pretty much what I've found so far on my own after my Dell 2209WA developed distracting non-uniformity artifacts just a year after its 3 year warrantee expired.</p>

<p>One thing I can't seem to get a thorough answer online even from Dr. Soneira of DisplayMate.com is how to detect whether some of these newer (and cheaper) sRGB gamut LED backlit displays have 8 bit or 6 bit +A-FRC panels. From what's been mentioned over at tftcentral some of the Asus and Dells you've listed come with 6 bit panels.</p>

<p>Got tired of buying displays online hoping I'ld get one that came with a decent IPS panel and decided to try out the LG 27ea63v-p from my local Best Buy ($330). I did the typical gradient tests and can't see any dithering or severe banding. I also developed a lowered saturation complementary color purity gradient test on 3 layers inverting and changing blend modes to try to see any signs of the LG being a 6 bit and couldn't.</p>

<p>It's gamut after profiling with the Colormunki Display is a bit less than sRGB in blues, cyan and red but juts out beyond in greens and yellows. All my images look the same as on the Dell so I guess I got me a good enough monitor for photo editing.</p><div></div>

<p>Speaking of sRGB and Adobe RGB coverage, you can get the wider gamut coverage in your phone if you want:<br /> http://www.phonearena.com/news/UL-certifies-the-4.99-FHD-display-on-the-Samsung-Galaxy-S4_id43874<br /> Doug</p>

<p>From Doug's link...</p>

<blockquote>

<p>According to UL, the screen on the Samsung Galaxy S4 offers one of the <strong>best color reproductions</strong> in the mobile industry with the broadest color gamut of up to 97% for the Adobe RGB color space.</p>

</blockquote>

<p>If they don't implement color management technology their version of "best color reproduction" will look like a circus side show. I'm so glad I don't own a mobile phone.</p>

<p>It's not like Samsung doesn't know about this issue with wide gamut color devices if they can include it somewhat in their sRGB HDtv's like their UN32EH4003 I got at Walmart. They call it "Wide Color Enhancer". What wide color are they enhancing for on an sRGB tv? Is HD content in a wide color space? Not from what I'm seeing. Or is this specific to BluRay content?</p><div></div>

<p>Tim wrote: "From what's been mentioned over at tftcentral some of the Asus and Dells you've listed come with 6 bit panels."<br>

Absolutely <strong>not!</strong> All the monitors on my list are 8 bit or higher.</p>

<p>Can you prove it, Frans? Really! I mean it! I don't have a slew of panels to compare against. That's why I resorted to buying local so I can take it back without the shipping hassles and waste of time. Do you have a test for determining bit depth on these monitors? I'ld really like to try it out.</p>

<p>I can't find anyone online to prove whether those displays come with an 8 or 6 bit panel. The manufacturers aren't talking and in fact according to the REALLY long discussions at AVSforums they're now deliberately concealing the manufacturing codes that indicate each particular model's panel technology. </p>

<p>This cnet review of the Dell U2410 doesn't suggest it's not an 8 bit panel but the descriptions of the visual anomalies indicates the reviewer must have gotten a bad "not so perfect" build version of this model in the review posted below...</p>

<blockquote>

<p>Static dithering in the shadows?! Pinkish hue in whites?!...</p>

</blockquote>

<p>Huh?! I don't even get that on my $300 Dell and current LG 27"!</p>

<p>http://reviews.cnet.com/lcd-monitors/dell-ultrasharp-u2410/4505-3174_7-33772972.html</p>

<p>I wish congress would do something about forcing these display manufacturers to be truthful and forthright about the most important component they're including in these displays...THE FREAKIN' PANEL TYPE AND BIT DEPTH! HOW HARD CAN THAT BE?!</p>

<p>Look it up in product reviews, manufacturers' specifications and websites like prad.de/en and tftcentral.co.uk.</p>

<p>Frans, I believe I already indicated I did that. Yes, their specs say they're 8 bit panels for the copies they reviewed. That's not what I was talking about which I thought I made clear with the Dell U2410 cnet review.</p>

<p>Over at AVSforums they're calling it the "Panel Lottery" issue when purchasers who are not established online reviewers get their copy of the same model only to find the panels have been swapped for either better grade (like the IPS panel that came with my Samsung HDtv) or less desirable MVA and even TN panels.</p>

<p>http://www.avsforum.com/t/1406151/the-official-2012-samsung-eh4000-eh5000-eh5300-owners-thread/1440#post_23412205</p>

<p>Okay, I took your advice and recalibrated my monitors to 5000K. This is gonna take some getting used to; we'll see how it works out...</p>

<p>I couldn't take it. I've already backed off to 5800K.</p>

<p>Tim,</p>

<p>Do you have any more specific references to this so-called lower-bit-depth-panel-swapping than a 49 page thread? In the past Dell was notorious for swapping TN and IPS panels in "their" monitors, but that is as far as I can tell way, way back in the past. It would be nice to back up your claim about the Dell and, I believe the ASUS monitors, with more than hearsay.</p>

<p>Mark,</p>

<p>You make my point exactly about people developing bad habits. 5000K is way more in line with the real world than anything higher than that. Did you compare your 5000K screen image with a printed version? How did that compare?</p>

<p>I didn't get as far as printing anything; in the past I've felt I had a reasonable match at 6500K. Just couldn't take how muddy everything looked. I'll let myself adjust to 5800K for a week or so, then maybe I'll try 5000K again, along with some print testing. I've never felt that the lighting in my work environment really matches the (highly variable, not controlled by me) lighting where my prints are hung anyway...</p>

<p>Mark,</p>

<p>The mismatch I'm talking about doesn't apply between the computer screen where you do your image editing and a viewing environment outside of your digital darkroom. It's the mismatch between your screen and the print when viewed in your digital darkroom lighting.</p>

<p>I was persuaded by the argument that the lighting where my prints are hung (again, not under my control) is typically going to be considerably warmer than 6500K.</p>

<p>The CCT K value you calibrate to is all over the map. For one, you can calibrate with instrument A, software package A and get one result, use the <strong>same instrument</strong> and package B and get another result. There are a number of reasons for this.<br>

Next, the only value that makes any sense is the one that produces a color appearance that matches whatever you want to match (usually a print). As a print is illuminated by something, that plays a huge role so the value you enter into the software is going to vary. Next, the type of illuminate used in the panel will play a big role too. I just read a fascinating article by Abhijit Sarkar's recent submission to CIE regarding his work on Standard Observer. I don't think it's out for public consumption yet but in the piece, there's mention of an experiment where users were asked to match via a display using differing back light an image they are viewing in a booth. Two displays were both calibrated using <strong>the same product to 5200K</strong>. Both displays had quite different SPDs as one used CCFL, the other LED and the spectrum of the two are vastly different. They saw colors on one display being much warmer than the other, the results of the differences of the match they made was between 3 and 13dE (76), the average for all users was a dE of 8! Mostly in the aStar. All viewers were checked for color blindness and scored high in the FM100 hue test. </p>

<p>Anyone that tells you to calibrate to this CCT or that cd/m2 without having the print viewing conditions in mind, and based on this piece, the type of display technology isn't providing much useful in the advise. The numbers DO NOT MATTER unless you've got a group of really high end reference display systems in a collaborate group and everyone is expected to see the same results. And this doesn’t mean the numbers everyone agrees on is correct, but at least consistent. </p>

<p>Andrew,<br /> My article makes very clear that the color match I'm discussing is the one between the image on the screen and the print of the same image viewed in the digital darkroom lighting. I also emphasize that the brightness of the screen and print need to be matches as well, and I describe how to check for those matches. I also advise to use 5000K for the monitor with gamma 2.2 (very important as well to compensate for the distortion caused by gamma encoding) and the use of SoLux 5000K bulbs.<br /> Your comments, while I know that they are not ment to be intepreted as such, can easily be interpreted by the reader as "anything goes", something that is absolutely not the case.<br>

The numbers matter a lot when you want the image on the screen to closely match your print, so there is a strong correlation between the two; without such correlation it becomes difficult to do any meaningful editing. The print provides feedback and if there is no good match, that feedback becomes meaningless at best and misleading at worst.<br /> I personally don't put much value on any study that uses fluorescent or LED lighting because of the horrendous peaks in their power spectrum.</p>

<blockquote>

<p>My article makes very clear that the color match I'm discussing is the one between the image on the screen and the print of the same image viewed in the digital darkroom lighting.</p>

</blockquote>

<p>Every lighting condition, even the intensity of the same illuminant and every backlight will affect what is possibility an effect upon viewer metamerism and again, the software and instrument you use has to be identical to the reader if you (and they) expect the same results.</p>

<blockquote>

<p>I also advise to use 5000K for the monitor with gamma 2.2</p>

</blockquote>

<p>Again, 5000K is a range of colors and pretty meaningless based upon the context of what I wrote! And TRC gamma is meaningless in ICC aware app's. Further, there are advantage to some display systems using a native gamma setting IF the software supports it. <br /> Bottom line is advising people to use fixed settings without them using absolutely identical components (display with defined backlight, instrument+software, viewing conditions) does far more harm than good. YMMV! So will theirs.</p>

<blockquote>

<p>Your comments, while I know that they are not ment to be intepreted as such, can easily be interpreted by the reader as "anything goes", something that is absolutely not the case.</p>

</blockquote>

<p>Actually that's spot on unless we are discussing identical hardware and software. Further, ignoring viewer metamerism, even each individual's perception of color isn't a good idea. This is exactly why the piece I referenced exists! It points out the folly of trying to suggest simplistic settings and advise without specific hardware, software and viewing conditions is a waste of everyone's time.</p>

<p><a name="pagebottom"></a></p>

<blockquote>

<p>I've never felt that the lighting in my work environment really matches the (highly variable, not controlled by me) lighting where my prints are hung anyway...</p>

</blockquote>

<p>This begs the question, if the print isn't anywhere near the display, how would you know there's a match between the two? </p>

<p>Andrew,<br>

Instead of critisizing what I wrote as meaningless and harping on what all could go wrong, maybe you should follow my advice and see for yourself how a 5000K monitor compares to the print viewed with SoLux 5000K lighting after the brightnesses of the monitor and lighting have been matched as described in my article. I'd be very interested to hear how that works out for you.</p>

<blockquote>

<p>Instead of critisizing what I wrote as meaningless and harping on what all could go wrong, maybe you should follow my advice and see for yourself how a 5000K monitor compares to the print viewed with SoLux 5000K lighting after the brightnesses of the monitor and lighting have been matched as described in my article. I'd be very interested to hear how that works out for you.</p>

</blockquote>

<p>I never said your piece was meaningless, it has some very good points. Next, I do have Solux viewing conditions and <strong>no</strong>, 5000K doesn't work on this end with an NEC SpectraView II anywhere as well as the settings I had to find by trial and error to produce a visual match. Next, if you read what I wrote, you'll see we are in agreement in a number of areas including how differing SPD's of both print viewing and display backlight can affect this match. You also may wish to consider the piece I referred to and the results whereby depending on the SPD, a dE difference of as high as 13 was seen on two differing display technologies. That's due to viewer metamerism failure. Based on both our understanding of just how the backlight of a display can alter the results, how can you say that everyone should calibrate to 5000K? Are you certain that every instrument used to calibrate a display, mated with all the various software products that have a setting for 5000K produce identical results? You certainly understand that 5000K is correlated, the only way to really get a display to produce that would be to heat it until it was a pool of molten plastic. That the blackbody is a theoretical object and that yes, your mileage will vary, a lot and that simply expecting everyone to set 5000K and get a match is impossible. IF that were not the case, why is it that higher end systems like Eizo and NEC provide such precise control over differing white point settings? CCT Kelvin values in one value increments, the ability to enter chromaticity values, even the ability to measure with an instrument the viewing conditions? Why not have a single setting in such software products: 5000K? Because that one size fits all WP doesn't work for everyone, not even close. <br>

<br>

The answer is clear. 5000K is a range of colors, doesn't ensure any kind of match to anything and varies by instrument and software package. You ask: "<em>anything goes", something that is absolutely not the case. </em>When in fact, that is the case. The correct numbers, which vary are the ones that produce a match for you (or anyone else). <br>

Do you have a Spectrophotometer that can read the Solux bulbs? I assume you do since you show SPD's in your article. What's the Spectrophotometer say the CCT value is of the Solux bulbs? 5000K on the money? </p>

<p>Andrew,<br>

What do you mean by "SoLux viewing conditions"? Do you use the SoLux "5000K" bulbs? Do you have any other lighting in your digital darkroom? Is there no outside light getting in? Do you have a neutral "color" scheme in your digital darkroom? Does the brightness of the monitor match that of your prints? And what are the settings that you use with your SoLux lighting?<br>

<br />I don't have a spectrophotometer; the SPD curve is straight from the SoLux website, with permission. The datasheet on the SoLux website states the CCT for the "5000K" bulbs as 4901K at 12 Volt and the CRI as 98. And yes, when I calibrate my monitor I select 4900K and a gamma of 2.2.</p>

<blockquote>

<p>Do you use the SoLux "5000K" bulbs?</p>

</blockquote>

<p>No I don't use them nor recommend them! Do I own and have I tested them? Yes. I have probably every Solux product you can buy and some you can't. You'll find my name on their web site (http://solux.net/cgi-bin/tlistore/infopages/color-proofing.html). I've written about them for nearly a decade: In my book (2005), including how to build a print viewing system with them. I've written about them in print (http://www.ppmag.com/reviews/200604_rodneycm.pdf) and in presentations (Epson Print Academy). I tested their PAR's before release and was asked by Solux to write a letter on their behalf about my needs as a industry expert for such bulbs to petition congress to allow them to continue to make and sell them!</p>

<p>The only reason Solux introduced more recently a bulb labelled 5000K is to make people like yourself happy with a <em><strong>number</strong></em>! It is the same bulb as their 4700K with an increase in voltage which shortens the life and cost you more to buy. This is why <strong>I do not recommend</strong> them but I understand how buyers would strongly wish to have a bulb with such a value, especially after reading articles like yours. There is no need for this bulb and in fact, depending on the proofing work you are doing, you're better off with the CCT 4100K blub! When I was teaching for Art Wolfe in his studio/gallery/class, we did testing with both 4100K and 4700K and the former was far better for viewing prints and proofing. The 4700K's are too cool appearing for many. For my needs, I use Solux CCT4700K (and a GTI Fluorescent booth).<br>

<br /> There are Fluorescent bulbs ranging from $1 to far more that have 5000K on them. Do you believe they are identical and produce 5000K? But enough questions here, I answered yours about my use of Solux products, how about you answer the questions I posed to you above.</p>

<blockquote>

<p>I don't have a spectrophotometer; the SPD curve is straight from the SoLux website, with permission.</p>

</blockquote>

<p><strong>I do have a Spectrophotometer</strong> and the plots you see in the article were measured by me. You're not equipped to evaluate color temp if you don't have such an instrument, sorry. As for my studio/lab, I have a very well designed color reference environment. I've setup multi-thousand dollar systems for clients who's viewing and proofing needs require precision to a level of less than 1dE! <br /> Sorry you're put off by peer review. Maybe you would consider sending the PDF to say Chris Murphy or Phil Brandfield at Solux. But the <strong>bottom line</strong> is, suggesting a fixed value to set for display calibration without taking a number of important factors into consideration does more harm than good. Those factors have been presented to you more than once, I would ask you look them over and reply. Or ignore them, doesn't matter to me, might to anyone reading your updated piece.</p>