This blog post has been some time coming. A couple of weeks ago I read a discussion post on Leanne Cole’s blog about the difficulties of choosing the right monitor and calibrating it. I left a note there saying that I’d be writing up a blog post about this because it’s such an important topic and most people wouldn’t even know where to start. I don’t think I’ll be saying anything new here, but the information may be hard to find. And in any case, this is my take on it, and how I have applied it to my own photography.
So, the big question is: if one wants to edit photos, what monitor should one get and how important is calibration? Of course, there is no correct answer here. Much depends on budget. If money’s no object (read: if the cost of gear is simply part of the business expense) I guess one would go with a monitor designed specifically for the creative industry. Things like Eizo’s ColorEdge range or NEC’s Spectraview. I won’t be talking about those here, simply because they’re probably beyond the budget of non-professionals.
For mere mortals, this does not mean one cannot get a decent display. But you’ll have to do your research. The biggest difference will probably depend on the display technology itself. Assuming you’re going for a flat-screen, it’s important to realize they’re not all the same. You may still find some plasmas, though mostly it will be LCDs. Then these can be backlit with CCFL (i.e. flourescent lights) or LEDs. The latter are more efficient. And even LED backlighting can be direct or side-lit. Then there’s the matter of whether the LCDs (I assume TFT, though there are others) are TN, IPS, etc. You can find a good summary of these technologies on Wikipedia. This affects colour rendition, update speed (affects ghosting), angle of view, etc. Frankly it’s a bewildering choice, and the best thing to do is to read up and do your research on specific models within budget.
Personally I went with a couple of Dell UltraSharp U2312HM, a 23″ IPS display, that based on the reviews I had read, performed well for the price. I was lucky enough to have tried them out at work, too, and was impressed. The key thing about the choice of monitor is that this is the primary thing that will affect the quality of what you see. A bad monitor cannot be fixed with calibration, and its limitations cannot be overcome that way either.
What calibration does is to ensure that you get the best out of what you have. With equal importance, it also ensures that what you see is repeatable on another good calibrated display (or printer). Calibration actually involves two steps: setting up the device (correctly, this is the calibration step) and profiling it to create a software profile. Both are best done with the aid of a hardware device that measures the colour and brightness of a patch of screen. Personally, I went with a Spyder2, simply because it works on Linux and used ones are rather inexpensive. Since it’s a solid-state device that is not normally used for long periods of time, there isn’t much to go wrong. I use Ubuntu as my primary OS, so I couldn’t simply use the Spyder2 software directly. Instead I went with dispcalGUI software, which does an excellent job (even better than the software that comes with these devices, in my view).
The idea behind the first step is to adjust any parameters that the monitor allows you to (at least you’ll get contrast and brightness, and on many you can also play with the balance between R,G,B channels) to get them as close as possible to the ideal. In my case I set brightness and contrast until I had the screen brightness I wanted (this is also measured by the hardware) and then set the RGB channel gains to get as close as possible to the target 6500K white point (this is the standard reference daylight). It bears saying that this process should be done after the monitor was left on for sufficiently long to reach a stable operating temperature (read: probably a few hours).
Once that was done I simply let the software profile the monitor. This creates a table (and/or curves) to load into the graphics card, which inverts any non-linearities in the monitor output. In effect, for every colour within the monitor’s range, it tells the graphic card what it should put out so that the displayed colour is what it should be.
For example, for my monitor, the graph above shows its gamut (the colour triangle) against that of the sRGB colour space (the dotted line). The curvy triangle is the range of all visible colours. This clearly shows two things: a) that my monitor pretty much covers the sRGB colour space, and b) that this is much smaller than the range of all possible colours.
Note that the size and position of the monitor’s triangle has nothing to do with the generated profile. But it will be affected by the calibration step, i.e. by the settings on the brightness, contrast, RGB channel gain, etc. It is also worth mentioning that if the range of the monitor is much smaller than sRGB, no profiling will fix that.
The limitations of sRGB space are well known, and this is why many creative pros would not use it as a default space. For example, the Adobe RGB (1998) colour space, which can be used on most modern DSLRs, allows a considerably larger gamut. You can see how my (sRGB) monitor compares with this extended colour space above. Note the much larger range of Adobe RGB on the green side of the spectrum. I mention Adobe RGB as an example, and because in all likelihood you’ll have heard of it at some point. There are other, even wider, gamuts, but for the purposes of most people, these are of no practical use.
So what can we gather from the above? That if you care about editing photos in such a way that someone else can see them exactly as you intended, you need both a good monitor (i.e. one with a good range of reproduction), correct calibration, and a proper profile. You also need to expect every one of your viewers to do the same, which frankly won’t happen. The best you can hope is that most people will have monitors which are reasonably close to sRGB, and export your images for that colour space.
Another possibility, of course, is that your target is a printed photo. It seems to me that not many people do that often, which is a pity. The detail and presence of a good large print cannot quite be equalled on screen. In this case, thing are a bit more complicated. Most printed media have a considerably different gamut than an sRGB screen. For example, the figure below compares my monitor with the ISO Coated v2 profile, which is a standard profile for offset printing.
You can see right away that the printer loses out in the pure green and pure blue corners, as well as some of the violet range. Deep blue, in particular, is notoriously difficult to print well. However, since we’re talking photography, I’ll end with a profile for the Fuji Crystal Archive paper when exposed with a Polielettronica laser (this is the printer and paper used by PhotoBox for their larger prints). This is shown against the sRGB space.
In this case, observe how the printable range effectively covers the sRGB space. But what you cannot quite see is that the brightness at the corners is diminished. It is almost impossible, for example, to print fluorescent green on such a printer/paper combination. It’s also worth noting that the outline is highly nonlinear.
The best way I found to print is to first convert my images from Adobe RGB (my default working space) directly to the printer profile using a relative colorimetric intent with black point compensation. This ensures that the file I upload for printing is already at the target printer’s profile; I then simply ask them to print as-is. Using this method I managed to get a satisfactory print of Glistening Greens, which has some rather difficult colours.
As a matter of principle, I also apply the same process when exporting my images for sharing on the web (this includes everything in my recent posts). Of course, in that case, the target profile is simply standard sRGB.
My apologies for such a long-winded post. I hope it has been useful. Comments welcome, as always. And please note this is not a sponsored post.