When people initially get involved in color management, one of the first questions that pops up is what ‘gamma value’ (determinant tone reproduction curve) they should calibrate to. These days, 2.2 seems to be the most common answer, with an occasional 1.8 supporter thrown into the mix. On this page, I will discuss the reasons for picking one of these values, and I will discuss two more exotic options as well.
The tone response curve
Before delving into the details, it is important to establish the difference between display calibration and profiling, and their relation to the tone reproduction curve. In this light, calibration means getting the display chain into a state with specific tone reproduction curve. This is achieved by a combination of controls on the monitor and look-up tables (LUTs). Those LUTs can be located on the video card or inside the monitor (on advanced models only). The profile, on the other hand, describes the tone reproduction curve and makes this information available to the operating system and application.
A color-managed application will take the image data and convert it to the display using the information supplied in the profile. Part of the conversion process is the application of the inverse tone curve, to compensate for the tone curve that will be applied by the display system. By design, the net result of these operations is that the shape of the tone curve does not matter. In other words, for color managed applications it makes no difference whether you use a gamma of 1.8, 2.2 or an altogether different tone reproduction curve.
If the choice of tone reproduction curve has no effect on the output of color-managed applications, it would seem that there is no reason to even discuss this subject. However, there are two reasons why one could prefer one tone curve over another.
First, note that only color-managed applications compensate for the tone reproduction curve. The other applications and the operating system itself simply send their data to the display unadjusted. The choice of tone reproduction curve will therefore determine the look of these other applications.
Second, the careful selection of a tone response curve can be used to evade low-precision bottlenecks in the system. In particular, the DVI connection that is used to connect most flat panels only supports 8 bits of color information per channel per pixel, which limits you to 256 discrete values. By distributing these values (visually) evenly across the tonal range of the display, the effects of this bit-depth limitation can be minimized.
Armed with this knowledge about tone reproduction curves, we will now look at various candidates and discuss their relative merits.
A gamma value of 2.2 is the default champion. Many programs and interface elements are made with a gamma value of 2.2 in mind, so this is a safe choice. Is is also close to the tone reproduction curve of sRGB, meaning that non-color managed sRGB images are shown with approximately the correct brightness.
A gamma value of 1.8 used to be the default on the Mac platform. However, as far as I know (I’m not a Mac user) this is no longer the case. Unless you have a specific reason to choose 1.8, I’d advise against it.
L* calibration has become somewhat of a rage among higher-end profiling packages in recent years. The L* curve approximates the human ability to perceive brightness differences. By calibrating to L*, you make optimal use of low bit depth channel between the video card and the monitor. However, this only makes sense for monitors with an internal LUT. If you have to rely on the video card’s LUT, which usually has 8-bit precision as well, there is no benefit to this tone curve.
The final tone reproduction curve is a little unusual. For the sRGB color space, each channel is encoded by a tone reproduction curve that is similar, but not equal, to gamma 2.2. The main difference is that the shadows are a little more well-behaved in sRGB. The advantage of calibrating to this tone reproduction curve is twofold. First, the change in shadow behavior leads to slightly improved shadow definition when soft-proofing in Photoshop. Second, using the sRGB tone curve means that sRGB grayscale images also look as intended in non-color-managed applications. For these reasons, I am currently using the sRGB curve to calibrate my own display.
Author: Simon Tindemans