Sidebar 3: Technical information

All of the testing was done using a very large set of test patterns produced by a special version of the DisplayMate Multimedia Edition, our most advanced Windows-based video diagnostic product. This is the same product used by CNET and most major publications for testing and evaluating monitors, projectors, and HDTVs. The Multimedia Edition is able to produce accurate test patterns for all of the player's unique resolutions because it generates the patterns in real time from scale-free mathematical equations. Producing the test patterns at the native resolution of each player was essential for all testing.

We also used the NIST (National Institute of Standards and Technology) FPDM, the Flat-Panel Display Measurements standard, which is produced in association with VESA (Video Electronics Standards Association). The screen-reflectance measurements in Table 2 were done in accordance with FPDM 308-1, Reflectance with Diffuse Illumination, using an integrating hemispherical dome and a calibrated diffuse white reflectance standard. All of the players were tested with their backlight set for maximum brightness and connected to their AC adapter, so the battery state was not an issue.

All photometry and colorimetry for the article was done with a Konica Minolta CS-200 Spectroradiometer, which was provided by the Konica Minolta Instrument Systems Division. A Spectroradiometer was necessary because the more common and inexpensive colorimeters would have delivered inaccurate results. This Spectroradiometer also has a narrow 1-degree acceptance angle, which is very important when measuring LCD panels. The NIST-VESA FPDM calls for a maximum acceptance angle of 2 degrees.

Figure 1 shows the grayscale transfer function (also referred to as the gamma function) for each player, measured with the Konica Minolta Spectroradiometer. The open symbols plotted on the graph are the measured data points. They are plotted on a log-log scale because the desired relationship is a mathematical power law, which appears as a straight line on a log-log graph. The graphs are all staggered vertically for clarity. All of the players depart significantly from the desired grayscale and show significant compression (flattening) at both the bright and dim ends of the grayscale, which is typical for undercompensated LCD panels. The gamma (Table 4) was calculated from the logarithmic slope in the central 40 to 80 percent signal range. The optimum gamma values are in the range of 2.2 (the official standard) to 2.5 (high contrast). Most players greatly exceed this, which produces too much picture contrast and increases color saturation, which is actually beneficial because of their reduced color gamut (see below). The bright compression (Table 2) is calculated by extrapolating the gamma fit to 100 percent intensity and the dark compression (Table 4) by extrapolating the gamma fit to 20 percent intensity. In the 20 to 40 percent range, the Archos AV500 and the Cowon A2 both show dark acceleration instead of compression, both of which are bad.

We measured the primary colors for each player with the Konica Minolta Spectroradiometer and plotted them on a 1976 CIE Uniform Chromaticity Scale diagram with u',v' coordinates in Figure 2, together with the Rec.709 standard primary colors, which are marked by the black triangle. The white points for the players are plotted together with the D6500 standard (which is obscured and lies almost directly underneath the value for the Cowon A2). The white point for the Creative Zen Vision is far from both D6500 and the black-body line.

Table 4 lists the correlated color temperature for each player's white point. Table 4 also lists the color gamut for each player compared to the entire visible spectrum, determined from calculating areas in u',v' color space. They are used to calculate the color gamut values in Table 2 compared to the sRGB and HDTV Rec.709 standards.

Figure 3 shows two optimized grayscales for the Archos AV500 produced with the DisplayMate image optimizer that was used to produce the optimized photos. The native grayscale for the player is shown in black and is the same as in Figure 1. Note that it is shaped like an S instead of a straight line. The blue graph is an optimized grayscale designed to have a gamma of 2.20, and the red graph is an optimized grayscale designed to have a gamma of 2.50, which is a bit steeper. The open symbols plotted on the graph are the data points measured with the Konica Minolta Spectroradiometer for the optimized grayscales on the player. The optimized grayscales turn horizontal when they reach the player's black level. The optimized grayscales for the other players behave in exactly the same way. A gamma value of 2.30 was used to generate the optimized photos for each player.

More resources