Joined: 06 Mar 2006 Posts: 12404 Location: Ottawa, Canada
TV/Projector: JVC DLA-RS56
Link Posted: Thu Jun 10, 2010 9:05 pm Post subject: Colour Science: Video Calibration Myths
Colour Science: Video Calibration Myths
There are many claims about video calibration—mostly having to do with hardware—that get routinely repeated over and over throughout the years that simply do not stand up to careful scrutiny.
Myth #1: Filter-based colorimeters are acceptable for CRTs, but not for other types of more modern digital displays.
Fact: This myth probably has its genesis in the fact that many years ago colorimeters were based exclusively on internal calibration tables derived from measurements of CRTs only. Even if this was true at one time, it is certainly no longer true. Both of the X-Rite colorimeters include operating modes that are optimized for CRT and for other digital displays.
Myth #2: Filter-based colorimeters are acceptable for measurements of white, but not for measurements of color. Thus, if you want to do simply a grayscale calibration, then colorimeters are fine. However, if you want to calibrate a display's gamut using a color management system (CMS), then you must use a spectrophotometer.
Fact: Colorimeters are generally not as accurate or as stable in the long run as spectrophotometers, but there is no reason to believe that they are any more or less accurate with white than with color in general. In fact, often the largest errors seen when comparing a field colorimeter with a reference spectrophotometer is in white. Colorimeters also seem especially accurate with red. Admittedly, colorimeters tend to show the largest inaccuracies with green. But the average accuracy that colorimeters return with the primary and secondary colors is no better or worse than what you would generally see from a typical measurement of white.
This particular myth seems to have arisen from a marketing campaign designed to convince consumers that the only way to get meaningful benefit from the color management systems that are becoming more common is to use an affordable spectrophotometer, such as the i1Pro. The i1Pro is more accurate than X-Rite's standard colorimeter, the Display 2. However, it is more accurate for both white and color. Furthermore, X-Rite's premium colorimeter, the Chroma 5, offers accuracy similar to the i1Pro and this is again for both white and color.
I now have enough data on the Chroma 5 colorimeter to provide a fairly clear picture of its level of accuracy relative to a known reference. The table below shows the errors of the Chroma 5 when reading white and then when reading an average of the primary and secondary colors for four different types of displays.
White x White y Color x Color y
Plasma 0.003 0.005 0.005 0.005
Sony LCD 0.005 0.003 0.006 0.007
Samsung LCD 0.002 0.005 0.005 0.005
Front Projector 0.004 0.004 0.004 0.005
Plasma: The errors for white and color are, on average, identical.
Sony LCD: Color is slightly less accurate, 0.001 in the x axis and 0.004 in the y axis.
Samsung LED LCD: Again, color is only slightly less accurate in the x axis by 0.003. In the y axis they are identical.
Front Projector: The errors for color and white are virtually identical.
This data explodes the myth that a tristimulus colorimeter does well with white but poorly with color. This data shows that there is either no difference at all or in some cases only a very small difference. Again, this is for the average color reading. Any individual reading may, and often is, less accurate than white. But the average results are about the same.
This shows that the Chroma 5 is an excellent device, at least when factory-fresh. These errors will increase over time as the filters are exposed to changes in heat and moisture, which is why annual recalibration is recommended.
Myth #3: The luminance readings of filter-based colorimeters are inaccurate at very low light levels.
Fact: Colorimeters have a specified range of operation. The specifications for the X-Rite models claim that they read down to 0.02 cd/m2. Within their specified range of operation, their luminance readings are very accurate.
The only way to know if an instrument is returning accurate results (luminance or color) is to compare it against a known reference. Checking the luminance of the X-Rite colorimeters using a Minolta LS-100 as the reference on a 9th generation Kuro reveals the following.
As you can see, the luminance readings of both the Chroma 5 and the very inexpensive Display 2 are quite accurate down to their rated level.
Getting good luminance readings at reasonably low light levels does not require expensive equipment.
Myth #4: The qualities that make for a good display are the following in order of importance: Contrast, Color Saturation, Color Accuracy, Resolution
Fact: This list comes from ISF but has something of a murky origin. The ordering of the qualities are unexceptional. This is often a matter of subjective preference in any case. The problem with this list is two-fold. First, it is not clear what "Color Saturation" is supposed to mean in this context. For a given color, gamut, and gamma there is only one amount of saturation that is correct. Having "more" of it is not a good thing, but a problem that calibration seeks to fix. Thus, the second criteria is either false, or simply a confused restatement of #3. Second, this list ignores sharpness, clarity, uniformity, and freedom from artifacts, any one of which is at least as important as resolution. Most people would much rather have a 720p display with perfect convergence, white field uniformity, zero chromatic aberration, and superior deinterlacing than a 1080p unit that performed poorly in these areas.
Myth #5: Tristimulus colorimeters, such as the Chroma 5, require special calibration tables to work well with LED-backlit LCD displays.
Fact: Not true. The claim that colorimeters, such as the Chroma 5, require special calibration to work with LED displays is an empirical assertion that stands or falls on what the evidence shows.
Recently, we conducted some careful tests under controlled conditions to determine what, if any, special attention the Chroma 5 would require for measurements of LED-based displays. We used two pairs of displays, the conventional Sony XBR7 and the LED-backlit Sony XBR8. We also tested the conventional Samsung B650 and the LED-backlit Samsung B6000. We chose these displays because they represent mainstream popular choices among consumers and the displays in each pair were as similar as possible to each other except one used conventional CCFL backlighting and the other used LED backlighting. Then we measured the entire spectrum of white, red, green, blue, yellow, cyan, and magenta on all of the displays with a reference 5nm spectroradiometer and then with a factory-fresh Chroma 5 colorimeter. Finally, we compared the differences between what the reference device measured compared to what the Chroma 5 measured. If the Chroma 5 had any special difficulty with the LED displays, then we would expect to see a significantly larger discrepancy between what the reference instrument reported and what the Chroma 5 reported on the LED displays than we would see between what the reference instrument reported and what the Chroma 5 reported on the conventional displays. The results are shown below.
The xyY values on the left in black show the accuracy of the Chroma 5 on these four displays. The xyY values on the right in red show the differences in accuracy between the Chroma 5 on the LED display compared to the conventional LCD display.
As you can see, the differences between the reference instrument and the Chroma 5 are generally quite small, indicating that the Chroma 5 is an excellent colorimeter. You can also see that the Chroma 5's accuracy when measuring LED and conventional displays is virtually the same. In fact, to the extent that there is any difference at all, at least with the Sony displays, the Chroma 5 was actually somewhat more accurate (x0.004, y0.007) when measuring LED-based displays than it was when measuring the conventional LCD (x0.005, y0.009). With the Samsungs, the Chroma 5 was slightly less accurate overall with the LED display (x0.004, y0.008) than it was with the conventional LCD display (x0.005, y0.005). On average, the errors in luminance were negligible.
When comparing the differences in accuracy between the Chroma 5 on conventional displays and its accuracy on LED displays we can see that the differences are very small and on average they are below the threshold of visibility.
The evidence is pretty clear. A standard Chroma 5 colorimeter does not perform significantly worse on LED-based LCD displays than it does on conventional LCD displays and may even perform slightly better.
To order ChromaPure software & meter packages at CurtPalme.com exclusive prices, see our ChromaPure order page. Same product, same support - just more money in your pocket at the end of the day!
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum You cannot attach files in this forum You cannot download files in this forum