If you walk into a Walmart, Best Buy, or any major TV retailer, all of the TVs are showing identical videos, but their colors will be noticeably (to obnoxiously) different on every TV on the wall—even in Best Buy's high-end Magnolia showrooms. Why is that?
One reason is that the TV picture controls have been played with—but this would still be true even with factory-fresh TVs right out of the box. It's also true with smartphones and tablets that as a rule don't provide any color controls, which is probably better because visual tweaking generally makes matters worse. And that is exactly the root cause of the problem that originates right at the factory—the displays are not individually fully adjusted and calibrated with instruments, and instead depend on visually tweaking at some point during manufacturing.
Some manufacturers and models provide better color accuracy than others. We have taken the six best mobile displays from our Display Technology Shoot-Out article series over the last year and compared their color accuracies all together side-by-side with detailed and very revealing measurement results. Since we only test the best performing displays to begin with, they were already known to have fairly good color accuracy, so we'll learn which are the Best of the Best, and the reasons why.
But why is color accuracy important? Poor to mediocre color accuracy has been the rule since the dawn of color TVs in the 1950s, and people are also accustomed to seeing mediocre color prints from their film and now digital cameras. But the technology is already available that makes it possible for today's consumer displays to be as color accurate as the best studio production monitors that cost $50,000 ten years ago. And once you get used to beautiful accurate colors on a display you won't want to go back.
One reason why color accuracy is now especially important is that most internet content is loaded with images and photos, and it's nice (and sometimes important) to know that you are actually seeing what the images and photos really look like. A more practical (and sometimes critical) reason is when you are buying online merchandise—you want to be sure that the colors you see on the display are accurate, so you'll have a good idea of exactly what you are buying and are less likely to return it. And for many, an essential reason (and the clincher) is that you want to see accurate colors for your own digital photos, and those from family and friends, which is especially important because you often know exactly what everything and everyone should actually look like.
Currently the cameras on smartphones and tablets (as well as consumer compact cameras and pro digital SLR cameras) are better calibrated than the displays you view their photos on because displays are more difficult to accurately calibrate. And finally, there are many specialized and professional applications that require or would benefit from much better color accuracy, such as in sales and marketing presentations, and especially medical imaging, where it can improve diagnostics.
In this article we measure and analyze the Absolute Color Accuracy of each display in four different ways and then rank them in each category:
Entire Color Gamut: First, for each display we'll measure and analyze the entire Color Gamut covering the complete range of colors that the display produces to see how the color accuracy varies throughout the entire Gamut.
Facial Skin Tone Colors: While the eye is sensitive to a very wide range of colors, some colors are more important than others. In particular, it is especially important to accurately render facial skin tone colors (for people of all races and ethnic groups worldwide). So second, we will measure and analyze the accuracy of Facial Skin Tone Colors independently.
Organic Colors: Most organic colors that occur in nature are heavily weighted in the red to green parts of the spectrum, which also encompasses browns, oranges, and yellows. This includes most foods, fruits, vegetables, and plants—so it is especially important to get those correct because we all carry accurate visual memories of what they actually should look like in the real world. For example, we generally evaluate the quality of most foods by their color. Many displays don't even do a very good job with ordinary green leaves and grass. So third, we'll measure and analyze the accuracy of Organic Colors independently. This same region of vibrant red to green colors is also frequently used to get your attention in advertising and signs, to clothing, to familiar products and everyday objects—another reason to get these colors right in a display.
Blue Region from Cyan to Magenta Colors: On the other hand, the accuracy of Blues covering the entire range from cyan to magenta are generally less critical for visual color accuracy. While the eye can still detect color differences and color errors in them, for the most part we are less likely to notice or be troubled by color differences and discrepancies with colors in the blue region. So fourth, we'll measure and analyze Blues from cyan to magenta independently as well.
While Color Accuracy has been steadily improving, it still has a long way to go because the color differences between these flagship displays are still easy to see. We'll explain the causes and the solution. We'll cover these issues and much more, with in-depth comprehensive display tests, measurements and analysis that you will find nowhere else.
We have taken the six best mobile displays from our Display Technology Shoot-Out article series over the last year and compared their color accuracies all together side-by-side. Since we only test the best performing displays to begin with, they were already known to have fairly good color accuracy. The displays in this Color Accuracy Shoot-Out include (alphabetically):
- Amazon Kindle Fire HDX 8.9 2013 model: we haven't yet tested the recently released 2014 model.
- Apple iPad Air 2
- Apple iPhone 6 Plus : the iPhone 6 has very similar Color Accuracy performance to the iPhone 6 Plus.
- Microsoft Surface Pro 3
- Samsung Galaxy Note 4: set for the Basic Screen Mode.
- Samsung Galaxy Tab S 10.5 set for the Basic Screen Mode.
Note that the Basic Screen Mode for the Samsung models, which we test here, is just one of four available Screen Modes that is selected under display settings (which many consumers and reviewers seem to be unaware of).
To examine the performance of these six displays we ran our in-depth series of Mobile Display Technology Shoot-Out Lab tests and measurements. We take display quality very seriously and provide in-depth objective analysis based on detailed laboratory tests and measurements and extensive viewing tests with both test patterns, test images and test photos. To see how far mobile displays have progressed in just four years see our 2010 Smartphone Display Shoot-Out, and for a real history lesson see our original 2006 Smartphone Display Shoot-Out.
In this results section we provide background information and highlights of the lab tests and measurements. You can also skip these Results and go directly to the Color Accuracy Conclusions.
If you've ever wondered why colors are off on a display—there are many contributing factors and causes including the Color Gamut, the calibrated White Point, the Intensity Scale, and sometimes "advanced" dynamic picture processing that instead introduces color errors. It should come as no surprise that in order to produce accurate color everything needs to be done just right. It's possible to accurately measure and map the absolute color accuracy and color errors for any display by using a spectroradiometer and DisplayMate proprietary test patterns, which we do throughout our Display Technology Shoot-Out article series.
Virtually all current consumer content is based on the standard sRGB/Rec.709 Color Gamut, which generates colors using a specified set of red, green, and blue primary colors. A given display can only reproduce the colors that lie inside of the color triangle formed by its primary colors. Extremely saturated colors seldom occur in nature so the colors that are outside of the Standard Gamut are seldom needed and are unlikely to be noticed or missed in the overwhelming majority of real images. Note that consumer content does not include colors outside of the Standard Gamut, so a display with a wider Color Gamut cannot show colors that aren't in the original and will only produce inaccurate exaggerated on-screen colors. The Standard sRGB/Rec.709 Color Gamut is shown in Figure 1, with explicitly calculated colors that accurately show the real colors within the Gamut—the colors shown in most published Color Gamuts are wildly incorrect.
The on-screen absolute color accuracy for any display can be measured using a spectroradiometer together with our proprietary set of DisplayMate Test Patterns. The accuracy of the colors can then be calculated using the 1976 CIE Uniform Chromaticity color space and compared to the eye's sensitivity to differences in color. Note that the older 1931 CIE Diagrams that are published by many manufacturers and reviewers are highly non-uniform and are meaningless for Color Accuracy. Many reviewers also incorrectly evaluate color accuracy by using a metric called dE, which is useful for display calibration, but is meaningless for Color Accuracy because it includes Brightness (Luminance) in addition to color (Chromaticity). See this regarding Bogus Color Accuracy Measurements.
We present the color accuracy and color errors here in terms of MPCD (Minimum Perceptible Color Difference) or JNCD (Just Noticeable Color Difference), where 1 MPCD = 1 JNCD = Δ(u'v') = 0.0040 on the CIE 1976 Uniform Chromaticity Scale in Figure 1. Color differences less than 1 JNCD are visually indistinguishable, while values greater than 1 JNCD are visually noticeable when the two colors are touching on-screen. When the colors are not touching and are further apart, the visual threshold for Just noticing a color difference is higher.
In order to deliver very good color accuracy, a display must have a Color Gamut that is very close to the Standard sRGB/Rec.709 Color Gamut. All of the flagship displays tested here come very close. Older and lower performance LCD displays typically have Color Gamuts in the range of 55-65 percent of the Standard, which results in very large color errors greater than 20 JNCD.
In order to evaluate the Color Accuracy throughout the entire Color Gamut we defined 21 Reference Colors, which are shown in Figure 1. We measure the accuracy of these Reference Colors for each display, which tells us how accurately the Full Color Gamut is reproduced. The numerical results are listed in Table 1 below and the individual data points for each display are shown in Figure 3a. The color spread in the measured Reference Colors between all of the displays is quite large around 10 JNCD.
The Samsung Galaxy Note 4 [set for the Basic Screen Mode] has the best Full Gamut Color Accuracy with just 1.5 JNCD average error. The other displays are listed by increasing error, with the iPad Air 2 the largest, with 3.9 JNCD. Next we examine the Color Accuracy for different color regions.
Table 1 courtesy DisplayMate
Accurately reproducing the subtle differences in skin tone and complexion in people's faces may be the single most color critical application for a display. In fact, some manufacturers actually tweak the display calibration for some countries to make sure that face colors come out just right for the local population. That's understandable, but a much better approach is to accurately calibrate the display so that the faces for people of all colors automatically come out just right without resulting to tweaking, which always introduces other color errors.
We measured the skin color for a wide selection of people of all races and ethnic groups in our photo library using a spectroradiometer and a very accurately calibrated display. The results are shown in Figure 2. What is especially interesting and significant is how they all fall along a well defined narrow line of color for people of all races and ethic groups, from the lightest caucasians to the darkest Africans (which is perhaps not surprising given that we all just have varying degrees of melanin and blood capillaries). Note that we are measuring the actual underlying skin color (chromaticity) not the skin brightness. From this spectroradiometer data we defined three Reference Colors that accurately describe the range of skin colors, which we then use to evaluate Skin Tone Color Accuracy. We measure the accuracy of these Reference Colors for each display, which tells us how accurately the entire range of skin tones are reproduced. The numerical results are listed in Table 1 above and the individual data points for each display are shown in Figure 3b. The color spread in the measured Reference Colors between all of the displays is again quite large around 10 JNCD.
What is particularly interesting and significant is that the Apple iPad Air 2 performs considerably better in Skin Tone Color Accuracy than Full Color Gamut Accuracy, while three of the other displays perform considerably worse in this critical region. The Galaxy Note 4 and iPhone 6 Plus also perform better with Skin Tone Color Accuracy. We'll examine this further in the Conclusions section.
Most organic colors that occur in nature are heavily weighted in the red to green parts of the spectrum, which also encompasses browns, oranges, and yellows. This includes most foods, fruits, vegetables, and plants (except flowers). There are only a relatively small percentage of exceptions, such as blueberries, egg plants, radishes, some plums, and purple cabbage, for example—just about everything else falls in the red, brown, orange, yellow, and green categories, which is a well defined and relatively small region of the human visual color space as shown in Figure 1. In addition, humans have a much lower sensitivity to blue light. Most fruits have vibrant and saturated colors in order to help attract the attention of animals that eat and then scatter their seeds. We even evaluate the quality of most foods by their color. Not surprisingly, these same food colors that are essential for our survival have also been incorporated and used to highlight and get our visual attention: such as in advertising and signs, to clothing, to familiar products and everyday objects—another reason to get these colors right in a display. Flowers, on the other hand, need to attract the attention of insects for pollination, whose vision is weighted towards the blue and ultraviolet portions of the spectrum, which accounts for the more varied range of colors in flowers that includes many saturated blues, purples, and violets.
In a similar fashion to skin tones, we measured the colors for a wide selection of colorful foods, fruits, vegetables, and plants (but not flowers) in our photo library using a spectroradiometer and a very accurately calibrated display. The results are shown in Figure 2. The most saturated reds include tomatoes, strawberries, apples, and red peppers. The most saturated greens include (chlorophyll) leafs of all types, green peppers, and limes. In between are oranges, carrots, lemons, and bananas. This range also includes cooked and uncooked meats. From this spectroradiometer data we defined a set of Reference Colors that accurately describe the range of these organic colors, which we then use to evaluate Organic Color Accuracy. We measure the accuracy of these Reference Colors for each display, which tells us how accurately the entire range of Organic Colors are reproduced. As discussed above, these same saturated food colors are also common in everyday (inorganic) objects. The numerical results are listed in Table 1 above and the individual data points for each display are shown in Figure 3c. The color spread in the measured Reference Colors between all of the displays is again quite large around 10 JNCD.
The Apple iPad Air 2 again performs considerably better in Organic Color Accuracy than the Full Color Gamut Accuracy, while three of the other displays perform slightly worse in this especially important color region. The iPhone 6 Plus performs slightly better and the Galaxy Note 4 maintains its excellent color accuracy. We'll examine this further in the Conclusions section.
The more important Skin Tone and Organic Colors all fall in the red to green parts of the spectrum and CIE Color Space. On the other hand, the Blue region covering the entire range from cyan to magenta shown in Figure 1 covers about half of the entire sRGB/Rec.709 Color Space, so it is still very important visually. However, as we have discussed above, while the eye can still detect color differences and color errors in the blue region as small as 1 JNCD, for the most part we are less likely to notice or be troubled by much larger color differences and discrepancies with colors in this range compared with skin tone and Organic Colors. The numerical results are listed in Table 1 above and the individual data points for each display are shown in Figure 3a. The color spread in the measured Reference Colors between all of the displays is again quite large around 10 JNCD.
The Apple iPhone 6 Plus (with 3.8 JNCD) and iPad Air 2 (with 5.4 JNCD) have the largest errors in the blue region. All of the other displays have significantly better accuracy, about 2.0 JNCD in the blue region.
All display colors except the three full saturation red, green, and blue primary colors explicitly depend on the defined color of white, which is called the white point, so it is especially important for the display to have a very accurate white point. The white point clearly affects all of the lower saturation colors because they are relatively close to white. However, even full saturation colors like cyan, yellow, and particularly magenta change considerably with even a minor shift in the white point because they are the complementary colors to the red, green and blue primaries, so they "reflect" through the actual white point set for the display. Since magenta is the furthest away from the white point it changes the most, but all colors (from low to high saturation) are affected by the exact location of the white point.
All of the relevant Standard Color Gamuts (sRGB, Rec.709, Adobe RGB, for example) use the Standard D65 white point, which is essentially the color of outdoor natural daylight at noon, with a Color Temperature of about 6,500 K. D65 is needed to produce accurate colors for digital photos, videos, TV, and internet content. However, many displays are set to a bluer white point with a higher Color Temperature from 7,000-8,500 K. Many consumers are simply used to that white, and many actually prefer a bluish white for the background on text screens. Unfortunately, that shifts all of the display colors and adds a bluish cast to all images, which may be quite noticeable with some Facial Skin Tones (people will look more pale) and with many Organic Colors. One reason why the Kindle, iPhone, and iPad have lower Full Gamut Color Accuracy is due in part to their bluer less accurate White Points, which are listed in Table 1 and plotted in Figures 3a-c.
With large color variations of up to 10 JNCD between the displays shown in Figures 3a-c, it was fairly easy to see some significant visual differences between all of the displays on many test photos and test images. None-the-less all six of these flagship displays for the most part have fairly good color accuracy for most casual viewing applications. But for careful or intensive viewing of most images and photos, and for special applications like sales presentations and medical imaging, for example, they may not be good enough.
The primary goal of this Display Technology Shoot-Out article series has always been to point out which manufactures and display technologies are leading and advancing the state-of-the-art of displays by performing comprehensive and objective Lab tests and measurements together with in-depth analysis. We point out who is leading, who is behind, who is improving, and sometimes (unfortunately) who is back pedaling, all based solely on the extensive objective measurements that we also publish, so that everyone can judge the data for themselves as well. See the main Display Color Accuracy Comparison Table for all of the measurements, and the Results Highlights and Introduction sections for background information and details.
We have taken the six best mobile displays from our Display Technology Shoot-Out article series over the last year and performed an in-depth analysis of their Color Accuracy. Since we only test the best performing displays to begin with, they were already known to be fairly good. There are none-the-less significant differences between them. With color variations of up to 10 JNCD between the displays shown in Figure 3, it was fairly easy to see some significant visual differences between all of the displays. So, while we learned which ones are the best of the best, there is still plenty of room for future improvement, which we discuss below.
We used the Color Accuracy Measurements from Table 1 above to generate an ordered 1 to 6 Ranking of the displays in each Color Accuracy category to help identify the Best of the Best, which is shown in the Table below. When the color accuracy values are fairly close we labeled them as a Tie with abcd according to their actual pecking order. Here are the results:
1) The Samsung Galaxy Note 4 is the winner in Absolute Color Accuracy, coming in first place in all categories for its Basic Screen Mode setting.
2) The Microsoft Surface Pro 3 and the Samsung Galaxy Tab S 10.5 are tied closely for overall second place.
One particularly interesting and significant result is that the Apple iPhone 6 Plus and iPad Air 2 perform considerably better in both the very important Skin Tone and Organic Color Accuracy categories, with the iPad Air 2 coming in a solid second behind the first place Galaxy Note 4.Its seems likely that Apple has concentrated on the important Red to Green part of the Color Space, which includes both the skin tone and Organic Colors. On the other hand, both the iPhone 6 Plus and iPad Air 2 are in last place for the Full Gamut Color Accuracy. This is partly the result of an over saturated Blue primary that distorts almost the entire Blue Region, which accounts for about half of the half of the entire Color Space and increases the Average Color Error, and also partly due to the less accurate bluish White Point. The iPhone 6 has very similar Color Accuracy performance to the iPhone 6 Plus as shown here.
The Amazon Kindle Fire HDX 8.9 winds up being squeezed between these two major Color Accuracy trends and comes in between fourth and sixth place. When we originally tested it in November 2013 it captured first place in Color Accuracy. It is still very good, but has slipped because overall Color Accuracy has been steadily improving over the last year. We hope to test the new 2014 model in the near future.
Table courtesy DisplayMate
Display Color Accuracy will continue to rapidly improve, particularly for mobile displays, now that all of the leading products are providing a fairly good match to the sRGB/Rec.709 Standard Color Gamut.
Up until now this has been accomplished almost exclusively by adjusting the materials and chemistry of the backlight LEDs, OLEDs, and LCD color filters, which is a very difficult and involved process. The Next Step is to use Color Management implemented in software and firmware to tweak the color mixtures to produce a perfect match for the red, green, and blue primary colors and the white point. When that happens (together with an accurate power-law Intensity Scale) the Display Color Accuracy will become visually indistinguishable from perfect.
In fact, this is how Samsung has risen to first place in Color Accuracy—because their OLED displays have a very wide native Color Gamut they had to implement Color Management in order to be able produce varying multiple Color Gamuts on a display. The very accurate Basic Screen Mode, which we tested here, is just one of four available Screen Modes on the Galaxy Note 4 that is selected under Display Settings (which many consumers and reviewers seem to be unaware of). In addition, manufacturers will also need to finish implementing fully automatic display calibration for each individual unit using instruments (rather than partial or spot calibrations with visual tweaking).
Another major stumbling block for high Color Accuracy is the bluish white points with 7,000-8,500 K that lots of manufacturers and consumers seem to like for their text backgrounds. That produces a bluish color cast for all images that significantly degrades Color Accuracy, which needs a 6,500 K white point. Color Management can also solve this by automatically switching between different white points for text and images, even when on the same screen at the same time.
The leading manufacturers are close to implementing much of this in the next generation of displays. This will provide not only better Color Accuracy for your precious digital photos and online purchases, but will also provide much needed improvements for many specialized and professional applications that will benefit from much better Color Accuracy, such as in sales presentations, advertising, and especially medical imaging, where it can improve diagnostics. Follow DisplayMate on Twitter to learn about these developments and our upcoming display technology coverage.
This article has been republished with permission from DisplayMate.com, where it can be read in its entirety.
About the Author
Dr. Raymond Soneira is President of DisplayMate Technologies Corporation of Amherst, New Hampshire, which produces video calibration, evaluation, and diagnostic products for consumers, technicians, and manufacturers. See www.displaymate.com. He is a research scientist with a career that spans physics, computer science, and television system design. Dr. Soneira obtained his Ph.D. in Theoretical Physics from Princeton University, spent 5 years as a Long-Term Member of the world famous Institute for Advanced Study in Princeton, another 5 years as a Principal Investigator in the Computer Systems Research Laboratory at AT&T Bell Laboratories, and has also designed, tested, and installed color television broadcast equipment for the CBS Television Network Engineering and Development Department. He has authored over 35 research articles in scientific journals in physics and computer science, including Scientific American. If you have any comments or questions about the article, you can contact him at firstname.lastname@example.org.
About DisplayMate Technologies
DisplayMate Technologies specializes in proprietary sophisticated scientific display calibration and mathematical display optimization to deliver unsurpassed objective performance, picture quality and accuracy for all types of displays including video and computer monitors, projectors, HDTVs, mobile displays such as smartphones and tablets, and all display technologies including LCD, OLED, 3D, LED, LCoS, Plasma, DLP and CRT. This article is a lite version of our intensive scientific analysis of all types of displays – before the benefits of our advanced mathematicalDisplayMate Display Optimization Technology, which can correct or improve many of the display deficiencies. We offer DisplayMate display calibration software for consumers and advanced DisplayMate display diagnostic and calibration software for technicians and test labs.
For manufacturers we offer Consulting Services that include advanced Lab testing and evaluations, confidential Shoot-Outs with competing products, calibration and optimization for displays, cameras and their User Interface, plus on-site and factory visits. See our world renownDisplay Technology Shoot-Out public article series for an introduction and preview. DisplayMate's advanced scientific optimizations can make lower cost panels look as good or better than more expensive higher performance displays. For more information on our technology see the Summary description of our Adaptive Variable Metric Display Optimizer AVDO. If you are a display or product manufacturer and want to turn your display into a spectacular one to surpass your competition then Contact DisplayMate Technologies to learn more.