Individual members - don't forget to link your account to an IEEE account to access the SMPTE Journal, and any library discounts. Creating an account is free. Log in at My Account at smpte.org to get instructions.
Hot Button Discussion
Expanding Display Color Gamut
By Michael Goldman
As digital image acquisition technology evolves, among the improvements it has the potential to offer the industry is the capture of wider color gamut on imaging sensors. How to represent that improvement in a meaningful way on the other end, through modern display devices, however, is a question without any standard answer. Color gamut typically refers to the range of colors that can be reproduced on a display device—typically, the range of saturated colors across different hue angles, but also including the luminance dynamic range. Historically, color gamut for displays is based on primary colors—by changing combinations of each primary color's output, mixed colors are produced. Since the dawn of the CRT display, this concept has traditionally been built around mixing red, green, and blue (RGB) primary colors on such devices. Consequently, display devices have never been able to reproduce the full and complete color range that the human eye is ideally capable of perceiving. However, with the arrival of LCD displays, laser displays, and other digital imaging breakthroughs, this paradigm has shifted. While it's doubtful any device will ever be able to reproduce nature's full color gamut range, this area has become rich with research, improvements, and strong opinions as it relates both to consumer display devices and professional/theatrical displays.
Many such opinions were debated in April at NAB 2011 during a panel discussion about wider color gamut. That panel was moderated by Scott Daly, Senior Technical Staff member for Dolby Laboratories, and featured representatives from major display device manufacturers. The discussion largely focused around design approaches and signal processing issues for pumping wider gamut into new devices, as well as the creative and philosophical questions of how important, how visible, and how meaningful wider color gamut is for viewers.
Daly recently discussed the color gamut display situation with SMPTE Newswatch, and points out as it relates to theatrical displays specifically, that "the pragmatic state of the art has been SMPTE ST431-1, sometimes referred to as the P3 gamut specified by DCI. However, there have been laser projection systems for quite some time, which have a much wider color gamut. But it was only recently that they have been cost-competitive with traditional systems."
"With digital cinema, there are actually now a number of approaches to achieve a wide color gamut, but lasers are the easiest to use," he adds. "They have some disadvantages, but the main reason why lasers can produce wide color gamuts is that primaries using them lie directly on the spectral locus (the boundary of the famed horseshoe CIE diagram), thus enabling fairly wide coverage of visible colors. The traditional primaries lie somewhat inside the spectral locus. However, due to the curved shape, you cannot cover all visible colors with just three primaries, no matter how close to the spectral locus."
As it relates to consumer display devices, Daly points out that "TVs have been available for a number of years that exceed the NTSC color gamut, and more importantly, the Rec 709 gamut, which is the current signal bottleneck for HDTV. The main approaches include multi-primary colors, such as having RGBCY (cyan and yellow added to red-green-blue) sub-pixels instead of the usual RGB, or by using narrow spectral backlight, such as RGB LEDs. There have also been some rear projection approaches using lasers, but these have not been as successful."
Daly says that among the disadvantages of lasers for home television applications is the lack of a wider color signal format. "What happens is that the Rec 709 signal input is stretched out to the wider gamuts," he explains. "If this is done simply, as was done in the early wide gamut TVs in the early 2000's, the saturation of all colors is boosted. While this can be pleasing for many colors, it can cause problems with skin colors, and other subtle colors near white. Therefore, advanced nonlinear stretching algorithms are used, but these cause distortions from the real scene or the creative intent, depending on the application. Even with a standard gamut TV, manufacturers often boost colors to give the impression of wider color gamut. However, this causes various problems such as clipping near the gamut boundary, which can result in the loss of spatial texture, and often, reflective objects look like they are glowing."
All that said, impressive work on various approaches to the multiple primary color concept is clearly widening color gamut range on today's state-of-the-art LCD viewing devices from many manufacturers. Representatives from some of them participated in the NAB panel discussion, including officials from Sharp Corporation in Japan. At press time, Sharp was also preparing a presentation for Siggraph 2011 later this summer about the company's exploration of device independent imaging systems for high-fidelity colors that includes extensions of its previous work in recent years on multi-primary color (MPC) display systems. The company is also currently demonstrating new prototype MPC-based monitors at the "Love Letters from Vermeer" art exhibit in Japan (running June 2011 to March 2012).
Sharp's current work in the area of wider gamut for LCD consumer display systems involves building on the work it showed the industry in 2010 on a system it called QuintPixel, which added yellow and cyan primary colors to the traditional RGB approach. Now, company officials say, they are attempting in current research to add to the QuintPixel display system six sub-pixels (including doubling the red area) in order to achieve 4k equivalent resolution (via high-resolution rendering) in 60-in. display screens.
According to Akiko Yoshida, a senior researcher at Sharp's Audio-Visual Technology Laboratories in Japan, who specializes in color science, Sharp wants to break free of typical RGB limitations.
"The basic design concept of a display device is still the same in many cases as what was defined 60 years ago to fit the technologies of the time," she says. "This means (some manufacturers) obsess over three primary colors. But RGB was defined as the phosphors on CRT displays, and there is no longer any special reason to be fixated on that. So, at Sharp, we first set our goal as 'let's create an LCD display which can reproduce all existing colors of real objects efficiently and faithfully.' So we introduced (with QuintPixel) additional primary colors to RGB and called it MPC (Multi Primary Colors). Our (QuintPixel) prototype is RGB, yellow, and cyan, and it can reproduce over 99 percent of real object colors."
These are exciting developments, but at the end of the day, the industry remains far from standardizing the expansion of color gamut for consumer home viewing devices. That is largely because there is no compelling reason to do so. Getting close to the gamut of most real objects, and how the human eye perceives it, for instance, is part of Sharp's goal relative to its current research, but that is merely the parameter Sharp has selected. Yoshida adds that other manufacturers may have entirely different parameters in mind.
"There have been discussions for standardization for MPC-type displays, but it is not going in the direction of pre-defining the primary positions of MPC displays like RGB," she says. "It's going to make a definition something like XYZ in the DCI spec, instead of fixed color space. For our MPC devices, the MPC primary colors solution is being fitted to match with our own goal, which is to achieve real object colors. But someone else could have other goals and create an MPC device based on their goals. So defining fixed primary locations for MPC displays does not make that much sense. Instead, people in the MPC standardization community are mainly discussing how to standardize the measurement method for MPC display devices."
Indeed, one of the main conclusions emphasized at April's NAB panel, according to Daly, relates to the measurement issue. Relative to modern display devices, Daly says panelists agreed that familiar 2D gamut area plots "can be misleading, and 3D volumetric analysis is needed for proper color gamut assessment."
Triangular 2D gamut area plots in the XY or UV domain, he says, worked fine to compare display color gamuts when all available displays had similar maximum brightness and contrast ratios during the CRT era. However, he insists, as dynamic ranges and brightnesses vary widely now, color gamut comparisons need to be made in a 3D plot to provide accurate gamut analysis.
"A common 3D plot is to use the LAB (color space) domain, which is a reasonable and well-used model of color perception," he says. "A display's color gamut usually looks like a football in this space, coming to points at white at the top and black at the bottom. The volume of the 'football' is important for conveying the ranges of perceived colors throughout the entire range. Guarav Sharma, an associate professor at the University of Rochester, has done some nice analysis of two displays with nearly identical gamuts when plotted in XY, and he shows what happens to full 3D gamuts under different conditions. The two displays had a similar contrast ratio, but one was five times brighter than the other. When both displays are viewed in a pure dark room, they have similar gamut volumes. However, in a living room situation, the dimmer display's gamut volume reduces from the bottom up due to flare light and viewer adaption. For side-by-side viewing, the dimmer display's gamut volume shrinks from the bottom up, from the top downward, and from the sides inward. This is because, in a side-by-side viewing, the visual adapts upward to the brighter display, as well as having a white reference point set by the brighter display. Under these conditions, it's possible for a display with even a smaller 2D color gamut than another display to end up having a larger perceived gamut volume if it is brighter, or if its range is larger than the other. These consequences fall directly out of the properties of the equations underlying the CIELAB color scale, as well as more recent advanced and more complex models of color perception. So, to truly compare perceived color gamuts across the widely differing displays of today, 2D gamut plots can be misleading, and we must (start using) 3D gamuts."
There is also the issue of how important the expansion of traditional color gamut space for viewing devices is in terms of the ability of the average viewer to discern it, in terms of how this change can impact the creative process for creating content for display on such devices, and in terms of other important applications that would benefit from viewing wider gamut.
Yoshida, for instance, says Sharp views the medical industry as a major potential market for future MPC displays. "It would be useful to surgeons because our display can reproduce the difference between arterial blood and venous blood, reproducing differences that cannot be seen on conventional three-primary color displays," she says. "Anyone else who needs accurate color reproduction on displays, such as an interior designer or a vehicle designer, would also benefit."
As far as the creative process goes, there is no question that new camera acquisition technologies are capable of capturing a wider color gamut on sensors than ever before. So failing to offer viewing devices capable of accurately representing those images, Daly suggests, would be a major shame. Further, he is convinced that improving color gamut on display devices of all types will, in fact, open up new horizons for content creators. For the first time, he suggests, they could consistently and accurately present viewers with "the rarer colors."
"Of the various dimensions being explored (in viewing devices) at NAB this year, such as high bit depth, high frame rates, and so on, color is the hardest to sell in a showroom situation," Daly says. "Most colors in the real world are not very saturated. This means you would have to wait to see when the statistically rarer wider gamut colors appear on the display. However, rarer colors have more significance.
"For example, cyan tropical waters largely lie outside standard gamuts, as well as the deep blues of special places like (Oregon's) Crater Lake, and deep red oceans and desert sunsets. For this reason, out-of-gamut colors are sometimes referred to as 'vacation colors.' People spend money and time to see them in the real world, and they cause memorable experiences. For many people working in wide gamut color, it is believed that if we tone down color boosting for average colors, and let wide gamut colors sing out when they make their rarer appearances, the overall viewing experience will be more rewarding. They are definitely perceivable in a side-by-side comparison."
Therefore, he adds, creative possibilities for content creators can grow exponentially as they did in the fine art world, where saturated, rare, out-of-gamut colors tend to be over-represented.
"From oil paintings of the masters to the painters of the French Academy, the impressionists, and modern abstract painters, we see expressive use of extremely saturated colors," he adds. "So turning to the creative process for broadcast or movies, the main point is that a wide palette allows more creative flexibility, which can be either carefully or chaotically employed to instill memorable images in the viewer."
Opinions expressed in SMPTE Newswatch do not necessarily reflect those of SMPTE. Reference to specific products do not represent an endorsement, recommendation, or promotion.
The last issue of SMPTE Newswatch stated that Red's Mysterium-X imaging sensor utilizes "pixel shifting techniques." This is incorrect—Mysterium-X generates a 5k image that can be sub-sampled to any desired output resolution, but does not involve pixel shifting, which is a technique used on three-chip systems.
back to top
Overcoming Shutter Limitations
Over the course of his extensive coverage on the Pro Video Coalition web site of technology innovations on display recently at the Cinegear Expo 2011 in Los Angeles, industry analyst Adam Wilt threw in this important nugget—his view that a new filter on display at the Expo could well be, in Wilt's words, "nothing less than the most important advance in camera shutter technology since motion picture cameras were invented." Wilt is referring to the Time Filter from a new company called Tessive—a dynamic, front-of-lens filter that synchronizes with frame acquisition. Wilt feels this way because, in his view, the Time Filter addresses traditional shutter limitations on digital cameras specifically, resulting in smoother motion that mitigates problems like aliasing and others things.
As virtual production technologies and techniques proliferate in "Avatar's" lengthy wake, numerous post-production companies are offering a wide range of mobile services and/or proprietary, on-set workflow options for productions. One of the most recent approaches in this arena took the "mobile" concept to a new level this year by essentially building from the ground-up a rolling post-production facility in a trailer, all networked through a massive super-computer. The trailer is called Mobilviz and comes from a company called Silverdraft LLC. It allows different functions—such as digital image capture, motion-capture data capture, and 3D image processing—to simultaneously process data through the same, central brain, thanks to what Silverdraft is claiming are new super-computer methodologies designed with a modest footprint to fit into the trailer and bring massive computing power directly to a set or location. Silverdraft officials insist the trailer's super-computer can process up to 350 teraflops of data through 20TB of solid-state storage via a networked core of over 1,500 computers. At press time, Silverdraft was testing Mobilviz and planned announcements for several projects to utilize the technology later this year.
The era of the personal robot is well underway, even if most folks don't realize it in our mobile device-driven society. A number of companies are developing, producing, and selling to consumers robotic technologies that actually can perform particular tasks, as well as entertain and amuse. The MIT Media Lab's Personal Robotics Group has been working on a variety of prototypes for various tasks over several years, and Honda's humanoid robotic technology, dubbed ASIMO, recently celebrated, believe it or not, its 10th anniversary. The TechZulu site recently ran an interesting interview with the CEO of another company pursuing this market, RoboDynamics. That company recently debuted its robot LUNA (see video of him in action here). CEO Fred Nikgohar had some interesting thoughts on where this technology is going and how to make it work better to become something more than a novelty item for consumers.