SeanTech

New TVs - Are six colors better than three?

A new kind of TV is quietly sneaking to market - one that handles color differently than any model since the color television was invented.

The TVs -- both front- and rear-projection types -- will use a snazzy-sounding technology from Texas Instruments called "BrilliantColor." What that marketing-speak means is: three primary colors and three secondary colors, for a total of six. The promise is that it will deliver richer colors -- especially in shades where digital technology is kinda weak, like those golden sunsets you get with Kodacrhome film.

Now at this point, you might be wondering, what's a primary color? OK, quick lesson in color science.

Remember when you were a kid and your art teacher (back when schools had funding for art teachers) showed you how to make green by mixing blue and yellow paint? That amazed me for weeks, seriously. The idea of mixing two or more colors to make a third is the central concept behind color printing and video.

In the print world, the colors you mix are usually cyan, magenta, and yellow. (See the three ovelapping circles at the beginning of this posting.) Printers also throw in black ink to get those really dark shades. This gives us the CMYK system (in which "K," for some reason, stands for black). In video, the three colors are red, green, and blue, giving you the RGB system.

By more than just coincidence, the human eye has three types of color receptor cells called cones. They don't exactly match up to CMYK or RGB, but the RGB match works better for video. So, by varying the amount of red, green, and blue light, a TV can stimulate the human eye to, in theory, perceive any color. (You get black by turning off all the light.)

That's the theory. In practice, no TV can produce the ideal shades of red, green, and blue to match the full color gamut of human vision (as represented in the image at left). Printers have a similar problem with CMYK. So to compensate, they started adding extra colors. For example, I've been using a Canon iP8500 printer with eight colors of ink to make test prints from cameras I'm reviewing.

"BriliantColor" does something similar with TV. It takes RGB, known as the primary colors (see image at left), and adds CMY, known as the secondaries.

Now, why is this appearing first in projectors? Because it's easiest. To save money, many projectors use a single imaging chip and illuminate it by shining light through a spinning wheel with color filters, thus flashing the primaries in rapid succession: RGBRGBRGBRGB, etc. It alls blurs into a color picture in your mind. (Unless you move your eyes too quickly, in which case it goes a little psychedelic for a split second.)

So, going to a six-color system is a relatively simple process of adding three more colors to the wheel and retiming everything. That's a lot easier than doubling the number of pixels in a plasma or LCD screen.

Will it work? Hard to say. It was almost tried about a year ago. Philips, together with a company called Genoa Color Technologies, was on the verge of launching a six-color projection set, but then backed out. The main barrier may have been that the set used a somewhat problematic projection technology called LCoS.

BrilliantColor is backed by Texas Instruments, which makes the Digital Light Processing (DLP) technology that goes into almost all the high-end projectors and rear-projection TVs. In fact, one product is already out, the Mitsubishi HC300U projector. Someone at TI told me that she expects more TVs to be announced soon.

BrigthSide’s Super LCD TV

Another interesting find at the recent Flat Information Displays conference in San Francisco was a company called BrightSide that is making a turbocharged LCD TV. Its technology, dubbed “High Dynamic Range” purportedly produces a 200,000 to 1 contrast ratio (the difference in brightness between a screen’s darkest and lightest state). Contrast ratio is a big deal, because it helps you see slight shading variations and therefore more detail and dimensionality in screen images. In comparison, a regular high-end LCD TV, such as Sharp’s LC-37D7U AQUOS has a claimed contrast ratio of 800 to 1. Now, all these numbers are probably fudged to some degree, but the difference between the BrightSide TV and typical LCDs is still remarkable. And I say that with some confidence because I’ve seen BrightSide in action.

So how do they do it?

Instead of using fluorescent bulbs to provide a constant light source behind the LCD panel, BrightSide uses an array of light emitting diodes (LEDs), which can be switched on or off faster than the frame rate of the video. So they can custom-adjust the lighting for every frame. Plus, they can control each individual LED to adjust the brightness for specific regions of the screen. (They call this technology "individually modulated LEDs" or IM-LED.) This gets around the main problem of an LCD panel. LCDs use fast-moving, light-blocking crystals to regulate how much of the backlight comes out the front of the screen. In theory, if you tell the crystals to go black, they will block all the light. But in practice, a considerable amount of light “leaks” thought the screen. So black on an LCD is really just dark gray. With such a wimpy black, LCDs just can’t produce a very high contrast (black to white) ratio. (Plasma and CRT televisions do far better blacks.) By turning down the backlight where needed, you can make the screen much darker.

And BrightSide uses a lot of LEDs. Its 37-inch DR37-P model has approximately 1,400 of them. That’s enough to get pretty specific about which areas are bight and which are dark. You can think of it as a low-resolution display with 1400 pixels, sitting behind a high-resolution display of 2,073,600 pixels (a 1920-by-1080 LCD). But this doesn’t lead to light and dark splotches on the screen, because BrightSide can also tweak the LCD settings to further refine the image resolution. Say for example, you have a feature a few pixels wide that you want to make incredibly bright, and around it are details that are supposed to be dark or just medium bright. BrightSide’s driver technology can compensate for the intense LED backlight behind that area of the screen by closing down the pixels around the bright feature more than you normally would on a regular monitor with a constant backlight level. So, even with some light leaking through, they look a lot darker than the highlighted item. As I said, this really works. I’ve seen it.

So what’s the catch?

I’ll be buying a house before I buy this TV. The DR37-P lists for $49,000 and so far has sold in only limited quantities to film producers and designers. Oh, and it uses an obscene amount of electricity – up to 1680 Watts, depending on the video content. (A regular 37-inch LCD consumes about 200 Watts.) In fact, the panel needs a liquid cooling system to keep from melting.

However, a representative of BrightSide tells me they are working with an unnamed TV maker on a deal to produce a slightly less-bright and cooler model at a “significant reduction in price.” The screen size will probably be somewhere between 37 and 50 inches.

Acronym Attack - IBM's 3D HDTV

Sorry I’m a bit late posting this, but here is the first of some interesting tidbits from last week’s Flat Information Displays conference in San Francisco:

IBM gave its first public demonstration of a new technology that allows projectors or rear-projection TVs to switch into a 3D mode. The system requires that users wear glasses with polarizing lenses, as viewers do for modern 3D movies, such as Spykids 3D: Game Over or Aliens of the Deep. The big deal is that IBM representatives say the technology can be implemented on projectors for a small amount of money - about $50 or less, and that the TVs could be set up to switch from 2D to 3D mode at the push of a button.

The IBM rep wouldn't tell me exactly how it works, but here’s my theory:

• The system uses filters to produce polarized light output that alternates in rapid succession. One frame of video, representing the right-eye view, is polarized in a way that it can pass only through the filter on the right side of the glasses. Then comes a frame of video with the left-eye perspective, polarized so that it passes through only the left-eye filter. Because the images flash in such rapid succession, and because all images persist in the mind for a split second, you “see” the alternating right- and left-side perspectives at the same time, creating the illusion of depth.

• This system can be easily implemented using off-the-shelf graphics hardware. 3D games and graphics cards have long-supported the ability to create stereoscopic (right- and left-side) views. One old-time application is in LCD shutter glasses, which alternately darken to block the left and right eyes. The alteration is synched to the refresh rate of a computer monitor, so that it displays the left-eye image when the right eye is blocked and the right-eye image when the left eye is blocked. This helps explain why the IBM solution is so cheap, and also why it’s easy to switch from 2D to 3D. Their main innovation is simply to use glasses with polarizing filters in place of LCD shutters. And graphics cards have the built-in ability to switch between the shutter-glass setup and a standard “2D” monitor setup.

• IBM may be using a spinning wheel made up of polarizing filter segments. As it spins through the light path of the projector, it alternately polarizes the light waves in one direction, then another. I originally thought that IBM had modified the color wheel of a DLP projector, which (in its simplest implementation) has segments of red, green, and blue color filters that allow a DLP projector with a single imaging chip to produce the three primary colors in such rapid succession that the mind sees them all at once. I envisioned a new, six-segment color wheel with a red left-eye polarizing filter, a red right-eye polarizing filter, a green left-eye polarizing filter, etc… My suspicion was heightened when the IBM rep told me that their system does not work with LCD projectors, which have separate chips for red, green, and blue and therefore don’t use color wheels. But the vague, one-page handout that IBM provided states that “The light engine based data projector has not been changed in any way.” So that would rule out installing a new color wheel and recalibrating its timing. My new theory is that IBM has some type of switching polarizing filter that goes on the front of the projector. So why won’t it work with LCD projectors? LCDs control light output through the liquid crystals by using polarizing filters. The light coming out of an LCD is already polarized, with the light waves all aligned in one plane. So you can’t re-polarize it in a second plane to create an image for the other eye.

Does anyone care about a 3D TV?

Probably not, but you never know. IBM (which is seeking to license the technology, not produce actual products) is initially targeting use with video games, which already contain 3D video information. Otherwise, there isn’t much content. Nearly all movies are produced for 2D, although big-time directors such as Steven Spielberg and James Cameron are taking an interest in 3D. The other option is to “upconvert” regular movies to 3D by analyzing the video, guessing at the depth information, and converting it into left- and right-eye images. A company called Dynamic Digital Depth in fact has created software to do this in realtime. I tried it out earlier this year on a Sharp Actius AL3D notebook equipped with a 3D LCD screen. It wasn’t perfect. The images sometimes blurred or went double, what a representative of the company calls “tearing.” But it worked a lot better than I would have expected.

3D has been around since the invention of stereoscopic photography in 1838. And many companies are working on modern 3D technologies using LCDs, plasmas, projection sets, and headgear. Some products require glasses, others don’t. So far, 3D has caught on only with researchers and designers who need to view complex items such as airplane designs or organic molecules in 3D. It could catch on for entertainment, but only if it works so smoothly that people don’t perceive it as some kind of dorky kludge. We’re not there yet.

The Future of the Internet - What the Times didn't tell you

The New York Times recently published a special section on Networking. As sometimes happens, budget problems caused them to cut material, including my report on the coming of super high-speed Internet service to the US. Here is that story.
-----------------------------------------------------------

Now Arriving - Faster Internet Connections

THE halcyon days of the late 1990’s spawned visions of telecommunications companies  spinning out fiber-optic cables like cotton candy — opening the way for instantaneous downloads, high-definition television and videophones in every home.

Instead, Americans found themselves simply praying that the cable TV provider would add Internet service to their neighborhood, or that they lived close enough to a telephone company’s central office (about three-and-a-half miles) to get a digital subscriber line. The lucky ones with cable or D.S.L. have been happy to get a primitive broadband link to the Internet — with typical data rates from about 0.5 megabits a second to 3 megabits a second — allowing Web pages packed with graphics (and advertising) to load easily, and making music downloads and Internet phone calls possible.

But now, after the dot-com bubble has gone, super-fast Internet access is finally arriving — whether through new faster connections into homes or new network technologies that make those connections perform even better. Verizon, for example, has already brought fiber-optic cables to the front yards of one million homes and businesses, and it plans to reach three million by the end of the year. Verizon is offering data rates of up to 30 megabits per second on the new lines — fast enough to download about 20 six-megapixel photos in a half minute. It will also provide telephone and television service over those fiber lines. SBC Communications, another telecom, is undertaking a more modest, but more widespread fiber-optic expansion that aims to bring connections of 20 to 25 megabits a second to 18 million customers by  early 2008.

Beyond the spread of fiber-optic cable, another factor in the speedup has been the realization by telecom companies that they can mimic how the Internet works to make their own internal networks simpler and more efficient. Using the technology known as Internet protocol, data is broken into small chunks that can travel via different routes — favoring the most efficient at the moment — to reach the same destination, where they are reassembled.

In phone service today, most calls are still circuit-switched: each call gets an exclusive connection for the length of the conversation. But telecom companies are quickly moving to packet-switching, in which multiple calls are converted into parcels of data that travel together over the same line. Verizon, for example, estimates that this move to voice over IP makes its system about twice as efficient.

Telecom companies can likewise use IP to combine phone calls and Internet traffic. And several companies, from the regional giant SBC to smaller companies like SureWest in California and Utopia in Utah, are challenging the cable and satellite TV providers by also using IP to provide TV service. (Verizon’s strategy is slightly different. Instead of converting television to IP, it will transmit standard digital and analog television signals through its fiber-optic cables.)

But to provide TV using IP technology is demanding: depending on compression technologies, a single high-definition program runs at about 3 to 15 megabits a second — meeting or exceeding the total capacity that  most cable or DSL Internet connections typically provide. And to compete with cable and satellite, service providers must be able to stream several shows at once to feed multiple TVs in each home. That demand takes home fiber-optic connections from being a luxury to a necessity.

Whether telecoms succeed as television providers, millions of consumers will get the windfall of a high-speed network that will vastly improve Internet service. They may also benefit from another behind-the-scenes technology called Multi-protocol Label Switching, a mouthful that essentially means “traffic control.”

One drawback of IP technology is the possibility of deteriorating quality. For example, mixing data from voice calls and e-mail messages may lead to delays that cause the audio to break up. But the new traffic control systems can recognize voice or video packets, for example, and fast track them so that conversations or movies stream smoothly. Less time-sensitive packets, like those for e-mail, can be made to wait a split second while the others pass.

Currently, individual telecom companies are putting out the new system to streamline data moving within their networks, but not necessarily between them. The next step, in the works, is for the telecoms to agree on a common method of using the technology, so that a packet given a certain priority label in AT&T’s network, for example, retains the same priority when it passes to, say, Verizon. A standard could emerge within two years. If and when companies choose to follow it, that would enable a new fast-track system for moving data across the Internet, thus allowing important innovations to take hold.

The biggest gain for most people could be a redefinition of television. Independent video producers are already gaining a following on the Internet through video blogs and download sites like iFilm.com. Several new sites and services are emerging, including Akimbo, Brightcove and Dave.tv.
The combination of fiber optics and new data sorting techniques may eventually provide enough speed and reliability to dispense with the click-and-wait download process and stream high-quality (even high-definition) video instantaneously — making Internet television similar in quality and experience to current cable and satellite offerings. As news blogs continue to challenge “mainstream media,” Internet video may challenge Hollywood, offering many more styles and choices of programs than cable or satellite TV can.

The Future of TV - What Wired Didn't Tell You

Wired magazine’s September issue covers “TV of Tomorrow.” I contributed a series of sidebars on new TV technologies, but they represent just a tiny sampling of all I learned while working on the piece. Here are some of the cool items that didn’t make it in.

Superchips that will Change TV

The next generation of TVs won’t simply be screens with tuners. New chips will make them into supercomputers.

Take the PWBSP-16 from Pixelworks and the SMPA8634 from Sigma Designs. In addition to the MPEG-2 video of today’s DVDs and HDTV, the new chips handle formats like H.264 and VC-1 – leading contenders for high-def DVDs, next-gen HDTV, and IP television. And they can plug into every video source - cable, antenna, satellite, and the Internet – making TV’s (or set-top boxes) with them into omnivorous media collectors. The PWBSP-16 can even download updates over the Internet – allowing it to pull in new video formats as they develop.

But Hollywood’s offerings will be more controlled than ever. The SMPA8634 uses military-grade encryption technology to protect against pirating. “It’s mind-boggling what they are putting in place,” says Sigma Designs’ VP of strategic marketing Ken Lowe.

Wherever the video comes from, it will look a lot better when it hits the screen, thanks again to military technology. The messy compression required to squeeze HD video down a wire results in some fuzziness and blockiness. Today’s video processors do a fair job of cleaning it up, but that pales in comparison to what the Realta from Silicon Optix can do. Based on a $60,000 component originally used for missile tracking, the Realta brings its one-Trillion-operations-per-second capacity into a chip costing under $100. That’s enough power to perform up to fifteen calculations on each of the over two million pixels of top-quality HD in real time. And the Realta can download new algorithms to address new video processing technologies.

Shopping the Video Bazaar

“It’s like eBay for video. It’s self-service and self-published,” says CEO Jeremy Allaire of his emerging company Brightcove. The same eBay comparison crosses the lips of Josh Goldman, his counterpart at rival Akimbo. They and other competitors, such as DAVETV, are all onto the same idea. Sources of video are virtually limitless, high-speed Internet access keeps growing, and bandwidth is dirt cheap. When he ran the mySimon price comparison site in 2001, Goldman was thrilled to pay $30 per gigabyte sent from his servers. Now he’s paying under thirty cents, and better compression technologies squeeze three times as much data into each gigabyte.

That lets Akimbo host, for free, anyone who wants to post video on its servers. Publishers range from big media companies like CNN and A&E to extreme sports channel High.TV to personal video blogs. Each publisher sets its own price terms – subscription, pay-per-view, ad-supported, or free – and the host takes a cut of any revenue. This gives them access to the Long Tail – a portion of the market comprised of many niche sectors that are small on their own but huge in aggregate. No one video niche may be big enough to get its own spot on channel-restricted traditional cable, but on-demand access via the Internet provides the equivalent of limitless channels (and limitless broadcast times) at virtually no extra cost.

Colorful Screens Big and Small

There’s no middle ground in TV screens. Twenty-seven inch versions are passé, but people go equally gaga for handheld models and cinema-size displays. One technology makes both of them possible.

Light-emitting diodes are moving far beyond the power indicator lamp on your TV. Their greatest attribute is astounding color. In the 1950s, the National Television Standards Committee drew a line around a big chunk of the colors that the human eye can see and set that as a goal for the American TV system. A half-century later, TVs still haven’t been able to cover much more than seventy percent of the NTSC color gamut.

But LEDs produce far-richer colors than the phosphors in CRTs and plasmas or the lamps and color filters in LCDs and projectors. New LED-based LCDs from Sony and Samsung are hitting over 100 percent of the NTSC colors, and upcoming front projectors and rear-projection TVs may hit 130 percent – displaying colors that have never appeared in video.

And LEDs work for displays of almost any size – from a 46-inch Sony LCD TV with 450 LEDs to a handheld projector from Mitsubishi with three.

TelcoTV versus Cable

Some people bitch about their cable providers, others about their phone companies. And many hate both. They should all enjoy the coming slugfest over TV service. The hoopla is about IPTV, which transmits video as a stream of data packets using Internet Protocol, but not using the Internet itself. Currently, cable companies send out from dozens to hundreds of channels – of which viewers filter out all but one at a time. Though the number of channels is high, they are limited to offerings with big enough appeal to earn a spot in the broadcast. With IPTV, viewers select just what they want – from a theoretically limitless supply of options - as if downloading a Web page.

A few dozen regional telecom companies already provide IPTV. SureWest, in central California, has been offering standard-def TV service since 2003. But the big companies are just now stretching their fiber optic networks close enough to homes to provide the capacity high-def IPTV. To guarantee smooth play, telcos will set aside a good chunk of the total bandwidth as a direct link between their video servers and people’s homes: Internet access and voice will be kept separate from TV data. SBC Communications is in the middle of a $4 billion expansion that aims to bring 25 megabit-per-second connections to 18 million customers by mid-2008, and it has a $400 million deal with Microsoft to offer IPTV to all those customers.

But telcos have a long way to go in catching up to cable companies – for example, in making deals with hundreds of content providers. And cable companies offer their own broadband services (though not with the data rates that telcos are promising). So they have the option of mixing both digital cable and IP-based video services for customers.

Will Microsoft be the Big Show?

Launched in 2002, Windows Media 9 has already taken over Internet audio – being the file format for most music download services. WM9 also handles video up to and beyond today’s high-definition resolutions, plus it’s only one-third as bulky as the old MPEG-2 standard used for today’s DVDs and HD broadcasts. And to please Hollywood, it has strict digital rights management to prevent copying. It also has a built-in system for embedding metadata -- making it easier to search for and filter video content.

WM9 is already dominating Internet video. It’s the format for movie download services such as Cinema Now, and it’s used by aspiring new video services, such as Akimbo, Brightcove, and DAVETV. Even the antiestablishment Participatory Culture Foundation supports it (along with most other media formats). And Crown Castle will use WM9 for its upcoming TV broadcast service for cell phones.

Windows Media 9 is gaining respectability because the Society of Motion Picture and Television Engineers is adopting an open version of it called VC-1. Standardization makes it more likely to be supported in non-PC applications, such as set-top boxes and DVD players, and Warner Brothers Studios plans to use VC-1 for high-def DVDs of its films.

BYOD (Be Your Own Distributor)

As SETI@home uses hundreds of thousands of humble PCs to do the work of a supercomputer, Bit Torrent pools many wimpy Internet connections to form a mondo download network. When you request a file, the network polls all the computers that have any portion of the file, and several of them send out chunks that add up to the finished product. Using multiple servers prevents any one file source from getting swamped. In fact, the more popular a file is, the more people are likely to have it, and thus the more servers are available to contribute to the upload. It’s almost a perfect technology for distributing indy videos – hefty downloads that may start out as unknown titles but have the potential to grow into blockbusters.

The one drawback to Bit Torrent is actually using it. Putting up files for sharing requires some rather techy operations, such as setting up a database to track its changing whereabouts on the network and creating a Torrent file that people use the initiate the download. So the Participatory Culture Foundation created Broadcast Machine - a free application that automates the process. All you need is a basic Web site (the “My Home Page” variety will suffice). Broadcast Machine installs the database and allows you to simply point and click on the files you want to share. PCF also has a free application for searching out and playing videos.

Find that Film

If you think it’s overwhelming to channel surf through a hundred or so offerings on cable TV, imagine trying to get through thousands or even millions of Internet channels. Search engines made Web pages navigable, and similar tools are needed for video. Some approaches to video search look a lot like old-fashioned Web page indexing. For example, both Google and Yahoo crawl the Web and identify videos using descriptions on the hosting pages or links to the videos from other sites.

Rather than looking for the text already associated with video, a company called Blinkx sucks text from the audio feed via speech recognition. And it uses context clues to improve the accuracy – for example, guessing at a mumbled word based on what people were discussing before and after it was uttered. Blinkx has also developed visual analysis tools, such as face recognition, that it may use in the future.

But artificial intelligence doesn’t always trump the real thing. Bradley Horowitz, who developed the original technologies for Blinkx while a grad student at MIT, says he experienced a transformation when he first saw Flickr. The photo-sharing site lets people specify keyword tags that make their photos searchable. It also allows visitors to add comments and allows members to join groups based on similar interests.

Flickr is now a part of Yahoo, where Horowitz is now the director of technology development. Yahoo hasn’t launched a video Flickr, but Horowitz leaves little doubt that something along those lines is coming. In addition, Yahoo recently developed a video tagging system called Media RSS – which allows creators of material to provide Yahoo’s video search service with descriptive text such as film credits or transcripts.