Post a Comment
News and New Products
Small displays present unique problems at SID
By Ron Wilson, Executive Editor -- EDN, 6/14/2006
| More from SID: Motion video blurs picture for big panels Displays become more than dumb panels |
When small display panels topped out with low resolution, their inflexibility forced content to change. Web protocols arose specifically to provide low-resolution images to handsets, and user interfaces grew out of the assumption of a 3-cm display. But with market forces ramming NTSC video and HDTV down the tiny throats of handheld devices, and with Windows, with its numbingly inefficient screen use, moving into the handheld arena, displays and their supporting circuitry now must evolve. This has set off a technology scramble.
Two obvious approaches exist. One is to accept the incoming video stream in whatever its native resolution happens to be, and then scale it to match the physical resolution of the display at hand. But this approach, so sensible in console televisions, is fraught with danger for handheld devices (even if one assumes that sufficient bandwidth exists to get the full-resolution stream to the device in the first place).
Downscaling to produce an appealing image is not a simple matter of horizontal decimation, which could be done on the fly with trivial hardware. It requires two-dimensional, and sometimes two-plus-time-dimensional, interpolation, with nonlinear adjustments such as edge antialiasing, preservation of fine-line detail, and motion enhancement. These requirements translate to relatively large local memory, substantial processing power, and, just as harmful to the budget of a consumer device, the possible need to license algorithms and IP from third parties.
According to Steve Atwood, CTO for Capstone Visual Product Development, these factors have many palmtop system developers pushing for tiny displays with large pixel counts. Even if the resolution of the display exceeds the commonly accepted norms for the resolution of the human eye at a reasonable viewing distance, it may be simpler, less expensive, and easier on battery life to use a VGA-class display only a few centimeters across than to scale an image for a lower-resolution display. Sharp, for instance, is producing both VGA and XGA resolutions in small panels today.
However, this approach has its own problems. It's hard to keep shrinking the pixels in an LCD (or to keep making smaller OLEDs). As LCD cells get smaller, the supporting stuff around the cell doesn't scale very well, so the aperture through which the light passes becomes a smaller percentage of the cell area. The result: A dimmer display for a given backlight intensity. With the backlight already gobbling a third of the total energy budget for many handheld devices, this quickly becomes a problem.
One way around that is to use a transflective panel. This doesn't increase the actual brightness, but in bright ambient light, where most users feel the need for added brightness, the reflective crystals in the panel assist in keeping the image bright enough to be useful. NEC, for one,is betting heavily on this technique as a way around the costs—for both panel vendors and systems designers—of the quest for ever-stronger illumination behind handheld panels.
However, vendors are finding that moving to higher panel resolutions doesn't entirely eliminate image-processing algorithms. Image processing may still be necessary to enhance apparent contrast and to improve the rendering of fast-moving objects.
An intellectual-property vendor called Clairvoyante is touting a striking use of image preprocessing. Based on years of research into not panel design but perceptual psychology, the company has developed a novel LCD panel layout and a set of image-processing algorithms to support it.
The approach inserts white cells along with the red, green, and blue cells in the LCD matrix, creating an RGBW LCD. This allows Clairvoyante to cover essentially the same luminance/chrominance space as that covered by conventional RGB cell arrays, but with significantly greater perceived brightness, at least on most images. This translates, according to the company, into the system developer's choice of either twice the perceived brightness or half the backlight power at the same apparent brightness.
Furthermore, the company's rendering algorithm uses, on average, only two subpixels (that is, for instance, one red and one blue, or one blue and one white) per pixel in the data stream, compared with the mandatory one each for red, green, and blue in ordinary rendering. This can result in apparently higher resolution for the same subpixel pitch.
Like other scaling techniques, the Clairvoyante approach requires substantial preprocessing of the data stream before it reaches the panel drivers. This, however, consumes much less energy than a stronger backlight, according to the company.
If techniques to render high-resolution images on small screens depend on image processing, where can engineers put all the requisite signal-processing hardware and memory? The emerging answer is COG (chip-on-glass) technology. In fact, the back of the panel represents the only remaining real estate in many handheld devices. "For the smaller displays in handheld devices, we are already seeing the use of COG," affirmed Sriram Peruvemba, Sharp's senior manager of product marketing.
Thus small displays, even as they climb toward invisible pixel pitches, are becoming COG digital subsystems with their own complex software loads, varied power supply requirements, and DSP hardware. The panel is becoming a platform.
