No illusions: toward brighter flat displays

EET: Are all these algorithms turnkey, or do you still do a lot of customization?

Elliott: Everything needed to drive a PenTile RGBW panel is in the chip. We have a lot of flexibility in our standard processor core in the driver chips. We can adjust to correct for different display gamma, even correcting for different gamma on each color, to balance the color. We can adjust the way that we convert from RGB to RGBW color, putting more emphasis on brightness vs. color saturation for certain colors. Even the amount of sharpness can be adjusted by register settings. We work with the customer to determine the best choice of layout and algorithm to use for the given application.

EET: You also say that gray-scale levels are dependent on subpixel color. How does that work?

Elliott: The psycho-physics of the eye shows us that slight errors in the blue channel are more easily tolerated than errors in the red and green color balance, which means that for a given set of gray levels — say, 16-bits — you need the same resolution for red and green in order to maximize luminance contrast and color accuracy, and need less accuracy in the blue. So, for 16-bit systems we recommend using 6, 6 and 4 bits for red, green and blue [respectively], rather than the industry standard of 5, 6 and 5 bits [respectively]. But industry standards are just that, and we work with whatever the customer has already in their system.

EET: In your lectures you warn of the dreaded "Island of Death," where flicker and dot crawl can result if display designers fail to avoid certain parameter combinations.

Elliott: Yes. For instance, we advise display designers never to create time-varying brightness levels near three cycles per degree at three cycles per second, no matter how shallow, because people are especially sensitive to both flicker and dot crawl at that rate. This is important in designing drive schemes, such as dot inversion vs. line inversion, or different spatiotemporal dithering systems. We make specific recommendations for these issues so that our new layouts will work without introducing unwanted artifacts.

EET: You also say that PenTile accounts for higher levels of visual processing done by the visual cortex of the brain, something called the Gabor wavelet response. What does that mean, exactly?

Elliott: The brain uses Gabor wavelets, which look like wave groups on the ocean, to very efficiently identify patterns and textures by mapping out spatial frequencies in several directions. That makes it easier for the brain to connect the dots in patterns, so people can identify patterns and textures more quickly. We make sure our layouts and algorithms are compatible with these wavelets. JPEG2000 image-compression algorithms use similar wavelets to more efficiently compress images with less detectable artifacts.

EET: How many licensees do you have?

Elliott: Well, almost all of the top 12 LCD suppliers are looking at our technology, but even if they were to become licensees, many of them won't want to verify that publicly. We do have announced licensees, such as Samsung, the world's third-largest producer of mobile phones, and Chunghwa [Picture Tubes Ltd. in Taiwan], which is producing low-power, high-resolution 2-inch and 2.2-inch displays using our PenTile driver chip from Silicon Works. Sitronix is also making driver chips.

The Chunghwa display is amorphous silicon, but an even higher-resolution polysilicon display has also been announced by AU Optronics, which is currently the third-largest maker of large flat-panel LCDs. AUO has announced a 2.8-inch display with VGA resolution that extends battery life for mobile applications by using half the power of the equivalent [RGB] stripe-version LCDs.

Candice Brown Elliott

Born:
June 6, 1957, in Denton, Texas

Education:
BS in physics and psychology, with a minor in biology, University of the State of New York, Albany, 1982

Current position:
Chief technology officer, board chairwoman, Clairvoyante Inc., which she founded in 2000

Career: