Better video at the Chip Level
By Cliff Roth
And then, a small revolution in video processing occurred. Instead of offering a tweak on picture tube performance, or an optional frill (like on-screen graphics that disappear after a few seconds), suddenly advanced video processing was required to generate the picture, constantly.
Fixed pixel displays—that is, practically everything that isn't a CRT, including LCD, DLP, PDP, LCoS and so on—are the reason. Fixed pixel displays could handle only a single resolution—the display's "native resolution"—without video processors capable of scaling.
Scaling is arguably the most important, and visually apparent, task that today's video processor chips perform.
There are specialized and general-purpose video processors, and they come in all kinds of prices and with all kinds of capabilities, from low-end chips intended for cell phone video displays to super-high-end chips intended for professional broadcast equipment, medical imaging and defense applications.
Anyone who has ever seen very funky-looking text on an old 640 x 480 laptop display screen has seen the effects of really bad scaling. Within the realm of chips intended for home theater equipment, scaling—converting a signal from one set of dimensions to
another—is one of the most demanding and critical jobs required of a modern digital video processor.
The quality of scaling can be critical to the overall viewing experience, and fixed pixel displays offer certain challenges that CRTs don't. With CRT picture tubes, the video scanning process actually offers great inherent flexibility to change speeds and the number of scan lines. With fixed pixel displays, however—plasma, LCD, DLP, LCoS—each pixel is a discrete point that must be individually addressed. So if a plasma screen has a resolution of 1024 x 768 and it is fed a signal from a DVD player with 704 x 480 resolution, the display must convert from one format to another—in this case, scaling smaller dimensions to fit larger dimensions.