Today’s analog TV broadcast standard was developed to fit the TV cameras and picture tubes available at that time. Now, modern high-definition, flat-panel TVs must be adapted to this old television standard in order to provide a good picture with these sources.
In principle, high-definition TV (HDTV) should be the solution to all image problems, since HDTV transmits an image with four to five times higher resolution. Ideally one broadcast pixel should correspond exactly to one pixel on the screen. However, the ATSC (Advanced Television Standards Committee), used in North America and Korea, provides 18 different combinations of pixel resolution, frame refresh rate, and field/frame transmission.
Every ATSC-compatible TV must receive, decode, and adapt to the respective display characteristics. Thus, HDTVs need scalers and deinterlacers as well as MPEG-2 decoders, and these HDTVs must still remain compatible with the old standard-definition TV for the foreseeable future.
Moving images present the biggest challenge with regard to correct deinterlacing. Broadcast TV uses the interlaced method; the CRT displays the image precisely as broadcast by the transmitter (because the TV standard was developed for CRTs). Thanks to the short persistence time of the phosphorous, it delivers a short, bright, stroboscope-like snapshot of motion. The brain reconstructs the original sequence of movement from this sharp flash and is able to perceive a fluid movement.
LCD and plasma displays don’t work with interlaced signals. They require conversion of the two-field interlaced signal to a progressive frame. As the screen size increases and the speed of image movement accelerates, the disturbing side effects associated with deinterlacing become more and more annoying.
The simplest method for field-to-frame conversion—line doubling, or line repetition—only delivers acceptable results on small screens, up to about 17in. This is due to the halved vertical resolution and the line flicker. Line-based interpolation brings slightly better horizontal resolution, good enough for sets up to about 21in. For larger displays, more complex methods are required so as not to impose any image degradation or compromise on consumers.
With an image memory, each field pair is assembled to form a frame. Frame-based deinterlacing delivers full vertical resolution of the PAL or NTSC source. However, this method creates jagged images (“jaggies”).
A state-of-the-art method combines line interpolation for fast- moving areas with frame-based deinterlacing for other regions. Called motion-adaptive deinterlacing, it’s become the standard in 28in. or larger TVs.
VIDEO OR FILM, NTSC OR PAL?
Motion-adaptive deinterlacing would be completely satisfactory, yet there’s another problem in the mix. In Europe, TV and DVD have a 50Hz frame rate, but LCD and plasma displays prefer 60Hz. The frame-rate conversion required is 1.2 (60/50), and is performed in the TV by repeating every fifth field. Sharp-eyed viewers may notice irregularities with fast movement or panning motion.
Film-source conversion is more visible. Originally captured at 24 frames/s, film in Europe is broadcast or recorded on DVD at 50Hz. When deinterlacing, it’s crucial to compile the respective associated fields to one frame, and not two fields that originate from different frames. Because this field-reassembly information isn’t included in the analog or digital broadcast signal, or the SCART signal, a film- mode detector is required in the TV. Here, field contents are compared to identify the pull-down mode used (2:2 for PAL/ SECAM, 3:2 for NTSC), so they deinterlace correctly.
To achieve the 60Hz frame refresh rate required for flat displays, the reconstructed frames are repeated alternately two or three times (reverse 3:2 pulldown). Consequently , a “juddering” occurs, which is clearly perceptible on 32in. panels. This phenomenon is also referred to as “film judder” (Fig. 1).
Chips available today, however, offer an intelligent alternative to simple 3:2 pull-down: conversion of 24 or 25 frames to 50 or 60 or more frames by generating new inter mediate frames. These interpolated frames give a constant sequence of movement. This method, known as motion-compensation, gives striking results with motion in front of a static background or with camera pan and zoom.
Since the intermediate frames aren’t transmitted from the signal source, they must be calculated in real-time in the TV video processor. This requires a well-tuned algorithm that compares two original frames, computes a motion estimation of all objects found, and then calculates one or more intermediate frames with motion-corrected object positions.
Addressing the film judder problem, Micronas developed the “Real Motion HD Film Dejudder” technology (Fig. 2). It increases the frame refresh rate (from 24 fps to 50/60 fps or even to 100/120 fps) by generating additional frames.
PERSISTENCE AND INTELLIGENT DISPLAYS
LCDs and other sample-and-hold type displays (LCOS, DLP) generate a bright, flicker-free picture by holding the color and brightness value of every pixel constant for the frame’s entire duration. This isn’t optimal for fast-moving images. The fast stroboscopic effect of a CRT fools the eye, the slower LCD does not. In the case of a sample-and-hold-type display, each pixel appears on the screen longer than needed; the brain perceives motion blur, which can’t be reproduced through display measurements (Fig. 3). This effect often occurs with current LCD models that have a response time of around 8ms, since this is one half of a single frame period.
In principle, a correction can be performed in two ways. To make an LCD-image subjectively sharper, you can shor ten the on-time of the backlight via synchronised deactivation of the background lighting for half of the picture duration (backlight scanning).
Or, instead of inserting a dark- scanned picture, you can insert a (new) motion-compensated intermediate frame, which contains precisely those supporting points that the brain expects at that moment (Fig. 4). Called motion-compensated deinterlacing, this preserves the image’s brightness.
The development of an image- enhancing IC that manages this process at 120Hz requires high technological competence. The variety of motion situations that must be handled correctly makes time-intensive optimization necessary, and that requires high data rates.
GLOBAL, COST-OPTIMISED, MODULAR TV CHASSIS ARCHITECTURES
High-end television architectures today are still based on multichip solutions, whereby separate modules are still used for deinterlacing and scaling, MPEG decoding, and audio processing. Innovative manufacturers already rely greatly on new, powerful components with full two-channel capability (for PIP, PAP, PAT applications), as well as a variety of picture-enhancement measures.
Because consumers want state-of-the-art picture quality, Micronas grouped its image technologies under the single brand-name truDHD. Included in this group are:
- Full-HD video processing (1920 x 1080)
- Real Motion HD (HD film judder cancellation that provides steady movements)
- Motion blur removal (removal of motion blur using 100/120Hz conversion)
- Smooth diagonal lines (elimination of the typical “staircase” contours flat-panel screen)
- Sharpness enhancement (recovers well-defined contours)
- Contrast enhancement (optimised contrast for LCD and Plasma screens)
Micronas offers the truDHD brand as a license by the company to television manufacturers.
The picture-enhancement options available with the latest generation of video processors lead to a broad spectrum of optical styles. These styles can vary greatly, depending on the television manufacturer as well as the market.
For Europe, besides being HD- ready, a flat-panel TV must also excel when displaying a conventional analog or digital SDTV picture. For the television manufacturer, the correct choice of picture-enhancement ICs is rather decisive.