REV. 0

AD723

–16–

One well-known distortion of composite video images is called

dot crawl. It shows up as a moving dot pattern at the interface

between two areas of different color. It is caused by the inability

of the monitor circuitry to adequately separate the luminance

and chrominance signals.

One way to prevent dot crawl is to use a video signal that has

separate luminance and chrominance. Such a signal is referred

to as S-video or Y/C-video. Since the luminance and chromi-

nance are already separated, the monitor does not have to perform

this function. The S-video outputs of the AD723 can be used to

create higher quality pictures when an S-video input is available

on the monitor.

Flicker

In a VGA conversion application, where the software-controlled

registers are correctly set, two techniques are commonly used

by VGA controller manufacturers to generate the interlaced

signal. Each of these techniques introduces a unique charac-

teristic into the display created by the AD723.

The artifacts described below are not due to the encoder or its

encoding algorithm as all encoders will generate the same dis-

play when presented with these inputs. They are due to the

method used by the controller display chip to convert a nonin-

terlaced output to an interlaced signal.

The first interlacing technique outputs a true interlaced signal

with odd and even fields (one each to a frame, Figure 10a). This

provides the best picture quality when displaying photography,

CD video, and animation (games, etc.). However, it will intro-

duce a defect commonly referred to as flicker into the display.

Flicker is a fundamental defect of all interlaced displays and is

caused by the alternating field characteristic of the interlace

technique. Consider a one pixel high black line that extends

horizontally across a white screen. This line will exist in only

one field and will be refreshed at a rate of 30 Hz (25 Hz for

PAL). During the time that the other field is being displayed the

line will not be displayed. The human eye is capable of detect-

ing this, and the display will be perceived to have a pulsating or

flickering black line. This effect is highly content-sensitive and is

most pronounced in applications where text and thin horizontal

lines are present. In applications such as CD video, photography,

and animation, portions of objects naturally occur in both odd

and even fields and the effect of flicker is imperceptible.

The second commonly-used technique is to output an odd and

even field that are identical (Figure 10b). This ignores the data

that naturally occurs in one of the fields. In this case the same

one-pixel-high line mentioned above would either appear as a

two-pixel-high line, (one pixel high in both the odd and even

field) or not appear at all if it is in the data that is ignored by the

controller. Which of these cases occurs is dependent on the

placement of the line on the screen. This technique provides a

stable (i.e., nonflickering) display for all applications, but small

text can be difficult to read and lines in drawings (or spread-

sheets) can disappear. As above, graphics and animation are not

particularly affected although some resolution is lost.

There are methods to dramatically reduce the effect of flicker

and maintain high resolution. The most common is to ensure

that display data never exists solely in a single line. This can be

accomplished by averaging/weighting the contents of successive/

multiple noninterlaced lines prior to creating a true interlaced

output (Figure 10c). In a sense, this provides an output that will

lie between the two extremes described above. The weight or

percentage of one line that appears in another, and the number

of lines used, are variables that must be considered in developing a

system of this type. If this type of signal processing is performed,

it must be completed prior to the data being presented to the

AD723 for encoding.

1

2

2

1

33

44

5

6

5

6

7

7

=+

NONINTERLACED

ODD FIELD

EVEN FIELD

a. Conversion of Noninterlace to Interlace

2

1

3

4

5

6

7

=

+

NONINTERLACED

ODD FIELD

EVEN FIELD

1

2

3

4

5

6

7

b. Line-Doubled Conversion Technique

2

1

3

4

5

6

7

=+

NONINTERLACED

ODD FIELD

EVEN FIELD

1

2

3

4

5

6

7

c. Line Averaging Technique

Figure 10.

Vertical Scaling

In addition to converting the computer-generated image from

noninterlaced to interlaced format, it is also necessary to scale

the image down to fit into NTSC or PAL format. The most

common vertical lines/screen for VGA display are 480 and 600

lines. NTSC can only accommodate approximately 400 visible

lines/frame (200 per field), PAL can accommodate 576 lines/

frame (288 per field). If scaling is not performed, portions of the

original image will not appear in the television display.

This line reduction can be performed by merely eliminating

every Nth (6th line in converting 480 lines to NSTC or every

25th line in converting 600 lines to PAL). This risks generation

of jagged edges and jerky movement. It is best to combine the

scaling with the interpolation/averaging technique discussed

above to ensure that valuable data is not arbitrarily discarded in

the scaling process. Like the flicker reduction technique men-

tioned above, the line reduction must be accomplished prior to

the AD723 encoding operation.

There is a new generation of VGA controllers on the market

specifically designed to utilize these techniques to provide a

crisp and stable display for both text- and graphics-oriented

applications. In addition, these chips rescale the output from the

computer to fit correctly on the screen of a television. A list of

known devices is available through Analog Devices’ Applications

group, but the most complete and current information will be

available from the manufacturers of graphics controller ICs.