| Video Noise August 2001
Interlaced Video
Explained
Long ago and far away, television was invented. Everything
was broadcast "live" from television camera feeds or from film because there was
no videotape. Electronics were slow back then compared to what is possible today, so ways
had to be devised to produce an acceptable image that wouldnt be impossible to
reproduce using television sets priced so that average people could afford them. And
television cameras...it was hard enough to build one, let alone at a price that networks
or local television stations could afford.
The interlaced image became the NTSC broadcast standard for the U.S. and a number of
other countries that adopted the same specification. Interlacing the image essentially
allowed the capture, broadcast, and display of two half-resolution images in such a way
that they looked like a single full-resolution image on the television set. Your brain doesnt notice, too much, that the images are half
resolution as long as they are knit together accurately as 1/30-of-a-second,
full-resolution images. Breaking each image into a series of "lines" does this.
These lines are horizontal across the image. The cathode ray travels across the picture
tube, which is coated with a phosphor that glows when hit by the energy of the cathode
ray. The phosphor glows for a little while after the cathode ray is gone. This is an
important thing for television because it helps the low-resolution images look acceptable
to your brain. As the cathode ray moves across the picture tube it gets stronger and
weaker so that certain portions of the phosphor are brighter or dimmer.
It takes two half-resolution images that are displayed for 1/60 of a second each to
cause your brain to see one "full resolution" image that is displayed for 1/30
of a second (1/60+1/60 = 2/60 = 1/30). How you create the half-resolution images is
critical to making this system work. Using the 480 line DVD standard as an example, the
first 240 lines displayed are the odd numbered lines (1, 3, 5, 7, etc.). The second 240
lines displayed are the even numbered lines. You see the first low-resolution image for
1/60 of a second and the second low-resolution image for 1/60 of a second. But your brain
sees the two half-resolution images as a single high(er)-resolution image that is on
screen for 1/30 of a second.
In Figure 1 below, the numbers 1 through 10 represent half-resolution portions of the
image. The resulting images your brain thinks it sees are represented by Image A through
Image E. Keep in mind that this is how typical NTSC video cameras capture images and how
those images are recombined on your television set at home.

Motion and resolution artifacts
If the image you are showing is stationary, the interlaced images are excellent because
both 1/60-of-a-second images are identical. But if there is motion, you see the motion as
jagged edged and/or blurred advancements. You may know from 35mm still photography that it
takes shutter speeds of 1/125 of a second or faster to "freeze" motion -- the
faster the motion, the less time the shutter can be open. Fast moving objects may require
1/1000 of a second or 1/2000 of a second to freeze. By comparison, 1/60 of a second is
relatively slow. Even a walking person would have blurred feet, legs, hands, and arms at
best if they were photographed at 1/60 of a second.
Jagged edges from motion occur because the object is in a different location every 1/60
of a second. The even lines show the object in one position while the odd lines show the
image in a different position. When you knit the odd and even scan lines together, you see
ragged edges around moving objects, especially if you can stop the motion of the image and
look closely at the picture tube. Furthermore, thin horizontal lines in the original
image/set/stage that are the width of a single scan line (or smaller) will flicker on and
off as the image is panned vertically or if the object with the horizontal lines moves
vertically when the camera is not moving.
Motion artifacts and horizontal "line twitter" are the most notorious NTSC
artifacts. The closer you sit to your video display device and the larger the video
display device appears, the easier it will be to see NTSC artifacts in images. Some newer
television sets employ powerful image processing that can make NTSC artifacts very
difficult to find. HDTV (high-definition digital television) includes standards for
higher-resolution progressive scanning, which eliminates the video image artifacts we have
endured for over 50 years. Unfortunately, many HDTV products have chosen the higher
resolution 1080i format (1080 lines interlaced) to use to convert everything regardless of
how it was broadcast or recorded. This is unfortunate because interlace artifacts remain
quite visible even in the 1080i format.
On the threshold of better broadcast video
The most common television camera standard has been the topic of this article up to
this point. But there is another type of video camera that produces superior images
because it is not interlaced. Each 1/60 of a second, it captures a full
high-resolution image (not a coarse, half-resolution image like interlaced video cameras).
However, professional progressive video cameras are rarely used today because they are
expensive and because broadcast, cable, and satellite TV are all still 99.9% interlaced
mediums. To say nothing of the hundreds of millions of interlaced televisions that are
used around the world. HDTV and DTV (digital television) are able to make use of cameras
like these because of the supported progressive scan modes offered by each (480p for DTV
and 480p and 720p for HDTV sets). DTV and HDTV broadcasting are happening now in larger
cities. If you have an opportunity to see a demonstration, it is definitely worth seeing
what your television is going to look like in the future. Already there are a fair number
of network prime-time series being broadcast in HDTV and some of the premium
cable/satellite movie networks have one or two HD channels showing mostly movies at this
point.
What about film-to-interlaced video?
When you go from film to video, interlacing gets annoyingly difficult because modern
films are shot at 24 frames per second. This does not mix particularly well with the
1/60-of-a-second half-resolution images or the 1/30-of-a-second
"full-resolution" images of the NTSC system. When you start with 24 images
within one second, the easiest way to get to the 1/30-of-a-second NTSC image rate is to
add six images per second. The easiest way to do this would be to duplicate six of the 24
frames every second. Obviously this will make motion ever so slightly uneven because every
fourth frame will be played two times. Using this technique could make the motion too
jumpy, so a different way of achieving the same thing is used. You show movie frame one,
not for two 1/60-of-a-second half-resolution images, but for three 1/60-of-a-second
half-resolution video frames. Movie frame two has the expected two 1/60-of-a-second video
images. Movie frame three goes back to three 1/60-of-a-second video images, and the cycle
repeats itself with the 3-2 cadence. You can see this 3-2 pattern in the red "Film
Frame" row in Figure 2 below. This 3-2 pattern repeats throughout the movie, usually.
It is called "3 2 pulldown." There is a problem with using this unaltered
pattern for an entire movie though. Sometimes the contents of a frame are so different
from the neighboring frame that interlacing in the normal 3-2 cadence produces very
obvious distortions on the monitor.
In Figure 2, the video images, which could have content problems, are purple/magenta.
In these cases, the producers may alter the 3-2 cadence. Sometimes in editing, the 3-2
cadence is intentionally changed to avoid more severe artifacts. Other times, interruption
of the 3-2 cadence is an unintentional error that may or may not be visible in the movie
when viewed at home. Notice how the 3-2 cadence causes some of the "full res"
frames to be displayed with half information from one movie frame and half information
from another movie frame. If there is motion on screen when this happens, motion is
definitely not as crisply rendered as it could be with a different video standard.

Movies on DVD
DVDs have movies stored on them at 24 frames per second. Each movie frame is stored as
two half-resolution images already "split" for interlaced video standards.
Half-resolution images that will be repeated as part of the 3-2 playback cadence have a
special MPEG "repeat" flags attached to the digital data. The DVD player reads
the flag and repeats the frame at the correct time. Therefore, the DVD player reconstructs
the 3-2 pulldown cadence following the MPEG "repeat" flags for interlaced
display systems. The problems DVD players have in reconstructing good images can be
related to how well the DVD is assembled and mastered. There are well-known discs with
mastering errors, sometimes involving the "repeat" flags. Fortunately these are
not noticeable in most DVD players.
Now you understand interlaced video. Progressive scanning will be discussed in a future
Video Noise column.
...Doug Blackburn
db@hometheatersound.com |