NEW STANDARDS FOR LIGHT SENSITIVITY
HELP YOU COMPARE TODAY'S CAMCORDERS
The first color TV cameras in 1951 were blinder than bats. For
the cameras to register an image, the poor performers were parboiled
under thousands of watts of light. In fact, the very first TV cameras
in 1927 required so much light that live performers could not be used;
a bake-resistant Felix the Cat figurine was TV's first star.
Cameras have changed from bats to cats and some of today's
camcorders can almost see in the dark. When buying a new camcorder,
low light sensitivity is a high priority. The less light a camera
needs to make a picture, the better your picture will look when you
shoot in a church, at a dance, or to the light of birthday candles
(without extra video lights). Under moderate light a sensitive camera
can make a gorgeous picture. Under minimal light a sensitive camera
may produce a hideous picture, but that's still better than 1000
words. Often we find ourselves shooting in places that we cannot
control, using whatever light is available, so low light cameras are
How Bright is Bright
Albert Einstein brightness was measured in IQ. The brightness
of light, however, is measured in footcandles, lux, lumens, or
candela. Physicists, including Dr. Einstein, have defined a candela
as "a unit of luminous intensity equal to 1/60 of the luminous
intensity of one square centimeter of a blackbody surface at the
solidification temperature of platinum." For the less precise among
us, it is the brightness of one candle.
Of course the brightness of one candle depends on how close you
are to it. If you held a white piece of paper one meter (39.37
inches) away from a standard candle, that paper would be illuminated
to a brightness of one lux. Physicists technically demand that the
sheet of paper be fashioned into a perfect globe with a radius of one
meter surrounding the candle, and that you measure the brightness over
the entire inside surface of that globe. Goodness knows how many
paper globes they set afire trying to get this measurement. Suffice
it to say that one lux is pretty dim.
TV cameras designed to yield a good picture in low light will
list in their specifications words like "sensitivity: 3 lux." That
means the camera will deliver a picture if the subject is illuminated
to a brightness of 3 lux (perhaps three candles were lit one meter
away). Another camera sporting a sensitivity of 1 lux should give you
the same picture when illuminated with only 1 candle.
So far, so good. We know that the lower the lux specification,
the more sensitive the camera and the more events we can shoot and
still get a picture.
You'd expect a 3 lux camera to yield a picture about twice as
bright as a 6 lux camera, right? Well, almost. The laws of physics
stop where the laws of the marketplace begin. The lux rating should
tell you the illumination required for an acceptable image --- but
pictures that are acceptable to advertising executives might not be
acceptable to you. For years there has been no industry-wide standard
for measuring camcorder lux ratings in the United States, and this has
left lots of room to stretch the truth. In Japan, lux ratings are
standardized by the government making it easier to compare one camera
to another just by comparing the numbers.
EIA to the Rescue
The Electronic Industries Association Consumer Electronics Group
(EIA/CEG) after two and a half years of study has developed a
standard which will help buyers compare camcorder low light
specifications. This standard, called EIA-639, establishes a unified
method of measurement for camcorder manufacturers to use in their
literature. Now camcorders of all types (VHS, SVHS, 8mm, HI8, and
digital camcorders) can list in their advertisements the low light
performance as "measured by the EIA Standard". Consumers can look for
these words when buying a new camcorder and be assured that the
numbers are meaningful. Samsung, JVC, Philips, Sony, Hitachi, Canon,
Panasonic, and Thomson Consumer Electronics (RCA and GE), have
indicated they will be using the new EIA Standard in their advertising
right away. Others will surely follow.
What Does it Take to Make a Camera Sensitive in Low Light?
First you need an efficient lens. A lens with a low f-number
(such as f1.4) focuses much of the light from your scene onto the
camera's sensitive CCD chip. A lens with twice as high an f-number
(f2.8) will pass 1/4 as much light through it, making the camera 1/4
The camera's CCD chip can be manufactured in a way that is super
sensitive to light. Seen through a microscope, a chip looks like a
city viewed from above; the rooftops are the light sensitive parts.
In order to catch the most light, the city planners need to keep the
streets (support circuitry) narrow and hide other circuitry
underground and make the rooftops as big as possible. Placing lenses
on the rooftops (microlenses that look like an insect's eye on the
surface of the chip) concentrate the light even further.
The signals from the CCD chip are then amplified. The more they
are amplified, the brighter the picture, but also the noisier the
picture. It is like turning up the volume to hear a distant radio
station; you hear more music, along with more static and more
interference. This electronic noise appears as graininess and color
splotchiness in your picture. Improved video circuits amplify the
picture signal while adding very little noise. The result is measured
as SIGNAL-TO-NOISE RATIO (S/N) and is measured in dB. The higher the
S/N ratio, the better. Some cameras in order to operate in very low
light have GAIN BOOST switches that increase the amplification of the
signal maybe 6, 12, or 18 dB. The pictures will become brighter, but
the graininess will become more pronounced.
Halving the camera shutter speed is another way of increasing
camera sensitivity. Normally TV cameras make 60 pictures per second.
Thus they have 1/60 of a second to collect the light and convert it
into a video signal. If the cameras made 15 pictures per second, the
CCD chip would have 4 times as long to "look" at the picture and absorb the
light. It's a little like taking a time exposure with a film camera.
Switching to 15 frames per second may quadruple the camera's
sensitivity, but it will smear the picture more when objects move
(just as it does with a film camera).
Remember that city of rooftops representing the light sensitive
parts of the CCD chip? If each roof were twice as big, it could catch
twice as much light making the chip twice as sensitive. One way to do
this is to join two adjacent roofs into one electronically as if it
were one big roof. Doing so will increase the sensitivity of the
camera, but reduces the resolution of the picture by 50%. Instead of
tiny roofs capable of sensing tiny details in the image, there are fat
roofs that miss those details. In the end you get a brighter, fuzzier
A wide angle lens (or a lens that is "zoomed out") captures
light from all over the scene. A close-up lens (or zoomed in lens)
catches the light from only a small part of the scene. Naturally, a
lens that is zoomed out gives a brighter picture than one that is
zoomed in. Normally you do not see this difference but in very low
light situations, the focal length of the lens (how far it is zoomed
in) affects the camera's sensitivity.
How Professionals Measure Camera Sensitivity
Professional videographers take camera specifications very
seriously, and so do manufacturers of professional video products.
For this reason, professional camera specs are very precise. They
list MINIMUM ILLUMINATION which is the smallest amount of light that
allows the camera to "get" a picture, and then lists SENSITIVITY which
measures the camera's low light ability while guaranteeing a "decent
When professionals compare cameras, they make sure that the
lenses used are at the same f-stop and focal length. Professional
specifications also include the S/N ratio for which the measurement
was made. They will also tell if the gain was boosted in order to
make the measurement. The number of TV lines of horizontal resolution
will also be mentioned. A typical professional TV camera might have a
spec that reads "2000 lux at f/8.0 with 63 dB S/N, 800 lines
resolution with gain at 0 dB boost." This tells you that you get a
bright picture at 2000 lux illumination with the camera set at f/8
(which yields excellent depth-of-field), and see a very smooth picture
(63 dB S/N ratio) with high image resolution (800 lines) and no
boosting circuits added.
Professionals define the brightness of a picture in terms of
numbers and voltages. On a professional camera, "minimum
illumination" indicates the minimum amount of light falling on a white
surface that will produce .5 volts (70 IRE) of picture information
with maximum lens opening and maximum gain boost. IRE stands for
Institute of Radio Engineers and represents a measure of video signal
strength observable on a special oscilloscope that graphs the
electronic video wave. In a perfect picture, black should be 7.5 IRE
and white 100 IRE (.714 volts). As an image becomes dimmer, the
voltage from the camera decreases and when it reaches 70 IRE (.5
volts) that's where the engineers peg the minimum brightness for the
How Consumer Cameras Used to be Rated
Unfettered by standards, manufacturers would do whatever was
necessary to make their cameras appear to be sensitive in low light.
They would open the lens all the way (low f-number) even though it
produced poor depth-of-field. They would boost the gain (video
amplification) even though it made the picture grainy. They would
lower the shutter rate, even though it smeared the picture, and would
pair the light sensitive CCD elements together, even though it reduced
camera resolution. Then they would boast that they had a picture even
though it was dim and grainy.
And who could blame them for these shenanigans? When there were
no standards and the other manufacturers were touting fantastic
see-in-the-dark specifications, each would have to fight hyperbole
with hyperbole. For this reason we have seen many camcorders sport a
1 lux rating here in the U.S. while the same camcorder lists a 7 lux
rating in Japan where standards of lux measurement have been in place
for some time.
How New Cameras Are Tested
The EIA Video Systems Committee set five measurement parameters
that would be evaluated under low light conditions.
1. Luminance level
2. Black level
3. Luminance signal-to-noise ratio
4. Chroma level
Tests were developed to measure these parameters and minimum
limits were set on the measurement results. To be judged acceptable,
the camcorder would have to meet the minimum limits on all five
*The Actual Tests
Each camera is aimed at a test chart with its zoom lens adjusted
between wide angle and half-way zoomed in. This achieves respectable
light transmission. The chart is illuminated with 3100 degree Kelvin
(3100° K) video lights adjusted to spread the light evenly over the
chart. A digital light meter is used to accurately determine the
illumination of the test chart. The camcorder circuitry is exposed,
if necessary, so that the video signal can be measured directly using
a video waveform monitor, video noise meter, and high resolution video
The camcorder is switched to its full AUTOMATIC mode for
exposure and color balance, and its GAIN UP control is set to the
"normal" position. (The manufacturer is permitted to perform these
tests in the GAIN UP position as long as it indicates in its
specifications that this position was used during the test).
The camera is manually focused and set to 1/60 second shutter
speed. The electronic image stabilization circuits are switched off,
the digital zoom is switched off, the on-screen display (OSD) feature
is switched off, the RF adapter (TV channel 3/4) bypassed, and of
course, any built-in accessory lights are deactivated.
*Test 1 Luminance Level
The camera is aimed at a logarithmic gray scale chart (a chart
with two horizontal rows of eleven progressively darker gray bars and
a white bar in the center) and illuminated to 1000 lux, as measured
with a digital light meter. Under normal light, the white chip will
appear white on a TV monitor and will trace a plateau at 100 IRE on
the waveform monitor scale, representing pure white. The progression
of darker bars appear as stairsteps on the waveform monitor, with the
bottom step at about 7 IRE. The light level on the chart is then
reduced. As this happens, the TV monitor picture gets darker and the
trace on the waveform monitor gets lower. Eventually, the trace that
represented the white bar only reaches 50 IRE, the cutoff point. Put
another way, when white things look 50% white (50 IRE), experts
consider the picture to have minimum acceptable brightness. The
technician, fumbling in the dark, locates his light meter and takes
another lux reading.
*Test 2 Black Level
The darkest bar on the chart is 2% white and appears as a black
bar on the TV monitor. On the waveform monitor the signal makes a
trace at approximately 7 IRE on the scale. Cameras measuring below 4
or above 10 on the scale for black level do not qualify for further
testing (these cameras, since they are not making correct black,
cannot have whites that are trustworthy).
*Test 3 Luminance Signal-to-Noise
Next the gray scale chart is removed and an 18% gray card is put
in its place. The camera now sees nothing but dark gray. The video
noise meter then measures the video signal and a calculation is made
which determines the signal-to-noise ratio. If the S/N ratio is 17 dB
or greater, the picture is considered acceptable. Incidentally, the
lower the S/N ratio, the grainier your picture, and a number below 17
represents a picture that is quite snowy. Remember earlier how
professional TV cameras sported S/N ratios around 63 dB? That's one
reason why professional video looks so good compared to what the rest
of us make at home.
*Test 4 Chroma Level
The gray chart is then removed and replaced with a Macbeth color
checker chart, a chart with little colored squares. The chart is
illuminated to 1000 lux and the chroma signal (the color part of the
video signal, the C in Y/C signal) is sent to the waveform monitor.
The waveform monitor is adjusted so that it measures just one square,
the pure red square on the chart. The technician notes the signal
level of the red square and then decreases the illumination down to
where it was set earlier in test number one. With less light, the
chroma signal becomes weaker on the waveform monitor. The minimal
acceptable chroma level for the red square with low light is 25% of
the chroma level when the red square was fully illuminated.
*Test 5 Resolution
A test pattern with converging thin lines replaces the previous
chart and the illumination is cranked back up to 1000 lux. Where the
lines are far apart, they are easy to see on a TV monitor. Where they
are close together, they merge into gray mush. At some point the
technician can tell where the lines are so close together they just
begin to comingle. Numbers on the chart tell what resolution that
part of the chart represents, so the technician just reads the numbers
off the chart. Next the illumination is decreased to the low light
level determined in step number one and the technician again checks
for where the lines blend together and reads the new resolution
numbers off the chart. These new numbers must be 70% of the value
he/she read at full brightness.
If the camcorder passes all of the above tests, then the light
meter readings in step 1 can be used as the minimum lux ratings for
the camera. If the camcorder does not meet all of the minimum
performance levels in tests 3, 4, and 5, then the illumination level
must be increased in step 1 and the whole process repeated.
Eventually it is possible to apply enough light so that the camcorder
passes the minimum requirements of each test, and that chart
illumination level is the one used to describe the camera.
This may seem like a lot of technobabble, but now you know how
honest light sensitivity measurements are made. Thanks to the EIA and
the participating manufacturers, buying a camcorder just got easier
for those consumers who know to look in the specs for low light
sensitivity "measured by the EIA Standard".
|About the author||About Today's Video 4th. ed.||Return home|