Camera's NEED Light....!
Having now moved many of the 50+ short films my students have produced over the past few months through post-production and begun screenings and reviews a dominant issue has arisen. Exposure
in shooting. A great many of the recent films were under-exposed or
shot too dark and this has created difficulties in post-production,
limited colour grading options and a less than optimal final screening
image.
These problems as very easily avoided using some simple ideas,
techniques, tools and processes and so I set about producing a set of notes to help re-inforce good exposure practice for the next projects.
First we need to revisit a little background info.
Exposure
Exposure is simply the amount of light let through to the camera sensor to make the resulting image.
Every camera has a limited range of exposure, known as dynamic
range. This range covers from black to white and any image captured
will have a range of tones across this spectrum. The Exposure therefore
of a shot is the amount of light let through to the camera sensor and
captured as a range of brightness levels from Black to White.
As well as the lights placed onto a scene or subject, the camera
possesses a set of systems to control the exposure to limit or expand
the amount of light let through to the sensor; these include the
Shutter, the size of the Sensor and the Resolution in pixels. Most
significant however is the Iris or Aperture. How wide open the Iris is
to let light in is measured in f-stops. Every Stop you open the
iris up you are letting in twice as much light as the previous 'stop'.
Clipping
The camera's dynamic range is limited and so exposure that moves beyond
the range is said to have 'Clipped'. Under-expose with not enough light
and the shadow areas will clip to black. In simple terms the camera
will cease to be able to display shadow detail and so will just render
the dark parts of the image to a harsh full black.
At the other end, over-expose the image and the highlights in the
picture will fall outside the dynamic range and be clipped, or cut off.
The result is no detail in the highlights and instead being cut off to
an unnatural white.
Obviously the key to a good image is neither Under or Over exposing
but getting the most even spread of light across the dynamic range as
possible. But there is another factor to be considered..
Signal to Noise
Image noise is a fact of life. The better camera technology gets the
less noise is inherent but every digital image has an amount of noise,
artifacts and distortion in the picture. If the image is well exposed
then the noise will be virtually invisible but expose poorly and that
inherent noise becomes visibly evident.
And this where we see the distinct difference between light and dark
in the digital image. The light exposure in a digital image is not
evenly distributed across the dynamic range. Colour and Effects guru
Stu Mashwitz uses this image below to show the uneven distribution of
information between light and dark on the sensor.
Each swatch segment from black, through grays to white represents an
f-stop - a doubling of the light level; from left to right each swatch
is double the amount of light than the one on its left. The orange
columns represent how much pixel data is used to represent that part of
the dynamic range; every camera has a foxed amount of data to dedicate
to the image. Notice how disproportionate this is? how far more
"orange' column value is dedicated to the brights than to the darks..
To quote from Stu:
you're using half of your sensor's dynamic range to capture the
values between the rightmost bit and the one next to it! By the time
you start looking at the mid-gray levels (the center two bits), you're
already into the bottom 1/8th of your sensor's light-capturing power."
In other words; your camera can only capture a fixed amount of
visual information and it captures the vast majority of it in the
highlights and lighter areas and leaves precious little for the dark
areas. Digital camera sensors are highly biased to brighter areas than
darker.
What this means is that noise in the dark areas of an image is far
more apparent than noise in the highlights. Raising the gain in an
image will very quickly make the shadows become visibly noisy because
youre trying to turn up information that simply isn't there. Where as
the same gain will be unnoticeable in the highights.
The rule of thumb that comes from this is that even if you want to
have a dark and moody image you should still aim to shoot with good
even exposure and then lower the image brightness in post. An image can
be easily made darker; the shadows get pushed to crisp black and the
highlights do not show visible noise because they have way more image
data dedicated to them. But a dark image cannot be made brighter
without amplifying the noise in the dark area which are starved of
visual data from the beginning.
Shooting an image deliberately dark is a very risky proposition
because its forcing all your visual information to be collected in the
smallest allotment of the camera sensor.
Monitoring on Set
In an ideal world the image in the viewfinder, LCD, or field monitor
on-set would be a perfect replication of exactly what the camera sensor
is recording in terms of exposure. Alas this is almost never the case.
Camera LCD screens have their own brightness controls which can be
wildly different to the image being recorded, often they are
deliberately turned up very bright so they can be seen in daylight on
exterior shoots. You can attempt to match them by eye to the scene but
this is very much guess work. The camera LCD is really there for
compositional framing and does not do very well for a true judgment of
exposure. For that most cameras offer two, more accurate, tools that
are based on mathematics rather than eyeball judgment.
Histogram and Zebra Pattern
A Histogram is simply a graph that shows the levels of light from Black
to White. Cameras such as the Sony Z1 and Sony EX1 can display a
histogram graph in the corner of the LCD viewfinder which gives you a
live reading of the exposure of your image.
The shape of the histogram immediately tells you whether your image
is biased too dark or too bright without having to rely on the
luminance accuracy of the LCD display itself. The three images below
show three histograms similar to those on the camera. The Left end of
the graph is Black, the Right end is White. The first image is
under-exposed with too much in the dark and no highlights. The second
is potentially over-exposed (or has a lots of light areas and almost no
dark areas). The third is well balanced with an even distribution of
light and dark across the picture.
1. 
2. 
3. 
Where your LCD display may lie to you because of its own settings
and innate inaccuracies; the histogram, being based on mathematical
values, wont deceive you. It can help you greatly in making a more
informed reading of your exposure and ensure you get as much detail in
the picture as possible without losing visual information into the
noisy dark.
The other tool that helps greatly is the Zebra Pattern. This feature
brings up a diagonal striped pattern in the Viewfinder and LCD on any
areas of the image that are approaching over-exposure. Its a common
mistake to think that you do not want any Zebra showing in your image.
In most cases (largely for the reasons above related to noise) youll
want your image to show Zebra on parts that should be white. The two
images below show this. The first shows Zebra on the window and the
face highlights. the second shows Zebra on the White snow highlight
faces of the mountain.
You can also adjust the percentage at which the Zebra will display -
70% or 80% instead of the 100% which indicates full white and the edge
of the images dynamic range. When using such a setting having the
Zebra show up on the Highlights of the image tells you that youre
getting a good level of exposure and not under-exposing. If you had you
Zebra set to 70 or 80% and it wasnt showing on the image highlights,
you'd know for sure that your image is probably too dark despite how it
looks on the LCD.
WYSIWYG..?
One of the common traps many inexperienced DoP's fall into
is the desire and attempt to have the image you see in the viewfinder
be exactly that you want in the final image of your film. In computer
terms this is known as WYSIWYG (What You See Is What You Get); the
attempt to bake-in the image of the movie in-camera.
WYSIWYG is not only near impossible, its also highly undesirable in
the digital age. Impossible because unless you have extremely well
calibrated field monitors and perfect, dark viewing conditions on-set
with limited ambient light you will never get an accurate view of
exposure and colour. Undesirable because you will inevitably want to
colour grade the image, to manipulate it in post production for feel,
mood, tone and style; to interpret your film with colour just as you
interpret it with sound.
So rather than trying to shoot WYSIWYG you should be aiming to shoot for LATITUDE.
Latitude
This terms refer to having an image in post that is well exposed and
has lots of detail and so will allow you lots of options for
manipulation in colour grading. If you shoot for a very specific
exposure, especially dark exposure, you will severely limit what you
can do in post. if you get it exactly right in camera then theres no
problem. if you dont get it exactly right in camera (which will be 99%
of the time) then youre left with no options and no control.
A well exposed image with lots of detail gives lots of latitude for
creative choices in post. Room to move, room to control exactly how
bright, or how dark, or how contrasty you want the image. The DoPs job
should certainly be to craft an image to communicate the story but also
to ensure that they provide raw image material that allows the Director
and Editor flexibility and creative options in post.
Increasingly Colour Grading is seen as an extension of
Cinematography rather than editing and many DoPs are becoming more and
more involved with the grading of their images. This likewise leads to
the idea of aiming for image Latitude rather than WYSIWYG.
Camera
The absolute truism that defines a good film is good planning. For the DoP that means test-shoots.
Testing the particular traits of the camera, testing the location and
the space, testing the lighting, testing the design, colours and
costumes that will appear on screen. Far too often however a crucial
element is missing from the test; the colour grade.
Its seems plainly obvious but its amazing how often test shoots are
done and decisions made without testing the footage with the colour
grade process that's intended for post. In order to know how to Expose
your shots you absolutely need to know what youre going to want to do
to them in post-production.
These before and after images below from the excellent short RedWhitePanic, show this thinking and planning.
The original shot is deliberately exposed for maximum dynamic range.
The image is rather drab, low in contrast, light is even, all the
colours are balanced, its outdoors under diffused light of an overcast
sky, possibly with reflectors to ensure lots of detail on the face.
This Before image is a million miles from the After image but it has
absolutely been shot to provide the best possible basis for the
colour-grade to follow. It is an image with lots of latitude and, as a
result, the graded image is dynamic and vibrant and beautiful.
Each camera has its own particular look, its own particular feel and
there are a multitude of factors that contribute to this. Number 1 is
of course the lens; any camera is only as good as the lens and the
quality of the light delivered to the image is wholly dictated by the
glass up front. But beyond that understanding the other elements of the
camera can go a long way to helping shape the shooting process to get
the best possible exposure.
The Sensor
Aside from the glass of the lens the next most important camera element
is the Sensor; the element that the light is focused onto by the lens.
There are two key elements to the sensor that relate directly to
exposure...
The first is the number of Photsites on the sensor. The Photosites
are the light sensitive diodes that capture the image and relate into
pixels and resolution. The more photosites there are the more detail
the image can pickup and also often the more light sensitive the camera
is.
Some cameras work with a native number of photosites meaning that
the number directly relates to the resolution (For example the JVC201/251 has
1280x720 native resolution and the Sony Ex1 has 1920x1080 native
resolution). Other cameras use up-sampling and pixel-shifting to make
up a resolution greater than the number of photsites (the Sony Z1 has a
native sensor resolution of 960x1080 and up-samples this to 1440x1080, the Panasonic HVX202 has a native res of 960x540 upsampled to either 1920x1080 or 1280x720 depending on what mode you're shooting.)
The second, and in many ways more important factor, is the physical
size of the sensor itself. The bigger the sensor the larger the surface
area the light can focus onto; known as the imaging plane. (this
phenomena is also known as the Circle of Confusion and if youre up for
it you can read about it here http://en.wikipedia.org/wiki/Circle_of_confusion)
Larger sensors allow for greater control over depth of field (ie can
create better shallow-focus often cited as the film look), sharper
images, better motion and much better light sensitivity. (Wikipedia has
a great article on Depth of Field if youre up for a bit of hard
science here http://en.wikipedia.org/wiki/Depth_of_field)
The main digital cameras we use at the International Film School Sydney are the Panasonic DVC30, the Sony
Z1, the JVC201/251 and the Sony EX1. All four of these have a different
combinations of native resolution and Sensor size. The DVC30 DV camera
has 1/4inch sensors. The Sony Z1 and JVC201/251 has 1/3inch sensors and
the Sony Ex1 has 1/2inch sensors.
When in production or doing test-shoots these specifications can
make a big difference to the Exposure you are able to obtain and the
amount of light you will need. At one end the DVC30s have the smallest
sensors and the lowest resolution and so need the greatest amount of
light to get even exposure and have the least depth of field control.
At the other end of the spectrum the Sony EX1 has significantly larger
Sensors than the other cameras along with the highest native resolution
(1920x1080 which is more than double the number of pixels of JVC210/251
which shoots 1280x720)
The greater the number of pixels and the larger the Sensor the more
light sensitive the camera. The better it is able to re-produce a wide
dynamic range and give good exposure under a variety of conditions. The
greater the control the camera will have over Depth of Field.
For these reasons currently the SonyEX1, despite being a small-form
hand-held camera with a fixed lens, it stands out as technically the
best camera in the schools armory. The Ex1 has the largest sensor, the
most pixels, the lowest image compression, the highest data-rate and
the largest lens barrel.
The disadvantage the EX1 compared to the JVC201/251 was that it
couldn't use prime-lenses being a fixed lens camera. But just this past week our Technical Supervisor Christophe Healy has been able to rig the Ex1 to a RedRock Micro M2 system and use
Nikon and P+S prime lenses with the Ex1. The Ex1 now has the same P+S and Nikon lens options as the
JVC201/251 but the light from those lenses is being conveyed to a much
bigger sensor with twice as many pixels.
Conclusion
Light, sensor size, dynamic range, histogram, zebra pattern,
resolution, clipping - all these factors are at play when youre trying
to get get good exposure. The way to master these ideas is simply to
PRACTICE PRACTICE PRACTICE.
- Get familiar with all the different cameras and their different lens systems.
- Do test shoots that include a test colour grade.
- Plan ahead for light sources
- And remember that because the digital sensor uses much more
information to record the image in the brighter parts than the darker
it's much better err on the side of More light rather than Less.
Posted at 12:00AM Jul 23, 2008
by Mike Jones in video |
