Week 2 - Expose Right!

Week 2 - Expose Right!
In the days of film photography (remember film?), the biggest technical hurdle was learning proper
exposure. If you messed up on exposure, your photo was destined for the trash, end of story! Film
photographers spent much of their time discussing things like gray cards, the sunny 16 rule, incident
light meters, the zone system, and spot metering. Hundreds of books, courses and magazine articles
were dedicated to the arcane skill of exposure. If you couldn’t get exposure right, you would never make
it as a photographer.
But digital photography changed all that. Now, very few photographers worry about exposure. Why
should we? Most digital photographers just let the camera determine exposure for them automatically.
If things turn out a little too dark or a little too light, then a simple tweak in imaging software will fix
things. How simple and easy! Now everyone can be a photographer because exposure is no longer a
technical hurdle to overcome. Or so it seems….
Proper exposure is just as critical with digital as it was with film especially if you are looking to get the
best files possible. Why spend $3000 for a camera only to get $3 results? The good news is that getting
proper exposure with digital cameras is much easier than it was with film and is a simple two-step
process. Let’s see how we can get the best information possible on our digital camera sensors so we
have noise-free and detailed files.
How Digital Sensors Record Light Levels
Most digital camera sensors can record about six stops of dynamic range from pure black to pure white.
A ‘stop’ is defined as the doubling or halving of the intensity of light. Many modern cameras have 12-bits
(or more) of information per colour channel with recordable values ranging from zero (black) to 4096
(pure white) in 12 bit capture. Because digital sensors record data in a linear fashion, the distribution of
the various tonal values is uneven over the dynamic range of the sensor. For example, the brightest stop
of highlight data contains half of all the tonal levels in the image, and the second brightest stop contains
half of the remaining levels of tonality and so on. Figure one shows the tonal information on a sensor
divided into six stops of dynamic range. Here we can see that 75% of the levels of tonality that an image
holds are within the brightest two stops of light intensity. The last four stops of light intensity contain
only 25% of the tonal levels!
Figure 1:
So what does all this number stuff mean in a practical sense? It means the more image information we
can capture in the ‘bright’ half of the sensor, the more levels of tonality we have to play with in postproduction. The more levels of tonality we have, the less likely that our images will get artifacts like
banding, posterization or noise. In short, we should expose our images with a bias to brighter side of the
sensor so we get more data to work with. How do we do this? Simple. We use the camera’s histogram
feature to help us judge exposure.
The Histogram
Digital photographers rely on the LCD display of a digital camera so they can review the photos just
taken. Many photographers also use the LCD to determine if their photo is properly exposed. If the
photo looks good on the LCD, then they assume it is exposed correctly. Nothing could be further from
the truth! The image displayed on the LCD is the camera’s interpretation of the finished image projected
onto a TV-like screen. Look at the LCD image under a dusk sky or under bright sun and the photo will
look very different. How can we judge proper exposure when the image displayed varies so much
depending on the light source we view it in? The real tool we need to use to help us determine correct
exposure is the histogram. All digital SLR’s allow us to review photos on the LCD with the histogram
option turned on.
Where to find Your Histogram
How do you get your histogram to display on your LCD? If you shoot with Canon cameras just hit the
playback button on your camera to view a photo on the LCD. Now press the ‘display’ or ‘info’ button
several times until the image comes up with the histogram. You will get a screen that looks like this
(photo 1):
Photo 1
If you press the display button one more time you will get a composite histogram that shows the pixel
information for each colour channel (red, green and blue channels) like this (photo 2):
Photo 2
If you shoot with Nikon you can access your histogram by first playing back an image and then using the
back thumbwheel to cycle through the histogram display options to give you LCD displays as shown
below (photo 3):
Photo 3
If you own a different camera brand or if you are having troubles accessing your histogram display, then
consult your camera manual. To get the most out of your camera you have to be able to see the
histogram to make proper exposure decisions.
What is the histogram?
Essentially the histogram is just a graph displaying the distribution of pixel tones from dark to light in our
photos. Figure 2 shows the components of the histogram. On the x axis are the levels of tonal
brightness and on the y-axis are the numbers of pixels falling into each level of brightness. The range of
brightness levels where pixels are distributed is called the tonal range. The shape of the histogram will
vary depending on the visual content of the image. We don’t really care about the shape of the
histogram; what we care about is the distribution of tones. In a perfect world we want more of the
tones to be towards the right or bright side of the histogram (highlights). If the pixel tones slam up
against the right wall of the histogram then there are tones that have not been captured by the sensor.
Lost tones are ‘‘clipped’ and this information provides no useable data for an image. We want to push
the pixel information to the right but without clipping useful data.
Figure 2
Real World Results
Ok, you think to yourself, “this all sounds interesting in theory but how does it translate to the real
world”?
Photo 4
In photo 4 above, I have taken three different photos of the same scene. The left image was
underexposed with a histogram biased to the shadows. The center photo looks ‘properly’ exposed with
the histogram biased to the mid tones and the image on the LCD looking the best. The photo on the
right looks overexposed with the histogram biased to the highlights.
I took all three images in Photo 4 and brightened the underexposed (left) image and darkened the
overexposed (right) one to give me equal brightness for all three images (see photo 5 below). I also
added contrast to each image so the photos had more ‘snap’. The results are predictable; all the images
in Photo 5 now look the same. So if I can make a dark image brighter and a light image darker why do I
need to worry about precise exposure?
If we enlarge each photo and look closely at the details, we see that there is significantly less noise and
artifacts in the image that had the brightest histogram (the image on the far right). The most
‘overexposed’ image in this series has the best quality! Of course, if we overexpose too much and the
highlights get clipped, then we lose information. Notice also how noisy the darkest image (far left) has
become once it is lightened to acceptable levels. Even the ‘mid tone’ image shows some image
degradation. If you get consistently noisy images from your digital camera, chances are you are under
exposing your images. The key to get best quality is to always expose as far to the right as possible
without clipping the highlights.
Photo 5
The 2-Part Exposure Workflow
Those who follow my articles know that I almost always use aperture-priority mode when doing outdoor
and nature photography. I pick the aperture I want for the depth-of-field I desire and let the camera
automatically pick the shutter speed. I take a photo at these settings and then check the histogram on
my camera (step 1). I look for a histogram that is biased as far to the right as I can get without clipping
important highlights. If I need to make a different exposure to get a better histogram, then I set my
exposure compensation dial to add more or less exposure as the situation dictates and then retake the
photo (step 2). Once I have the best exposure possible, I move on to the next composition.
For example, photo 6a shows the initial capture I made using the camera in aperture-priority exposure
mode. The histogram was mostly to the middle and to the left side of the histogram but the image on
the LCD looked rich and beautiful! I knew not to trust my LCD to judge exposure but rather that I needed
to judge the histogram. The histogram in photo 6a tells me this image needs more exposure to be
properly exposed.
Photo 6a
To get more exposure I went to my exposure compensation dial on my camera and added +1. Here is my
camera display for the original photo above (photo 6b):
Photo 6b
Notice I picked an aperture of f16 for good depth-of-field and the camera gave me an exposure time of 5
seconds. On the bottom of the camera display is the exposure compensation setting which was set for
“0”. To make the histogram move more to the right, I need to add more exposure, so I set my exposure
compensation to +1 as seen here (photo 6c):
Photo 6c
Now my camera increases the exposure time to 10 seconds but the aperture remains at f16. When I
retake the photo with the new setting (10 seconds at f16) and check the LCD my photo looks like this
(photo 6d):
Photo 6d
Although the photo looks washed out and colourless on the LCD (yech!), the histogram is perfect with as
much information to the right as possible without clipping. It’s hard to feel good about taking a photo of
an amazing sunrise only to have it look washed out on the LCD! But trust me, you’ll be more happy with
the end results with this more properly exposed image.
I can take both the images, the underexposed photo 6a and the properly exposed photo 6d and make
them look exactly the same with quick moves of the sliders in Lightroom or Camera RAW. Just like
before, we can brighten a dark image and darken a light one very easily in post production. But where
the rubber meets the road is in the details of each file. Compare the image detail on the left (the
underexposed image) with the image detail on the right (the properly exposed image – photo 6e). The
properly exposed image has no noise and is a much cleaner file (right side).
Photo 6e
So in the end, I always ignore how the photo looks on my LCD – don’t trust your LCD for determining
exposure! I care only about getting a good histogram. For me, a good histogram is weighted to the right
whenever possible but with no clipping of important highlight detail. With a good histogram you can
darken and adjust the photo and retain the best quality your sensor can give you.
The Blinkies – Helping You Find Clipped Highlights.
Many digital SLRs have a feature where you can turn on a ‘highlight alert’ to warn you of overexposed
highlights. Generally the highlight areas will flash with grayed out or coloured ‘blinkies’. Many
photographers find this feature to be helpful, and they will adjust exposure of the photo until the
blinkies go away. Be careful not to get trapped by those flashing pixels though. Often, by reducing
exposure to retain highlight detail the photo will be darkened so much that most of the histogram is too
far left and the quality of the whole image suffers. Sometimes the blinking highlights can simply be
ignored because they represent areas such as bright specular highlights that we would not expect to see
detail in anyway; things like sparkles on water, glint on metal, highlights in eyes, or the sheen of ice are
examples. If the highlights are just bright looking spots to your eyes, they should remain that way in our
photos. So yes, the highlight warnings can be useful; just don’t get too focused on eliminating them.
For example in Photo 7, my camera warned me of clipped highlight detail in the water drops on the leaf
(the red areas). If I underexposed my photo to get rid of these overexposed highlights I would drive my
histogram further to the left and risk having no shadow detail. For me bright sparkles on the water drops
look natural and so I left the exposure the way it was realizing that I can’t see detail in the sparkles so
the camera does not need to either.
Photo 7
Or check out this photo (photo 8). This image shows a spike of information jammed right up against the
right side of the histogram which means I have highlight information that has been lost in the photo.
With my highlight warning set on my camera (the flashing blinkies), I realized all the lost information
was in the sunburst, an area which the human eye does not see detail anyway.
Photo 8
So remember if you get some highlights clipped to the right, just ask yourself if these are important
highlights or ones that do not need information anyway. For example, in the photo 9a, I noticed I blew
out some details in the clouds – these cloud details are important to me so I needed to adjust my
exposure by dialing in negative exposure compensation until I got an image with a good histogram
which is seen in photo 9b. The final finished image (exposure and contrast adjusted) has lots of cloud
detail (photo 9c).
Photo 9a
Photo 9c
Photo 9b
Some Tips
Remember it is not the size or shapes of the histogram that you need worry about, only that you place
the histogram as far to the right without clipping important highlights. Check out the histogram in photo
10. The bulk of the image is dark with the only light area being the lichen. The histogram has a long
highlight tail. You want to place the tail as far right as possible without clipping the detail in the lichen.
The fact that the bulk of the shape is still to the left means nothing other than you have lots of dark
tones in the photo. Do not be confused in a case like this and push the main peak of the histogram way
over to the right. If you do you’ll lose all detail in the lichen.
Photo 10
Notice in photo 11a that the photo has a high peak and then a long highlight tail. The histogram is not
pushed as far to the right as it needs to be for the best files. In photo 11b, I added +1 exposure
compensation and got a better histogram with the highlight tail near the right edge of the histogram. A
few quick edits in Camera Raw (more contrast, colour correction and darkening of the photo) gave me a
nice finished image (photo 11c).
Photo 11a
Photo 11b
Photo 11c
Remember the shape of the histogram is not important! All that you are trying to do is push whatever
shape the histogram is as far right as you can without clipping the highlights. In photo 12, the histogram
is u-shaped, but all that matters is that the highlights are not clipped here. I wanted snow with detail.
The trees are silhouettes without any detail in the black so it is OK to have the blacks up against the left
side of the histogram. I do not want to pull any detail out of the rich blacks. With silhouettes it is OK to
have the blacks push up against the left side of the histogram.
Photo 12
When Shadow Detail Matters
If you want images with good shadow detail and very little noise in the shadows, then you have to get
the shadows away from the left side of the histogram as far as possible. In a perfect world you’ll push
the shadows right without clipping the highlights like we see in the bear photo below (photo 13a).
Photo 13a
Photo 13b
The original RAW image in photo 13a looks flat and washed out, but with just a few tweaks in Camera
Raw or Lightroom I can make a finished bear photo that is full of life and has lots of shadow detail with
no noise (photo 13b).
In photo 14, I could barely get all the tonal information to record on the sensor. I wanted detail in both
the shadows and the highlights in this photo and by carefully checking my histogram I saw that the
exposure I chose (-2/3 exposure compensation) gave me the details in both areas.
Photo 14
In summary, your most important lesson is not to trust the image on your LCD to judge exposure.
Instead use your histogram as a tool to help you get the best exposure possible from your camera. Often
your images will look washed out on your LCD, but that does not matter because you can darken the
photo easily in post-production without worry of increasing noise. Anytime you lighten a photo in postproduction you risk increasing noise and lessening the quality of your files especially in the shadow
areas.
The Final Problem
So far we have looked at images where the range of contrast in the image was recordable by the camera
sensor. That is, the camera was able to retain shadow and highlight detail simultaneously. In all the
cases shown above I had ‘room’ on the histogram to alter exposure to move the image more to the right
or left as needed to give optimal exposure. But often in nature we photograph scenes with our cameras
that are beyond the sensor’s ability to record the entire tonal range. In photo 15, I made the best
exposure possible for the scene but both the shadows and the highlights are clipped.
Photo 15
In cases like photo 15, we have a ‘broken’ histogram where detail is lost in both the shadows and the
highlights. In future lessons we will look at how to fix a broken histogram by either adding light to the
shadows with fill flash or with a reflector or by darkening the highlights using filters. To get great images
you need to know how to fix a broken histogram.
In the end fine quality files are achieved by getting our histograms as far to the right as possible. Do this
consistently your photos will sing with clean detail!
Assignment
Never believe anything you read or hear without testing things out for yourself. Go and take a photo of
the same scene with one exposure pushed to the left side of the histogram without clipping the
shadows. Now try the same scene but bias the exposure to the right side of the histogram without
clipping the highlights. Process the two photos and adjust them so they look the same (brighten the dark
photo, darken the light photo). Now compare detail by enlarging the two photos on your monitor or
making a large print of each picture. Can you see any differences? If you can (and I bet you can) then you
should be trying to bias your exposures to the right side when possible.