In a previous installment we saw how proper exposure of skin tones is as close as the palm of your hand. Today we will apply this approach, comparing what a D700 does in Matrix metering vs. our take-control metering approach. Take note this is a D700, not a D80, D90 or D7000. The last 3 cameras are reputed to over-expose when used in Matrix metering mode. You might be surprised to see that the D700 doesn’t really fair any better.

These two shots are taken in Aperture priority mode, letting the camera’s Matrix meter pick the exposure. Under-exposed subjects and blown backgrounds are the result. Yes, a D700 did this.
These next two shots used Manual mode, using spot-metering on the nearest cheek to bring up the exposure. The result is still under-exposed, especially around the eyes, where the deepest shadows reside.
The solution would be to spot-meter on those darker areas, with a goal to bring them into the midtones. Of course, that’s not what I did. I took one look at the histogram and knew immediately that opening up by about one stop would do the trick. Yes, the background is severely blown, but I chose what’s important in this exposure: the skin tones.
As it turns out, center-weighed metering gives the same exposures.
For this next set of shots, we use the post-processing method we discussed in prior articles: apply negative compensation (here with ViewNX), then add shadow protection to keep the skin tones where we want them. The result compresses overall dynamic range (and contrast), taming down those bright highlights.
For the final set of shots, we used the 2nd set and applied only shadow protection in ViewNX. In terms of exposure look and feel, the results are nearly identical to the previous edited images, except that pixel-peeping will reveal additional noise creeps in the shadow areas around the eyes.

If none of these results seem ideal to you, you’re probably right. Shadow pushing has really taken a toll on those two last pairs of photos. If you need to grab a quick snapshot where your subject is properly exposed, this approach will do in a pinch, but nothing can make up for bad light. This brings up a key point: this is not a metering or exposure problem, as much as it is a bad light problem. Moving your subject into a better lit area, or augmenting the foreground lighting with flash or reflectors will not only produce better exposures, but better photographs overall.

When we looked at how to meter using the usual suspects, we saw that out in open sunlight, spot metering on clear blue sky gives us well exposed mid-tones. For me, this works out really well when I shoot outdoors, especially when capturing landscapes in those non-golden hour times of the day.

Now that I’m doing more portrait work, however, with a key goal of keeping skin tones well exposed, things get a little tricky. As it turns out, for skin tones that lay somewhere along the midtones (middle gray in black and white), the solution is right in the palm of your hand. The next few sample shots demonstrate this approach.

The first sample used matrix metering in Aperture priority mode, to let the camera choose its own exposure.

At first glance, it looks like Matrix metering did a fairly good job. Upon closer examination, however, we can see that the reds are clipping (check out the histogram for yourself), seen here in the bright skin areas.

What if we use the spot-on-clear-blue-sky trick? The results aren’t great, now heavily clipping the reds.

Spot-metering on the palm of your hand, however, solves the problem. Though if you examine the histogram it will look like the highlights and the shadows are missing, the midtones are where they should be: in the middle.

Will this work if we step back into the shade?

The background may be blown now, but the skin tones are where we said we wanted them to be: in the middle. If you were to take a portrait under the same light of someone whose skin tones are approximately the same as the palm of your hand, their facial features would be well-exposed – not necessarily well-lit, but at least not blown or in deep shadows. Try this next time Matrix metering lets you down.

One of the first articles on this blog demonstrated a technique to favor the midtones through spot metering on clear blue skies in Manual mode. This strategy works fairly well when shooting subjects under bright daylight, and it was Bryan Peterson calls “The Blue Sky Brothers” in his book, Understanding Exposure. While this technique is great for the midtones, it can lead to severe issues where near pure whites are concerned, especially if detail in that color is important to the shot.

A while back, I read an Outdoor Photographer article on this topic titled The Digital Zone System in which the author recommends protecting highlights by placing them in Zone VII. In this blog entry we will see how this advice works out in the field, and we will compare its results to what one would get with straight Matrix metering and Spot metering on clear blue skies.

Our first set of photos show how Matrix metering handles challenging situations that include brightly lit whites.

ISO200, f/5.6, 1/250 sec

ISO200, f/5.6, 1/2000 sec

ISO200, f/5.6, 1/800 sec

Since Matrix metering did not really do a very good job of protecting our whites, let’s see what spot metering on clear blue skies will give us.

ISO200, f/5.6, 1/800 sec

ISO200, f/5.6, 1/640 sec

ISO200, f/5.6, 1/800 sec

Except for the first sample scene, Matrix metering actually handled this situation better. Why? Primarily because these subjects are not about the midtones, so spotting on the midtones totally misses the point, and we shouldn’t be surprised that our subject suffers in the process. Matrix’s guess is off, but in these three cases happens to be better. Still, we need to meter on what’s important, the whites, so let’s see how we go about applying the advice in the Outdoor Photographer article.

Essentially, we still spot meter, but this time on the whites. If we zero out the meter, this will render the whites as middle grays, which is not what we want. Middle grays are in Zone V (5), and since the recommended approach places the whites in zone VII (7), that means we must open up our exposure by two stops (or +2EVs). In my Nikon D300′s viewfinder meter read-out, this means opening up shutter speed (or ISO or aperture, but I chose shutter speed) until the reading is 2 full stops to the left of center. I can then click and get these shots.

ISO200, f/5.6, 1/2000 sec

ISO200, f/5.6, 1/2000 sec

ISO200, f/5.6, 1/2500 sec

If you examine the histograms of each of these shots, you will see no clipped highlights. The whites are well contained, and a simple eye-ball inspection of the white tones in these three shots will confirm that fine textures and details are preserved, perhaps with the exception of the white rose’s detail, which is really hard to make out. You will notice also that in the other two lily shots, shadow areas are compromised, i.e., tend toward being blocked to the point they are hard to see. This illustrates that even though we did what we needed to do in the field with regards to exposure, we are not done. We need to finish out our goal (preserving white detail while not totally losing the shadows) in post-processing (PP). Through a bit of tweaking of shadows (in the lily shots) and exposure compensation plus shadow protection (in the rose shot), we get closer to our goal.

ISO200, f/5.6, 1/2000 sec PP: shadow=50

ISO200, f/5.6, 1/2000 sec PP: -0.5EV, shadow=50

ISO200, f/5.6, 1/2500 sec PP: shadow=50

In parting, I’ll note that the challenge in these shots isn’t just getting the right exposure, but rather managing harsh lighting that one should try to avoid if possible. For these types of shots, I prefer lighting of a softer quality.

[This is another oldie but goodie article on exposure, written just before the blog went live in 2009 and reproduced here for reference and archival purposes.]

When I first started shooting with my Nikon D80, I soon developed an incredible phobia about over-exposure. Soon I was either applying -0.7EV for Matrix metering, or Spot-metering and applying -0.3EV just to make sure. Anything to avoid those dreaded blinkies! As it turns out my reward for such conservatism was noise in shadow areas (and the sky) when I would correct for the numerous under-exposed images this approach yielded.

It took the arrival of the Nikon D90 to get me to reconsider my approach. By now I was feeling a little more comfortable with ViewNX and RAW editing, and I started noticing that I could safely over-expose with the D90 by as much as 1 stop in some situations, and still recover all highlight detail in ViewNX, with the added benefit that I could draw out shadow detail with Shadow protection without piling on the noise. From experience, I developed a rule of thumb to avoid over-exposure of more than 0.7EV, and I became very good at judging this from the histogram. Anything at or under 0.7EV of over-exposure, I told myself, was fair game. I could always get back to a beautifully exposed final product with ViewNX.

The following samples try to put some evidence behind my new approach. In addition, I also wanted to know how my D80, which I still use profusely, would compare. In other words, how much can I over-expose with it and still recover highlight detail in ViewNX? To be thorough, I bracketed from the centered exposure at various -EV and +EV settings to also answer the question of how much noise creeps in when one applies +EV in ViewNX (or equivalent technique, like Levels or Curves is PP) to an under-exposed shot.

Setup and other details: The D80 samples were taken with the Sigma 10-20mm, and the D90 samples were shot with the Nikkor 10-24mm, with both lenses set at a focal length of 16mm and aperture of f/16. The Picture Control on the D90 was set to my own custom emulated control of D2XMODE1 (PORTD2XM1), and since the D80 has no native Picture Control functionality (only Picture Modes), I switched its samples to PORTD2XM1 in ViewNX. To keep things as simple as possible, the D80 was set at 1/3 stop higher shutter speed to account for the slightly brighter Sigma performance at this aperture. All other things, such as focus point and camera position through the use of a tripod, were kept as equal as possible.

Some points regarding over and under-exposure to keep in mind…
As we have seen in the samples above, an under-exposure strategy may leave you with too much noise in the shadow areas, especially if you brighten the image in PP. Applying a given under-exposure of, say -0.7EV across the board, will guarantee that some of your shots under-expose by far more than that. Correct for that in PP, and you will have to deal with noise.

In addition, regarding the blinkies, sometimes they belong in the shot. If there is too much range to cover or if in order to properly expose the subject under available lighting conditions calls for it, so be it: let the blinkies flash all they want. For example, in a nighttime shot with street lamps, you will have to decide whether it’s okay for the lamps to glow, or whether your subject will be too dark for the shot to amount to anything. In that case, let the lights bloom and blink all they want.

Finally, over-exposure is not the end of the world! I don’t need to feel like a failure because I had to dial back exposure by 0.3EV in ViewNX or apply Highlight protection of 25 and/or bring out the shadows with Shadow protection of 80. In some cases, given the dynamic range of my D80 and D90 (especially the latter), over-exposing by as much as 0.7EV and applying correction through EV compensation and/or highlight protection in ViewNX yields preferrable results.

The moral of the story seems to be, yes, do avoid too much over-exposure, but know also how much over-exposure your camera’s sensor can handle given its dynamic range capabilities. The D90, and even the D80, can handle an amount of over-exposure and still capture highlight detail. Experiment with your DSLR, get comfortable with RAW and ViewNX, and you too may no longer suffer from over-exposure phobia.

Update: No doubt the new D7000′s improved dynamic range will help greatly with this. I plan to post some results here as soon as I have time to test one.

[This is an oldie but goodie post from my old website, posted here for archival purposes. Stay tuned for future articles on metering.]

The following write-up summarizes a comparison experiment I ran using my D90 and D80 to follow up on discussions regarding metering performance. As I’ve done before (see Sunny-16 vs. Metering), I first baselined against the Sunny 16 rule to see how close either camera comes to that age-old guideline. Then, I compared results for matrix metering and spot metering against the actual scene to see which came closest to representing what my eye saw. I will first summarize the setup and test results, then propose some preliminary take-aways. As always, we should do more testing to learn how best to expose various scenes.

Setup:

Equipment: Nikon D80 & D90, with Sigma 30mm f/1.4, mounted on tripod to produce consistently framed shots.

Conditions: The scene was a brightly lit 10:30am Southern California morning, camera facing due west, sun shining from South east (i.e., side/back-lit). AF mode was AF-Single. ISO was set to 200 to match the D90′s base ISO, hence Sunny 16 is 1/200sec @ f/16.

Image processing: All shots are RAW, down-sized and saved as JPG in Faststone 3.6, with no other adjustments or modifications.

Results:

Experiment	D80	D90	Notes
(1) Sunny 16 Manual @ f/16	1/200sec	1/200sec	Focus point on dark criss-cross glass of house’s front door.
(2) Matrix Metering Aperture Priority @ f/16	1/80sec	1/125sec	Focus point on dark criss-cross glass of house’s front door.
(3) Spot Metering Manual @ f/16	1/160sec	1/125sec	Focus point on dark criss-cross glass of house’s front door. Spot metering performed on gray section of pavement, pre-judging that to be the mid-gray for this scene. Same result achieved when metering off green grass (brothers green and mid-gray agree).
(4) Matrix Metering Aperture Priority @ f/16	1/100sec	1/100sec	Shot re-composed to bring focus point on palm fronds to the left of the front door.
(5) Spot Metering Manual @ f/16	1/160sec	1/125sec	Focus point on palm fronds to the left of the front door. Spot metering performed on gray section of pavement, pre-judging that to be the mid-gray for this scene. Same result achieved when metering off green grass (brothers green and mid-gray agree).
Additional experiments…	…	…	with VR16-85mm @ 16mm
(6) Matrix Metering Aperture Priority @ f/16	1/50sec	1/50sec	Focus point on Flag.
(7) Spot Metering Manual @ f/16	1/80sec	1/60sec	Focus point on flag. Spot metering performed on green (small) palm fronds to the left of the flag. [D80 spot metering missed?]

Parting thoughts

And now, at the risk of sticking my neck out too far, some thoughts about these tests results.

• Use of the Sunny 16 rule without some adjustments remains an unreliable practice. The D80 under-exposes by about 1 stop, and the D90 under-exposes by about 2/3 stops. As others have pointed out, Sunny 16 isn’t so much a rule as a guideline to get you in the ballpark. Please note that results may vary from camera to camera due to component variation, and more importantly, built-in tone curves applied against your JPG output. Bottom line: test your camera and judge how close it gets to Sunny 16. I know that when shooting with my D80, Sunny 16 becomes Sunny 11 (f/11, 1/200 sec @ISO200). I’ll need to do more testing to convince myself that Sunny 16 becomes Sunny 13 on my D90.
• Matrix metering on the D90 does not seem to have the heavy bias toward the focus point that the D80 exhibits. Experiments (2) and (4) produced pretty consistent results for the D90, even though in test (2) the focus point is in the shade whereas in test (4) it is in open sunlight.
• Moreover, Matrix metering and spot metering where also within 1/3 stop of each other for experiments (2) through (5) for the D90, whereas the D80 varied by a full stop under the same conditions.
• Somewhat subjectively, I judged the right exposure to be 1/125sec. Spot metering on the D80 only missed this by 1/3 stop, and a review of the histogram will show very minor variation. Exposure is acceptable between 1/100sec and 1/160sec, though at 1/100sec, highlight details begin to blow out. I realize this is a judgement call, and pure examination of the histogram might suggest to some that 1/100sec is the right exposure, but as a matter of personal taste, I would disagree.

Once again, we are left with the question of whether metering will reliably help us obtain the “right” exposure. I don’t know how others want to approach the issue, but for me, metering is, like Sunny 16, a tool or guideline to get me in the ballpark. I, based on the conditions and the results I want to achieve (more detail in the shadows, for instance), get to decide what the “right” exposure will be. With a little work and care, Spot metering based on pre-judged mid-gray points remains the best way to “nail” the exposure.

After making some disparaging remarks about Active D-Lighting (ADL), the data caused me to reconsider. Primarily, I now see I had the wrong expectations.

How I thought ADL worked: I thought ADL would auto-magically compensate for Matrix metering’s (MM) tendency to over-expose. Indeed, in some test cases I saw how ADL applies negative compensation, but not enough to make a difference consistently.

How ADL seems to do best: If you apply negative compensation to MM or otherwise meter to avoid clipped highlights, ADL helps avoid under-exposed shadows through adjustment of the tone curve (or however it does it), effectively pushing out shadow detail and producing a balanced shot.

To elaborate, I assumed that ADL would adjust MM’s tendency to over-expose, much in the same way I sometimes purposefully expose to the right by +0.3 to +0.7EV from what the Spot meter indicates as the centered exposure, then back-off the exposure in ViewNX and apply shadow protection to recover shadow detail. Why I thought ADL would do something similar I can’t really explain, but to move on and learn something new let’s compare some samples of my approach vs. ADL and see which does better. Along the way, we will examine the effect each approach has on noise, and to illustrate the effects of each of the approaches, we will use ISO800.

The first capture uses Matrix metering with ADL=OFF. The bright back-lighting against our dark subject produces the typical Matrix metering clipping in the highlights.

[Click for larger image]

For the next capture we set ADL to Normal. ADL tames down the highlights a bit, but not enough, and pushes out shadow detail in the vase. Start paying attention at the 100% crop on the right and look at whether noise is becoming more evident there.

[Click for larger image]

For the next two exposures, we use -0.7EV and -1.0EV respectivealy. Now the highlights are under control, but what is happening in the noise department?

[Click for larger image]
[Click for larger image]

To arrive at this final image we use the first exposure, thankfully captured in RAW, and we apply -0.7EV adjustment in ViewNX and shadow protection = 30. (For another example of this technique, see When Over-exposure happens). Compare the resulting image to the previous 4 samples.

[Click for larger image]

Which adjusted exposure achieves the best noise performance should be self-evident. Which of these achieves a better balance in the overall exposure is probably largely a matter of preference. My perspective is that the last image balances highlights, middle tones and shadows best. I wish ADL or some other algorithm performed similar simple adjustments on the fly in-camera (hint, hint, Nikon).