What in the world did my meter just do!?
by eNoBlog on Aug.03, 2009, under Techniques
If you are even half human, you probably have asked yourself what your meter has just done to you after you review a photograph in your camera’s LCD. I know I have. And reading about how meters try to turn the shade they measure into 18% gray may just be as confusing, in particular if as Thom Hogan likes to point out, it’s more like 12%.
Perhaps some hands-on work in is order. Let us photograph three cards, one white, one gray and one black, and let’s see what our cameras do under their various metering modes. A simple setup as shown here should do.
To make this interesting, we’ll test 2 cameras, the Nikon D80 and the Nikon D90, two alleged bafflers when it comes to metering. For Matrix and Center-weighed metering, we’ll aim at the center cross hairs and in Aperture priority mode, let the camera pick the shutter speed. For spot metering, we’ll switch to Manual mode, move the focus point off the cross hairs so we can spot on a solid shade, then bring the focus point back to the cross-hairs before taking the photo. To make it fun, let’s show the results and see if you can pick out which card was photographed for sets 1, 2, and 3. Ready, set, go!
| Set 1 | Matrix | Center | Spot |
| D80 |  |
 |
 |
| D90 |  |
 |
 |
| Set 2 | Matrix | Center | Spot |
| D80 |  |
 |
 |
| D90 |  |
 |
 |
| Set 3 | Matrix | Center | Spot |
| D80 |  |
 |
|
| D90 |  |
 |
 |
Have you stared at these long enough? Okay, here is the answer. Set 1 is for the black card, set 2 is for the white card and set 3 is for the gray card. Don’t be surprised that none of the metering methods got close for the the white and black cards, though I’d like to point out how the D80’s matrix metering made the most valiant effort. Only for the gray card do we get somewhat reasonable results, though I’ll confess quickly that the gray card I generated for our experiment is an approximation of middle gray (RGB=128,128,128). Obviously, this test was rigged.
The take away here is that faced with extremes in whites or blacks, we need to learn how to compensate against our in-camera meter’s tendency to push the exposure toward middle gray (whether that’s 18 or 12%). After all, the first sample image in this write-up does show the correct shades of white, black and gray, so it must be possible to arrive at the right exposure. Did the photographer perhaps spot-meter against one of the usual suspects? I think you know the answer to that question.
Additional resources:
Understanding Camera Metering and Exposure
Accurate Exposure with your meter
Past related blog entries:
Spot-metering the usual suspects
Understanding the Exposure triangle
The quest for the correct exposure
August 3rd, 2009 on 8:05 am
Interesting how the D80 has the more accurate metering results with all three cards. Two of them using matrix metering too. You would think the newer camera would have been improved in this area knowing the problems the D80 had. Looks worse though, based on this test. Spot metering on that last card means a lot more because matrix is always in question but the D80 sure outperformed.
August 3rd, 2009 on 9:39 am
Carl – Yes, the D80 got “closer” but the point here is that the way meters work, neither camera got it 100% correct, which still leaves us with an incorrect exposure. Actually, the real point is that when faced with light-dark scenes, you will have to compensate/adjust somehow. Examples of this are scenes with lots of snow under sunlight and scenes with extremes of shadows and highlights.
August 3rd, 2009 on 10:03 pm
This was interesting. I wonder if the FX cameras are better at metering than these two DX cameras are. Or maybe the D300 is better at metering than the D80 and D90.
What was the white balance setting of the cameras used?
August 3rd, 2009 on 10:37 pm
Simon: whether the metering is better is a matter of opinion, but the general consensus is that the D200, D300 and higher level cameras offer more “consistent” metering, with lesser likelihood of over-exposure. These actually implement a more sophisticated meter. As for White balance, I’m afraid I muffed that in camera, but I corrected all images to the “Shade” WB in ViewNX before I converted them to JPG — that, BTW, was the only edit other than resizing I did on these images.
August 4th, 2009 on 3:54 am
Looking at your results, i am guessing that the Active D-lighting function in the D90 menu was switched ON…..
this explains why all the D90 cards look similar.
(Had the Active D-Lighting function been OFF, then the D90 results would have been better than the D80.)
…..I am not going to get into the Workings of the Active D-Lighting function…but it is to do with contrast and shadows. boosting the shadow but preventing the highlights from being blown out……
….shooting cards like that with it ON, would have confused the exposure
August 4th, 2009 on 5:04 am
Adrian: interesting thoughts here, but I turned off Active D-Lighting in my D90 the day I got it. Though Active D-Lighting may be a useful feature, I didn’t want to confuse myself nor affect my RAW file… I much rather do the tweaking myself in PP.
August 4th, 2009 on 7:08 am
Yes, I understand the D80 got it closer, but when you look at your results you have to admit it got it WAY closer. The difference is drastic.
Was your white balance taken off the gray card and on both cameras?
August 4th, 2009 on 11:12 am
Carl – I adjusted WB in ViewNX to “Shade”, so the in-camera setting is ignored.
August 4th, 2009 on 2:49 pm
First, thanks for doing the test.
I am confused by your MM results from the D80. Assuming that the card filled the frame for each shot, I would have expected the camera to render each shot as middle gray. However, it seems to have recognized, at least to a degree, the shade of each card. What information did the metering have to make that determination? If for example, you decreased the intensity of your light source sufficiently, the brightness of the white card could be made to match that of the gray card. How could MM tell the difference between a brightly lit gray card and a less brightly lit white card?
The D90’s MM, OTOH, produced roughly equivalent exposure for each card, which is the result that I would have expected, given the uniform nature of the targets and the lack of anything to “evaluate.”
(I am open to the possibility that I am missing something obvious.)
August 4th, 2009 on 4:01 pm
You raise some interesting points, and I’m not sure I can address them in full… One possibility is that since I printed those cards, the greater amount of ink on the black card reflected more light. Another possibility is that at least for Matrix metering, the D80’s greater emphasis on the focus point took into account the shade of the cross hairs? Whatever the case, I can’t come up with explanations that account for all the results, especially since center-weighed results are somewhat askew as well.
August 5th, 2009 on 8:20 am
It is puzzling. The fact that the D80’s MM produced results that seem sort of correct for each card makes it hard to come up with a simple explanation.
In any case, I’m pretty happy with my D90’s MM. It does tend to give more exposure than I’d choose for scenes with a dark foreground and a bright sky, but that’s correctable with EV compensation and/or PP, especially if you shoot raw. I have also experimented with ADL on Auto. It will reduce exposure to protect highlights in high-contrast scenes. Shooting raw (and not using Nikon software for PP), that’s all it seems to do.
August 8th, 2009 on 3:46 pm
Jeff,
The beauty of Auto ADL is that if the scene has no problems it will do nothing. It uses the exposure sensors to determine if contrast is excessive for normal JPG rendering. If it is, it will lower the exposure the appropriate amount using Matrix Metering. After the scene is captured by the sensor it analyzes the high and low values and then sets the high value contrast and low value contrast as needed to preserve the white point. It may set them higher or lower for 1 to 3 segments of the histogram curve. It tries to resolve the shadow side and the highlight parts of the picture and reduce the difference where they are next to each other. Each increment of ADL, High, Low, or Normal seems to have different rules, not just degrees of adjustment. eNo and I will be independently looking at that, soon. He works faster than I and it is very difficult to stage a still life scene to reverse engineer what is happening. My best opportunity exists in real nature and must be done with direct sunlight with a variety of clouds at a certain time of day. We have had a week of storms here in Idaho. ENo and I may be taking different routes. Right now we need a still life scene that satisfies the requirement to have measurable results over a 10 stop exposure scale. That means looking at the raw values before conversion and special raw software which will allow us to judge the headroom before the JPG conversion takes over. I keep saying we, but our efforts are completely independent. We share a common interest.
August 8th, 2009 on 11:50 pm
Leon, look for next Tuesday’s (August 11) blog entry. I’m still scratching my head a little, so expect it to be “rough” and rudimentary. I’ve been experimenting with some additional test shots which I may feature on Wednesday to show what I’ve learned — which so far is, ADL helps a little, but it doesn’t save you from lousy light.
August 7th, 2009 on 2:50 am
Hello. Thank you for this great info! Keep up the good job!
August 9th, 2009 on 3:35 am
thank you! I really liked this post!
August 10th, 2009 on 6:14 am
thank you! I really liked this post!