Nikon D80 metering - An experiment to show what actually happens

OK, I've done it - actually done a little experiment to see what happens. All

images are of an 18% grey card. Metering on all shots is matrix & using the

central focusing point. The first shot of the grey card on its own puts the

histogram dead centre - exactly where it belongs. Please see subsequent shots

with explanations as to what is happening.<div></div>

Here is a shot with a white piece of paper in the top left corner. Again matrix metering off central focusing point. Note no real difference in exposure.<div></div>

Same again, only with black(ish) paper in corner - matrix central focusing point. Note slight tendency to increase exposure, but not huge impact.<div></div>

Now the fun starts. White paper in centre, central focussing point. We now have some under exposure of the grey card.<div></div>

Finally, black paper in the centre, matrix metering & central focussing point. Now we see some serious over exposure of the grey card.<div></div>

And the point of this is?

I tried experimenting with different focusing points, but the results were not as conclusive. In summary, The D80's meter meters correctly for average scenes - hence the success that people have with centre weighted metering.

 

The matrix metering algorithm is unduly influenced by the tone that lies under the active focusing point - particularly the central one.

 

Therefore, if using matrix metering and the central focusing point, pay careful attention to the tones at that point and where the other tones are for your major areas of interest.

Yana - you should have waited for the final comment! D80 metering has caused some interesting threads with a lot of hot air and little hard evidence. The point is that D80 users need to work a little harder to get what they are after and some understanding of D80 exposure metering characteristics will help.

Simon, so your conclusion is that your D80's matrix meter is essentially acting like a center weighted meter. I assume that the AF point is set to the center one in all cases.

 

BTW, I see pretty clear vignetting in all shots. That is probably due to the lens or perhaps flash?? limitations.

Simon, thank you for the effort. This another good data point to have.

I am grateful for your work in sharing your data.

 

I'm not sure I agree with your conclusion. The camera doesn't know what 18% grey is. It's trying to assign values to whatever luminescences are in front of it that will differentiate them in such a way that they are accurate relative to each other, and within dynamic range if possible. I think all of your test shots did that.

 

Still, the way it did it is interesting indeed, and most informative.

Interesting test!

Shun, yes, I think that's where I'm leaning. Certainly Thom Hogan has come to the same conclusion in his D80 review. I'd do some more playing with different active focus points & smaller white / black paper, but I've satisfied myself that I know what's happening.

 

DHG: Thanks. As understand it, most camera meters try to render a scene so that, if averaged, it would come out as an 18% grey tone. Matrix metering attempts to interpret a scene to determine 'where the important bits are' based on a library of 'typical' images. How this then translates into overall exposure then depends on the algorithms used to calculate the exposure value. The D80 has one view of the world & other bodies have different interpretations. It's simply a case of understanding & working with what it does.

Simon: I think that's exactly right, except I'm not sure it's neccessarily trying to average to 18% grey all the time. If it did that all the time we'd have a lot of trouble making "high key" shots with digital cameras!

 

One thing that doesn't seem to get stressed much in discussions of digital cameras is that, unlike film, much of the result depends on what the software decides to do, which is a different issue than metering. I think there's a lot of logic in the software that's second guessing my intentions, (why else have a huge library of images stored in the processor) and sometimes I like the result and sometimes I don't.

I think the results of this test show that under difficult exposure situations you need to spot meter. If you average the world together, it's about 18% grey so that's what the camera tries to do. If you have a light subject in a dark environment the camera is going to even the entire picture out by over exposing the light subject. The reverse is true if you have a dark subject in a light environment such as backlit scene. The camera will under expose the subject to try to get the whole picture at 18% grey. If you spot meter your subject or an area in the scene that is 18% grey, the exposure will be correct. There was a thread recently with a picture of a hallway that was difficult to expose. If you look down, there is a perfect 18% grey in the concrete at he photographers feet. Spot meter the concrete, recompose and shoot. Much quicker than playing with the exposure compensation.

Shun, forgot to day that I was using the 18-70mm @ f5 - not the best arrangement I know! Thinking about it, the metering characteristics in matrix mode are more 'focus point spot' than centre weighted, but the non central focusing points seem less sensitive to what's there. When I have more time, I'll repeat the test under better conditions and with smaller pieces of black / white paper and with other active focus points.

 

One of these days, I'll get my website up & running where I can put all of this stuff up properly. Then I can pretend I'm an English Ken Rockwell and ask for money.....

Simon, very interesting test. I'm not exactly following though, are you suggesting the matrix pattern will shift with AF point, or that the matrix meter is just heavily weighted towards the central region? Certainly it would make sense if the meter becomes center-weighted if focusing on the central region, maybe there are additional 'hot' nodes where the meter can become biased.

 

I believe all new Nikon bodies have color sensitive meters, so I wonder if there is additional bias towards certainly colors and/or contrasts of colors?

 

Nice work, very intersting!

The camera doesn't actually store a library of images for the matrix meter. The algorithm used by the matrix meter to determine exposure is adjusted in the design phase to give good results on situations shown by the library of images. But the actual image data is not in the camera.

 

In any case, according to Nikon the matrix meter (at least a few years ago) works by taking a centerweighted meter reading and adjusting the reading by a variety of additional data. So it is no surprise if the results are weighted by the center area.

 

To make sense out of the matrix meter one would need to run a lot more comprehensive tests. A typical scene is not a card in white background. Thus the meter isn't expected to give good results in such a case.

Simon - what you have done is similar to what Sekonic has here: http://www.sekonic.com/classroom/classroom_2.asp showing the difference between incident and reflective metering.

 

One of the real differences between the matrix metering of the D80 and other Nikons is that typical matrix metering "figures out" that there is something unusually like or dark in the frame and factors that in and gives a pretty good exposure. The D80 gives a lot of weight to the brightness of the subject that is being focused on.

 

BTW, the camera does not store a "great number of images" in the camera for matrix metring. What Nikon has done is to analyzed a great number of pictures with a "grid" placed over it and then calculated what the exposure compensation should be for a given pattern of bright, dark and medium toned areas. The relative brightness pattern that the camera meter measures are then matched to a simple look up table to find the appropriate exposure bias.

<i>BTW, the camera does not store a "great number of images" in the camera for matrix metring.</i>

<br /><br />

 

I took the following from the information on Nikon's website concerning the D80 (emphasis mine):

<br /><br />

" Nikon's 3D Color Matrix Metering II instantly and accurately evaluates brightness, color, contrast, selected focus area and subject-to-camera distance information,<b> referencing the results against an <u>onboard database</u> of 30,000 scenes from actual photography.</b> "

<br /><br />

Are they lying?

While there has been a lot of talk about the D80 metering system and pattern, this isn't at all inconsistent with how matrix metering has worked in the past. Nikon has published, I believe, the percentages it gives to certain parts of the pattern, and explicitly states that, say, bright areas in the corners are ignored because it's thinking it's a sky scene and that, if you have something dead center that is darker, it's the dead center object that matters (not illogical, though a pain in the ass with backlit settings).

 

The question, which has been asked, is what happens if you put that white or black card under one of the other focus points, rather than just in an upper corner. I'd be interested in that.

 

allan

From what I've read in the past, Nikon's Matrix Metering system also tries to compare the pattern and degree of tones to an internal database of patterns that describe common photographic scenes. I think it is too simplistic to say that the system is just trying to arrive at an exposure that can encompass the highest and lowest tones. This database of patterns will also be used to throw in a bias that will give you what might be viewed as inconsistent results.

 

 

larsbc

Thanks Simon! I think this kind of thing is both interesting and very helpful.

 

From reading various posts/articles, it seems those who complain about the D80 Matrix Metring is its UNPREDICTABILITY (which, it appears, Simon's tests are moving to demystify).

 

Therefore, what I don't get is: why doesn't Nikon offer a special guide to Matrix Metering (using actual pictures and saying, "here's what MM did, here's why, and here's what it's likely to do (given these parameters). Also, why doesn't Nikon offer a firmware download to allow D80 users to use a D200 style formula? I recall Minolta using flash insert cards some years ago to achieve something like this option.

 

Lastly -- is there a 3rd party that might want to make a few $ offering such an option?

How does the D200 react to such a test? Is it very different? If it also

meters the average, it is suposed to do the same.

regards,

Martijn

<i>Therefore, what I don't get is: why doesn't Nikon offer a special guide to Matrix Metering (using actual pictures and saying, "here's what MM did, here's why, and here's what it's likely to do (given these parameters).</i>

<p>

A matrix meter is not something that a user can understand or second guess easily. It is meant to be used as it is, when you can't or don't have time to make intelligent creative decisions about exposure. When you can, you use the predictable centerweighted and spot meters.

<p>

Also, since Nikon tweaks the algorithms from one camera to the next, it is unlikely that an individual would learn the quirks of one meter while using multiple different bodies over the years.

<p>

My advice is that you should learn to use a spot meter reliably, and use the matrix meter when you use low-to-medium ISO settings, shoot RAW and can adjust the tone curve in postprocessing. You can of course also use the matrix meter and bracket.

<p>

No point in blaming the matrix meter for anything. It is not the only method for exposure determination after all.

I have noticed the same thing on a F100 a long time ago by framing a landscape and see the exposure change as I moved the focus point around.

 

There's nothing magic to matrix metering. In the end, it isn't much more than an average of the whole frame with a emphasis on the focusing point.