Nikon Matrix Metering and Exposure Compensation Query

[Steve Smith]

Another thought I have had against using matrix metering is the use of a polariser. From what I have read, it seems that the camera knows what aperture it is working at and works out the actual light level of a scene. Therefore, if the majority of zones give a bright reading i.e. snow about two stops above mid grey, it knows that it should compensate. If you now put a polariser on, these bright zones are now two stops darker - or mid grey! Therefore, I don't think it will compensate as it thinks that it is a fairly normal mid toned scene.

I think I will send her off with a mixture of slide and negative films!


Steve.

 

[dslater]

Another thought I have had against using matrix metering is the use of a polariser. From what I have read, it seems that the camera knows what aperture it is working at and works out the actual light level of a scene. Therefore, if the majority of zones give a bright reading i.e. snow about two stops above mid grey, it knows that it should compensate. If you now put a polariser on, these bright zones are now two stops darker - or mid grey! Therefore, I don't think it will compensate as it thinks that it is a fairly normal mid toned scene.

I think I will send her off with a mixture of slide and negative films!


Steve.

That's a very interesting thought. You could very well be correct there, although it is possible that the scene matching algorithm in the camera could save you there. It would be a very interesting experiment to take 2 pictures of the same difficult snow scene - one with the filter and one without and see how the results compare.

On the other hand, even with proper exposure, I wonder if the polarizer would kill so many reflections that the snow would lose it's sparkle?

 

[Sirius Glass]

That's a very interesting thought. You could very well be correct there, although it is possible that the scene matching algorithm in the camera could save you there. It would be a very interesting experiment to take 2 pictures of the same difficult snow scene - one with the filter and one without and see how the results compare.

On the other hand, even with proper exposure, I wonder if the polarizer would kill so many reflections that the snow would lose it's sparkle?

The light meter measures through the polarizer => this is not at problem. I have taken many snow scene photos this way. HOWEVER dislater is right, while the polarizer cuts out glare, it also cuts down the sparkle.

I would recommend shooting with and without the polarizer.

 

[Steve Smith]

I think I will do some experimenting tonight using a lightbox and a polariser and see what the metering differences (if any) are with matrix and centre weighted metering.

Steve.

 

[Lee L]

That is not the entire story. While it is true that part of the matrix metering process involves matching against a pre-programmed set of scenes, the other part of matrix metering is that Nikon lenses with a CPU chip tell the meter what the aperture of the lens is. With this information, the meter now knows what the actual luminance of the scene is. Since the luminance is known, thing like snow and bright sand can be detected and the meter can place them on zone 6 or 7.

If the metering system is making this kind of comparison, how does it detect snow (or a mostly high key scene like light sand, etc.) under varying conditions like sun, shade, or on an overcast day? In other words, how would it know snow in shade from a gray card in sun? Or does that adjustment only happen in full sun? Seems that a TTL reflectance meter would need something like a comparison from a non-TTL incident meter (comparing illuminance to reflectance) to make the judgement that a scene is mainly white in anything other than full sun. Is that the case with Nikon matrix metering, or have I just given away a patentable idea?

I'm not a Nikon user, but since the system was introduced, I've wondered how you make intentional adjustments (other than the usual bracketing) to a computer adjusted reading that isn't "explained". I guess users get an instictive feel for this after a while. How many stops will matrix metering shift the exposure from medium gray in it's most extreme deviation? Are adjustments geared for the dynamic range of transparency film, or for negatives? Yeah, I'm overthinking this, but the promise of computer matching to one of a few kabillion standard scenes begs the question. These are just a few questions that have come to mind. No need for a long explanatory reply unless you know how the matrix metering is programmed.

Lee

 

[Steve Smith]

No need for a long explanatory reply unless you know how the matrix metering is programmed.

No. But I will post my findings after some experimentation.


Steve.

 

[dslater]

If the metering system is making this kind of comparison, how does it detect snow (or a mostly high key scene like light sand, etc.) under varying conditions like sun, shade, or on an overcast day? In other words, how would it know snow in shade from a gray card in sun? Or does that adjustment only happen in full sun? Seems that a TTL reflectance meter would need something like a comparison from a non-TTL incident meter (comparing illuminance to reflectance) to make the judgement that a scene is mainly white in anything other than full sun. Is that the case with Nikon matrix metering, or have I just given away a patentable idea?

I'm not a Nikon user, but since the system was introduced, I've wondered how you make intentional adjustments (other than the usual bracketing) to a computer adjusted reading that isn't "explained". I guess users get an instictive feel for this after a while. How many stops will matrix metering shift the exposure from medium gray in it's most extreme deviation? Are adjustments geared for the dynamic range of transparency film, or for negatives? Yeah, I'm overthinking this, but the promise of computer matching to one of a few kabillion standard scenes begs the question. These are just a few questions that have come to mind. No need for a long explanatory reply unless you know how the matrix metering is programmed.

Lee

Hi Lee,
I think it is the combination of knowing the actual luminance and doing a pattern match against some 30K internal stored scenes that allow the matrix metering to handle these varying situations. In the case of snow on an overcast day, I would not be surprised to see the snow come out darker than on a sunny day as it is indeed darker - however, I don't think it will come out as dark as it would using a traditional averaging meter.
I haven't read anything that would indicate the meter takes dynamic range into account. With only 5 metering zones, that would pretty tough to do reliably.
As to making an intentional adjustment, you are absolutely correct. People who use matrix metering most of the time develop a feel for the amount of correction a scene will need. I generally only use matrix metering when the lighting is not too difficult and I am shooting quick snapshots of my family. When the lighting is difficult and the shooting slow enough, I will switch to the spot meter and take separate readings of shadows and highlights to determine exposure. If things are moving too fast for that, then I will take a reading of a scene with the matrix metering, then use the spot meter to read the same scene, note the difference in exposure I get, and use that to determine an appropriate exposure compensation for matrix metering. I will also tend to do a fair amount of bracketing as my F80 and F100 have some nice auto-bracketing features.

 

[Lee L]

Dan,

Thanks for the reply. I still can't see how matrix metering could distinguish between say a dark asphalt road in full sun and a white adobe wall in shade, i.e. place fairly low contrast scenes at their "correct" places on the scale, or how it could infer incident light levels correctly from its reflectance readings in similar circumstances. But this is all hypothetical for me anyway. My mom's remarried and her husband shoots Nikon (don't know which body), so maybe on my next visit I'll ask if I can try out his camera.

I do much the same with TTL metering as you in difficult lighting, but with center weighted averaging (rather than matrix) and spot, which in my case reads the central split image and rough microprism ring. I also carry along a Gossen Digiflash, especially with rangefinders, over the last couple of years for fast and accurate incident readings or a "reality check".

I'll be interested to see what Steve finds.

Lee

 

[dslater]

Dan,

Thanks for the reply. I still can't see how matrix metering could distinguish between say a dark asphalt road in full sun and a white adobe wall in shade, i.e. place fairly low contrast scenes at their "correct" places on the scale, or how it could infer incident light levels correctly from its reflectance readings in similar circumstances. But this is all hypothetical for me anyway. My mom's remarried and her husband shoots Nikon (don't know which body), so maybe on my next visit I'll ask if I can try out his camera.

I do much the same with TTL metering as you in difficult lighting, but with center weighted averaging (rather than matrix) and spot, which in my case reads the central split image and rough microprism ring. I also carry along a Gossen Digiflash, especially with rangefinders, over the last couple of years for fast and accurate incident readings or a "reality check".

I'll be interested to see what Steve finds.

Lee

Well, I think that's where the patter matching comes in - I'm sure that if you moved in on the wall or the road so they field the frame, you could easily fool the meter, but that wouldn't really be a typical shooting situation. - Matrix metering is definitely not perfect - it just tends to get it right more often that simple center-weight metering.
The F5 and newer

 

[MartinB]

Steve,

Helen's advice to keep it simple is best.

I have an F5 that I have used for frequent mountain photography (along with non-analogue Nikons) I ski most weekends in the Canadian Rockies. I find that the F5's matrix metering is mostly accurate and if I don't have time to think about metering I will use matrix with +2/3 of a stop. The matrix meter in the F5 does not seem to be "fooled" by a polarizer. I have no first hand experience with the F601 variant of the matrix meter.

I find that there are a large number of factors when it comes to photographing snow scenes in the mountains and it can drive you nuts trying to account for them all. While the contrast range can be extremely high on a clear sky sunny day (i.e. between bright snow and dark subjects such as bare ground or evergreens), the same scene can be much lower contrast on a bright hazy day. It seems counter intuitive because the light levels on a hazy day are still very high (so much that you still must wear dark sunglasses) and you would think contrast levels are still high but they are not. On a bright hazy day you will likely retain detail in the darker parts of the scene whereas in the first case the dark areas tend to black out.

There will be cases where you will not be able to hold detail in both the brightest and darkest parts of the scene and other times where you will. For outdoor portraits, matrix metering seems to work pretty well and makes the right choice of letting the snow "blow out". I think if you try to work out a set of guidelines for your wife to follow based on percentage of snow in a scene, you will end up with many exceptions to the rule and that will take the fun out of it.

If your wife is going to take both slide and print film, I would recommend for scenics:
1) print film - use matrix metering and dial in +1 stop compensation (or set ISO lower when loading film and leave it there)
2) slide film - use matrix metering - take 2 shots of every scene - first shot at matrix meter reading and second with +1 stop (if you can set up an auto bracket sequence that will do it great - you don't need to bracket minus compensation)
3) For people/portraits - just go with the straight matrix meter reading.

Don't worry about the polarizer from a matrix metering point of view.


I believe that Nikon has at least 5 variations that they call matrix metering which may explain varying opinions on its accuracy. The simplest is 5 segments with a simple algorithm and the most complex is in the F5/F6 (and D2..) that has 1005 segments that read colour in addition to the luminance and distance and compare to a database of over 30,000 photos. Some pro-sumer models have 1005 segments but no colour info so read the spec sheets carefully.

Moose Peterson has an interesting method for testing exposure that may be useful if you are still thinking of testing. I would use any cheap slide film rather than wasting good expensive stuff since you are mainly interested in the relative differences between the exposures.
http://www.moosepeterson.com/techtips/teddybear.html

I will see if I can dig out any exposure information from past snow photos.
Good luck

 

[eddym]

If the metering system is making this kind of comparison, how does it detect snow (or a mostly high key scene like light sand, etc.) under varying conditions like sun, shade, or on an overcast day? In other words, how would it know snow in shade from a gray card in sun? Or does that adjustment only happen in full sun? Seems that a TTL reflectance meter would need something like a comparison from a non-TTL incident meter (comparing illuminance to reflectance) to make the judgement that a scene is mainly white in anything other than full sun. Is that the case with Nikon matrix metering, or have I just given away a patentable idea?

I'm not a Nikon user, but since the system was introduced, I've wondered how you make intentional adjustments (other than the usual bracketing) to a computer adjusted reading that isn't "explained". I guess users get an instictive feel for this after a while. How many stops will matrix metering shift the exposure from medium gray in it's most extreme deviation? Are adjustments geared for the dynamic range of transparency film, or for negatives? Yeah, I'm overthinking this, but the promise of computer matching to one of a few kabillion standard scenes begs the question. These are just a few questions that have come to mind. No need for a long explanatory reply unless you know how the matrix metering is programmed.

Lee

No, you're not overthinking this; I've been thinking the same thing. How the hell does a Nikon meter, or any meter, know that there's snow on the ground and not gray ash, or black coal? The aperture has nothing to do with it. It has to have some other point of reference in the scene, like maybe a clear blue sky in the background, so it can think, "hmmm... lots of light down below, 2-3 stops less above... either there's snow on the ground, or this idiot is holding the camera upside down."
I have an F100, and admittedly, we don't shoot a lot of snow scenes here in Puerto Rico. But I can assure you, I have "fooled" the matrix meter plenty of times. It's not infallible, and I am using spot metering a lot more often than I did when I first got the camera.

 

[Karl K]

This is my second bit of two cents when using the Nikon F100's Matrix Metering System: if the scene's brightness to darkness ratio is too great for the film in question, then the Nikon seems to underexpose by about one stop. This is most often seen using Velvia or K64 in bright sunlight. It has happened to me shooting high contrast street shots in NYC and in Philly, where the sun streams onto streets lined with skyscrapers. It also reared its ugly head shooting the frozen Paterson Falls several years ago, using Ektachrome 100. This situation is really not just a Nikon Matrix Metering problem, it's a simple fact of life. That's why I suggest that color negative film is the best answer.

 

[dslater]

This is my second bit of two cents when using the Nikon F100's Matrix Metering System: if the scene's brightness to darkness ratio is too great for the film in question, then the Nikon seems to underexpose by about one stop. This is most often seen using Velvia or K64 in bright sunlight. It has happened to me shooting high contrast street shots in NYC and in Philly, where the sun streams onto streets lined with skyscrapers. It also reared its ugly head shooting the frozen Paterson Falls several years ago, using Ektachrome 100. This situation is really not just a Nikon Matrix Metering problem, it's a simple fact of life. That's why I suggest that color negative film is the best answer.

If the scene brightness range it too great for the film, then the problem isn't under-exposure - it's that the brightness range is too great. You're either going to have blown out highlights or dark shadows. For example, suppose you have a scene with bright snow and dark shadows - say your brightness range is something like 7 or 8 zones. If the exposure you get places the bright snow on zone 7 and your shadows fall on zone 1 or 2, then that's not really underexposure - your brightness range is simply too great. In this situation, you as the photographer need to make a decision - do you retain detail in the highlights and have dark shadows, or do you maintain detail in the shadows and have blown highlights - you can't really expect the meter to make this decision for you.

 

[dslater]

No, you're not overthinking this; I've been thinking the same thing. How the hell does a Nikon meter, or any meter, know that there's snow on the ground and not gray ash, or black coal? The aperture has nothing to do with it. It has to have some other point of reference in the scene, like maybe a clear blue sky in the background, so it can think, "hmmm... lots of light down below, 2-3 stops less above... either there's snow on the ground, or this idiot is holding the camera upside down."
I have an F100, and admittedly, we don't shoot a lot of snow scenes here in Puerto Rico. But I can assure you, I have "fooled" the matrix meter plenty of times. It's not infallible, and I am using spot metering a lot more often than I did when I first got the camera.

Knowing the aperture does have something to do with it. Since the meter knows the actual luminance and sun can only get so bright, then snow under clear skies will give a luminance value that can only be white - gray ash or black coal simply can't have as high a luminance value.

 

[Karl K]

Well, of course, but....

If the scene brightness range it too great for the film, then the problem isn't under-exposure - it's that the brightness range is too great. You're either going to have blown out highlights or dark shadows. For example, suppose you have a scene with bright snow and dark shadows - say your brightness range is something like 7 or 8 zones. If the exposure you get places the bright snow on zone 7 and your shadows fall on zone 1 or 2, then that's not really underexposure - your brightness range is simply too great. In this situation, you as the photographer need to make a decision - do you retain detail in the highlights and have dark shadows, or do you maintain detail in the shadows and have blown highlights - you can't really expect the meter to make this decision for you.

...what I'm saying is that the NIkon Matrix Metering System on my F100 consistently under-exposes in the above-mentioned situations. I am aware of the limitations of all films in their ability to reproduce very contrasty scenes. What you did not address is the Matrix Metering's tendency to under-expose in these situations. "-you can't really expect the meter to make this decision for you" is not the answer to the original question as posed by Steve Smith.

 

[eddym]

Knowing the aperture does have something to do with it. Since the meter knows the actual luminance and sun can only get so bright, then snow under clear skies will give a luminance value that can only be white - gray ash or black coal simply can't have as high a luminance value.

But the aperture (alone) has nothing to do with the luminance of the scene. I can shoot at f2.8 in bright daylight if my shutter speed is fast enough, or at f22 at night if it's slow enough.

 

[dslater]

...what I'm saying is that the NIkon Matrix Metering System on my F100 consistently under-exposes in the above-mentioned situations. I am aware of the limitations of all films in their ability to reproduce very contrasty scenes. What you did not address is the Matrix Metering's tendency to under-expose in these situations. "-you can't really expect the meter to make this decision for you" is not the answer to the original question as posed by Steve Smith.

So are you saying that in this situation your snow comes out on zone 5 or 6 instead of 7? Because I have not had that happen to me.

 

[Karl K]

Yes, it underexposes the "whites."

So are you saying that in this situation your snow comes out on zone 5 or 6 instead of 7? Because I have not had that happen to me.

I just ran another test on the Nikon F100's Matrix Metering. Since there is no real snow around NJ right now, I set up large pen and ink drawings (16x20) with mostly white backgrounds and some fine black lines drawn. The camera was set up on a tripod and the lens zoomed in to include only the drawing. Slides were shot at normal 100 ISO and processed by a pro lab. All slides were about one stop under, BUT the single standard drawing, which was not a line drawing, was exposed perfectly. I've had my camera checked by Nikon and they say it's spot on. So...I always over-expose one stop in white snow scenes. Doing that insures that my snow comes out white.

 

[Christiaan Phleger]

I am not surprised that a poster from Puerto Rico could consistently fool the Nikon Matrix Metering, which still has a EV limit of 16 2/3. I live in Hawaii and its frequently bright enough to exceed that, think white wedding dress on the beach, clear sky. Makes your retina's ache (not the kodak ones). Generally speaking it does a good job, one way I can fool it consistently is overcast, voggy day in the lava field, or a bright day at the beach on lava rocks.

 

[Sirius Glass]

Photographing snow

If you want the shape of the snow - drifted snow or moguls on a ski slope use the meter reading from the snow. If you want people/objects take a reading from your hand with the sun on the hand for sun light on the subject; tilt the hand so that it is shaded for backlit subjects.

Steve