D700 Underexposure Puzzle

<p>I want to first show the scene that I used for this test. I think you'll agree it easily qualifies as the proverbial "average scene;" the kind that exposure meters are calibrated for. I imagine it probably reflects somewhere close to 18%. Blue northern sky; fully frontally lit red brick wall; white garage doors; medium gray asphalt; black car. Picture taken maybe 10:45AM on a sunny day. The fire station wall faces South.<br>

This picture was taken at f/2 with a 35mm MF lens with a CPU. I adjusted the exposure with the exposure compensation until the histogram was as close to the right as possible without crashing into the right edge. It took 1.3 stops of exposure comp to get this. OK that's just a little background. I haven't told you the problem yet.<br>

Standby while I try to get the image inserted.</p><div></div>

<p>The next point is that I discovered this phenomenon by accident. I wanted to see how a couple of lenses performed at their wide apertures. I set the camera to "A" and the ISO to 400. Trusting the camera's auto-exposure, I shot the picture at f/2. f/2.8, etc. up through f/8 or f/11. When I downloaded the pictures, I saw that the ones at the widest apertures were badly <em>underexposed. </em> The wider the aperture, the worse the underexposure.</p>

<p>So I went back the next day and re-shot the experiment, this time at ISO 200, and previewing and setting the exposure comp for a perfect histogram. Here are the settings needed:</p>

<p>f/2: +1.3<br /> f/2.8: +1<br /> f/4: +.7<br /> f/5.6: +.3<br /> F/8: +0<br /> F/11: +0</p>

<p>I then repeated the experiment with my 35mm f/2/8 PC-Nikkor and got the same result.</p>

<p>Then I tried my 35/2 AF-D Nikkor and got the same result, except the AF-D required rather less exposure comp.</p>

<p>Obviously the D-700 meter was overcorrecting for the additional light at wide apertures. But why?</p>

<p>I tried with both matrix metering and center weighted. Not much difference there.</p>

<p>Ever seen this one? Any ideas?</p>

<p>Was shutter speed max/min out, say, via auto ISO or the like? At the scene, why not use manual exposure to figure out the problem right then?</p>

<p>Just using sunny 16 (at EV16), the camera at f4 or more open, the SS would have been top over 1/8000...but then, it would have been over exposed, right?</p>

<p>Some random thoughts:<br /> Half underexposed is not as bad as double overexposed. You could argue that filling the histogram only to the middle just loses one bit out of the 10 or 12 or 14 density/exposure/ADC bits you have. No big deal.<br /> (Some theoretician at U.of Chicago has a pretty good web page about exposure and noise and all that; not bad to read at all. Not sure it will answer your question, but it might make you feel better about the underexposure. On the other hand, that guy still hasn't found superstrings, so maybe his photography page isn't all that great either. I keed I keed....)<br /> Second random thought: besides the normal inherent imprecision in exposure measurement and the slight (or larger) fluctuations that come with cloud coverage etc., there are also errors having to do with what exactly makes it into the scene. Those errors might be harder to quantify. A slight change in direction might exclude a few very bright pixels that change the (matrix) metering situation significantly. This 'systematic' error could be minimized by doing a test with a tripod.<br /> And third it could also be that it's not the metering but the stop down that's not exact. I don't know how exact AIS and better lenses are, I know that non-linear stop down is supposed to be a problem with AI lenses.<br /> <br /> I know that the viewfinder screen in my D70 doesn't really get darker until I stop down to f/4 or so. But that's the opposite end of the mirror box from the meter, and (supposedly) the texture of the screen, or lack thereof, is a perfectly good explanation for this. I'm only mentioning it because I've been told there could be some feedback from the viewfinder to the metering. Any special screen in your camera?</p>

<p>Well, let's hope that one of the actually knowledgeable people here answers soon.</p>

<p>The shutter speed was probably maxed out at f/2. It was 1/8000. But that wouldn't explain the progressive incremental 1/3 exposure reduction from f/8 down to f/2. I don't guess auto-ISO was enabled. I set 200 manually. I could check the camera, though.</p>

<p>Nope, auto-ISO is off. This camera must have 42,969 menu things to set up, though. I wonder what else could be set wrong?</p>

 

<blockquote>

<p>The shutter speed was probably maxed out at f/2. It was 1/8000. But that wouldn't explain the progressive incremental 1/3 exposure reduction from f/8 down to f/2.</p>

</blockquote>

<p>Bracketing enabled?</p>

 

<p>Bracketing. No, I don't think so. If it were, I guess I would be able to see all the bracketed shots in my download, no?</p>

<p>If you're taking one shot at a time, this could happen; changing lenses doesn't reset the bracketing. By default, bracketing is tied to the FUNC button, which is the lower button on the front of the camera near the lens. Press that and see whether it says "0F" on the top display. If not, bracketing is enabled; rotate the main control dial (the one on the back) until it says 0F then test again.</p>

<p>Maybe you didn't insert (or change) different (AIS) lens info i.e. largest aperture of each lenses into the d700?</p>

<p>Bracketing makes sense...</p>

<p>Leslie, that would only be an issue with the 35mm PC-Nikkor. The other two lenses are chipped. And I did have the correct non-CPU data entered and selected.</p>

<p>Mark, my FUNC button is set up for non-CPU lens data. Could the culprit be lurking somehow in that choice?</p>

<p>Leslie, that would only be an issue with the 35mm PC-Nikkor. The other two lenses are chipped. And I did have the correct non-CPU data entered and selected.</p>

<p>Mark, my FUNC button is set up for non-CPU lens data. Could the culprit be lurking somehow in that choice?</p>

<p>Leslie, that would only be an issue with the 35mm PC-Nikkor. The other two lenses are chipped. And I did have the correct non-CPU data entered and selected.</p>

<p>Mark, my FUNC button is set up for non-CPU lens data. Could the culprit be lurking somehow in that choice?</p>

<p>Leslie, that would only be an issue with the 35mm PC-Nikkor. The other two lenses are chipped. And I did have the correct non-CPU data entered and selected.</p>

<p>Mark, my FUNC button is set up for non-CPU lens data. Could the culprit be lurking somehow in that choice?</p>

<p>Leslie, that would only be an issue with the 35mm PC-Nikkor. The other two lenses are chipped. And I did have the correct non-CPU data entered and selected.</p>

<p>Mark, my FUNC button is set up for non-CPU lens data. Could the culprit be lurking somehow in that choice?</p>

<p>Leslie, that would only be an issue with the 35mm PC-Nikkor. The other two lenses are chipped. And I did have the correct non-CPU data entered and selected.</p>

<p>Mark, my FUNC button is set up for non-CPU lens data. Could the culprit be lurking somehow in that choice?</p>

<p>Mark: my FUNC button is set up for non-CPU lens data. I wonder if that is somehow related to the problem?</p>

<p>Leslie: I did have the non-CPU data selected: 35mm, f/2.8 for the PC-Nikkor. The other lenses have CPUs.</p>

<p>Rob, I'm not sure I should enter this debate about aperture inaccuracy, but you're not the first person to experience this phenomonen.</p>

<p>Repeat the test in manual, ONE stop down on the Aperture, ONE stop up in shutter speed. they should be identical exposures.</p>

<p>Can you borrow a modern G lens and run the same test. That would remove an chipped-or-not data errors and any aperture ring problems. If it still does it, it a metering issue. If it doesn't, it's a physical issue.</p>

<p>I think every lens I have used in my life show different densities wide open; they all require compensation to get the same exposure as stopped down. Some just a bit, others a little more.<br /> I don`t know why. I use to think that the nominal aperture could not be the real one. Maybe lens testing must be done under laboratory conditions. Sincerely, I never care so much about it.</p>

<p>Well, I guess that comparing the in camera jpg histograms at different apertures may not make sense. There are too many settings that can effect on the exposure.<br>

Method: use all manual settings: time aperture, ISO. Make comparisons then.<br>

Jose, Mike, I agree.</p>

<p>OK. I will repeat the experiment on manual later this week.</p>

<p>Rob, even if bracking isn't tied to your FUNC button, bracketing could still be your problem. I don't remember how to get there through the menus, but one way or another you should ensure that bracketing is disabled.</p>

<p>Rob, perhaps returning the body to 'default' settings and begining again with your own would eliminate some of the variables. </p>

I have a number of comments.

 

First, I'm not sure that I understand the objective of your experiment. Could you please state that?

 

Secondly, it's not a mid toned scene. White doors and clock, large shadow on dark pavement, reflective windows.

 

Auto exposure in some cameras is linked to the location of the selected focus point. Did that change?

 

1/8000 is at the limit of the shutter's capability. Systems don't perform optimally at their limits. It looks to be performing better closer to

the middle of the range of shutter speed values.

 

Auto exposure systems are somewhat random in my experience. Take ten shots in a row of the same scene in the same light, and there

will be some variation.

 

Did you set the aperture for your manual focus lenses in the menu system as recommended in the manual?

 

Histograms depend on settings such as contrast and saturation.

 

Auto exposure does not use the histogram, as the histogram is calculated after the shot, not before.

 

There's no such thing as a perfect histogram.