That's been my experience as well. I certainly find D70 under exposes - but only a little.

- Pradipta
Speed is significant and interesting but accuracy is downright fascinating!
My Nikonians Gallery

I'm not looking for a debate so I'll simply try to answer your questions.

The data I submitted shows a .2EV discrepancy that compares well with your "about 1/3 stop" comment.

You may have missed it but the test was done with Spot metering.
(But I have gotten the same results with the other metering modes.)

ADUs are the linear (raw) 0-4095 value out of the sensor.
Therefore, gamma etc. are not relevant.

Finally, FWIW, in addition to the "charts and figures", I too have used the camera for over 1 year and over 14,000 exposures; and I don't find that it under exposes.
Perhaps our techniques, or more likely our tastes, are different?

C'est la vie!
(Or more appropriate perhaps Viva la differance! )

Bill

Visit me, info and galleries at: Nikon Photos and Information

The D70 acts almost like Fuji Velvia slide film. The ISO rating is not quite right. But in both cases you can learn to live with it. I do not see any meter adjustment knob in my D70s as there is in my Minolta Spotmeter F. Maybe Nikon's techs have a hidden calibration knob or software value?

Gamma absolutely plays a role in this discussion. The metering of an 18% gray is only correct if that output image delivers the same perceived gray. The sensor doesn't perceive as the human eye does and without a profile that compensates for perception that raw data looks like a very badly underexposed and flat image.

You don't mention how you're measuring the output gray level and comparing it to the gray card reading. You may as well disclose the entire procedure if you want any kind of intelligent discussion...

In any case, I don't see how you can devise an single camera test and control for variable like lens diaphragm action. I have some Nikkors that consistently underexpose (the iris closes too far) compared to others. I have one with enough slop at the mount that there's a half-stop variation in exposure depending on which way you twist the lens. You may have structured your tests to deal with all these variables, but I'm guessing not.

Sorry for the delay. I'll answer most of your questions here.
I expect to make a new post to this thread shortly that I think will answer the others.
(And create new ones )

Our cameras could easily vary by 1/3 stop for meter calibration.
I doubt there's that much variation.

The metering of an 18% gray is only correct ...
I didn't meter an 18% target.
But I did assume that the meter would treat a uniform gray target as if it were 18% reflectance.

You don't mention how you're measuring the output gray level
I'm using the linear values straight from the raw data.
In my other tests I have found the sensor to be very linear.

In any case, I don't see how you can devise an single camera test and control for variable like lens diaphragm action.
I admit I only used one D70 so "your mileage might vary".
However, my opinion is that the sample variation is quite low.
FWIW, I did perform the test at least 4 times with 2 different lenses.
The results were consistent.

You may as well disclose the entire procedure
I thought my original post was pretty clear.
(Although I miscommunicated something if you and others think I was shooting an 18% target)

Never-the-less, here is a somewhat expanded recap:

I shot a computer screen, calibrated at 6800K, of a white target; varying the intensity from 0 to 255 in steps of 1.
200 of the 256 possible images were within the metering range given I was shooting manual at f/1.8 and 1/800s with a 50mm f/1.8D.
The lens was against the screen, focused at infinity; obviously I was using manual focus.
This type of image is called a "flat field" image because the field (or frame) is uniform (or flat).

FWIW, a large number of random shot would have been sufficient; this just seemed like a fun way to collect a wide range of data.

The meter will interpret what it is given as if it has an average reflectance of 18% and compute an EV accordingly.
This EV, in turn, controls exposure.
In this case the only "free" variable was ISO since I was using ISO Auto.

If ISO had a much finer granularity than 1/6EV then we would expect a straight horizontal line at 18% of the 100% ISO 200 value.
Since granularity is not that small we get a series of line segments that are constrained by 1/12EV above/below that line.

The linear value of 643 for 18% reflectance is consistent with a sensor that has a native sensitivity of ISO 174 (placing 100% reflectance at 3572).

Regards,
Bill

Visit me, info and galleries at: Nikon Photos and Information

Every photographer is a visual historian

Photography is an art, craft and science and when each contributes with the appropriate balance, one achieves an excellent picture. It is the photographer that will achieve this balance not the camera.

Unfortunately, there are a number of people who's motto is:

"Don't confuse me with the facts, I have made my mind up." H. Simpson

George

If I use a gray card and spot metering, for instance, for an indoor shot in an available lighting situation - perfect exposure (and perfect histogram) every time. So maybe the metering is OK after all?

Presumably under these conditions if the "mid gray point" is at the centre of the histogram then no pixels are 2 stops lighter or 2 stops darker than mid gray?

If I try to do the same excercise outdoors I very often get clipping in either the highlights or shadows or both indicating that the exposure latitude is greater than 4 stops? If I adjust the exposure so that the highlights are no longer clipped then I get a "darker" image which can be adjusted in PS, for instance, by use of the mid slider in levels.

This last situation is surely what people are "complaining" about? The metering very often gives you an exposure for the highlights which means that everything else has to move to the left - hence "dark" exposures.

Is what we need, NOT more accurate metering, but sensors with more exposure latitude? Maybe six or eight stops instead of the four or so that we have now?

DaveG

George

This article gives a good overview of http://www.acecam.com/magazine/gray-card.html The Tool for Better Exposures How to Use an 18% Gray Card .

A more technical article about http://www.cameraguild.com/technology/gray_card.htm|HOW TO EFFECTIVELY USE THE GRAY CARD>. This is for cinematographers, but should work since their films use the same basic principals of still camera films and digital cameras try to match those characteristics.

George