Testing your exposure metering

Started Nov 29, 2007 | Discussions thread
Shop cameras & lenses ▾
ForumParentFirstPrevious
Flat view
gollywop
gollywop Veteran Member • Posts: 7,999
Testing your exposure metering

There has clearly been a lot of whoopla of late on this forum regarding exposure metering in the D300. Many images have been posted as examples of over-exposure that have not actually been overexposure, but merely high on the tone curve.

The standard metering test that is suggested is to shoot a gray card and check the peak of your histogram. But what gray card? The standard standard is the so-called 18% reflectance gray card: shoot this card and check to see if the histogram peaks in the middle. If so, your exposure meter is working properly.

But, as many of you know, there is some controversy as to whether this should be an 18% gray card or a 12% gray card. At http://www.bythom.com/graycards.htm , Thom makes the strong case that the ANSI standard is 12% and that Nikon engineers (in Japan, though not necessarily in the USA) follow this standard. To test your meters, says Thom, you should use a 12% gray card.

So I took my 18% gray card and made three exposures, using matrix, center-weighted, and spot metering. All came out essentially the same, except a bit anomalous. All three showed the histogram peak about 1/2 EV high, except in my first shots, the red was measurably higher than the green and blue. Then I realized that I had a custom WB in play in d-0. So I redid the WB for the particular light I was using at the moment, and all three channels came out the same, all about 1/2 EV high of center.

I mention this WB issue only to point out that the histogram readings reflect the WB setting and do not really tell what is in the underlying RAW data. I wish there were a way to get a 3-color histogram reading for RAW that disregards the WB. If anyone knows how to do this, I would appreciate their chiming in.

So -- to continue. I don't have a 12% card, but I did a little on-line research to determine that the 4th gray on the GretagMacbeth card (Neutral 5) is a bit high at a Lab L of 51.61 and the 5th is a bit low at 35.96. Apparently 12% (actually 12.5%) would have an L of about 41. I just happened to find a piece of gray stock that looked to be roughly between these two shades of gray, and so must be near 12%. I redid my tests with this gray stock and, lo!, the histograms for all three color channels peaked very, very closely to the center.

I suspect, then, Thom is correct. The proper gray for testing your metering should be 12% reflectance, not the so-called standard of 18%. If you don't have a 12% gray card, it is determined that the 18% gray card will indeed produce a reading that is roughly 1/2 EV high. So shoot your 18% gray card, but expect the histogram to peak at about 1/2 EV.

I should also mention that the results were virtually identical for all three exposure modes: matrix, center-weighted, and spot.

What have others of you found?

gollywop

ForumParentFirstPrevious
Flat view
ForumParentFirstPrevious
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow