Did you just read the first sentence? I don't know the D90. You just quoted a very small fraction of what I said. The most important is:
At least for the cameras I know of (Olympus E-500, E-3, Canon G10 and Nikon D300s), her statement is incorrect.
In what way is my message touching your D90?
As for evidence, well, there is no way to proof anything like that because what is seen in the display is only seen by me. I can only show the final result, the image, so I don't know what kind of evidence you are looking for. When is the last measurement taken according to you and how can
you proof she is right? Just because the D90 has inconsistent exposure it is no proof of anything other than inconsistent exposure. The camera can still make the measurement the way I describe and still miss the exposure.
C'mon, I'm kidding, and you're overreacting again
Your assertions are no more reliable and substantiated if you write them out in bf - it shows nothing but the desire for domination. For what I've read from you two and what I've seem myself I'd put a bit more trust in Marianne's. I don't care
who is right though, but I'm curious how things work, considering the exposure issues I've seen. So far, the aperture calibration seemed like a plausible excuse, but I'm open to other explanations.
I don't know what kind of evidence would be suitable, but you could try a more constructive approach, at least some more specific circumstances and numbers. Perhaps any docs for the metering sequence somewhere? Unlike with preflashes, events of meter readouts during the sequence should not be obvious to detect from outside.
There obviously are differences among your D300s and my D90, but I guess they are likely to use a similar metering procedure.
I've done some aperture "measurements" myself some time ago:
http://forums.dpreview.com/forums/readflat.asp?forum=1030&message=35099206
In figs 1 and 3, the same-exposure combinations should result in the same rhs edge position of the histogram, but they obviously do not. These are made in M mode, with the meter value only read-out. There is an odd thing that the meter value is off-set a notch at widest apertures with the Tamron, even though this meter readout is made with mirror down and lens wide-open (half-press only). So the meter appears to correct for the lack of aperture a bit in A mode, without having to take another meter reading with the aperture stopped-down. I've noticed the same odd behaviour with the 18-200VR, but couldn't see it with the 50/1.8D. But I admit this does not prove there is no second reading and correction yet.
I have the impression that the folks at Tamron are not fully aware of the metering procedure either
. After service the Tamron does not underexpose generally anymore - only wide-open, where it does not appear to get any more light than stopped down one or two notches. But at (nominal) F/2.8 and F/4 in A mode for portraits it now seems to produces the same brightness as the 50/1.8D at half the shutter time. I've been too lazy to cut&paste histograms, since I'm not sending it to service again. The underexposure in A mode wide-open remains; I assume such thing should get compensated with a longer shutter time under the second reading theory, which does not occur in practice. This is why I find your statement unlikely, as much as you find hers. ...Unless the compensation is limited, and the final exposure is based on both readings with some formula?
I'll try to check if I get the mismatch between the meter readout and the image exif that you mention one of these days. So far I've only noticed such a thing in LV (both with the VF covered or not).
