Originally posted by RiceHigh I think you're the one who has got the whole point!
But, the official samples are all taken with a K-7 with firmware 1.0! Anyone interested can inspect the EXIF!
@RiceHigh:
Thanks for having a careful eye on the exposure topic. I see your reason for being cautious. As for the four images, I had an examination as well. So, let's share our results:
RESULTS:
All four images: K-7 firmware 1.0. JPG filters and shadow processing unknown.
Img #1: Matrix, +0.3EV, histogram slightly clipped in the highlights, with some room in the blacks. Would have been "perfect" at 0EV. But +0.3EV gives it more "popp".
Img #2: Center-weighted, +0.7EV, histogram has some room on both sides. Would have been perfect at +1EV.
Img #3: Matrix, +0.7EV, histogram minimally clipped on both sides. A perfect "to the right" exposure would probably have been at +0.3 or 0EV, with shadows recoverable in Raw. But +0.7EV gives it more "popp".
Img #4: Center-weighted, +0.7EV, histogram has some room on both sides. A perfect "to the right" exposure would probably have been at +1.3EV.
Let's have a closer look at Img #4 as it seems the most "underexposed". There are 77 metering zone read-outs. Unfortunately, PhotoMe only display the first 16 of them, thinking it is a traditional Pentax meter
These first 16 readings (probably the upper edge) is between 8.8 and 9.3 EV. The effective EV of the exposure is 11EV (incl. +0.7EV compensation -- would be 10.3 without).
So, the K-7 "metered" about 1 and 1.5 EV "brighter" than the image's upper edge, i.e., a lot brighter than the center. But as we know, it should have metered 3 EV brighter
Not much more we can say here before we know all 77 metering values.
So, what must be our verdict here:
For Spot Metering:
The test would have been if a 18% grey card would have come out 11% grey (with the sRGB profile, I believe, the right number was 11%). No test case supplied in the samples. This is an industry norm. A vendor has no freedom here (But some vendors are known to break the standard for non-pro cameras here. So far, Pentax (D)SLRs have always adhered to the standard, though.
For Center-weighted metering:
Two test cases supplied (incl. Img #4). Manufacturers have more freedom in interpreting metering values. However, there is only a single metering value available: the weighted average over zones. Here, it doesn't make any significant difference if the average is over 16 or 77 values. Almost identical result. If the average metering would be interpreted as 18% grey, most images would be heavily underexposed (think of snow
) So, most vendors go for some arbitrary figure higher than 18%. If you know this figure for various vendors, you can cross-compare vendors by adjusting EV compensation for the respective target luminance. Pentax is known to be more conservative than some other vendors here which, in the first place, is a good thing. No final conclusion for center-weighted metering can be reached because the center meter readings are not known.
For Matrix metering:
Two test cases supplied. Manufacturers here can combine position and value of all metering values, not just a single value. E.g., they could guess the scene type, or use the variance between zones to guess the 0.01% clip-off barrier, etc. This is the only mode yet of interest with the K-7. Both supplied samples would have been absolutely perfectly exposed without EV compensation applied.
Final verdict:
The exposure in the 4 samples leaves nothing to be desired, from what can be told from available data.
RH, do you agree?