Originally posted by simico Why don't you just meter with the lens wide open and then do you calculation (in an exposure calculator app or in your mind) for the desired f-stop?
Mostly because the meter is at its worst behavior wide open.
Like Lowell and dlacouture, I have done a fair amount of research and experimentation on this subject. I have also done exposure series similar to Lowell's, though with a little different methodology. (Used a A-contact lens meter readings as the gold standard and plotted the variance from those readings for each f/stop. No need for Photoshop.) I did all of the lenses in my collection that have an aperture ring and came to the following conclusions with the K10D:
- Severe underexposure at apertures wider than about f/4
- Fairly good linearity down to about f/11
- Moderate to severe underexposure narrower than f/11
- Actual curves varied by lens
- Behavior was identical with the stock screen, Katz Eye w/ Optibrite, and Katz Eye w/o Optibrite
From what I can tell, metering is closely matched to the viewfinder light environment in wide-open mode and is NOT based on a simple meter measurement. Correction factors are applied based on the lens maximum aperture as communicated through the A-contacts to account for the non-linearity of light passed by the focus screen. If the maximum aperture is unknown (M42 and non-A-contact lens), a default correction is applied. That default factor is simply wrong when shoot at other than moderate apertures.
The issue is compounded somewhat when the light is dim and the aperture narrow. Meter sensitivity is at fault here. EV 0 is easily attained at the sensor when ambient light is dim and the aperture is small. This is normally not a problem when metering wide open, but is historically an issue for stop-down metering in general. Cameras used to come with a chart that showed the range of acceptable shutter speed/aperture combinations at ASA 100 (ISO now days).
As expected, my tests is relatively low light showed a flat meter response at narrower apertures. The sobering thing was that the meter is not linear to EV0, but begins to lose sensitivity at about EV 2.
I would post the test plots, but I recently did some house cleaning and purged them from the hard drive.
All this being said:
The above issue is typical of the K10D, K20D, and other Pentax dSLRs of similar vintage. The other models have a pentamirror system so the behavior is a little different, but the general principles are the same.
On recent Pentax dSLRs the situation is much better. Starting with the K-7, Pentax implemented a linear meter program for M-mode when aperture information is not present. I have yet to upgrade from my K10D, but a quick test with the K-7 when it was first introduced indicated linear meter response in M-mode using stop-down metering. This is the procedure recommended in the camera manual. If you use other modes (e.g. Av) YMMV.
Steve
(BTW...I shoot mostly with vintage and non-A glass and generally either test/chimp/shoot or use a hand-held meter with those lenses.)
(BTW, BTW...Pentax is not the only camera brand with stop-down metering issues.)