i have found some time today to take some more shots with my smc-m 50/1.4, in _daylight_
. the behaviour is very consistent, and it confirms the previous breakdown posted here: some underexposure at 1.4, pretty much square at 2.8, allready one stop over at 5.6, by 8 it goes to almost two stops, and from there, up to 22 or so, it seems to stay "together" (about 2 stops over). note this was an informal test, and i had a hard time, as the light was changing very fast (cloud, sun, etc), a really great day for shooting, and bad day for being in the office
i am sad (it does kind of take the jam out of the cake, for using the smc-m), but i am even more intrigued. i cannot figure out what might be the cause. this happens consistently on various degrees of lighting (so the determining factor is somewhere inside, from the aperture blades going in
), but how? the way i see it, the camera will probably internally settle on a fixed "lensindicated aperture" (so as not to change the metering algorithm too much), and act as if everything is as usual, with a normal (modern) lens, reacting to the aperture closing as if it was a variation in ambient light. if this was it, the exposure should be spot on, as it allways was for me with both the k100d and now the k10d. so, either, uppon stoping down, the camera, for some reason, changes the "assumption" of indicated aperture (i could accept this might be a bug, but how can it tell how much the lens was stopped down, so it can decide how much to err? -- damn, it sounds sick
), or something else happens when stopping down. which brings me to the focusing screen: remeber those good-old split-screen mf focusing screens?, and how they become erratic (dark, but in various positions, depending on how you put your eye on the vf) when you stop down? maybe the metering cell (i will assume just one, for centerweighted) has the optics in front of it alligned such that a similar effect would occur (perhaps it is designed to be influenced less by "out of focus light", i need to test this by shooting completely oof, that might be interesting), the fact that osme people have had great luck with switching the focusing screen seems to enforce that idea, allthough only slightly, as i, for one, am not sure how the metering system is designed, and to what extent it should be influenced by the focus screen. and then comes the infor from katz eye, which state quite significant variations in exposure reading for their custom focusing screens on the k10d..
i think this must be addressed, somehow. right now, the only thing that comes to mind, is to try to research "A-converting" my smc-m, for that i would need to understand exactly how it works, though (for instance, if it is just a variable electrical resistance which the camera reads, shouldn't be so hard to do. if there's some sort of "chip", i am afraid i would be lost..), the idea being to just enable wide-open metering (no control from camera to lens, too much trouble). as things are looking now, i cannot really see how a fix in firmware would be doable, as the camera has no way of knowing what the f stop on the lens is (well, actually it does have a way, but not very reliable: on press of the green button, take a reading, stop down, take another reading, it is trivial to compute how much the lens is stopped down, but without knowing the wide-open aperture, impossible to know the actual aperture set. but than again, if the reading is wrong on stop-down, one would end up with the same problem..)
i will download my card, empty it, and quickly try to shoot out of focus, see what happens. btw: thank you all for your oppinions and relevant feedback so far, makes me proud to be here