As I spent the weekend playing lens/camera comparison I noticed a couple of fairly consistent trends:
1) The 16-45 does tend to underexpose, even in P
2) I get different exposures (EV, not just a different shutter/aperture selection) when shooting the same scene in P, Av or Tv. With the two AF lenses (43 ltd, 16-45) it's usually less exposure when out of P. With MF lenses (A 28/2.8, Tamron SP 35-80) there isn't a large difference. The MF lenses tended to expose more 'correctly' than the AF did, in Av and Tv.
I noticed this fairly consistently when I used the 43/K100D as light meter for my Diacord, across many conditions. I verified it with a handful of shots. This was consistent whether I shot RAW or jpeg.
3) I use a Gossen LunaLux, in incident mode. When using the reading on the 43/K100D in manual exposure, the result is dark but very saturated. Not unpleasant, but it tells me - assuming the LunaLux is accurate (it did confirm the sunny 16) - there's something else going on in the camera/lens interface.
Some results are here:
lens_testing Photo Gallery by jussi at pbase.com
The exposure info for the bookcase shots (all camera metered in Av f6.3):
A 28: 1/2s f/6.3 iso400
16-45: 1/15s f/6.3 at 28.0mm iso400
43mm: 1/25s f/1.9 at 43.0mm iso400
35-80: 1/10s f/6.3 iso400
the 16-45 in pgm mode: 1/50s f/4.0 at 45.0mm iso400
I realise there may be other things going on, such as the metering pattern changing between lenses or modes (something I'll have to take a look at), but what's going on? Anyone else notice these patterns?