I've been searching this forum but in vain, at least so far. My question (no.1) is: how does k10d decide to change the ISO when set to AUTO let's say 100-400? In other words, can I set the minimum speed or is there a criterion at which k10d will switch to higher ISO? I've noticed that though in AUTO ISO 100-400 mode, it didn't switch to 200 even at speeds of 1/45 or so. Other conditions: Av mode, AF and NR on.
And just curious question: does anybody here have any experience with Tokina 28-70/2.8 ATX-pro and k10d lens? To me it seems surprisingly less sharp than the 18-55. Perhaps because it is recognized as 28-70/3.5-4.5? Does this affect something?

Well I've had the same feeling when I bought the camera. Then I got used to it and I do MUCH more exposure correction than I was used to with MZ-5n shooting slides. By default I don't correct the exposure metering but quite often the highlights are blown so I correct it. In average more to underexposure than to overexposure.

1.30

I was playing with the same config at noon and I'm not sure what to think about it: everything powered up, flash in HS/P-TTL, Av mode but the speed stays at 1/180s, ISO is not changing automatically though it was set to, no under/overexposure indication. Then I switched to Tv, the same, speed could not be increased above 1/180s. But then, either because of changing to TAv and back to Av, it started to react correctly, i.e. changing of aperture and corresponding change in shutter speed. The same for Tv. When I find some time I'll play with it. Perhaps tonight, at my daughter's first birthday

Well I've tried some fill-in shots yesterday and it turned out quite well except for one thing (K10d, Tokina 28-70/2.8 and 360FGZ): flash set to HS, camera to Av. The shutter speed remained at 1/180s even though I changed the aperture. ISO set at 100, no automatic. No indication of overexposure or underexposure. Both camera and the flash switched to HS (well the flash was just 'showing' HS) when I used Tv and set manually the shutter speed above 1/180s. So Av was not working the same way as without the flash.

Yes, I tried to play with the exposure correction and it seems that I'll have to use it much more often than with the slides. Basically pushing it to longer exposure than the multi segment metering is suggesting - but not too far to get the highlights saturated (in most cases) nor to have no dark areas. Of course there are exceptions when one wants to have either dark or bright areas saturated.
Anyway, considerably more time will be necessary for image processing...

Yesterday I finally saw the 'blown highlights' warning - the corresponding parts of the image (sky at sundown) flashing red.

I also understood one more thing: if there is a big difference in exposure values for different parts of the image (sky at sundown at 1/640s and dark trees at 1/50s both with f/2. then the exposure must be corrected so the sky is not white. But this leads to the trees becoming very dark. I don't remember having similar problems with slides... do they have higher dynamic range that k10's sensor (fuji sensia 100)?

Well I'll need to have a closer look at the ETTR theory calculations...

About the 'underexposed' photos: they're not underexposed, they just look underexposed when looking at less than 50% of its size. Though there were some 'very dark' areas on the image (with no exposure correction) the sky showed shadows and was not completely white nor grey (rainy day). I'll try to post the photos to my picasaweb in PEF format.

Definitely there is a lot to learn.

I'm using multisegment metering, EV steps set to 1/3 too and PEF+JPEGs... Lots of new things to learn with the digital!

Thanks for your help and explanations. I'll try the kit lens though I don't think there will be substantial difference in exposure under the same conditions.

One secondary question about over/underexposure with digital SLRs: overexposed photo is then not containing any extractable info in white (saturated) areas. But what about the underexposed areas that are all black? Can some information be revealed from them rather than from the white ones?