PTTL - how its done??

OK, almost hate to bring this one up again but I've been playing around a bit with the 540 flash on my K10D and still having trouble getting proper exposures. I think I've narrowed down the types of shots that cause the confusion but need some clarification/advice, or at least a shoulder to cry on...

I've taken dozens of shots even with bright specular hilights all over the image. Even with things like chrome objects I can get decent images, but put a non-glare person or dog with a tiny specular highlight and the photo is heading for the bin.

OK... so no matter whether I have the exposure setting to spot, center or full screen (whatever the name), the effect is nearly identical.

What I seem to have found is that PTTL metering is somehow done within the inner more-squared rectangle area of the view finder.
All brightness within that area is averaged or otherwise evaluated regardless of other settings.
This really becomes a major problem for high contrast scenes more that for scenes with consistent luminosity.

So... if I have a dark dog laying on a dark carpet (to use my previous examples), and a sliver of white wall creeps into that rectangle the photo is lost.
If I am able to adjust the angle such that the wall fall just a hair outside this rectangle the image is vastly different than when it breaks that line.

Of course being that the sensor isn't square, the possibility to do this and how your image results will varies whether the contrasting area is top/bottom v.s. the sides. (Do I move the frame 15% or 30%.


I suppose it could be that the entire frame is always involved, but the averaging crosses some thresh-hold as the contrasting area breaks those boundaries.

I guess I am rather confused why spot metering doesn't reduce this to any degree. Is PTTL metering of the frame done that differently than metering for a non-flash photo?

Does that make sense?
Am I making sense?
If so, is that behavior normal and can anyone explain why?

Sorry for rambling on. Long day at work, tried to relax with photos, but only ended up frustrated debugging my camera rather than getting shots I wanted.
Hmm, guess that is part of what photography is all about anyway, eh? ...experimenting until you get it right? Guess I'll shut up now and live with it.

Anyway, thanks in advance if anyone can bring any further info on how this works and how to create more consistent photos.

Shots of dark objects will always cause a problem with metering.

The metering aims to get the subject exposed at middle grey - so a full frame of black will end up grey, while a full frame of white will end up grey as well!!
No camera can tell the difference between black that's brightly lit and white that's dimly lit

You always have to think just a little about the subject and use exposure compensation.
(All cameras will do this )

Understandably that's difficult when shooting ad-hoc, so your best bet is shoot RAW and adjust exposure after the event. When I shoot outdoor subjects with fill in flash the P-TTL works pretty well all the time.

However, if on the other hand you are shooting in a known environment you will find it much easier to switch to fully manual - both manual flash and manual camera. Usually takes me 3 or 4 shots to get the exposure right, then every one after that is bang on. There's a thread or two I've posted on with the "how to"...

AFAIK the metering mode does not affect the P-TTL metering. From what I can tell it's evaluative (the "full screen" always.

But I'll give your comments some thought and see if there's anything else I can add....

Matt

Thanks. I am aware that the frame will always be evaluated to middle grey.
But that is partly what surprises me so much by the results I'm seeing.
I just do not see enough white to appear to make up the difference. Especially when the delta between two frames is a hair's difference of white and it causes a two stop difference in metering.

Anyway, nice to confirm that PTTL uses the full frame for metering. I can work from that and hopefully learn to better predict what will happen.

I do have to agree, in a studio type shoot, I almost always use manual.
1. no pre-flash to disturb the subjects -- thus no half shut eyes.
2. more consistent exposure due to movement that might cause difference in metering.


Thanks

The other thing that may be worth thinking about is wavelengths...

I'm not sure what the spectrum output of the flash is, nor what the sensitivity of the light metering electronics is and how that is affected by the spectrum light. Perhaps the camera light metering is seeing UV or IR being reflected that the CCD does not detect?

It would be interesting to know how your little test changes when you bounce the light...
Matt

Does anyone know if it work correctly with the *ist DL it says it should...