Originally posted by dms Instead of saying the camera got it wrong--you got it wrong because you did not properly interpret the needs of the scene as you wanted to capture it. No offense please
Maybe you could go back to what I actually said - I prefaced everything with a statement that I made the
mistake of leaving the camera in multi-segment mode. Normally I would use centre-weighted. My post relates to why the multi-segment metering allowed the highlights to be blown out. If you don't want me to take offense, you could have addressed my question without referring to an error which I already admitted.
Quote: Although your posting is about the "inadequacy of matrix/multi-segment metering' in fact the metering as special K commented--is fine.
Maybe I should post a photo for comparison of another kangaroo in full shade. I disagree that the metering is fine. Looking at the image produced, the camera has exposed for the background not the subject. I'm not sure for that equates to the metering being "fine".
Quote: Actually the interpretation of the various segments exposure values is not so straightforward as to say that one proper exposure value follows. In fact the same scene can have many proper exposure settings--depending on the photographers choice--i.e., what he/she "wants to say."
Of course. The camera can't know what's in my head. At this point, I'm questioning the logic of the Pentax programmers who apparently decided that it's ok to let the highlights be blown out - even knowing that with a digital image, it means that the highlight detail is lost.
Quote: Actually the camera has some kind of logic it follows (in combining the various segments into a exposure value)--indeed one problem in using this metering mode--is the photographer is never sure what the camera logic is--and thus how to fix it (this is the reason may of us choose to not use this mode--as one cannot easily "accumulate the experience" to immediately/intuitively properly bias the exposure).
If indeed nobody knows the logic used by the camera in multi-segment metering, my question becomes moot and I'll simply avoid it. I'm trying to figure out what the logic may be, obviously not precisely but at least to get a better idea of it. In the example I posted, the exposure chosen by the algorithm seems to have exposed correctly for everything in the background, while allowing the foreground in the middle to be blown out and overexposed. The thing which shocked me wasn't that the middle was overexposed, but that it was blown out and detail was lost.
On pages 117 and 118 of the manual, the only specific difference mentioned with respect to the multi-segment metering "interpreting the scene" is backlighting. The manual specifically states that centre-weighted metering doesn't compensate for backlighting. That actually suggests the opposite of what I imagined for backlight compensation - it appears that the camera will underexpose the shadows so that it doesn't overexpose the highlights. If that's the case, and you had a person in shadow in front of a bright background, the multi-segment metering will probably result in a silhouette. Ordinarily, I would think that "backlight compensation" meant correcty exposing the subject and ignoring the backlight. I'll have to try it and see. I've used a combination of multipoint AF and multipoint metering on a Nikon (oh no, I mentioned the "N" word) which somehow correctly identified the person in the photo, correctly focussed on them and exposed for them.
Quote: The point is you must look at the scene and choose what to meter and how to adjust the meter reading you get. Since the camera uses reflected light (and not incident light) it starts of at a huge disadvantage--it doesn't even know what the exposure is to make a middle gray look middle gray! This is what you need to tell it--along with what end of the exposure range you want to emphasize or ignore.
I'm perfectly well aware that I have options like the use of the exposure lock, centre weighted and spot exposures. Please focus on the subject. If the intention of multi-segment metering really is simply a multi-segment average, and the metering algorithm doesn't care what detail is lost in the highlights and shadows, its usefulness seems questionable unless it's based on an assumption that each scene doesn't contain a wide range of brightness. It may be useful on a dull day, but not in the full glare of the afternoon Australian sun. It could have been far more useful if the exposure chosen
limited over-exposure in a situation where the middle of the image was bright and the borders were dark. In the reverse situation it would make more sense to allow the highlights to be blown out to avoid a dark subject with a bright background being lost in the shadows.