On 29 January 2012 12:36, Larry Colen <l...@red4est.com> wrote: > > On Jan 28, 2012, at 7:59 PM, Paul Stenquist wrote: > >> >> On Jan 28, 2012, at 10:31 PM, Larry Colen wrote: >> >>> >>> >>> On 1/28/2012 7:07 PM, Paul Stenquist wrote: >>>> >>>> On Jan 28, 2012, at 9:49 PM, Larry Colen wrote: >>>> >>>>> >>>>> >>>>> On 1/28/2012 6:29 PM, Paul Stenquist wrote: >>>>>> You can "expose to the right" or anywhere you choose by using exposure >>>>>> comp in any metering mode. The metering isn't based on jpeg or RAW. It's >>>>>> based on the light and what's in front of the lens. No meter is smarter >>>>>> than a photographer who understands how meters work. >>>>> >>>>> I'm certain that if I'm wrong someone will correct me. The metering mode >>>>> in our cameras picks a spot to meter on, and sets the exposure for that >>>>> point at midpoint. This means that if you look at the histogram, you >>>>> usually get a bell curve right around the middle of the graph, expose to >>>>> the middle. This means that if you go direct from RAW to JPEG without >>>>> any compensation in post processing, most of the pixels in the photo will >>>>> be right around the midpoint of exposure. >>>> >>>> No. In multi mode, the meter uses a program to analyze the scene and tries >>>> to achieve a balance of highs and lows. If you don't like the histogram >>>> that results, you can move it right or left with exposure comp. You only >>>> get a bell curve in the middle when you have an average scene without >>>> extreme highs or lwows. >>> >>> Interesting. Then why is it that if you photograph something like a white >>> table, or snow, using normal metering, it comes out grey rather than white? >> >> Because the meter is dumb. > > That is exactly my point. There's no need for a meter to be dumb when our > cameras have more processing power than supercomputers of not that many years > ago. > >> It figures that if there's just one color, then it's midrange. However, >> modern meters do a better job than the older ones. The K5 only misses by >> about a stop. The older center weighted meters or averaging meters missed by >> close to two stops. If I shoot a snow scene with the K-5, I usually give it >> about plus one stop of exposure comp. > > > >>> >>> So, what you are telling me is that the metering in our cameras is >>> optimized to give the best performance when shooting in raw mode, rather >>> than in jpeg? >> >> No, it's not optimized for either mode in particular. It's just a dumb >> meter. It measures light and tries to guess what the scene looks like based >> on its firmware. >>> >>> Or, that unlike in film where you'd meter differently for negatives and >>> slides, there is no difference in metering for getting the best exposure >>> out of jpegs and out of raw? >> >> You only meter a bit differently for negatives and slides because of the >> processing. A slightly overexposed negative can still be printed rather >> nicely with more exposure in the enlarger, but an overexposed slide is junk. >> When my processing was set up right, I usually exposed about the same for BW >> film and transparencies -- or slides if you wish. >>> >>> >>>> >>>>> >>>>> What it does not do is look at the pixels out at the tail end of that >>>>> graph. If a bunch of them are off to the right, and you expose for the >>>>> middle, then you end up clipping on a lot of your readings, in other >>>>> words, you'll lose highlight detail. >>>> >>>> Then you bring that back in by dialing in negative exposure comp. >>>>> >>>>> Alternatively, if most of the readings are to the left of the point that >>>>> is metered for, then exposing for the middle will leave you with either a >>>>> lot of pixels that are clipped black, or a lot of your shadow detail lost >>>>> in the noise. >>>> >>>> Then you dial in positive exposure comp. Simple. >>> >>> Why not have a mode in the camera that does it automatically? Give me the >>> source code for the K-5 and I could probably implement it in a week. >> >> You'd have to invent a meter that could tell the difference between gray and >> white. Today's meters just read light levels then compare them to firmware >> that tries to predict shat part of the scene is sky, what's grass, what's a >> face, etc. They don't really know what color things might be or how much >> light they are capable of reflecting. But the human brain and the human eye >> can make that call with precision. So determining how much exposure comp you >> need based on your own intelligence is the best way, and it will probably >> remain the best method for many years to come. > > Or, you could have the meter read all of the metering points, take a look at > the brightest, and dimmest, as well as the focusing point, apply a few > thousand cpu cycles and come out with something a lot better than what we've > got. > > Or, you could have a special mode that uses the sensor as a light meter, when > you'd rather let the camera spend a couple of seconds making the corrections, > rather than going through a few test shots yourself. > > Or, at the very least, as Pentax, you could describe the actual algorithms > used so that the photographer wouldn't have to guess what the camera is going > to do.
Or the photographer could just learn how to meter a scene. I'm very glad that the 1st exercise on the first day of my photographic education was learning how a light meter worked & to interpret the scene with that knowledge. -- PDML Pentax-Discuss Mail List PDML@pdml.net http://pdml.net/mailman/listinfo/pdml_pdml.net to UNSUBSCRIBE from the PDML, please visit the link directly above and follow the directions.