Archived posting to the Leica Users Group, 2006/09/19
[Author Prev] [Author Next] [Thread Prev] [Thread Next] [Author Index] [Topic Index] [Home] [Search]On Sep 19, 2006, at 4:49 PM, Henning Wulff wrote: > At 4:18 PM -0400 9/19/06, telyt@earthlink.net wrote: >> B. D. Colen <bd@bdcolenphoto.com> wrote: >> >>> And of course, if one routinely underexposes by a stop, one is not >>> only >>> avoiding overexposure, one is also discarding about 20 percent of >>> all the >>> information captured by the sensor. ;-) >> >> ??? can you explain ??? >> >> I typically set the DMR to underexpose 1/2 to 1 stop (RAW files) and I >> rarely if ever see clipping at either end of the histogram. >> >> Doug Herr >> Birdman of Sacramento >> http://www.wildlightphoto.com > > This being digital, it generally works like this: > > Depending on how the chip/firmware are set up, the first (brightest) > stop or stop and a half have half the total bits of information in the > file, the next darker stop to stop and a half have a quarter of the > bits, etc. The darkest segment, where things start blending into > noise, holds only a fraction of a percent of the info. > wugh away half of your bits. Fortunately, if you start out with a 12 > or 13 useable bit depth, that's not that much problem because even > with fairly drastic maneuvering in Photoshop you won't get noticeable > banding in the dar areas, and besides, they're _dark_ areas, where a > bit of banding doesn't show up as much. > > No point in talking about some cameras having a 16 or even 14 bit > depth depth; true, useable bit depths really don't extend beyond 12; > possibly 13. Not even DMR's :-). Do you have empirical or experimental evidence of this? > > The moral is: for highest quality try to expose so your histogram is > as far to the right (bright end) as possible, and yet clip only > specular highlights. Make sure you're looking at an RGB histogram, > because in the common luminance histogram one colour could be clipping > without you knowing it, causing serious, non-recoverable colour shifts > in highlights. > > In practice, I generally set exposure compensation to -1/3, and keep > an eye on the RGB histogram. > In my experience with the DMR, setting exposure comp to -.5 or -1 sacrifices very little deep shadow detail if any. It's not what I'd call a 20% loss of data. The histogram (yes it's RGB) is nowhere near clipping at either end aside from specular highlights. Perhaps B.D.'s 20% estimate is based on his experience with the E-1? Doug Herr Birdman of Sacramento http://www.wildlightphoto.com