No, that's not the right way to think about it. Read what I wrote below carefully.
The sensor has an absolute dynamic range. Exposure settings establish the environment in which that dynamic range operates electically, essentially setting the gain "window" over which the sensor can capture intensity values. You can adjust what HAS been captured within certain boundaries mathematically after the fact, which is what you're doing when you modify the gamma correction curve in a RAW converter, but you cannot change what has been captured. That can only be adjusted by exposure at the time of capture. An HDR technique of taking multiple exposures at different settings than merging them together allows the sensor's absolute dynamic range to be extended, synthesizing a larger dynamic range through repeated and different 'windows' on the subject intensities. Godfrey On Nov 4, 2006, at 1:56 PM, Tom C wrote: > What you say makes sense excpet when I think of it like this: > > I was thinking of the data captured by the sensor as basically a > bitmap. If > all adjusting gain (up/down) on the sensor effectively does, is to > make an > individual pixel, lighter or darker than it would have been > otherwise, then > it *seems* that the same thing could be done post-capture, sans- > sensor. > > So is my thinking basically correct in principle, but not > necessarially so > in practice? > >> In simple terms: >> >> - Making one exposure and than adjusting it once out of the camera >> always locks you into whatever happens to be the maximum analog >> dynamic range of the sensor. If elements of a scene fall outside that >> dynamic range, you get black/noise or total saturation, no matter how >> much adjustability a RAW converter might have or how much data >> recovery it can do. >> >> - Making a set of exposures at different exposure settings and then >> integrating them together allows you to window the scene with a >> dynamic range wider than what the sensor can acquire in one exposure. -- PDML Pentax-Discuss Mail List [email protected] http://pdml.net/mailman/listinfo/pdml_pdml.net
