Archived posting to the Leica Users Group, 2002/03/26
[Author Prev] [Author Next] [Thread Prev] [Thread Next] [Author Index] [Topic Index] [Home] [Search]On 3/26/02 12:28 PM, "Michael Abbott" <lists@mabot.com> wrote: >> Which is 3 and a bit (sorry don't have slide rule handy) > > it's 3.6. For n bits your formula gives > > TDR = log10(2^n) = n log10(2) = 0.3 n > > This seems to work for the few scanners I know of. Do you know if it is a > universal practice? I always assumed they worked out the number of bits > needed after managing to make a device with whatever dynamic range. You trusting soul, you. In every case I know of, this is how the Dmax is derived. > I'm confused about what your formula means. It says that th. dyn. range in > *stops* = number of bits. No, that isn't true. You can't talk about 'dynamic range in stops'. Dynamic range factors in the tonal resolution you require. For example, you could measure a one-stop difference on film (an incredibly low contrast negative, say) with a dynamic range of (say) 3.6, ie 12 bits of data. That would give you a very fine tonal gradation of 4096 steps for that one stop. Similarly you could digitise a normal negative with say 10 stops of range using just 4 bits, which would result in horrible posterisation but hey. 256 shades of gray (8 bits) is usually considered sufficient to be perceived as continuous tone, so theoretically the maximum Dmax any scanner requires is just under 3. This would however require you to get the scan absolutely perfectly right on the first pass. If you don't agree with this consider the fact that *all* images output to consumer printers are reduced to 8 bits before being printed. The reason we want more detail from scanners is because of the possibilities that opens up in terms of image manipulation. You can bring out your shadows, bring down highlights, etc etc, without posterisation. Dmax is often thought of as a scanners ability to 'see into the shadows' of slides. And this is in a sense true since you can crank up the exposure in the scanner further and still capture detail in both highlights and shadows. However ultimately (at the moment) it all ends up as 8 bits when output to screen or page, so there is no inherent advantage in capturing at a higher bit depth, except for image processing. Finally, the other important factor is gamma. This determines how the image data is represented by the available bits. Do you give all brightness values equal weight (gamma=1) or do you weight the representation towards the highlights (gamma>1) or towards the shadows (gamma<1)? Human eyes are supposed to be better at discriminating brightness values in the highlights than the shadows, which is why we generally use gamma values of 1.8-2.2, so that the tonal gradation in the highlights is finer than in the shadows. There is a school of thought that says this is heresy and we are all going to hell in a handbasket and our shadows are posterized if we did but know it, but that's an argument for another place (and count me out). - -- John Brownlow http://www.pinkheadedbug.com - -- To unsubscribe, see http://mejac.palo-alto.ca.us/leica-users/unsub.html