acf@pinet.aip.org (Alex Feinman) (10/05/90)
Does anyone know the relationship connecting the value that the computer stores in RAM and sends to the monitor, and the actual intensity of the light emitted by the monitor? It is obviously non-linear, but then the eye senses light in a non-linear fashion. Is it the function of the Dto A converter in the monitor, or the phosphors, or what? Thank you in advance
jbm@eos.UUCP (Jeffrey Mulligan) (10/06/90)
acf@pinet.aip.org (Alex Feinman) writes: >Does anyone know the relationship connecting the value that the >computer stores in RAM and sends to the monitor, and the actual intensity of >the light emitted by the monitor? A power function with an exponent between 2 and 3 is usually a good approximation. > It is obviously non-linear, but then the >eye senses light in a non-linear fashion. The visual response is irrelevant to the question you posed, although it does bear on the design of the "gamma" of the system. >Is it the function of the Dto A >converter in the monitor, or the phosphors, or what? My best guess is that the DAC's are nominally linear (i.e. they're normal DAC's); note that the DAC's are not usually in the monitor, but rather on the graphics card. Another guess/folklore is that the phosphors respond more or less linearly to beam current, and that the nonlinearity comes in the relation between grid voltages and currents. There are several ways of estimating these paramters for a monitor without using a photometer which involve making visual matches. For example, You can make a mixture of the top level (white) and the low level (black) either by flickering in time, or by coloring alternate scan lines. (You could also dither along a scan line, but many monitors have additional nonlinear artifacts that make this problematical.) Then pick the level for an adjacent uniform field that produces a brightness match. This is then the level that corresponds to 1/2 the max intensity. By repeating this procedure, you can sketch out enough points on the curve to do a decent curve fit. It is also a good way to check that you have chosen the correct exponent if you just go ahead and use a curve to start with. -- Jeff Mulligan (jbm@eos.arc.nasa.gov) NASA/Ames Research Ctr., Mail Stop 262-2, Moffet Field CA, 94035 (415) 604-3745