Using measurement to improve the 10HT

The diagram above shows a typical chart representing the measured light output of the 10HT as a function of input level or IRE value. The measurement was made using the various IRE windows on the Avia test disk and photocell-based light detector in front of the screen. Details on building a similar detector are included in the section Build the measurement device.
Even thought we often speak of gray scale linearity, the output of the projector is clearly not linear, nor should it be. All video displays need what is called gamma correction. Gamma defines the exponent in a the power function that relates input voltage (IRE levels) and output levels. Gamma corrections exists for historical reasons involving the nature of CRT displays, and accurate gamma correction is essential to good video reproduction. Most of the images we watch are not entirely black or white, but consist of midtone levels somewhere in between. The gamma correction defines how light or dark the various mid-tones levels will be reproduced. The Avia Disk info has good information about this, but basically the ideal gamma for a display is 2.2. Too low a gamma produces a washed out image, too high a gamma a very "punchy" or contrasty looking image. At first, a high gamma might sound like exactly what people are looking for, but the downside is that the mid-tone levels are reproduced at a lower level than intended and the whole picture may look a bit dim. A gamma of 2.2 gives the intended distribution of intensity. In the above diagram, I have also shown the theoretical curve, labeled 2.2, for a gamma of 2.2. With most reasonable settings, the 10HT seems to exhibit a gamma fairly close to 2.2, but one can see deviations, if things are not properly set.

If we were dealing with a black and white display then our job would be done when we had achieved the desired gamma correction. With a color display however, we also want to achieve a proper balance of colors at the various IRE levels. This is what setting up a custom color temperature is all about. For this purpose we need to measure the intensity of the various colors separately.

The three curves in the diagram above therefore show a typical graph used with SMART in which measured light intensities for the separate colors, Red. Green and Blue are displayed. This chart is taken from a measurement made by one of the first people to try this method. (Thanks Dana!) Ideally all three traces would be identical and thus maintain the same ratio of colors at all IRE levels. In this case the tracking of the three colors is not too bad. To the extent that there is too much of any color at essentually all IRE levels, this can be farily easily fixed by adjusting the gain setting approproiate for that color in a custom color temperature. There is an even more straight forward way of adjusting the bias settings.

The analysis spreadsheet also displays color balance as the ratios of the various colors as shown above. The idea in this display is to get all of the traces close to 1.0 as possible. This is the same precalibration data as in the previous graph, and one can see pretty good color uniformity, except at the lower IRE levels. Optimizing the bias settings will improve this situation. One could also consider calculating actual "color temperatures" or xy positions on the CIE color map, and I may at some time, but at this point it would seem to be making too much of the data. The raw color intensity plots seem to contain all the information and so that is what I tend to use.