Significance
Contrast sensitivity measurements using a computer display have been reported to lack accuracy when displaying small contrasts. This report investigates if the characterization / calibration of display luminance may contribute substantively to these kinds of described inaccuracies.
Purpose
To investigate what errors in contrast sensitivity may result from characterizing a display by fitting a gamma curve through physical or psychophysical luminance measurement data.
Methods
The luminance function of four different in-plane switching liquid crystal displays (IPS LCDs) have been measured for all 256 gray levels (the actual luminance function). This has been compared to a gamma-fitted luminance curve (the gamma luminance function). Calculated are the errors in displayed contrast that may arise when assuming the gamma luminance function instead of the actual luminance function.
Results
The amount of error differs considerably between the displays. In general, for large contrasts (Michelson logCS <1.2) the error is acceptable (<<0.15 log unit). However, for smaller contrasts (Michelson logCS >1.5) the error may become unacceptably high (>0.15 log unit).
Conclusions
To improve the accuracy of testing contrast sensitivity with an LCD display, it is necessary to fully characterize the display, measuring the luminance of each gray level instead of fitting a smooth gamma function through limited luminance data.
