The behaviour of digital sun-sensors and associated super-resolution algorithms was explored. Using calibration data, a method was proposed to model the peak width of peaks across the image array. Using this with the non-linear least square algorithm gave improved performance across the field-of-view. A test was proposed that would measure precision for small sensor motions. Also, a method of accounting for local bias error was given. The small motion test defined limits at which the sensor detects motion, and the precision test gave metrics to measure how well the sensor renders motion. Finally, an extended kalman filter was developed that used sun-vector measurements, in addition to a new relative measurement. This was tested using a well-defined sensor as well as a generic sensor for which few error data were known. Results indicate that relative measurements only improve performance if random noise is low.