To compress such time series, therefore, we calculate a difference image that distinguishes each image from the one immediately preceding. Most of the values in the difference image can be expressed in one byte. For the few whose intensity difference exceeds one byte (absolute value > 127), we list the location and create an array of two byte differences. The final compressed file therefore contains a small descriptive header and the first time point, expressed as two byte integers. For each successive time point, we save an array containing the differences that exceed one byte, and an image array of the single byte differences for this time point. Figure 1 is a diagram of the file structure. Constructing the compressed file requires only addition and subtraction, and is thus exceeding fast on today’s computers.