Top left: raw image including overscan/bias strips. Top center: overscan removed, synthetic objects implanted. Top right: reprojected and background rescaled (median). Bottom left: bright pixels masked (masked pixels here are shown as having a value of 0 but in fact have Not a Number in the data array). Bottom center: template subtracted, using a “donut” template as described in Section 5 with a donut-hole radius of 40 images and a thickness of 10 images. Bottom right: Source Extractor background estimator subtracted. Note the changes in color bar scales. In the last image, most residuals have <2 counts per pixel. The image shown here is image 440 of the sequence and was deliberately chosen to demonstrate the advantage of also using the Source Extractor background estimator for eliminating background structure; for most images, the background already appears flat before the Source Extractor background subtraction. Zoomed-in versions of some of these panels can be seen in Figure 2. Pixel values are in electrons per second. The dynamic range of the color scale was chosen using the “z-scale” method commonly used for astronomical images (D. Tody 1986, 1993) for each panel individually.