In computer image generation (CIG) "spatial filtering" refers to the combining of tonal information from scene features inside and in the vicinity of a pixel to form the video for that pixel. Several investigators have recently proposed improved filters, validating their choices with pictures of sensitive test scenes. It can readily be shown that filters which produce the best static scenes generate serious artifacts when applied to dynamic field-rate update CIG. The investigations in the literature have not explored this topic. When some necessary conditions imposed by the temporal effects of interlace-scan systems are applied to the algorithms, the differences between simple filters and the more complex filters become quite minor, even on the static test scenes. On CIG training scenes, designed to simulate the real world, the differences in results of a variety of filters become imperceptible.