The scale of features shown on outcrop photographs can be critical to geoscience interpretations, yet little is known about how well individuals estimate scale in images. This study utilizes a visualization test in which participants were asked to estimate the absolute size of several boxes shown in outcrop images using high resolution, stitched photopanoramas (Gigapans). Participants viewed two different outcrops that highlight different kinds of photographic distortion, first using static images and then with "interactive" Gigapans that permitted zooming and panning. A test group was given basic scaling cues in the form of distance to and height of the outcrops, whereas a control group completed the test without any scaling cues. Other population comparisons were investigated (e.g., gender, age, experience level, and major) but no other statistically significant population difference was observed. Therefore, scaling cues seem to invoke a primary effect at least in the first part of the exercise. Results show that scaling cues increase accuracy overall, but with wider spread and a tendency to cause overestimation of size. The control group, which was not given any scaling information, was less accurate overall and tended to underestimate the size of features. Both groups gave more accurate scale estimates with smaller standard deviations for the extension-distorted photopanorama than the compression-distorted image. Participants also generally showed improved accuracy in the second part of the test, which probably reflects the impact of interactivity, although a training effect cannot be discounted. These results suggest that nonembedded scaling cues (as opposed to physical objects denoting scale in photographs) can be useful for some individuals to estimate the size of features shown in outcrop images. Results also underscore the importance of interactivity and multiple exposures in classroom applications.