Extended Data Fig. 9: Shape model assessments.

(a)–(e) show results from limb/terminator shape model assessments. Panels (a)–(c) show an example of (a) a reference DRACO image, (b) the rendered shape model with the same lighting and viewing geometry as the reference image, and (c) the difference between the model and the reference image. (d) and (e) show results from limb/terminator assessments from many DRACO images. (d) Sum of the absolute value of the image-model differences, normalized by the image perimeter. The median is the most relevant measure of uncertainty from this metric because the distribution is always one-sided and never gaussian. (e) The differences in the radii of the equivalent-area circles for the reference image and rendered shape model, respectively. The radius of the equivalent-area circle is the radius of a circle with the same area as the total area of lit terrain on either the rendered model or the reference image. The mean is the most relevant measure of uncertainty from this metric because the distribution should be symmetric. (f)–(i) show results from keypoint assessments. The colored lines in panels (f) and (g) connect features matched by the algorithm in the image and on the shape model. Most, but not all, matches are reasonable, so the median value based on all keypoints is used. (h) shows a metric derived from differences between corresponding keypoints across several tens of DRACO images. (i) shows a model-to-image scale factor derived by comparing the distances measured between all keypoints in the reference DRACO image to the distances measured between all keypoints in images of the rendered shape model. The arrows in panels (a) and (g) indicate the direction of Dimorphos north (+Z).