To investigate the influence of the shadow caster on the shadow image, we are looking for a gemetric shape composed as the average of all our silhouettes. This task can be approached either at the pixel level or at the vector level.
The pixel level function works by adding the gray values of all the images and then dividing by the number of silhouettes. The resulting pixel image looks like this:
The vector plane function works by finding the center of each of two vector points in 2D space. For two simple vector images the procedure looks like this:
Since our vectors do not have a uniform number of anchor points (i.e. there is not necessarily an equivalent for each anchor point), we calculated the average of two shapes with a similar number of anchor points. As a brief insight, the workflow looks like this:
In the tournament system we always calculated the average of the average of the average.
In the end, the result is a shape that is clearly different from the pixel version: