Shape

Here is an exploration of shape perception and the contribution of luminance, again with voiceover, another video letter to colleague. The point is that explicit shapes can be seen quite well even when defined by equiluminous patterns — here texture is used to demonstrate recovery of 3D structure from motion at equiluminance. But only if the shape is defined by a contiguous region — no missing, interpolated, or amodally completed parts please. If the shape has to be integrated across different surface qualities, it only works when luminance is present. The examples in this case include a rotating paper clip and a randomly checkered square moving smoothly across a similar background.

Luminance, what is it good for?