We see flow all around us. Liquids and gases flow according to some physical laws, heat flows as well, and then there’s also our brain flowing at certain points of time. But does light flow?

It does, in a sense. Though it’s not light itself flowing, the perception of movement in a given frame based on the paths followed by its individual elements provides a sense of flow and as an observer one can feel it by looking at consecutive frames. I came across this concept just a while ago, and it seemed quite fascinating - not only as a matter of academic interest but also how widely it is used for observing things all around us, especially for object tracking and path prediction. It’s being used in a project I’m working on at the moment, and I find it really intriguing how this simple idea of just noting differences in related objects in consecutive frames is applied to gather information about the paths followed by them (the math behind it is pretty interesting as well), and how well algorithms like this one actually perform in action.

Naturally, it is an image processing concept, and hence it is computationally heavy but that’s expected - after all, images are mostly humongous matrices no matter what kind of compression/standard is used. But GPUs are intended to be used specifically for such applications, and it’s amazing how the involved calculations are performed at the rate they are (processors typically run at GHz speeds, so that’s unfathomable to our consciously thinking brains, even though we may practically be doing the same thing albeit in a subconscious manner).

Amazing how all of it fits together and works.