Hewpiedawg

Here color and position are one object: a five-vector (r, g, b, x, y) mapped into itself. One transform rotates the (x, r) and (y, g) planes; another folds position together with luminance. The frame averages those maps with the seed sampled at the new (x, y).