It was a sunny summer day. I was walking to the bus stop when a rainbow, not in the clean sky, but on the gray asphalt of the road. I stopped to look better, and it almost disappeared. It was best visible in motion.
After a closer examination, it appeared that road workers were updating road surface marking and spilled some glass spheres, used to add retro-reflectivity to the marking. White sugar-like thing on the photo above is made of these small glass spheres. And just like water drops create rainbow in the sky, glass drops made rainbow on the road.
I wanted to make a better photo, but big contrast between the white glass "sand" and dark asphalt made rainbow hardly visible on a still photos. So I decided to try to average-out the noise, using the fact that rainbow observed position does not move when observer moves.
It's a common way to deal with noise in measurements: make many measurements and then average them. Since noise is random, it will become become weaker, as the number of measurements grows. (Noise amplitude is inverse proportional to the square root of the measurements number).
So I made 15 photos, moving a little between frames, and averaged them.
Here it is! There is something unsettling in this picture… I intentionally made all photos in the same pose to use the shadow for aligning images, and on the merged photo black shadow makes strange view on the blurred road.