I've noticed this particular frame for a couple of years now, off and on, and finally got around to photographing it a week or two ago. I might have shot it before, but I can't recall.
This is a purely random welter of twigs, wet from the rain, and yet the appearance of circles, of orbits, is quite definite. I had to think about it a little to work it out.
From the perspective we're looking at these twigs, the only surfaces which reliably reflect toward our eye are those oriented to the streetlight in a particular way. The only surfaces we "see" then, are the ones that follow the orbit. The rest are lost in darkness.
There is another tree I observe while walking the dog, with another welter of random twigs. This welter is lit by a security light on a building, which casts shadows on the tree's fairly white trunk. I don't have a photo of this one, it's not very photogenic despite being technically interesting.
The shadows cast on the tree trunk are entirely horizontal. Perfect straight horizontal lines.
This took me somewhat longer to work out. The security light is in fact a horizontal row of LED lights. It is "big" in the horizontal direction and therefore, as it were, wraps around vertical twigs and eliminates their shadows. The light is "small" vertically, and so does not do the same to horizontal twigs. It is, in effect, filtering out verticality. Mathematically, there's probably a convolution somewhere in here.
This led me to think a bit about strip lights. We often see them talked about as a way to create interesting catchlights in portraits (a "cat's-eye" effect, with a vertical strip light on each side) or to rimlight things, or whatever. I don't think I've ever noticed a talk about how a strip light acts like a big light
in one dimension, and a small light in the other.
In theory, you should get hard shadow edges parallel to the strip, and soft shadow edges perpendicular to the strip.
This is a preliminary and not very successful attempt to illustrate the effect. I think my faked strip light isn't very good. Don't even ask. I rather like the picture, though.
Work in our labs proceeds.
Further lab work produces these two photos, which make the point.