As the solar eclipse of August 21st 2017 approached, I was thinking about the effect it might have on shadows.  After all, the size and shape of a light source has a significant effect on shadow quality, and the size and shape of the sun was going to change (apparently) from the occlusion of the moon.

What would happen, exactly, I wondered.

Well, the smaller (or more distant) a light source, the sharper the shadows become.  The larger (or closer) a light source, the softer shadows become.  But the sun wasn’t going to get larger or smaller, exactly.  The sun was actually going to remain the same height, but vary in width.

As you can see in the image, the sun starts as a disk and gets narrower, like a curved line.  So the sun will get smaller on the horizontal axis, but not the vertical axis.  This suggests that shadows will become sharper on the horizontal axis, but not the vertical axis.

I decided to test this in 3D using a rectangular area light.  Here is what a shadow looks like when the height and width of the light source are the same:

We can see a normal, proportional penumbra (area of partial shadow) and umbra (area of total shadow) behind the cylinder.  Now here is what a shadow looks like with a light shape that is disproportional, like a line.  I have narrowed the area light across its horizontal axis, similar to what will happen to the sun in the eclipse:

Notice how the shadows remain soft along the vertical axis of the shadow (with distance from the cylinder) but are much sharper across-wise.

Inversely if I shorten the light on its vertical axis but leave it wide on the horizontal axis, the shadows are sharp on the vertical axis and remain soft on the horizontal axis:

The last test I could do is to create a light of approximately the same shape as the crescent sun at about 90% totality.  This should simulate what the real eclipse will look like:

So what will it look like when the sun gets thin on one axis and stays wide on the other?  To answer this question, I set up our Black Magic production camera in time lapse mode and filmed the video you see below.  Elegant proof.

We can clearly see that as the sun reaches 90% coverage, it gets darker and the shadows become sharper, but only along one axis, just like in the experiments.  Look at the shadow cast by a fluorescent tube some day.  It’s a similar effect.