Translucent Shadows with Directional Light

I am having some trouble with a translucent shadow test and I hoped someone here might be able to help me. I have managed to make something similar to the stained window described [here][1]. I managed to make it work well enough with a spot light (though it is a bit resource intensive to build the lights).

So I thought I would try it out with a directional light. Using the standard settings for a directional light however, resulted in a pretty blurry image.

The article I linked earlier does warn us about the light source angle of the light. Thinking this was the issue, I changed it to 0, but it doesn’t seen to have had much effect.

Neither does the issue seem to be the lightmap resolution. The plane on which the light is being projected in those images had its lightmap resolution overridden to 1K. Still, I tried doubling it and this is what I got:

Which seems pretty much the same to me. Am I missing something about how the translucent shadows are supposed to work? My glass window was made using a simple plane with a two-sided, unlit modulating material. It just connects a texture to the emissive input, multiplied by an “intensity” number, to make the projection more clear.

By the way, I tried rebuilding the lights with a source angle of 5 to see if it would make the image significantly more blurry. The result was not much different either:

The result is slightly more blurry, so it is not like the angle is not having an effect.

Thank you for reading this question through!

By the way, here is an screencap of a spot light with the same stained glass window. This is more or less what I expected with the directional light experiment.