Shadows backwards

Anyone know why shadows would be rendered backwards?

Your light is below the ground, put it above instead:

Below ground causes an inverse shadow?
Seems like there are other oddities also, like below??

There are some directions that make sense, like this:

But some not:

Well, when you use a direction that makes sense given the light position and the ground, the shadows should be the right ones, else you will get… What you get, which may depend on the light / the filtering method used :slight_smile:

For eg, I’m not sure what you expect from the last screenshot, but the shadow makes sense to me, it’s as if the rays of the light were going in both direction (which they do in reality), so the shadow is the right one for a direction which is the reverse of the arrows we can see in the yellow bulb.

I ran a test to confirm that case b is not as i would expect it. Unless the directional light is in some way different than my test case here (what would be a better light source to test?)? Is there a scenario in which the shadow would come back towards the light source if the light was going in the opposite direction?

Case A

Case B

ps. so sorry that Yoda doesn’t have a head. I really looked hard to find it for this test.

Torch is not a good way to test Directional Light, as its more of a Spotlight

I’m not sure what the “direction” sliders update, because I don’t have the same value when looking in the inspector:

If I invert the direction in the inspector, the shadow disappears:

1 Like

@Evgeni_Popov ah, that was b/c i wasn’t setting the initial values correctly (you would have needed to adjust all three values to have them sync properly. fixed that).

What i discovered is that the shadow seems to be wrapping! it’s going off one end of the ground and coming back on the opposite side. So there is a position where it’s gone as in your screenshot, but if you zoom out, you’ll see that it’s wrapping back around (in most directions - although there is one angle where it’s totally gone). Is that expected?

Here’s a case that doesn’t make sense at all, right?

It depends on the filter method. Only exponential / close exponential behave like that (I guess it’s a “by-product” of the method), other methods don’t produce shadows in this case in my testing.