First bug (missing shadows for nearby parts)
My first bug to report is that surfaces very close to the light (less than [or equal to in the negative direction] 0.25) do not cast any shadows.
Observe both the “less than” case, and the “less than or equal to” case here:
The +X axis is from left to right in this image. As you can see, the brick on the left has its surface exactly 0.25 studs away from the center of the point light. So does the brick on the right. Only the brick on the right casts a shadow when its surface is exactly 0.25 studs away from the light, while the brick on the left does not.
When moving the brick on the left any further (even by 0.0001 studs, via the properties window) away, it casts a shadow. I think that 0.25 is an exact constant somewhere in the system, and this is probably a limit that can be decreased.
Second bug (shadows leak light)
My second bug to report is that near the light, shadows are behind their surfaces by approximately 0.9 studs in all cases. (It could possibly be 1 stud, with some feathering at the edges, though)
This causes light leakage through any part thinner than 0.9 studs.
This part is semitransparent but still casts a shadow. You can see how its shadow only appears about 0.9 studs behind the surface that is supposed to be blocking light. The part is exactly 0.9 studs thick (left to right in that image).
Not only does this cause light leakage through thin walls:
But it also causes shadows to not meet up with the corners of parts, which looks pretty weird in my opinion.
This bug used to be more obvious, and the transition was instant:
It is less obvious now, but in my opinion shadows should still meet up with the corners of parts. And light should not leak like this.
The light leaking is a little quirky in that it’s uneven. Here I widened that thin wall:
Now you can see thin streaks of light that extend out.
I believe this is a clash between two cases, since the wall in the negative direction looks like this:
Indeed, it looks like we are getting two effects at once. When the wall is closer than 0.25, we get this:
When the wall is further:
And now when the wall is exactly 0.25 like before:
The fading effect of the first case is might be overlaying over the second case, for some reason… I don’t know what it is, or how to explain it, but something is happening here and it is definitely a bug.
I would recommend making this 0.25 threshold smaller, after fixing this problem.
Why I put these both in the same bug report
Because they both reference and involve each other heavily and I believe they’re both symptoms of similar issues. They are both annoying, they both have to do with shadows being Weird, and they both seem to have something to do with a 0.25 threshold.
Here is a place file for the last gif.
FiB_0.25_threshold.rbxl (20.8 KB)