Hi there,

It seems to me like a disk filter would be more appropriate for simulating shadows from the sun, right? If you look carefully at the soft shadow region from sunlit objects you can see an odd halo which you dont see with a gaussian filter shadow.

I tried to simulate it quickly with a disk illuminated by a GI light and you can see some difference in there.

I think it might make a valuable difference for sunlit scenes to have it in there - anyone feel the same?

Thanks,
Jason

Is'nt that “halo” supposed to be a perceptual anomaly? A trick our minds play on us. Can't quite remember where I read something about it

You'll get a halo, or visible edge, any time the filter you're using has a slope discontinuity. The best way to see this is in the Ramp COP. Throw down a few ramp points, and change the color on them. If the global interpolation is linear, you'll see vertical lines at each of the points' locations - an illusion caused by our eyes' sensitivity to edges (which are in essense discontinuities). Change the interpolation to ‘ease-in ease-out’ and they disappear.

Since gaussian is nice and smooth, you won't get an edge, but a disk filter, with its sudden dropoff, will produce one.

You may be able get this effect in post, by using a color curve with a sudden dropoff (even a small one).

Don't get me wrong, I think I want the discontinuity the disk filter will bring