Self shadowing & bump bug?

[fxnut]

Just wanted to raise the profile of this thread that's running at the moment (so hopefully Newtek see it and investigate):

http://vbulletin.newtek.com/showthread.php?s=&threadid=10793

It appears that if you have a raytraced light and turn the bump percentage up, you end up with some very odd problems with self shadowing. It's even more pronounced when you go above 100% to the extent that the self shadow disappears.

This ONLY happens if it is a continuous mesh, i.e. you'd be able to select the whole thing by selecting one polygon and doing a 'select connected'.

If you're curious, try making a ball, giving it a fractal noise bump map and turning the bump up to say 700%. Now render it with a raytraced spotlight lighting it from the side. The ball will look as if it's being uniformly lit - no shadows. (see below)

[Matt]

isn't it just the bumpy surface is catching more light because bumpy surfaces are inherently more diffuse?

[fxnut]

I know what you're getting at, but surely there's no way a surface should be being lit if it's pointing directly away from the light source?

[geoff3dnz]

Originally posted by fxnut
I know what you're getting at, but surely there's no way a surface should be being lit if it's pointing directly away from the light source? Maybe, because bump mapping's a hack anyway, there is a way ;)

[automan25]

I too have run into this problem. Hack or not, bump mapping is an essential tool in this business. If it isn't working correctly, then it needs to be fixed.