how sensitive is z buffer?

Greetings. I’m supposed to do a dental software project, and I’ll be handling plaster casts scanned in VRML. The thing is, I’m supposed to find out where the gum line is, and I’ve been thinking of ways to do it. Do you guys think using the Z buffer to test for sudden change in z values is a good way to detect the ridge between gums and teeth? I’m afraid it might not be too accurate. Or do you think there’s a better way? Thanks!

I would test for the orientation of normals…along a tooth, all normal vectors point in approximately the same direction, with very gradual changes. At the gumline, a normal may be turned by 30 degrees or more.

I think that what method you use will actually depend on the precision of your scanning equipment. You have to display the image as a cross section and then see what the characteristics of it are on your machine.