CompuServe Thread

Mike: If you get really stuck I could try running the boolean operation in my solid modeller, GDS. I am testing Syndesis Interchange and I think your problem mesh could be a good test for the program as well. If you want me to give it a try e-mail me the mesh as it is and zip it up with a description of how thick you want the crust and other pertinent details. – Bob Ritger

>>punching out the earth… Without seeing your mesh it is difficult to guess the problem, but if both meshes have topological integrity, the booleans should work fine. A few points to check for… Are the land-mass meshes completely "solid", i.e. back side faces with no holes? (booleans won't work consistently otherwise) Are all face normals facing outwards? With surfrev, did you check for integrity at the poles, welding if necessary? Are you trying to boolean too many things at once? Do you have any overlapping elements in either boolean object or do you have any long skinny faces that need to be fixed? I've found that r3/4 booleans almost never need tessellation to work correctly. It may help to simplify your land-mass meshs if they are overly complex. If you have created your land-mass objects by detaching faces from an existing globe, you can create solid objects by using Create/Face/Copy then Create/Face/Extrude. Flip the normals on the back side then attach and weld the extruded faces. (Don't try this with continents that curve around the earth too much – break them into smaller segments) You may even get the effect you want by selecting faces while still attached to the globe and extruding them inward, deleting them, then attaching a scaled down, normal-flipped copy to the inside if you need a solid inside. BTW, you can probably use a simple sphere primitive, or rather two spheres. I've used this method to hollow out several things with "windows" to the surface: union window plugs the to a central core, then subtract that object from the outer shell. If you mesh is not too big to email I would be glad to look at it to see if I spot anything – my kind of fun. <g> JKJ

Have you checked the axis of the revsurf for degenerate faces? These are faces that run from pole to pole and have only 2 vertices. This often occurs along with vertices that don't _quite_ meet at the poles. That is a common problem with revsurfs and can really screw up a boolean. I don't really understand why you cannot use a normal sphere. What do lights have to do with it? Terry Gilbert Forcade & Associates

The techniques described by JKJ have hit the nail on the head for this project. I found that when I took the original sphere and did a scale/copy to around 90% to create the inner core, booleans would not work. But, when I created a new sphere with less geometry and its poles were 90 degrees out from the outer sphere…things punched perfectly! Booleans seem to be a lot of trial and error. Looks like I am on my way to creating this unusual model. Thanks to all that helped. Mike