CompuServe Thread

The command "object unify" will not work if the only some of the faces in an object are in wrong normal direction. The solution is too use "face flip" to flip each face individually, When you perform a series of boolean operation on one object, such as a series of holes, grooves or whatever, the sequence of picking the object for boolean operation is tricky. It is always better to do those from the inside first and work outwards. If you still have problem at the final object, then you need fix the meshes by "edges\turn". Most of the time, you do not need to rely on other software.

Don, there's a big difference between the way ACAD (and almost every other program that does booleans) uses a CSG system to determine the boolean result, and the way 3DS uses an arbitrary face-topology system. Using deterministic CSG-based primitives, it's possible to always get deterministic results. But using arbitrary manifold topologies, it's impossible to be mathematically precise. We really try our best, however. – G

Don, Booleans as has been noted can get complex. And like you I've had problems with it but if its important enough I can usually get it to work. The dilemma is that to be able to do a good boolean you need a lot of faces. If you have a lot of faces conversely there is the potential to have many of them reversed. Craig's point about Optimize is a good one, but it also requires some planning which can't always account for all the problems you are likely to encounter. One thing which is really helpful is to assign a material to each object which is going to be booleaned. Then once the boolean is accomplished you can dismantle it based on faces assigned by material. This is useful for breaking it down into a more manageable size for clean-up. Not a solution per se but hopefully a few useful tips. 🙂 John

<<I, too, have been having a lot of problems with boolean and flipped normals. What I don't understand is why object unify doesn't seem to correct this. Because of this "bug" I tend to avoid using boolean operations altogether.>> Object Unify does not always fix the problem, not becuase of a bug, but simply because of the position of the face in question. The algorithim will look at a certain face, and based on its position and relation to the bounding box and center point, will determine whether it is facing inward or outwards. <<Also, I think it is a poor policy to suggest that the solution to a bug like this is to use ACAD. If boolean is in 3DS then lets just fix it so it works>> The boolean functionality between ACAD and 3DS are very different. The 3DS modeler was not designed as a full function Boolean editor. It is a question of using the right tool for the right job. If you need to create complex booleaned objects, you need to use a full function boolean editor. -Brian

Don, I can sympathize with your effects to do complex booleans in 3DS. It can be frustrating. I sometimes have to resort to using AutoCAD for some things (where possible). Maybe some background on the differences between AutoCAD and 3DS booleans will help you to decide when to use which program. There is a major difference between how programs like AutoCAD do booleans and how 3DS does booleans. This effects the chances of successfully generating the new surface. AutoCAD will only do boolean operations on objects created from solid primatives and simple extrusions. 3DS will do booleans on any objects (or at least attempt to). When doing booleans with solids, AutoCAD has an exact knowledge of the form of the original pieces and can easily predict the boolean result. Solids have know forms, volumes, and definate insides and outsided. They always have fully closed surfaces. AutoCAD cannot do booleans on arbitrary surfaces. When it comes time to put a mesh surface on booleaned solids, AutoCAD will seperatly generate meshes for each surface. The vertices for each surface may not line up and may not be connected. You will have to manually move and weld vertices. 3DS must analyze and attempt to find the surfaces of completely unkown forms and must figure out where the inside and the outside of the surfaces are. Sometimes it doesn't get it right. Several things can complicate the issue. Objects in 3DS don't have to be closed. They can be deliberately open or there can simply be inadvertant gaps in the mesh. This can confuse the routine that determines where the insides and outsides of a form are. Objects with multiple elements can also have this effect, expecially if the elements overlap. Any overlapping faces in the mesh will also confuse the boolean operation. Long thin faces are probably a problem because of rounding errors. Unify/Normals has similar problems with complex or damaged mesh surfaces. It will handle faces with mixed up faces. That is what it is designed for, but if there is a gap in the mesh or overlapping faces or multiple elements, then Unify/Normals will also get confused. 3DS's boolean function, when it fails, may create gaps or overlapping faces. If you do a subsequent boolean, it will compound the problem and Unify/Normals will not help. When you do a complex boolean in 3DS it is often best to do it in stages, one element at a time, checking the mesh carefully between steps for errors. Save a new version of the file between each step. This lets you go back to any given step and start over. You want to do this because you may not catch an error until some later step. I do hope that Yost Group will develop a smarter boolean function for R5 that has less tendancy to trash a mesh when it fails. I doubt that there will be a revolutionary improvement because of the difficulty of doing booleans on arbitrary mesh surfaces, but who knows. Terry Gilbert Forcade & Associates