booleans (2)

[holken]

ok now i'm fed up with theese stinking booleans that NEVER work!
i've tried deleting history.. using cleanup and all of that stuff... still does'nt work..

so..can someone please tell me how to make a cylinder shaped hole on a cube without using boolean?! i mean.. there's got to be another way.. right?

[blank-]

Hi,
You could project a circle on the cube & use trim surface. Then loft the circular isoparms after the trim that will create the inner surface. Where as Booleans are concerned it works well on a one piece object .

Good luck

[joojaa]

When doing booleans you have to be very carefull that you know exactly what you are doing, firstoff make sure the normals are the right way around (if even one is the worng way youll be screwed, use normals conform). Make sure no non manifols surfaces (some of the non manifolds get fixed by the provious step) exsist etc...
And offcourse this is for poly booleans only... Nurbs booleans are a total other ballgame...

Also make sure there isnt any coplanar surfaces anywhare (eg if you try to subtract a 1 lengt sylinder rom a 1*1*1 cube so that the ends are exactly on the same spot youll be at the mercy of the processors floating point accuracy)

Also cleanup isnt necceserily enough to make the sufficent changes... Since it doesent take into acount that you might have a number of different shells on the surface in question this confuses maya very much (atleast in poly booleans), so make sure your vertices are actualy merged. Same goes to Nurbs except maya allows you to do this as long as you use the group(s) (a nurbs cube is a group of 6 surfaces), instead of individual surfs, but be avare that this can be a mathematical horror at time s so not everything is doable.

Multible booleans many times mess things up a bit.

But i must say once i understud how they work only NURBS booleans some times cause some probs for me, anymore... but maybe thats expected since even very hardcore solid apps (like Catia, ideas and Pro/E) fail t the same nurbs stuff so....

Oh by the way i tried using a primitive cube and deleted some 50 different poly cylinders from it (with history intact), and they all worked fine...

[deke-kincaid]

Booleans have the same problems in every app. When I started on 3ds4, I was always bitching at kinetix. Then as I went to use other apps I realized that it is a constant in the universe. Booleans suck everywhere.

[joojaa]

Well if you want to modell in boleans id suggest you start using Pro/e or its like (eingeneers usualy sit on top of thees progs so dont go bongers whan you get a file with 2000000 trims) they hardly ever fail. But then again the entire prog is built around the concept of boolens, such apps are called solid modellers, and they excell at this but hey thats what they are built for (they consider surfaces to have a real in and outside so no such case where extra or wrong facing surfaces might come to pass in normal sircumstances, and this is what causes the probs). surfece modellers just cant cut it. But they have some other problems.

But if you are interested i just revised my list of things to check when booleans fail.

here goes:

Check that all vertices ar connected properly (use merge vertices with some very small amount). This is the hardest one to detect, even cleanup wont help you on this one, in fact in some instaces it creates this problem, so its better for it to show yoy the problem areas instead of fixig them. Also multiple shels and outside of bounds is the same problem a bit differently tought.

Make sure the normals are uniform and consistent (conform normals), also something that might be hard to detect is that if it encounters two out edges without passing a in edge it will fail. This is the reason a closed surface will work better, asuming it dont have rogue caces haging around inside it.

Avoid co-planar suraces (either in object or in relation to boolens), to avoid numeric inaccuracy.

Rest of the problems are actualy just derivates of theese three. And conditions 1 thru 2 dont happen in solid modellers, and they use higher accuracy... which mean the rarely fail.

Drerivates include:
[list=1]

No non manifold geometry allowed (2)

You end up outside a cutting plane (same cause as 1 but a special occurence)

fails but if ju nundge a bit it works (3)... Numeric accuracy is the hardest contender to cope with unless you can just make some distinct moves.