PDA

View Full Version : Boolean FAILS (8.5)



DrStrik9
12-16-2008, 12:25 PM
I know, I know, upgrade to 9.x ... I will when I have the dough ...

So this morning, I'm modeling a VERY simple extruded letter thing ... (Mac Pro w/ quad core Intel Xeon), I imported an eps from Illustrator ... and using boolean, or solid drill, worked perfectly ... ONCE ... but after that, it refuses to work, no matter WHAT I try ... I've seen the need to reboot Modeler regularly with this Intel processor, but that doesn't help today, and even with a total system RESTART, All the boolean-esque functions DO NOT WORK. Period.

Any ideas? (besides upgrading, that is) ... :-)

Thanks,
Michael

P.S. Does 9.5 also run in Rosetta on the Intel-based Mac? (If so, I'll wait till it runs natively to upgrade ... this is a TOTAL drag.)

Fadlabi
12-16-2008, 01:32 PM
some times I have problem like this, especailly Rail Extrude, I have work a round by copy my obects and open a new obeject file and paste in it, and it work well.
try a boolean on a new object file, it may help.

DrStrik9
12-16-2008, 02:16 PM
some times I have problem like this, especailly Rail Extrude, I have work a round by copy my obects and open a new obeject file and paste in it, and it work well.
try a boolean on a new object file, it may help.

I wish that DID help, but it doesn't. :-(

I created a new object in Modeler 8.5, copied only the single polygon I want a hole in to the top layer of the new object, copied the polygon I want as the hole in the first poly to the second layer, put the main poly in the foreground, and put the "hole" poly in the background.

Boolean subtract doesn't work, none of the Solid Drill functions work at all, and even Template Drill fails. :-(

Any other ideas?

Michael

plandrey
12-16-2008, 02:24 PM
Try merging points first... don't know why but sometimes the eps imports have duplicate points -- messes up the booleans and drills.

Cheers!

DrStrik9
12-16-2008, 02:26 PM
When trying Boolean, Modeler does NOTHING.

When trying Solid Drill, Modeler does NOTHING.

When attempting to use Template Drill, I put the primary poly in the foreground, and the cutting poly in the background, and choose the proper axis (Z), and ANY of the options (Core, Tunnel, Stencil, Slice), and Modeler chugs away for a couple of seconds, then I get the error that says, "Unable to regenerate contour. (This may be caused by highly non-planar polygons.)"

But there are NO non-planar polygons in this document: both polys (one in the foreground and one in the background) are simple, flat shapes.

:-(((

Is there no way to get 8.5 Modeler to work on an Intel Mac?

DrStrik9
12-16-2008, 02:30 PM
Try merging points first... don't know why but sometimes the eps imports have duplicate points -- messes up the booleans and drills.

Cheers!

Ooh! GOOD IDEA! Thanks ...

... naahhh, no points were eliminated by merging either polygon, and nothing in Modeler works at all to accomplish any boolean, solid-drill or template-drill functions.

:-(

Any other ideas?

DrStrik9
12-16-2008, 02:38 PM
I've also tried extruding both polys, in an attempt to use Boolean subtract (which should not be necessary) ... but this also fails to work -- it acts as though I didn't even select Boolean at all!

Interestingly, in the middle of all this failure, I tried a simple Template-Drill:Slice, with a 2-point poly in the background, and this WORKS (I just don't need it.) :-(

I've rebooted Modeler, and restarted the entire computer also, but nothing helps, so far ...

Any other suggestions? (And thanks a TON for all the suggestions so far!) :-) I just wish even ONE of them worked. :-/

plandrey
12-16-2008, 02:49 PM
Here are the steps I take when nothing works.

1. Merge Points (m)
2. Unify Polys (shift - I)
3. Open poly stats and delete 1 & 2 point polys. (w)

Sometimes it works... I wish I could see what you're working on... see if this helps.

DrStrik9
12-16-2008, 02:54 PM
OK, I killed all the polys and painstakingly rebuilt them, point-by-point, in order to build what Boolean or Solid Drill should have done. :-/

In the process of doing this, it seems that there is some kind of point limit in Modeler 8.5. After accomplishing this on the most complex poly in this simple project, all the other less-complex drilling worked FINE.

The polygon that could not be booleaned or drilled has a total of 528 points. Why would THAT (very low?) number of points STOP all booleans and Drill functions from working?

Giacomo99
12-16-2008, 03:52 PM
Unless you actually want to post the object, it's impossible to tell what the problem is. I guarantee if you post the object for scrutiny, someone here will be able to tell you why it's not working.

I will say for sure that it has nothing to do with the number of points. Objects created by extruding an EPS outline aren't automatically "well-formed." ("Well-formed" basically means that the form describes a "clean" volume—among other necessary conditions, a "well-formed" surface has no coplanar surfaces and no non-planar polys—that Boolean operations can be performed on with no ambiguity.) Boolean operations often fail when attempted on non-well-formed-objects. Many other 3D apps will give an error message— "object is not well-formed," but Lightwave doesn't.

Also: Boolean operations are designed to work on volumes, not surfaces. It's not a generally good idea to use them on objects (i.e, complex polygons) without a distinct volume.

DrStrik9
12-16-2008, 04:05 PM
Yup, I've used Modeler for long enough to know all the merge points, unify polys, get rid of any single- and double-point polys, etc. tricks to make clean geometry. Yes, Boolean is intended to work with volume, which is why I made sure the geometry was clean, then extruded it. I also know how to prepare paths in Illustrator so they come into LW nice and clean, and I know how to celan it up in Modeler if it doesn't.

If this strange problem wasn't an unplanned limit of numbers of points, then it's something I've never encountered before. In finally getting it to work manually, I deleted NO points, and it renders beautifully in Layout ... Did you notice that this is Modeler 8.5, on a Mac Pro with the Intel quad core Xeon running under OS X 10.5.5? THAT'S the real problem, it seems ... Modeler 8.5 is very flakey in general on this computer/OS. Many, many Modeler functions simply and inexplicable stop working for no apparent reason quite often, and Modeler must be rebooted. This is just one more in a long line of similar issues. I understand that the newest release of 9.x is far more stable and functional than 8.5 on this computer.

I still would like to know if 9.x runs under Rosetta on the Intel Macs under Leopard or not ...? :-)

Thanks,
Michael

cagey5
12-16-2008, 05:15 PM
Generally boolean operations perform better with lower numbers of points. Certainly quicker and sometimes more accurately. Try subdividing the ngons before the boolean operation. You can always merge them back together after the operation if needed.

DrStrik9
12-16-2008, 06:22 PM
Yeah, I know ... but in this simple model, after I finally got past that first strange hurdle, I subtracted 99,000 polys from around 5500 polys, which I fully expected would make Modeler 8.5 puke on this Intel Mac. But it worked FINE. Surprise, surprise ... :-)

Andrewstopheles
12-16-2008, 06:39 PM
If you triple your polygons before boolean it usually works - solved the problem for me, sort of :D

DrStrik9
12-16-2008, 07:06 PM
That's a good idea too, unless it's an 8.5 subpatch object. :-)

Giacomo99
12-16-2008, 08:32 PM
I still would like to know if 9.x runs under Rosetta on the Intel Macs under Leopard or not ...?

9.5 is rock-solid on my Intel Mac running Leopard--I'm assuming it's native, but if not, it's amazingly responsive and solid for a Rosetta application. (9.3.1 was, in my brief experience, unusably buggy on the same machine.)

If you want my advice: if it's at all possible for you to upgrade to 9.5, do so.

DrStrik9
12-16-2008, 08:43 PM
So 9.5 for Mac is OUT? Even Newtek's web site doesn't know that ... I just downloaded the 9.3.1 demo tonight ... hmmm, that's good news.

THANKS! :-)

Michael

Darth Mole
12-21-2008, 02:28 AM
No, but if you upgrade to 9, you can use the beta of 9.5.1 - which is simply the best LW for Mac ever. And, more importantly, there is a Universal Binary version which runs natively on Leopard on an Intel Mac. You might lose some of your older CFM plug-ins but this is absolutely the route to take. (And anyway, LW has so many new features now, and is so fast, I don't miss many of my old plugs.)

othornton
12-22-2008, 01:12 PM
That's a good idea too, unless it's an 8.5 subpatch object. :-)

That may be your problem right there. Booleans and Drill tools seldom work on sub-D objects, and sometimes are flaky even with normal polys. One solution to this is to turn of sub-d, make your boolean or drill, and stitch up the resulting wholes and n-gons by hand. Perhaps you have already done this; it is hard to know with only text details of the problem to work from.

Anyway, what I think Andystopheles meant was that you triple the area where the intersection/cutting will occur, not the entire mesh. This will still give you a more stable area of intersection without disturbing the parts of the mesh you are not working on and reduce the post-op clean-up greatly.

If you're using drill a lot though you may want to pick up LWCad. It's primitive booleans and drills are a lot more stable than normal LW tools. No idea why, they just are. The ability to add user-defined shapes to the shape libraries makes almost any kind of linear drill/boolean possible.