Boolean operation

Replies to This Discussion

Permalink Reply by george grimes on March 18, 2011 at 7:19pm

Hello group. This is my first post just a test. I am a self taught Rhino4 user and have designed an airplane and the engine for it with Rhino4.  I have an order of proceedures I use to avoid most boolian problems.

    To get reliable solids good curves are a necessity. If you can "extrude /tapered" your curves they are basically good but can be improved. If you can loft them but get segments rather than a single surface when you select it the curves need work.I first "project to c-plane or the current work plane and then explode the curve. Next I move all the segments but one away a little bit, then ove them back one at a time with only the end point snap on. About half way around if you can find a straight segment go to the other end or the unmoved curve and add the remaining segments back to the straight segment you quit on from the first end. erase that segment and replace it with a new line. rejoin all the segments except one and pres enter, enter and  if all you have selected select as one are joined add the last one and look for the curve to read closed. Step two is edit/curve/rebuild and add control points. I have very complex curves and rebuild to 360 points at the same degree. If  you are going to extrude tapered as I do a lot you can test with an extrude longer than you will be using to make sure the radius of the curve doesn;t go less than 0, or you can offset the curve to what it will be at the second surface and loft the surface/cap planer holes to get the solid.

         At this point I have all my solids assembled woth .005 interference on all the joining surfaces. and union the part which normally works, but sometimes I am grouping the parts rather than union and then want to boolian or slice something. Occasionally I will have an embedded solid such as an oil passage within one of the parts. This like the intersecting parts will result in slice operation failed.  I them move the embedded parts 30" X- positive and slice or boolian the remaining solids one at a time after ungrouping them. Then regroup the parts for each layer.  I am  generally outputting  to an Stl file for export to a CNC pattern  which doesn't seem to care that the solids have interference.

        If solids with  interference don't care to union that will often be with the perimiter of the parts being the same basic shape resulting in ambiguous points conflicting with the setting of tolerence. I can add a little around the edge of one of the parts with offset curve from surface extruded about twice the interference or just change the tolerence during that boolian. In my case I usually just use "group" and deal with future wire cut and slice problems on the individual solids after ungrouping them.  If the solid is going to be used and modified a lot in future work it is best to get it right. For a one-off to run through CNC time is money and I just group and try to figure out how to do better on the next component.

     If anyone knows about "manifold edges" solutions I would like to hear about it.

                                                                                                             george grimes (wizzardworks)

• ▶ Reply

Permalink Reply by Kunal Ghevaria on April 5, 2011 at 10:47am

I've given up on expecting Rhino Booleans to do what I expect them to. I just use the 'Boolean 2 Objects' command now, and click till I get the result I want, instead of imagining what a particular tool will do in advance. And it seems to be more reliable than using the individual tools. Just my experience.

• ▶ Reply

Permalink Reply by Charles Fee on January 30, 2012 at 9:25pm

Hi !

Actually Booleans work fairly well in RH-4. Now, don't misunderstand me, problems like yours are both more frequent than they ought to be AND the solutions aren't nearly as easy to find as they ought to be.

So-

Here's the scoop on booleans. RULE: Claims to the contrary notwithstanding, RH Booleans are only reliable one pair of component elements at a time.

FIRST, ungroup them. Groups are handled well in RH but RH's ability to differentiate discrete components collected in groups needs attention (attention it didn't get in the RH-5b releases). Remove any components you're about to boolean from any groups, no matter what the combination ...

SECOND, for any boolean, make sure that the edges of any surface involved don't touch. AND make sure that no faces are coincident. Tolerances are involved. Don't spend a moment on tolerances here. Your tolerances should already be set correctly if you're working from the proper opening template (LARGE OBJECTS - FEET for instance). The problem with tolerances insn't how big they are relative to the components. The problem is in the ways they're calculated. Ignore tolerances. Simply move the components to a position such that the way they intersect is clear and obvious. If its obvious to you, it'll be obvious to RHINO. For instance, extrude the circle you put on the surface, BOTH WAYS to depth or all the way through, to drill that hole. Just straight in won't work (reliably).

Third, If you get the "opposite" of what you intended (think UNION when you tried DIFFERENCE) your normals need fixing. There's ONLY ONE PRACTICAL WAY to to this. Go to each of your 2 parts. Check the normals. One of them has its normals pointing inward. Explode that one. Rejoin all but one surface. Reselect the resulting OPEN polysurface, and point its normals outward. Flip the normals of the remaining surface so they'll point outward too, and join them back to a CLOSED polysurface. Your booleans will then work ok. RH Tech support will claim that normals of closed polysurfaces always point outward. That's nice. Explode, flip and rejoin, and move on.

This should return the strong majority of your boolean operations to something more like predictable. Oh, btw- did you notice I was able to do this for you without a single You-Tube  "tutorial"? - Just sayin' ;)

Good luck, and post back here with results ?

C.

• ▶ Reply

Permalink Reply by george grimes on January 31, 2012 at 8:34am

C.   Thanks for the pointers.  Making  bolt holes  by extending the solid to difference through the surfaces  really is the way  to go.   I also did a solids/cap planer holes first on everything and it worked in a couple of seconds.  I then grouped  32 bolt hole solids which were three inches through the surfaces and differenced the entire group which also worked. having bolt holes that don't difference was a real pain before.  For solids that will not union I have been grouping them then saving as  .stl  file extension to export into  "meshcam3"  for the G-code  and the   CNC milling  machine runs just fine.

        I have made solids with the normals reversed.  This can happen if I loft two curves with one offset and raised to make a tapered part (draft angle 3 degrees) for a pattern. Then I use cap planer surfaces with the result that occasionally the normals are reversed.  There may be a correlation with choosing the smaller curve first in the loft so I now only choose the large curve first. I only do this when solid/extrude curve/ tapered produces an error even after I rebuild the curve to 360 points.

                                                                                                 wizzardworks

• ▶ Reply

Permalink Reply by John Brock on January 31, 2012 at 3:14pm

There's a good FAQ document here:
http://wiki.mcneel.com/rhino/booleanfaq

Here's a the general FAQ link:
http://wiki.mcneel.com/rhino/faq

 

John Brock
Tech Support
Rhinoceros
Seattle
USA

• ▶ Reply

Permalink Reply by Charles Fee on January 31, 2012 at 6:20pm

John-

GREAT background  info. I've found that when someone is having a specific problem, tho, a specific response works really really well. Assuming the model is set up based on an appropriate template, tolerances should be about right. Yes some unusual techniques creating closed polysurfaces can result in unusual normals, but when creating conventional multi-closed polysurfaces based on properly created sub-components, booleans ought to work. It really shouldn't be necessary to explode components and surface-edit them into submission. As long as care is taken that edges and faces aren't coincident, and normals are properly oriented, most operations should work. That Rhino ALSO offers the ability to take things further into the areas implied by the model in the FAQ page like necessary coincident surfaces is important. When encountering most operations when booleans fail, I've found that the 3 steps mentioned above ought to cover the majority of cases and should remove most stumbling blocks from progress returning the modeler to productive progress. 

John- THANKS for all your help, THANK the team for us for the improvements moving to RH-4, and THANKS for your continued support for the user base !

Charley.

• ▶ Reply

• ‹ Previous
• 1
• 2
• Next ›
• Page