Error message????

[cadlab126]

Hey all,

I made a polyshape by extruding and using the polyunion tool. For some reason, I can not subtract a cylinder from this shape. I already deleted history on everything, and the cylinder extends far past the object on both sides, so it is not the problem. It gives me an error message saying cannot find GiShader, or something equivalent to that. I tried exporting the object as DXF, DWG, Obj, but nothing would work.

thanks.
John

[joojaa]

check for nonmanifold faces! or post a picture, file atc and let someone have a looksee.

[cadlab126]

Hey,

I tried doing a Poly cleanup to eliminate the possibility of non-manifold geometry. That did not solve the problem. The exact error message is: "Error: No object matches name: .miGeoShader" The following reference shows the subtraction I am trying to execute.

http://www.enete.com/john/Post.html

Thanks.

[joojaa]

that error has again nothing to do with the boolean operation!

Try merging vertices first with a small radius... Se if theres two faces that are opposite to eachother but not connected you wouldnt necceserily se it but the booleand would.

Could you either send the scene scene with offending objects only so that we can have a looksee? In your picture show normals!

Still havent been sent a boolean that wouldnt have worked once i eliminated all the problem areas (altough not everything is booleanable).

[cadlab126]

Hey,

I am not sure if I left it out, but I meant to put the link: pictures
I tried merging vertices (2 merged at a .005 tolerance), and conformed normals. I tried flipping normals independently on each object to see if that was the issue. Any more ideas?? Thanks!

[joojaa]

move one object slightly!

[cadlab126]

Hey,

I tried moving each object in different directions, but alas.... the same error message comes back to haunt my progress. Any other ideas?? I really appreciate your help! Thanks.

[joojaa]

Send file. Its impossible to debug it unless i can se inans out of the geo.

The error has nothing to do with the bioolean operation failing (if the rsult just dissapers than its not in fact realy failing to do boolean jsut to compute result). Its ahsader info. In general i find that some MR connections yell error even if no error has occoured wich is annoying (i have a few examples of this).

[cadlab126]

Here is the file:

Press.mp

[joojaa]

Found it! illegal geomery! It has one of these hard to find bugs where the edges arent merged because on the other side there s 2 edges and on the other 1 causing a gap smack in middle of where the boolean happens. Its akin to a nonmanifold shape thats not possible in reality.

time for debug 2 minutes 28 secs.

se there are 4 of thees eoccourences in your modell. the face thats seleced is 5 sided (it should be 6) and has 1 ege on the boundary while on the otherside theres a 2 edged side now the vertice is not connected to the face. Causing a infitesimaly small gap between the faces (actualy no gap at all is a s bad!)

Be carefull with your geometry!

Hope this helps you in the future. its easy to fix tough just use the edge split tool to make a axxtrra edge here there (you can then delet the edge as long as you merge vertices). This was apparent from the way your smooth worked but i mistook it for a normal issue.

Normaly you can spot theese with the custm border edge display.

fixed a blooper in last paragraph

[campbell]

as joojaa prompted in first post, your geometry has non-manifold elements. to check for these, use the 'cleanup' tool. i often find it better (or makes me more comfortable) to clean up the geometry that is found rather than letting it do it for you.

ed: joojaa beat me - i was watching a race on the box whilst i was replying. i found it in 5 seconds. heh

[joojaa]

Technicaly its not a non manifold geometry!

[joojaa]

Yes i was looking for somehthing else and reading my mails at the same time

[cadlab126]

You rock! Thanks a bunch!

[campbell]

Technicaly its not a non manifold geometry!

hey? i see plenty o' faces sharing a vertex, but no edges.

[joojaa]

Im not sure of the purpose of this post?

point is that boolean needs sensible geometry, around its perimiter to know where intersections occour. MOstly booleans fail because of bad geo.

[campbell]

the purpose?

you made the point of making a post to state that, technically, the geometry is not non-manifold. believing that it is, i replied in a questioning manner, currious about a missmatch in understanding.

[joojaa]

Well a nonanifold geometry is something you couldnt fold of a sheet so to speak ie a place where 3 faces share a edge, or place whare the normals suddenly switch sides (the paper is one sided ).

Here the problem was a face that didnt share a edge with the face adjacent to it, this is not nonmanifold but still as bad.

[campbell]

sure, adjacent faces dont need to share eges, and doesnt make for non-manifoldness, of course, because thats a spacial concern, not topological.

but this is not the problem with the boolean failing.

the boolean is failing due to the mesh being non-manifold. non-manifold because there are dangling faces about the central vertices of the radial piece. fix these, and the boolean is happily performed (albeit due to a bit of luck on the part of the bool elem's placements, and it propogates the 'bad' geometry ...but this may be desired, and isnt what im commenting on).

[joojaa]

Oh blimey that true. Somehimes holes like this do however cause booleans to fail. Seems like the hole is the source of the nonmanifold geo tough.