Bug or TOUGH CASE with Booleans ...
As mentioned in a different thread ... Booleans feels pretty good on some tough or reasonable cases. One TEST I have been using for ManifoldLab / Carve CSG in past has been GeoDome high res minus a knot torus. I think it makes for a more interesting case than Cube minus cylinder.
My first try at above failed. And then I reduced the geometry a little to make it "
simpler" and the Wings Boolean then worked. I know the devil is in the details ... so after a time ... I can share a "model".
Fun stuff.
Since I don't want to SPAM Wings Forum about new Booleans ... I will ask here ... is there any way to UNION several objects at a time ? This seem maybe fundemental to booleans once we start using Booleans in organic or artsy workflow. It looks like for now ... it is just two objects unioned at a time.
Another Tough CASE might be WINGS 3D Text ... subtracted from maybe a grid-box. I tried it with difference ... subtracting LETTERS from a box. It was a touchy situation.
With manifold, I found that triangulating the objects before booleans worked better.
I didnt try much with 2.2.1 yet. Just a couple of spheres (which is what I previously had problems with ).
(12-02-2018 10:25 PM)ggaliens Wrote: [ -> ]One TEST I have been using for ManifoldLab / Carve CSG in past has been GeoDome high res minus a knot torus. I think it makes for a more interesting case than Cube minus cylinder.
My first try at above failed.
Just for evaluation purpose, how high res was the objects in that test?
I tried one before that crashed. After some changes made by
dgud, I now tested it without problem, then I smoothed it twice and still worked.
Even smoothing again it still worked. I hope now we may have more consistence in the results. But, for some cases we may need to add some loops to the object.
The memory used by werl.exe process in these case above (before | pick during | final):
154K Faces | 308K Edges => 213MB -> 730MB -> 336MB
616K Faces | 1,232K Edges => 600MB -> 2,7GB -> 978MB
A guaranteed failure in boolean:
Open the attachment and try to subtract any of the objects named "triangle" from the object named "support". Guaranteed failure, and very simple mesh. This is using 2.2.2 on Ubuntu Linux.
I've been
trying to use Wings as the base of 3D printing, since I like the interface far more than what Blender offers, but at this rate even on the latest version booleans fail far more often than they succeed, and without it Wings loses most value as a modelling tool.
And while I'm at it, where the option to rename the axes so X, Y, and Z correspond with the standards that 3D printers use?
[
attachment=544]
[
attachment=545]
Sounds like you are making progress Micheus. The memory numbers look good.
I still think some LOOPING variations would be nice. One object ... and then serially subtract other objects from it. I have fingers crossed that you guys might see the value in that sort of thing.
Thanks for taking care of Wings3D !
ggaliens, for the case of booleans I have doing nothing than provide some crash cases to help
dgud to workaround them. It's all his work.
And also ... as someone else has mentioned ...
"Subtract objects from secondary selection"
Is not logically correct. It should read ...
"Subtract objects from Primary Selection"
ggaliens, can you share your wings file, please?
The one I created surely doesn't fit in the case you create.