Failed support generation for point-overhangs

[Yona.Appletree]

It seems that s3d cannot find single-point overhangs, common with rotated pieces. Even at the best support generation resolution, it won't correctly generate supports for this model:



Resulting in this awesomeness:



This test model was generated with the following OpenSCAD code:

Code: Select all

cylinder(r=5,h=20);
translate([-5,0,20]) rotate([45,45,0]) cube([20,20,20],center=true);

One can of course manually add supports, but for my actual model, there are many and it's hard to find them all by hand. This basically makes me not trust the s3d support generation logic. Has anyone else had this issue or found a workaround that doesn't involve manually adding supports?

[Yona.Appletree]

Just for comparison, I sliced this in Makerware and it did what you'd expect:

[Rebekah_harper]

What's your support resolution at?

[jimc]

supports not being generated on an overhanging, downward facing point has been brought up before. it was awhile ago so not sure if that will be corrected in the next release or not.

[Yona.Appletree]

My resolution is at the minimum -- 1mm. I find it a bit strange that this needs a resolution, and I've seen some really strange results when changing the resolution around, where it will sometimes just completely misses large overhangs.

I think MakerWare uses a different (better) algorithm (and likewise has no resolution setting). It never misses overhangs and generally behaves more predictably in my experience.

Perhaps there is some external software that could be used to generate better supports?

[billyd]

Yes I find S3D's automatic support generation to be poor. But the fact that you can apply them manually is a great feature. I rarely use auto generated supports. I just take the extra time to define them manually so I know they will be the way I want them. It would be nice if S3D fixed this issue.