Inconsistent surface offset direction

Replies to This Discussion

Permalink Reply by Julian Buhagiar on August 2, 2017 at 11:27am

Interestingly if you use offset loose instead, you get a result that you might expect.

Permalink Reply by Julian Buhagiar on August 2, 2017 at 11:59am

Permalink Reply by Andrew Heumann on August 2, 2017 at 1:27pm

I think what's happening is that the surface offset component is offsetting in the direction of the Surface normal rather than the Brep Face Normal (which can be opposite the surface normal). Part of the common confusion is that what RhinoCommon calls a "Surface" is in fact always an untrimmed surface - whereas what Grasshopper calls a surface is a Brep with a single face. You can identify those GH surfaces that are the "opposite" direction with a short script: 

private void RunScript(Brep x, object y, ref object A)
{
A = x.Faces[0].OrientationIsReversed;
}

The ones for which this offset seems "wrong" are the ones which return a true value with this script.

Note that ultimately I agree that for consistency's sake in GH the surface offset component *should* behave the way you expect it. But hopefully this helps explain why it behaves the way it does.

Permalink Reply by Julian Buhagiar on August 2, 2017 at 1:55pm

Thanks for that. That helps, I hadn't realised the two (Brep and surface) could differ but I can see why it would now.

Permalink Reply by Olivier Suire on August 3, 2017 at 12:58am

Hi Andrew, thanks for the explanation.

If I follow you well, the surface normals as output by the "Evaluate Surface" component should be consistent with the "Offset Surface" component since they both deal with the "Surface" in the GH sense of the term.

That is not the case :

Permalink Reply by Hyungsoo Kim on August 3, 2017 at 2:18am

I didn't really understand what "Brep face Normal" means mentioned by Andrew, but anyway, when I apply his script, the directions of the offset are matched anyway.

Attachments:

• Inconsistentsurfaceoffsetdirection_re.gh, 22 KB

Permalink Reply by Olivier Suire on August 3, 2017 at 3:26am

Thanks !

I had a clunky workaround which offset surfaces in both directions, joined the resulting surfaces and sortes the resulting Breps by area...

This is much cleaner, although I still think that something is really screwy with those normals...

Permalink Reply by Andrew Heumann on August 3, 2017 at 8:50am

Read my reply - I agree that the behavior of offset surface is incorrect. 

Permalink Reply by Olivier Suire on August 3, 2017 at 9:09am

Yep.

How about you ?

Did you read my mail about retrieving object IDs as outputs of the "Place Block" and "Objects by selection" Human components ?

:)