BUG: intersection or boolean failing 'just because'

Hi Rhino commanders,

I ask here about a very common and very mind-perplexing bug regarding the failed intersection of two objects.

If you have a few seconds, please see what happens in this video:

…intersection-related commands { intersect, or booleandifference } fail to yield the correct result (all intersection curves or the boolean result) at a particular ‘location’ in the model.

Why? Why does this happen? And why would it ‘go away’ if I slightly move the geometry a micro-fraction? I don’t want to move the geometry, I want the intersection to occur where it is occurring.

image

Thanks in advance for your answers

  • edit:
    using Rhino WIP, if it matters…but the issue is the same on 6 - and these are all water-tight solids - one big, many thin, fully intersecting each other.

That free-form surface looks pretty wonky from the video - perhaps you can post the file, I have a feeling Rhino is not too happy with that object, the top surface may have some crossed isocurves or something like that…

hi,

when you select your “surface”, it sais “polysurface added to selection”. What result do you want exactly?

For me it looks like you have two surfaces lying on each other (under tolerance) and therefor you have two opened curves instead of one closed.
if you are looking for a closed Curve, you might want to try with a surface.
Explode your polysrf that looks like a surface and delete the double.

Ben

Thank you for responding @Helvetosaur!

What’s worse is that I am intersecting the same object with plenty more objects of same thickness - of which at least 5 show the same discrepancy.

They don’t have to be solid intersections for the issue to manifest - here I use the planes (before thickening):

I get other stuff like this:

Anyway, if I intersect with a plane instead, (plane from the surface) the curve is generated:


image

I imagined so, but no:

The solid I am intersecting is the result of another solid difference operation that came before it, which shouldn’t matter, I thought, as things look generally fine:



sorry for redundancies, I do not intend to put you to work - I’m just perplexed :slight_smile:

thanks @benedict,

right - because it is a polysurface, which is fine - if you see my reply to Helvetosaur you’ll notice exploding the object to get the single surface doesn’t solve it either :frowning:

Ok, thank you for the additional infos.
capture-20190925-023933
may you do a little pointsOn on that one and take a screenshot? It might have some fancy points just beside your point?
Can you bake the geometry and post it here as 3dm? it would help a lot

@corellaman Without an object to test, we’re just shooting in the dark here. My experience is that when these types of intersection failures happen, that there is generally something wrong with the definition of one or the other of the surfaces involved. In this case it’s most likely the twisted surface, but again, without being able to test, it’s only a theory.

Thanks again @Helvetosaur,

I agree with you that without an object to analyze you cannot provide further assistance. I was imagining this could point to a similar case that got solved in the past.

Currently, I have pursued multiple workarounds, all yielding the same failing result. Again, not asking to have you or anyone work on it for me.

I recognize a possible need to look at the untrimmed version of the shape, prior to the boolean operations. Let’s just say it won’t matter :slight_smile:

Using the initial geometries (untrimmed, before-thickness-and-before-booleans) results in the same anyway.

using the full closed parent surface, before booleans:

using an open untrimmed version of the parent surface, before booleans:

using the solid difference result:

if I recreate a partial (open) loft using only a few of the section curves passing through there, the intersection ‘works’ completely:

If I force that same partial loft into a messed-up closed loft IT STILL GIVES ME THE CURVE:

Unfortunately, I need the solid intersections, not these curve results. I was choosing to show/test them as redundancy tests. Plus I suspected originally it could be related to graphics or some other mesh preview settings, which I have had to edit in the past with this same project.

I’ll continue to work on it, making adjustments to the curves, the loft, or subsequently the cutting planes.

Thank you so much for your attention.

Hello - I do not yet see a file - there is not much to do unless you send the objects you’re asking about, either here, or to tech@mcneel.com, with a link back to this thread in your comments.

-Pascal

1 Like