Strange Sketchup behaviour. Bug or what?

[JoNoS]

I am using Sketchup Pro 2018.
This is a problem situation I seem to experience from time to time. The enclosed example has been maximally reduced to help illustrate the situation. Maybe anyone can explain why this occurs or if any tool or plugin can quickly remedy such a problem.

Test_4.skp (71.4 KB)

The actual problem here is the little triangle in the wall opening. If you select the triangle face, the the surrounding wall face is selected as well. This should imply that the triangle face is (at least) co-planar with the wall face, right?

Notice the two (blue) selected edges of the triangle. They are (as far as I can tell) actually CONNECTED to the window bottom edge. But if that were the case, clearly the triangle face should not be part of the rest of the wall face?

If the triangle edges were connected slightly ABOVE the window lower edge, the same should apply, the triangle face and the wall face should be distinct.

And if the triangle edges were connected slightly BELOW the window lower edge, then the two faces should also be distinct, because the triangle (actually it would then have four vertices) should be split in two by the window lower edge.

So it appears the faces are behaving as one single face, while in no case this should be so (given that the window lower edge is continuous, of course).
Is it just me being occasionally bugged by this type of phenomenon, or is this stuff bugging others as well? Anyone to explain why this is happening so?

[Wo3Dan]

A face (the triangle) created with one vertex on the perimeter of another face can indeed be buggy, depending on how it was created.

[whiterabbitdesigncompany]

this does come up from time to time and is a repeatable problem. If one vertex is touching, even though its co-planar with the original face it doesnt embed itself in the other face (remains bold), by drawing another line from the bold face to another line causes it to meld. Not sure if there’s a solution to this, but good to be aware of it and a way to remedy because it is a nuisance.

[JoNoS]

@whiterabbitdesigncompany

this does come up from time to time and is a repeatable problem

Thanks for your reply.
If I am not misunderstanding, you are suggesting to “heal” by drawing a line from the
triangle to the enclosing edge boundary.

However, in the included example this does not really fix the original problem. The triangle
bold lines disappear, but this is due to the fact the added line causes a face to be added into the
window part (around the triangle).
Still, if you select the face inside the triangle, the outside wall is still selected as well.
And if you delete the added (temporary) line and window face, you are back to the original
situation.

I seem to find these situations from time to time, and usually, they can be remedied by
fiddling around and redrawing parts. This time I thought I would ask for information from
the forum for a more “repeatable” solution (if any).

I apologize if I do misunderstand your suggestions.

[Forestr]

Draw a line, then erase it.

[Box]

It’s a long known bug from time immemorial that manifests itself in various ways.
I recently demonstrated it for one of the code developers at Basecamp. He expressed genuine interest and a desire t look into it.

[Geo]

One corner is slightly out of plane.
Moving it into plane resolved the separate faces issue.

The triangle sides are tiny … ~2mm
When modeling such tiny things; turn off Length Snapping and set Precision to the highest.

[Box]

@Geo George, I know you know this bug. The tiny error is part of it but not a result of user error.
You can draw it flat on a surface and it will still cause problems.

[whiterabbitdesigncompany]

In your case, if I draw what you have it isn’t an issue, so George’s note about being off planar might be part of it? coupled with the small size?

[Box]

I really hate these threads.

[JoNoS]

@Geo

One corner is slightly out of plane.

I apologize for continuing this thread, as I feel this item is not really fully clarified yet (at least not for me).
Thanks to Geo for notifying about the X-discrepancy. I did not really think of that
since Sketchup did draw everything as a plane.
This implies that Sketchup handles something which is not REALLY a plane, as a plane anyway, and by doing that it creates other oddities at the same time (two faces apperaring to be one face).
I wonder why Sketchup does not flag this thing with some kind of specific warning instead of just behaving strangely!

Please note that the attached .skp file was not drawn directly, because then you would not create the specific error case. This type of issue creeps up as a result of modifications on a complicated structure, by some tools or plugins, maybe, I am not quite sure why have observed a number of such cases lately. I wonder whether working on complicated solids for 3D-printing makes me vulnerable?

Anyway, I modified the file with the erratic behaviour to get all X-values the same. This did NOT resolve the problem in my case (see drawing below). So I guess there is more to this yet…
Any tools which would flag and/or correct occurences like described?

Test_5.skp (77.7 KB)

[Geo]

@JoNoS

Thanks to Geo for notifying about the X-discrepancy. I did not really think of that
since Sketchup did draw everything as a plane.
This implies that Sketchup handles something which is not REALLY a plane, as a plane anyway, and by doing that it creates other oddities at the same time (two faces apperaring to be one face).

I wonder why Sketchup does not flag this thing with some kind of specific warning instead of just behaving strangely!

SU has a certain tolerance to non-coplanar edges supporting a face.
The off-plane error in your example is below the tolerance and so the face remained intact.
Past a certain point, SU automatically triangulates faces that would otherwise fail.

The fact of my tweaking the errant geometry back into plane made the triangle face unique is merely a fluke.
The bug @Box points out has been around so long it’s more or less an accepted ‘feature’ users work around.

Bug fixes come along with every new version of SU.
But prioritizing them is up to the development team.

[tt_su]

All 3D applications have to deal with tolerances. It stems from floating point precision rounding errors in computers. Whatever you set the tolerance to be, when geometry is near the tolerance limit unexpected things can happen as the geometry is further modified.

@JoNoS

This type of issue creeps up as a result of modifications on a complicated structure, by some tools or plugins, maybe, I am not quite sure why have observed a number of such cases lately. I wonder whether working on complicated solids for 3D-printing makes me vulnerable?

In my previous job I would often run into these types of issues when importing from other file formats. I often imported DWG files and for some reason they tended to introduce there imprecision issues that cascaded when I modeled from the import.

[JoNoS]

@tt_su

All 3D applications have to deal with tolerances

Apparently, some responders appear to believe, that the described case has something to do with small values and tolerance limits. Currently, to me it appears that this has little or nothing to do with it.

In order for something to NOT be a “bug”, I would say that this is necessary:

• an action series (in Sketchup) must cause a result that is in correspondence with what the developers have intended
• the same action series must cause the same result when done multiple times (consistency)

In the specific case reported above, I think I need to understand whether my understanding of action/result is the same as the intention of the developers, so let us try to establish a baseline here:

Consider the first figure.

It was drawn by adding line segments as this: (The letter IDs name vertices)
A-B-C
C-D-E-F-C
E-G-I
What I get here is a face with three inner loops. This is what I would expect expect.

Consider the second figure.

In this case the line segements are added:
A-B-C-D-E-F-A
Note that I get two implicit vertices where lines cross each other in the plane.
Also in this case, I would expect to get three inner loops, but as the figure shows (I have selected the outermost face), I get only one.
Why is result of the second way of drawing (a very similar figure) so very different from the previous case.
Is this intended, and if so, what is the rationale for such very different behavior?

Now, if I repeat the second drawing sequence several times, the results can vary a lot as can be seen in the third figure.

I can get three inner loops, one inner loop with four vertices, or inner loops with eight vertices. By drawing lines in the exactly same sequence!
Which indicates that something here is not behaving consistent.
Again, is that intended or not?

From the discussions I get the clear impression that I am not the only one observing this (or similar manifestations) of this phenomenon.
It would be interesting to hear if others also get inconsistent results, similar to what I do?

[slbaumgartner]

These problems have been reported for many years in many versions of SketchUp, and I agree with you that they are bugs. I can’t believe that they are intended behavior.

What I don’t know, being just a volunteer here on the forum, is whether they represent inherent algorithmic limitations of the problem being solved (often called “corner cases”) or something else. It seems to me that if they were due to computer arithmetic limitations they would be sensitive to the size of the geometry. But I don’t think that’s the case, they seem to happen both for large objects as for small ones.