Triangulation and Convex Polygons
-
@mrapsouthern said:
Just to be clear, I want the hole to be present.
Is there some consistency to the mesh? Writing a generic tool would be pretty much impossible as there are so many mutations. But if the holes are always square - then the task is more predictable.
-
Oh, right - you posted that curvy thingy...
-
I have drawn a flat form like your. No Made any split !
Export in OBJ format by Tig OBJ ExporterNo Problem ! I have hole and only triangles in Wings3D for example!
I don't understand your little problem -
@Pilou Basically the exporter and 3rd party software I use cause the hole to be filled in. It looks like the OBJ exporter you used exported the underlying triangles defined by Sketchup.
I can achieve the same by using a triangulation plugin prior to export.
That is as far as I have got. As you have also shown, it works, but the issue is that the triangles are very skinny/narrow/spikey (however you want to describe them).
I think this is as good as it can be, but the skinny triangles are not desirable. The only way to avoid them that I can see is to approximate the hole with fewr polygons, at it is the curves that cause the thin triangles.
Thanks for your help. Further suggestions are welcome though.
Cheers
Alex -
what if... for each face with a hole in it one edge is drawn from the inner loop to the outer loop? (Even here there are cases where it could be difficult to do by script - for instance if there are multiple holes in the face.)
-
Got a solution in theory...
What I need is a plugin that will allow me to select a face with one (or multiple) holes in it and then divide that face in some way so that there would be multiple faces without holes.
i.e. for each hole in a face you only need to add two lines to split the face with 1 hole into 2 faces without holes.
Similarly,
a face with 2 holes requires 4 dividing lines and results in 3 faces with no holes.
a face with 3 holes requires 6 dividing lines and results in 4 faces with no holes.
a face with 4 holes requires 8 dividing lines and results in 5 faces with no holes.This solution would mean you can represent a complex geometry with holes without splitting the whole model into triangles.
Is there anything already out there?
Thanks
Alex -
Thomthom's idea should work great in case there is only one hole in a face, and if You make more vertices on the outside loop. The problem with 'spiky' triangles is that You have only 4 vertices to which the edges are drawn.
Assume that You have two sets of points:
[inner_vertices]
[outer_vertices] - for example 25 points per rectangle sideFor every point from [inner_vertices] find closest point from [outer_vertices] and draw an edge between those two.
(Possible problem) when two outer points have the same distance to inner point. Which point to choose ?
Cheers
Matt -
Can you post an image with result wished ?
-
I was thinking one edge from inner to outer was enough - in order to reduce the number of faces.
But thinking of it more, ultimately everything breaks down to triangles, so if you have a polygonal face in SU - internally it'll be made up of many triangles. The same thing goes for any other software. So converting a face into it's internal triangles doesn't really make the model heavier - in terms of total number of triangles the graphic card needs to draw.
-
Better to make a retoplogy in 3D party ?
-
Hi,
To update my previous post -
Any 1 face with ANY number of holes can be divided into 2 faces with no holes.
where the number of required dividing lines is defined by,
NumDividingLines = NumHoles + 1
please see the following diagram as illustration.
Alex
-
-
@gullfo said:
slightly off-topic:
http://gamma.cs.unc.edu/propagation/main.pdf
http://sketchucation.com/forums/viewtopic.php?t=36063It is off topic - but I am very familiar with that work - it's quite a coincidence you posted it....
-
@mrapsouthern said:
does the SU-Catt converter then create the correct output?
-
@mrapsouthern said:
It is off topic - but I am very familiar with that work - it's quite a coincidence you posted it....
i'm trying to re-work Jonathan's SU-ABEC to use ABEC 2.0 but it's tough because their model shifted a lot... http://www.jonsh.net/blog/?page_id=197
-
Glenn,
Small world. I know Jonathan well also. Have we met?
Anyway, Yes, if the faces with holes are partitioned as shown the output of SU2CATT is interpreted correctly.
Cheers
A. -
Will merging vertices help?
Your model reduced from 200+ to ~72
-
Merging vertices looks like it helps for that case. I can't check it for multiple holes though right now. Thanks
Advertisement