Hello,

first - I much like the MOI3D mesher, quality and speed are great. Thank you for this tool.

During tests and projects some times some meshes are created and I think, here could the mesher a little bit improved. I will collect this meshes in this thread.

Regards,
Micha

(1) Ellipsoid
From my tests with spheres I got perfect homogeneous meshes, but if I import an ellipsoid from Rhino, than I get irregular triangels:


At some mesh angles it works like expected


Maybe it's possible, that the mesh is perfect homogeneous allways.

(2) mesh #1
Here I wonder me, that at the big main surface no quads are created. Best would be, if the quad edges would match with the edges of the rounded edge at the sides.


(3) rounded cube
I'm not sure it's possible, but it could be nice, if all rounded edges of this cube would show the clean quad polygons.


(4) mesh #2
This is an object, that cause strong artefacts per Rhino mesher. At MOI3D it looks much better, but maybe it's possible that the triangles can be avoided.


(5) mesh #3
This is a plane, where I moved some control points at Rhino. The direct retry to mesh it cause some strange effects,

but after the control points are turned on and moved a little it looks like it should look

(6) mesh #4
The Rhino mesher create big kinks at this cylindrical surfaces. It is an old test object, but if I remember me right, it was from a project STEP file.


(7) mesh #5
Artefacts - that's new. ;) (angle 55)

But also the most cylindrical surfaces dosn't show a regular mesh of quads (finer mesh without artefact).




(8) mesh #6
At Rhino the opening in the middle is complete closed by a big artefact, at MOI3D the general shape looks a little bit strange at the sides (screenshot).



That's all that I found for testing at my hard disk today.

(9) rotation object
Once more ... I create a rotational object per Rhino and convert it per MOI - perfect, all polygons are quads. Than I add fillets to the source objects and convert it again, but now I got quads+triangles (should be possible without triangles or?). Interesting is, that the Rhino command _QuadrangulateMesh can be used to eliminate some triangles. I have the feeling, if _QuadrangulateMesh works, than the mesher created to much triangles, more than necessary.


(10) torus
Here it could be nice, if polygons around the torus profil could be in the same angle. At the moment the inner polygons are finer than expected. This could be interesting for using the mesh per T-Spline or Sub-D modeller.

Hi Micha, one note for you on triangulation.

There is an option that you can set in the moi.ini file to control whether triangulation will use that "radiate out from centroid" point style, which adds a center point to n-gons to triangulate them when possible, and then makes those kinds of "X" shaped stuff, or whether it will triangulate by only connecting points from the outside boundary of the n-gon and not add a center point.

The centroid style is the default since it can help to somewhat reduce skinny triangles.

But if you don't want the centroid style triangulation, there is a line you can edit in moi.ini to switch the modes.

Make sure to close down MoI before editing the moi.ini file, and then in notepad open moi.ini which you can find in XP here:
C:\Documents and Settings\ [your login name] \Application Data\Moi\moi.ini

Then in the [Mesh Export] section there is a setting:
CentroidTriangulation=y

If you switch that to say CentroidTriangulation=n then it will turn off that style of triangulation.

- Michael

Hi Micha,

> (1) Ellipsoid
> From my tests with spheres I got perfect homogeneous meshes, but if I
> import an ellipsoid from Rhino, than I get irregular triangels:

The ellipsoid is going through some changes in curvature in those areas, the base mesh was good enough for most of the shape but did not match the angle in some of those areas, so those areas got some adaptive refinement in them.

In general it is difficult for me to enforce a completely uniform mesh in every circumstance because it can easily lead to an extremely heavy mesh in other situations.

See this previous thread:
http://moi3d.com/forum/messages.php?webtag=MOI&msg=2379.1

For some discussion on this - this is the same issue that you are seeing here. If I always tried to make a completely uniform mesh, it would have a very bad consequence of shapes as shown in the other thread to have an extremely heavy mesh.

MoI's mesher does do quite a bit of work to try and get a very uniform mesh on shapes that are completely regular like a sphere or cylinder. But an ellipsoid is no longer regular like that, and has varying curvature.

Unfortunatley this is a tough area to improve without an unwanted increase in density in other situations. Many meshing things are like that, where with your human intelligence you can see that it would be a bit better to treat something with varying curvature in a different way but unfortunately from a coding perspective it is hard to make the code have "artifical intelligence" to do this kind of judgments.

- Michael

Hi Micha,

> (2) mesh #1
> Here I wonder me, that at the big main surface no quads are
> created. Best would be, if the quad edges would match with
> the edges of the rounded edge at the sides.

The reason for this will become more clear if you switch temporarily to Output:n-gons to see the n-gon structure, which looks like this:



During the process of meshing, MoI forms n-gons, which start out as quads that follow the UV structure of the surface, and then can turn into n-gons where they hit trim boundaries or also where they connect to the separate UV structure of an adjacent surface.

If you go into Rhino and set that object to display with a high number of isoparms, you can see this:



Notice how the isoparms for the fillet do not completely align with the other surface? That's because they have a different kind of UV space to each of them, that's one reason why they can generate some different initial UV grids.


I think it is going to be possible for me to do some work in the future to try and form the initial UV quads of a fillet after the "main" surfaces have been done and in some situations it may be possible for me to make a better initial structure for the fillet that matches an adjacent surface, but probably only when there is a plane to one side of the fillet since matching to pressure from multiple sides of curved surfaces is probably not going to work.

This is not going to be a completely easy thing to try to improve, so it probably won't happen very soon but at some point in the future.

- Michael

Hi Micha,

> (3) rounded cube
> I'm not sure it's possible, but it could be nice, if all rounded
> edges of this cube would show the clean quad polygons.

Usually rounded cubes like that should be getting nice aligned quads like you want.

Was that a rounded cube that you created in MoI, or was it an imported one?

It looks like this rounded cube model was not included in your attachments, could you please post it so I can test with it over here? Also do you see this happen at any kind of meshing parameters or is it only at a specific setting?

Here is a quick example of this kind of rounded box created in MoI and then exported using Quads & Triangles, I think it has the kind of structure you are asking for?



- Michael

> (4) mesh #2
> This is an object, that cause strong artefacts per Rhino mesher.
> At MOI3D it looks much better, but maybe it's possible that the
> triangles can be avoided.

It looks like the object has a pretty complex trimming curve in some of its edges, which is having an effect on the mesh generation.

Here I have duplicated the highlighted edge off to the side and turned on its control points:



Compare that to the control points of the surface itself:




As you can see, the edge has a complex curve and it does not match the simple structure of the surface - that edge is not an isoparm of the surface even though it looks like it probably could be.

Edges have a big impact on meshing since to make a "watertight" mesh, the mesher has to produce additional points along each edge. A complex edge like that is going make it harder for pieces to get lined up and handled in the most simple way.


I've attached an updated version of your object here as mesh#2_quads.3dm - to simplify the edges of this shape, I took your object, used Edit/Separate to break it into individual surfaces, then selected all the edges of those surfaces and did a Delete to "untrim" them and get the natural surface isoparms as a new simple trimming boundary, and then rejoined.

With this better edge structure you can now get all quads when you export from MoI.


In general if you have high complexity edge curves or other irregularites in the geometry itself, it will often have some kind of an impact on not making the most ideal meshes from it as well.

- Michael

Hi Micha,

> (5) mesh #3
> This is a plane, where I moved some control points at Rhino.
> The direct retry to mesh it cause some strange effects,

I can't seem to reproduce this problem, did you possibly attach the version after you have moved additional points where it works ok?

Also can you describe some more about the strange effects - is it the overly dense part, or do you not like the centroid style triangulation of n-gons - if it is the latter please refer to that previous message I posted for how you can disable that.

Otherwise if this is the fixed version that you posted here, can you please post the original version that has the problem in it, and please let me know if it needs any certain parameters to see the problem if it doesn't happen with the default parameters.

- Michael

Hi Micha,

> (6) mesh #4
> The Rhino mesher create big kinks at this cylindrical surfaces.
> It is an old test object, but if I remember me right, it was
> from a project STEP file.

No big kinks in MoI, right? So that is good! :)

Is your concern about getting only quads on this? If you do you'll probably need to simplify the edge structure of the shape, if you select the edges you can see that the boundary of those cylinder parts is split up into some additional edge segments rather than it having only a simple edge structure.

I've attached a tuned up version which will now generate all quads.

I did the tuneup by selecting the object, running "ShrinkTrimmedSrf" on it (which you have to set up with a keyboard shortcut to use in MoI, there is not currently a button for it), then Edit/Separate, then Select all edges, then Delete to untrim, then joined again.

The same comment applies as from that previous example - if the shape is set up to be overly complex like in this case with trim curves split up into more pieces than necessary, then that can often have some kind of impact on meshing as well.

If you have clean and simple geometry and topology it will help to generate clean and simple mesh results as well.

Complex geometry and topology will tend to make the mesh somewhat more complex as well.

- Michael

Hi Micha,

> (7) mesh #5
> Artefacts - that's new. ;) (angle 55)

This one is a bug! :) But it happens to be one that I've already fixed up for the next v2 beta.

It's a bug that got introduced with some of the changes to do faster meshing in the most recent v2 beta.

You can use MoI V1 right now for this one particular object to avoid the bug (just be patient while using v1, it is a lot slower of course).

Anyway the next v2 beta release will generate a proper mesh for this case the same as v1 did.


> But also the most cylindrical surfaces dosn't show a regular
> mesh of quads (finer mesh without artefact).

Basically all the same stuff I mentioned previously applies - the fillets in this case are between 2 curved objects, the fillets cause trimming edges that are not isoparms, they are curved. For example this fragment:




Notice how the trimming boundary there is not flat, it is curved? That means that the trimming boundary can have additional refinement points on it as well, same as the refinement for surfaces that I described earlier. When edges are refined it will generate n-gons rather than quads.

Additionally notice how the edges are in multiple segments...

- Michael

Hi Micha,

> (8) mesh #6
> At Rhino the opening in the middle is complete closed by a
> big artefact, at MOI3D the general shape looks a little bit
> strange at the sides (screenshot).

Are you using a very coarse meshing angle for that particular screenshot result?

When I tried your model at normal mesh densities it was looking fine...

At much coarser mesh densities the kinds of things that you show there are not uncommon and can just be a normal part of getting both a coarse edge boundary and a coarse surface quad density.

Generally something that is curved by quite a bit will look a bit strange when using only a small number of jagged polygons to represent it. Particularly if the interior surface happened to get something like no divisions on it while the edges got one division or something like that.

- Michael

Hi Micha,

> (9) rotation object <...>
> Interesting is, that the Rhino command _QuadrangulateMesh can
> be used to eliminate some triangles. I have the feeling, if
> _QuadrangulateMesh works, than the mesher created to much
> triangles, more than necessary.

Yeah it does look like those ones should be quads, I'll see if I can figure out what is happening there.


> Here it could be nice, if polygons around the torus profil
> could be in the same angle. At the moment the inner
> polygons are finer than expected. This could be interesting
> for using the mesh per T-Spline or Sub-D modeller.

This one has an easy workaround - it looks like it is only happening at a few particular very rough angles. If you adjust your angle slightly to 57, you should get the same mesh structure but without that one extra refinement:



When you have an angle value that happens to be right near a "boundary zone", it can pretty easily lead to this kind of irregularity. If you want to reduce refinements just a small adjustment in your angle can eliminate it. The mesher is not really very focused on doing such coarse things, I mean it will work but you can run into some slightly odd things at such extreme angles.

- Michael

Hi Burr,

> Wouldnt the avoid smaller than option fix this? or is
> that in the other direction.

It can help to reduce subdivisions, but more often on a more regular density mesh than one that is already being generated at a very coarse angle.

Basically the "Avoid smaller than" takes a distance, and any polygon that is smaller than that distance will instead shift to use a coarse angle of like 35 degrees rather than the regular angle.

This makes it possible for something like a tiny curved fillet in your model to get meshed at 35 degrees rather than say 5 degrees for everything else.

But if you are already doing everything at 50 degrees then it won't make any difference.

- Michael

Hi Micha,

re:
> (9) rotation object <...>
> Interesting is, that the Rhino command _QuadrangulateMesh can
> be used to eliminate some triangles. I have the feeling, if
> _QuadrangulateMesh works, than the mesher created to much
> triangles, more than necessary.

Some good news on this one, I was able to make a minor tune-up to preserve more quads in this situation.

In the next v2 beta this one will now get all quads, here is a screenshot of what it will look like:



- Michael

(3) rounded cube

The object I used is a Rhino object (copy&paste to MOI) and I attach it now. Could be nice, if the mesh could be same, dosn't matter it is a STEP import, a Rhino copy&paste or a MOI3D creation. ;)

(4) mesh #2

This object is a "o" from a Text object "Rhino". Interesting for me is, that you talk about independent edges and surfaces. I thought a surface or polysurface has edges, but you extract and delete them. That's I don't understand.
"I've attached an updated version of your object here as mesh#2_quads.3dm - to simplify the edges of this shape, I took your object, used Edit/Separate to break it into individual surfaces, then selected all the edges of those surfaces and did a Delete to "untrim" them and get the natural surface isoparms as a new simple trimming boundary, and then rejoined."

Could it be possible, that this is internal automatic done for meshing? Or can I do something at Rhino befor export an object to MOI? Are the edges needed for meshing? Is at MOI at command to select all edges?

(5) mesh #3

I attach the original surface, now it should be buggy. ;)

(6) mesh #4

Seems to be the same problem like at (4) and I don't understand it, why edges are independent from the surfaces and how to delete edges. I would like to avoid the effect per Rhino script befor I export the object to MOI. I afraid, at complex models the "separate-edge-delete-join" method could be a problem.

(7) I'm curious for the next beta. :)

(8) I used an angle of 35°. It's the "o" from a Rhino text object with a round.

This screenshot above was done befor I changed the CentroidTriangulation option at the ini. (Sidenote: the Rhino mesher create this overhanging peaks too.)
Interesting - now I tested the object again and it looks much better, also at a rawer angle of 55° - great! In regard to the easy usage of MOI it could be good, if the bad look of first screenshot could be avoided allways. ;)


Two other things to this - at Rhino I found, that some triangles can be converted to quads (maybe this is fixed in your internal version like you wrote at (9)) and could it be possible, that the CentroidTriangulation option is part of the general meshing parameters?

(9) :)
... one thought from me, as I bought MOI3D some days befor was, maybe it's a good time now for tests and improvement requests. :)
Thank you.

WOW, I increase the angle just for fun and the mesher dosn't create crap (case (8) - the Rhino "o" with rounds), it looks like it should look, also at this extrem raw angle. I'm impressed!



... but I found, that the result is not stable at all angle values - maybe it's possible that the mesher can be refined to avoid this effect. But's good to know, that it help to fine adjust the angle value to avoid some sideeffects.

Micha. You know all these type of shapes can be easily modeled in any poly modeler. You could also easily create directly in Rhino using combinations of rail revolve or sweep with polylines, with Surface Creasing plugin, then extract the control polygon. Weld the mesh, and send out ready for subd...

jonah

Thanks, the easy examples are examples only. I think, if the easy example works, than complex models should be good meshed too. Most important for complex scenes, where the polycount must be keeped low, but good looking - for example for rendering train / air plane interiors or large ships with many details.