Hi andrew - when you export N-gons to Modo, Modo will still triangulate them internally and actually renders the triangles.

It looks like probably in your case there Modo is generating a little skinny and maybe even reversed triangle on those particular N-gons.

I don't remember if there is a way to see or adjust the N-gon triangulation that Modo creates for the N-gon, but if there is a way to examine it take a look there and you might see that there is a little triangle there where you have those skinny render artifacts.

If you crank up the tessellation density a bit more then probably those areas will decrease in size to such an extent that you won't be able to see them.

Otherwise if that problem persists, I guess that you will have to use the "Quads & Triangles" output from MoI - MoI's triangulator is set up in such a way to try and avoid making little triangles along the far outside edges of N-gons (where there are things like 3 points of the n-gon nearly colinear like you have in various spots in your results here), and so that's why you get different results when exporting out from MoI as triangles.

It is possible to occasionally run into other problems with N-gon triangulation in Modo as well, but usually it is fairly rare though.

If you do crank up the tessellation density it will probably help you as well though, because when N-gons are slightly rougher and a little bit more twisted and non-planar it becomes more likely that Modo's N-gon triangulator might accidentally make a reversed triangle in a nearly colinear area.

I think that most people are exporting at a somewhat higher density than what you are showing there and you might try that to see if it helps if you want to keep things as n-gons rather than exporting triangles out from MoI.

- Michael

Also since it looks like it's a problem with Modo's N-gon triangulation mechanism, you should probably be able to repeat the problem if you manually created an N-gon with that same vertex structure.

Triangulation of N-gons is a fairly tricky business and there are certain kinds of shapes that tend to be more difficult to process and are more prone to making messy output, and one of those is areas where there are several nearly but not quite exactly colinear points in a row.

- Michael

But again though, easiest thing to try at first is to just crank up the density. What happens if you move the slider all the way to the right when exporting out of MoI, do you still see any of those kinds of artifacts or do they disappear?

- Michael

One thing to check is that weld points is switched on in MOI. In a couple of your renders, it looks like a vertex that should be attached to an edge isn't, leaving a long skinny hole (where you have highlighted it).

In Modo, try selected the edge where the artefact occurs. You should only have 1 edge selected, If is shows two, there is a triangular hole there.

Weld is on in all the examples. The think the reason things are going wrong is holes are made by non coplanar polygons. In this case there are 5 vertices on the fifth one is the pointed edge that is bridging the vertical fillet with the side of the knob.

Michael, I've searched through all the settings of Modo and can't find one that is going to fix this geometry problem. Can you please let me know, when tri + quads are selected as the export type do you generate any non co-planar quads? I suspect not, and this is why that export renders fine, but the n-gon doesn't.

Hi Andrew,

> Can you please let me know, when tri + quads are selected as the
> export type do you generate any non co-planar quads?

When you do tri + quads export, MoI can export non-planar quads - usually any quads created from a curved surface will be slightly non-planar.

But a non-planar quad can only be triangulated in one of 2 possible ways so there isn't really any way that the triangulation of a quad can be messed up.

Meanwhile an N-gon can be triangulated in many different ways, and unless it is done kind of carefully it can have glitches like some triangles leaking out of the boundary. I think that is what is happening to you in this particular case, Modo's N-gon triangulator is making a little triangle out of the 3 nearly colinear points along one side of your N-gons and that triangle may be oriented in the wrong direction.

MoI's triangulation is very careful to avoid making triangles that go in the wrong direction, so that's the reason why you don't see that particular problem when you do the triangulation inside of MoI.

Like I wrote above though, probably you will get better results from Modo's N-gon triangulator if you use a denser mesh - do you still have visible problems with the n-gons if you increase the overall mesh density, or do the problems essentially disappear when you use a denser mesh?

- Michael

Hi Michael,

Increasing the density of the n-gon doesn't help, and sometimes creates even more problems as shown in the attached images:




The only thing that works in modo reliably, as far as I can tell, is tri + quads. Any ideas how to get n-gons working? Is it possible to generate an all coplanar only n-gon geometry?

Hi andrew - it's only possible to generate planar n-gons on surfaces that are themselves planar.

Any kind of curved surface will have some amount of non-planarity to the n-gons that are generated from it. That's just what happens when vertices are placed at intervals along a curved surface.


It kind of looks like you've got a couple of separate problems there - your first screenshot there has the slight dark edge - that looks like it could be an n-gon triangulation problem.

I'm not really sure what the second problem that you are showing is though - that does not seem to be likely to be a triangulation problem but instead may be some kind of irregularity in your UV mapping or something along those lines.


I'm also not really sure why many other people typically use n-gons in Modo without running into problems. It looks like you are using some kind of heavy bump map or displacement map? How are you applying the UV mapping for that?

What about a totally plain non-textured render, do you see any artifacts at all on that style of rendering or not?

If the particular mapping/rendering technique that you want to use does not work well in Modo with N-gons then I guess that your best bet would be to use Quads & Triangles instead, if that solves the problem for you.

- Michael

Hi andrew,

There is really something fishy with modo here, I was able to get rid of some of the glitches by lowering the angle to 3, and using 'avoid smaller than' 0.1 to reduce the polycount ... but it reveals other reflection problem on the knob.



This only append once the reflection is activated, the shading (with specular) is looking right. If I triple the object inside modo, everything is rendering fine.

I've never seen something like that until now, but it's really weird.

I'm opening a bug report on luxo beta board right now, and I'll keep you inform.

I am doing the same UV mapping as stated in my first post, a cylindrical mapping around the knob in the direction of the lines shown. This is a one click operation in Modo using the "UV Projection Tool" and gives perfect results for tri + quad meshes from MOI, including the bump map for the brushed aluminium stock modo material, which I manually change the size of to 50 um bump depth - the whole knob is 14 mm in diameter. I've attached a closeup of the UV map of the area in question:



I was told that everyone was using n-gon export to Modo without issues, so I thought I would give them another and document it all in this thread for everyone to comment on. N-gon output still don't work properly with Modo for me no matter what I do. Even with a basic reflective surface only and no bump maps I still get incorrect lines rendering where non coplanar n-gons don't line up very well. If there are any other MOI + Modo users reading this thread can you please download the .3dm of the knob and give it a go with a basic reflective surface and zoom in on the same vertical fillet and let me know how you go? I always end up getting stuff like:



Michael, is it possible to split up non co-planar > 4-gons into a bunch of 4-gons? This may help for programs like Modo that can't seem to handle non co-planar > 4-gons very well. Any planar > 4-gons could remain to keep the poly count down for easier selection and UV mapping without going to the full mess that is tri + quads.

Mmm I'm looking a little more closely to the MoI tesselation, and I'm not really sure now if MoI is not doing something weird too on this particular model.
I have a 'deja-vu' feeling here, but I dont remember if it's 'normal' or not.

Here's a closup of the knob in MoI,



It's really strange that ngones and quad output dont give the same shading result in MoI either.

Hi Andrew,

There is definitively something wrong with the .lwo, and the way modo internally trianglulate ngones.
I think this is something that can be resolved, but from the luxology side.

For this knob example, you can try to export it in .obj, using ngones.
You will probably get some inverted normal in modo, just double hit the 'F' key to fix it ... but at least will not have this ngones shading problem, as the inernal triangulation will be fine.

At least I have concrete example now I can show in the modo bug report.

PaQ, thanks for letting me know you are having similar problems with rendering in Modo. We overlapped posts, thanks for already taking the initiative to go ahead and download the file and try it out with a reflective surface - really appreciated! Sounds like I need to join up the Modo bug reporting thingy too, and also report the issue with bump maps not working as expected with these n-gon type topologies as well.

I get identical results using either .lwo or .obj for export. Here is the same thing with a .obj file:

Hi Andrew,

Well you are welcome, I'm quite concern but the problem too ... and I'm wondering how do I miss that since 2 years or so :P

The bump is probably a side effect of this triangulation problem, not sure.
But a wrong (or not optimal) ngone tesselation produce UV stretching too ...

I've tried to isolate the problem as much as I can, like Michael teach me :)



So a question for Michael, because there is something I dont really understand :

In the .lwo created by MoI, is this internal triangulation of ngone write somewhere ? Why is this information lost, or no well translated from one software to an other ?

Well actually I have better result here, (using 601, using obj)



But ... it's not perfect everywhere, there are still some area when ngone create micro triangle that break the shading.



... better wait Michael now to have more advanced advice.

Between modo creating funky ngones, and MoI introducing strange glitches when using triangle export format, I never saw so many
problem for a so simple object :P

I get a perfect render from Modo 501 with this 3dm file. Export was ngons, 6 degree angle.




I also get a perfect render from 601 after changing the LWO option to Lightwave 10 to avoid the vertex nornmal issue.