What do you all think? Should 5 vertices faces be implemented in Blender like they are in Maya and Max3d?
yes sure…for highly detailed characters, cars etc. we always need faces with 5 or more vertices.it eases the pain of adjusting faces in the lesser detailed(smoother …ex: a fat belly in a highly detailed character) areas.it also helps reducing the number of triangles in a mesh and thus we can have less bumpy subsurfs.
but to achieve the this i think the whole mesh architecture has to be reconstructed…a major pain in ass :-? .
Kos: Regard the “problem” with subsurf and triangles, adding ngon support will not help, since those have to be triangulated anyway.
Non-coplanar quads are already a burden, I can’t even start to imagine sorting it out with an unlimited number of edges.
or even faces that aren’t even convex [simple example that isn’t too bad, the shape of a lima bean]
as for subsurfs, other applications would end up createing a vert in the middle with however many edges the face has coming out of it.
so, after smoothing once, you have all quads.
you have the same problem with 5+ sided faces as you have with 5+ edged verts with respect to smoothing.
I’ve used apps that had ngons and ones that don’t. imo, its not much of a benefit, ngons just hide the extra edges the software puts in, so it depends on how intelligently it does it.
On flat surfaces it doesn’t matter, but then, a few extra lines don’t either.
I’d rather see an edge data structure implemented, for true edge selection and editing, creasing and autounwrapping, which is something I am glad to see is in the works. (tks coders )
I tend to make most of my models in Wings these days, which has n-gon support. Obviously for using in Blender I then have to go over it making it all quads/tris but I find it comes out that way anyway. If most of your model (and I mean 99.999999%) isn’t quads you are doing something wrong. It takes experience to learn how to lay these things out, and how best to fill those hexagons you’ve just extruded etc.
3DS MAX is notorious for screwing up massively on “Meshsmooth” (the equivalent of Subsurf) if you don’t use all quads, it just becomes unpredictable, and nasty bumps/ridges appear. Maya does seem to handle these things quite well, although it’s a while since I’ve played with either, so things may be better.
So eh… why doesn’t blender do this neat little trick??? cause tri’s are a bit of a problem as is.
only need triangles.
triangles are perfect. all of them. other polygons are evil.
ngons are so helpful. They assist in speeding up the production of a model. Then after that you can make edge loops to have it animate better.
celeriac, tris are only ‘perfect’ for in game, i.e. unsubdivided models. when you go into subDs, it is quads that make for predictable subdivision.
I hope I am right… Ngons will be helpful in that cloth pluign. I think I remember the author saying that since blender does not supports ngons it can not work on the meshes directly.
Again I am not sure, but I though I had read it somewhere. %|
Any way as kos said we need them for more detiiled yet simplistic meshes, because some times it gets difficult to manage the surface smoothness due to nasty tri’s. But some can disagree as there is always a way to solve the problems :P.
Any way as kos said we need them for more detiiled yet simplistic meshes, because some times it gets difficult to manage the surface smoothness due to nasty tri’s. But some can disagree as there is always a way to solve the problems .
not always man.takes a LOT of time to figure out the right way…and thus not very productive.while modeling some very skinny areas of the human body…like the toe and hands where bones are quite pronounced, it is almost impossible to solve.
I am 99 percent certain that the catmull clark subdivision algorithm understands a 5 sided patch and n-gons for that matter. Given this the output of a subsurfed n-gon will be VASTLY different than if you tesselated it beforehand. So, the idea that we would ‘just be tesselating the n-gons anyway’ is a bit of FUD and dosn’t really amount to much IMHO.
Personally when I model something in a program that has n-gons I sometimes find myself with a lot of 5 sided patches and triangles to begin with even if I get rid of the n-gons/triangles later. It keeps things clean at the early stage.
Also something people seem to forget is that any vertex with a valence of < or > than 4 will generally produce pinching in the derived surface, regardless of whether or not you have all quads. You cannot have a closed polyhedron comprised of all quads with all vertices having a valence of exactly 4 unless it is topologically identical to a torus! Therefore it is sometimes convienent to use an n-gon/5 sided patch where you cannot tolerate ‘extraordinary vertices’.
5-sided patches do create problems with texture stretching on animated meshes from what I understand though.
Regardless, there are about a million reasons I can think of to have n-gons in Blender.
Just my 2cents.
The bottom line is that n-gons make things easier for the user, even if we assume that it doesn’t add any technical advantages. I would very much like to see n-gons in blender. However, it is not particularly high on my list. There are many other things that I think are more important.
I totally agree with Cessen here.
While n-gons may not be (3d animation) philosophically correct, the point remains that it would make the life of the user easier and that’s all that matters. As someone has mentioned before, n-gons would also be useful during the modelling process… this way Blender will not keep creating random tris as you cut faces, which you need to get rid of manually later anyway to keep with the edge-loop method.
Also another thing to keep in mind is this thread over at CG talk :
It’s very informative reading for all charater modellers, but the major point is that Stahlberg has experimented with an n-gon topology, and I’m sure everyone will agree that his results are “acceptable” to say the least.
So, it looks like the aesthetic stumbling blocks for subdivided n-gons might not be as strong as it was previously thought. N-gons shouldn’t be totally dismissed just yet.
I agree, n-gons are very confortable to work with…
blender doesn’t have tools well suited to ngons
you see, in wings3d you connect two vertices across a 4+gon to make it multiple polygons, that just isn’t the way things are done in blender now
to convert a quad to triangles you do just that, you don’t make the edge across the quad yourself.
however, I feel blender could be benefited by the workflow changes required to support ngons, but it isn’t there yet
yes, this joins the discussion there’s been around switching to winged-edge structures, some also talked of hybrid structures… anyway. These changes are needed I think, and of course they’ll imply workflow changes, but if there’s more benefits than disadvantages let’s go for it, but let’s go there slowly
I personally am against winged edges- a bit heavy a structure, I think. and making blender’s modeller a clone of wings’s will make me highly uncomfortable; I cannot work fast in wings, plus we already have that option, so why clone it? Stahlberg unfortunatly is fanboyed to death, he is very unconventional and I don’t think his modelling style is to be emulated as “textbook”. Plus, I don’t know if its his animation style or something about the model setup, but I have never seen a well animated Stahleberg model.
I don’t think blender will go to winged edges anyway, but something similar yet lighter and faster (I hope). Also I have no problem with having n-gon support, but I think having the speed and flexibility of blender is something not to be sacrificed for this.
There are other cool features a modeller can have besides n-gons, such as hierarchical or selective subdivision. Anyway I’m rambling a bit, forgive me.