PDA

View Full Version : Ngons probably won't be the answer some people think it will be



hrgiger
08-30-2005, 07:15 PM
Since we are getting the abilty to use ngons in our subpatch routines in Lightwave 9, I think something should be said about them. Now I haven't used them in other packages but I'm just making an educated guess that they will not be ideal in a lot of situations. Perhaps some people think "ah, finally I can now boolean a subpatch object and I won't have to convert everything back to 3 or 4 point polygons". Or even, "ah, I can make this high detail hand, and run all that geometry into a much lower detailed arm by just butting my lines up against what was orinally a 4 point poly."
I know for a fact that they will be very useful for me and I'm anxious to have use of them but I know that they are still going to be useful in a limited number of places.
I just think people should be aware that thought will have to be excercised to how they are placed and good modeling practices will apply. N-gons won't be a reason to get sloppy or lazy.
I would like to hear from others who have used them in other packages, what their experiences with them are and how they used ngons and where they were good and where they were not.

Surrealist.
08-30-2005, 08:17 PM
We'll I have had similar same thoughts. When I stopped and thought about it, I realized that there would still have to be deformation taking place and that more points on a flat poly willl present other problems when you want it to deform properly. And I think you are right. I think of ngons as similar to triangles. Use them is you have to is certain spots otherwise avoid.

One situation though where I think they may be invaluable is in the creation of non-organic shapes that have flat and curved surfaces such as furnature.

Will be interesting what responses you get from people who have actually used them.

Nemoid
08-30-2005, 08:44 PM
well i used ngons when evaluating -cough cough- modo and see its modelling workflow.
ngons come in handy in inorganic-but rounded objects which do not deform : so cars, design objects and so on.

so they can be useful, but not for organics, because they have to deform well. so far, subpatches algorithms do have probs with topologies including triangles sharing the same points, and even quads in many occasions, so i don't see easy deformation solutions for n gons as well.

this being said, design objects inorganic rounded ones and more can exploit this feature and allow fo less polygons and therefore if all is well optimized, faster rendering. :)

another advantage thet could suit you but depends highly from your modelling workflow, is that u could see every part of your organic model subpatched while using edge tools too. soactually, cut your mesh how you want while still subpatched. i dunno exactly if this would happen in Lw 9.0, but it surely is possible in Modo.
the downside of this is that u could get used to a bad modlling habitude, since the organic model has to be quads at the end.
dealing with ngons and getting rid of them depends mainly in if u're used to make box modelling slitting polys a la maya so using edge tools extensively.

i'd rely on traditional modelling techniques while leaving these possibilities to
fix little polyflow problems which will be easier and fun . :)

So welcome to ngons especially if used cleverly.

Surrealist.
08-30-2005, 10:36 PM
I would welcome using them in subpatch while I use add edges and so on as I do alot of that anyway. So yeah, that's a great potential right there for me.

Chazz
08-30-2005, 10:40 PM
I'll try and post some examples from Maya for you guys to see, but ngons can be used in organics without a lot of trouble. I wouldn't go using them all over the place in your joints, but in flat areas like the face or back, ngons are okay.

Surrealist.
08-30-2005, 10:47 PM
I'll try and post some examples from Maya for you guys to see, but ngons can be used in organics without a lot of trouble. I wouldn't go using them all over the place in your joints, but in flat areas like the face or back, ngons are okay.


Cool please show us, that would be great. That's kinda where I was going with comparison to tris so yeah, lets see.

BazC
08-31-2005, 01:06 AM
I use them all the time in Wings and Silo but I nearly always split them into quads as far as possible when I finish the model.

Mylenium
08-31-2005, 03:46 AM
Since we are getting the abilty to use ngons in our subpatch routines in Lightwave 9, I think something should be said about them. Now I haven't used them in other packages but I'm just making an educated guess that they will not be ideal in a lot of situations. Perhaps some people think "ah, finally I can now boolean a subpatch object and I won't have to convert everything back to 3 or 4 point polygons". Or even, "ah, I can make this high detail hand, and run all that geometry into a much lower detailed arm by just butting my lines up against what was orinally a 4 point poly."
I know for a fact that they will be very useful for me and I'm anxious to have use of them but I know that they are still going to be useful in a limited number of places.
I just think people should be aware that thought will have to be excercised to how they are placed and good modeling practices will apply. N-gons won't be a reason to get sloppy or lazy.
I would like to hear from others who have used them in other packages, what their experiences with them are and how they used ngons and where they were good and where they were not.

Well, I wouldn't see things as dark and gloomy. N-gons can indeed be a positive thing even on organic models if placed in strategically wise places. You can't overcome some of their limitations for mathematical reasons and that's not going to change no matter how sophisticated implementations may become in the years to come. Most people will try and avoid them even in the future, but in general it's agood thing to be able to resort to them.

Mylenium

hrgiger
08-31-2005, 05:06 AM
Mylenium, I hope you don't misinterpret what I was saying. For me, ngons are a godsend. I can see many different reasons and ways to use them. I was just suggesting that a lot of people that aren't the most adept modelers might think they can use ngons just as easily as a typical quad which I don't believe is the case. But then, I've always been a big advocate of using triangles in strategically placed locations and have always argued against the theory that you must model in all quads. There are dozens upon dozens of places on a organic model that experience little to no extreme deformation that ngons will be very useful.
Normally I am dark and gloomy, but with ngons, that's not the case.

Librarian
08-31-2005, 01:00 PM
For me, ngons are a valuable workflow enhancement. I use them in other packages. Not as an excuse for bad polyflows, but they help in the modeling process.
In the workflow I use, theyīre useful, and this workflow isnīt my prefered method in LW .
I usually start with a base mesh and refine the edge loops freely with freehand cutting tools like add edges. Itīs obvious that this produces >=5point polys during modeling.
You can model that way in LW and some people do.
I donīt prefer it cause I regulary need to check how the mesh looks subdivided.
Some modelers donīt need to check their subivided mesh, but Iīm not good enough to estimate the subpatch look.
I need instant visual feedback of the subdiv mesh without polys and subpatches mixed together.
Therefore I model a bit different in LW.

Furthermore you can use ngons sometimes to save polygons.

Would be nice to have. Not the most important feature on my list, but helpful.
Youīre right that ngon subs are not the Holy Grail and even in other packages, people are usually trying to keep most parts in quads in organic modeling.
I hope NT will implement them the way Mojo did. Subpatches and Subdivisions.
So that people who fear ngon subdivs cause of their 'bad attitude' nature will be satisfied, too :D

Nemoid
08-31-2005, 03:12 PM
I'll try and post some examples from Maya for you guys to see, but ngons can be used in organics without a lot of trouble. I wouldn't go using them all over the place in your joints, but in flat areas like the face or back, ngons are okay.


Now this is very interesting so pls post them ! i read in some CGtalk thread that is possible to use even ngons for organics. however maybe also its the case with maya subpatches which are particular and quite different than Lw ones. Lw deformation system is different than maya too. no lattices, constraints...

AFIK, even if i think some triangle in strategic places isn't the devil at all, i like more modelling in quads only(if i reach to) for organics.

i'm also wondering on how the new Lw subpatch algorithm will be. in facts since it supports ngons it should be quite a different beast from the current one. this could bring several good things. :)

Lewis
09-11-2005, 12:35 PM
For MY kind of work N-gons in SubDs will be great addition and i'll be faster in modeling 'coz i can CUT / bandsaw and KNIFE directly in SubDs mode without having HOLES when i get to N-Gon :).

But at end i'll probably clean all N-gons 'coz i do lot of modeling work for different render packages so All Quads/Tris is sure ticket ;). I do have automatic cleaning method of ngons which works in most of cases. Select all n-gons in statistic panel, hit triangluate and then click on free plugin called "MergeTrigons" which will make MOST of Ngons into Quads + few Trigons (depends of object) - so whole cleaning process takes 5-6 seconds :).

BTW Nemoid - Press release mentioned (AFAIR) Catmull-Crark method so i guess it should be same as in XSI (i'm not sure is modo using that same way beside LWs Subpatches method)

liquidpope
09-11-2005, 07:17 PM
http://vbulletin.newtek.com/showthread.php?t=33681&highlight=subdivide+n-gon

Did you know that you can already subdivide an n-gon?

mkiii
09-12-2005, 05:32 AM
AFAIK everything be it quad or ngon is going to be converted into tris at render time isn't it? So the issue is purely whether or not it enhances the workflow, and makes the mesh that you are working on appear to be that much simpler. Which it should of course.

I'm guessing that the same old issues that require you to manually triangulate the mesh if you want complete control over the way it deforms & renders are still going to be there.

This is the case now for characters I create for games. The mesh may start off as nice neat quads, but you always have to remember that the final output to the game engine is going to be tris.

wacom
09-12-2005, 09:48 AM
I'll try and post some examples from Maya for you guys to see, but ngons can be used in organics without a lot of trouble. I wouldn't go using them all over the place in your joints, but in flat areas like the face or back, ngons are okay.

Yes I have faith it will be just fine, as XSI also handles n-gons OK up to a point for deformation. I've seen 25+point n-gons working just fine for static geomtry though. The further you get away from 4/3 verts the more likely you are to have problems though- nothing a simple quading or tri tool can't handle though!

Lewis
09-12-2005, 01:25 PM
http://vbulletin.newtek.com/showthread.php?t=33681&highlight=subdivide+n-gon

Did you know that you can already subdivide an n-gon?

Heheh that is totaly different thing. Show me how tu subDivide N-Gon interactive with TAB key (on/off) :). This is just like freezing - not interactive and not too usable :).