Sun, Feb 16, 5:50 PM CST

Renderosity Forums / Poser - OFFICIAL



Welcome to the Poser - OFFICIAL Forum

Forum Coordinators: RedPhantom

Poser - OFFICIAL F.A.Q (Last Updated: 2025 Feb 15 11:01 am)



Subject: Quads Vs. Tris: Has anyone bought, or Sold, a Tri-Polyed Cr2?


Photopium ( ) posted Tue, 18 January 2011 at 11:20 AM · edited Fri, 07 February 2025 at 9:47 AM

I've been holding off selling clothing produced from Marvelous Designer/Clo3d because it outputs in Tris.  However, the few cr2s I've put together for myself seem to do just fine.

We've all heard, forever, that you need quads for conforming clothing and tris for Dynamic.  I understand quads in dynamic slow down simulations but what is it about Tris in conforming?  Is this maybe antiquated information from Poser's early days?

I'm seeing stuff in the store that looks like it came right out of Marv, and God Bless the vendors, but I'm wondering if anyone can confirm that and, if so, what sort of experience are you having with the products?

(Please note:  This is not a thread to discuss the pros/cons of Marv.)

 


PhilC ( ) posted Tue, 18 January 2011 at 11:31 AM · edited Tue, 18 January 2011 at 11:32 AM

I feel that it is mostly user perception. Poser should be able to handle both equally well. Does Poser effectively triangulate at render/simulation? Did I hear that somewhere?

Quads look prettier in wire frame and are arguably easier to edit because you can see the "flow" of an object easier.

Long thin polygons or tris and  any ngons are best avoided.

I think your best argument is to show what it looks like rendered. Proof of the pudding so to speak.


markschum ( ) posted Tue, 18 January 2011 at 11:51 AM

It really does not matter.  Depending on how they are arranged quads and tris in cloth simulations will give different folds. In conforming clothes there should be no difference .

If the item works in Poser then who cares. The only difference might be the polygon count , and I believe vertex count matters more for performance but I could be wrong.

I would rather have a nice triangulated mesh than one of non-planar quads :(


Photopium ( ) posted Tue, 18 January 2011 at 11:55 AM

Maybe you guys can correct me if I'm wrong, but I think the "No Tris" argument was back in the day, when Poly counts were fewer and system resources scarce.  Nowadays, you can really pack the polys in and, even if they're tris, they shouldn't distort noticably because they're so tiny.  Plus you can crank up shading and so forth.

I mean, do some people just insist on acting like it's the year 2000?

Planar Quads are delightful, to be sure, but IME nothing more "automated" produces them.  I have a retopo tool for max that almost works, but the resulting quads aren't quite planar, they are spiraled, which is maddening.


bob1965 ( ) posted Tue, 18 January 2011 at 12:03 PM

Game engines use tris almost exclusively with good results but then the textures tend to be better and hide artifacts that can accompany tri based meshes. By better I mean making full use of displacement, normal, bump mapping etc.

Depending on which school of thought you follow render engines convert to tris anyway.

Quads do seem to give better morph results though but I suppose with a sufficiently dense tri mesh even that wouldn't be an issue.

There is also a general preception that quads animate better but since most content ends up in static images I can't say if that's really a concern in general Poser usage.

I think it would be worth pursuing given that the automated grouping tools available would make set-up less of a headache than pure manual selection in the tri meshes produced by MD and if there are problems at the joint areas the construction of the pattern pieces can help alleviate those at the design stage.

Comes down to more or less what PhilC indicated...let your results answer the question.

 


PhilC ( ) posted Tue, 18 January 2011 at 12:06 PM

If I saw a well formed model that I wanted to purchase for animation and rendering it would not matter to me if it were tris or quads.


lkendall ( ) posted Tue, 18 January 2011 at 12:10 PM

PropViewer 3.2 by Kawecki (freeware) seems to have the ability to convert objects from Tri's to Quad's.

http://www.renderosity.com/mod/forumpro/showthread.php?thread_id=2819480

LMK

Probably edited for spelling, grammer, punctuation, or typos.


vintorix ( ) posted Tue, 18 January 2011 at 12:24 PM

 

Go for it! Quads are something that is approved by the designer/modeler. Easier to work with and looks nice. But whatever other pro & cons, the computer always prefer triangles and performs better with it. Remember that the render always must convert to tris before it can do anything.


bob1965 ( ) posted Tue, 18 January 2011 at 12:30 PM

There are several apps that can convert tris to quads even though that one is useful for it's export to Luxrender among other things.:biggrin:

The discussion here is whether the direct output of Marvelous Designer is suitable for use in creating conforming clothes for Poser as it runs contrary to the accepted knowledge.

Additionally Marvelous Designer produces a mesh that appears to be triangulated using the Delaunay algo or something very similar...it's better to leave a mesh like that as tris than to convert to quads, the results are beyond ugly on conversion. :scared:


MagnusGreel ( ) posted Tue, 18 January 2011 at 12:34 PM

NO

stop!

we just had a thread locked because of arguing over that!

Airport security is a burden we must all shoulder. Do your part, and please grope yourself in advance.


vintorix ( ) posted Tue, 18 January 2011 at 12:37 PM · edited Tue, 18 January 2011 at 12:38 PM

 

I big drawback with converting is that it destroyes the perfect UV maps you get with the Designer. Using PhilC's Obj2Cr2 for rigging I cannot see that quads function any better than tris.


bob1965 ( ) posted Tue, 18 January 2011 at 12:47 PM

Quote - NO

stop!

we just had a thread locked because of arguing over that!

 

Huh? What? Did I miss a memo? Site mail me if the explanation is something that will trigger another us versus them battle. :laugh:


vintorix ( ) posted Tue, 18 January 2011 at 12:53 PM

 

The Moderators are far too trigger happy. They destroy all interesting debates.

 


bagginsbill ( ) posted Tue, 18 January 2011 at 3:29 PM · edited Tue, 18 January 2011 at 3:33 PM

Quote -  

Go for it! Quads are something that is approved by the designer/modeler. Easier to work with and looks nice. But whatever other pro & cons, the computer always prefer triangles and performs better with it. Remember that the render always must convert to tris before it can do anything.

Stop repeating this misinformation please. REYES renderers convert everything to QUADS. Tiny ones. Smaller than a pixel. Firefly does not convert anything to triangles.

A non-planar quad renders as a membrane, not as two triangles.

The Poser cloth simulator shows evidence of converting quads to triangles, but it does not behave the same as a mesh made of triangles to begin with. Further, the triangulation is only for the purpose of doing collisions - not for rendering. This is one reason cloth sims poke through - different virtual geometries are used for sim versus render.


Renderosity forum reply notifications are wonky. If I read a follow-up in a thread, but I don't myself reply, then notifications no longer happen AT ALL on that thread. So if I seem to be ignoring a question, that's why. (Updated September 23, 2019)


alexcoppo ( ) posted Tue, 18 January 2011 at 3:57 PM

On the guerrillacg.org site there are two extremely interesting video tutorials subdivision-surfaces-overview and subdivision-topology-artifacts. Watch them carefully and you will add non-quads to the list of the no-no things in computer graphics (the other being booleans, which create trackloads of triangles).

If a renderer likes to convert quads to tris (like Terragen2) it is its business; when modelling go for quads.

Converting tris to quads modifies faces not vertices so UV mapping does not change (unless the converter is broken and mangles UV coordinates).

GIMP 2.7.4, Inkscape 0.48, Genetica 3.6 Basic, FilterForge 3 Professional, Blender 2.61, SketchUp 8, PoserPro 2012, Vue 10 Infinite, World Machine 2.3, GeoControl 2


vintorix ( ) posted Tue, 18 January 2011 at 4:10 PM

"no-no things in computer graphics"

The only thing that most definetly should  be no no in computer graphics is to render a human face in Poser Firefly. ;)


markschum ( ) posted Tue, 18 January 2011 at 4:30 PM

For the original question , I would test the model myself, then bundle it to a few beta testers. Then if no major problems pop up, submit it to a store , and see if it passes the store testing.  Add some notes to the readme if it handles the cloth room differently and otherwise dont worry about quads or tris.

I have quite a few models in 3ds format and they all work fine in Poser.


scanmead ( ) posted Tue, 18 January 2011 at 5:09 PM

There is a wide disparity between platforms and render engines on handling tris vs quads. Some don't bat an eyelash, others have fits.

The point is, this would be for Poser. Find 2 or 3 people who know how to push limits, and destroy things and have them test the heck out of one or two pieces. Test them as dynamic, or conforming, whichever you're trying to produce. If they work, they work. The final render, and how they conform (or drape) is what matters.

If you want to be really nice, you might have a couple of people try them is D|S, Bryce, Vue, Max, and Cinema, and include a caveat if there are any unusual glitches for those.

Personally, I drool over tidy meshes, but I use what I need, and as long as they render nicely, who cares?


Photopium ( ) posted Tue, 18 January 2011 at 5:27 PM

Very encouraging responses, folks, awesome! 

 


Photopium ( ) posted Tue, 18 January 2011 at 5:27 PM

Very encouraging responses, folks, awesome! 

 


vintorix ( ) posted Tue, 18 January 2011 at 5:45 PM

 

William_the_Bloody,

Don't forget to weld - get rid of the seams from the Designer before you use Obj2Cr2! In Cinema 4D you can do this with Optimise (Value 0.1-0.2)

 

lkendall,

Thank you for the tip of PropViewer 3.2, it does indeed convert to quads without inflating the polycount or destroy the UV map!


Photopium ( ) posted Tue, 18 January 2011 at 6:14 PM

No C4D for me, though I'm sure it's a delight.  I use Max.  I would like to be more into Zbrush but I feel that I'm just not ready for it. 


vintorix ( ) posted Tue, 18 January 2011 at 6:43 PM

Oki. Max is the best. I have it on my computer and would love to learn it but just haven't got the time. There are so many programs! I know some elementary things though.

 


DarkEdge ( ) posted Tue, 18 January 2011 at 6:56 PM

It kind of depends on the modeler, design and the maps (imo). The BEO2K10 figure (DAZ) is all tri's (at least the orig one was) and it renders out nicely. However, I have seen other tri models that look like sh*t.

Zbrush's decimation Master will convert to tri's and you usually can't tell, depending on how much you decimate that is. Sure if you look hard enough on an untextured model you will see some signs, but the maps usually cover that stuff up.

Generally speaking modeling in quads has been the mainstay and use tri's only when you need to and try to hide them when you do use them.

Comitted to excellence through art.


Terrymcg ( ) posted Tue, 18 January 2011 at 7:31 PM

I think even Apollo has quite a lot of tri's and he works just fine. I was also told to avoid tri's like the plague, but never really understood why.  I was under the impression that texture artists didn't like tri's. But I suck at texturing anyway, so I didn't really look into it.

Some of the very good poser stuff that I bought (clothing or props), has had lots of tri's in the geometry. Despite the tri',s the items work fine.

 

 

D'oh! Why do things that happen to stupid people keep happening to me?


kawecki ( ) posted Tue, 18 January 2011 at 7:55 PM

Quote - Remember that the render always must convert to tris before it can do anything.

You can render quads in a direct way witout converting to triangles.

In 2000 was much faster to render two triangles than a quad, in 2011 I think that this is not more valid and quads consume less memory than trinagles and the limiting factor of rendering speed is memory access and not CPU speed.

Stupidity also evolves!


vintorix ( ) posted Tue, 18 January 2011 at 8:08 PM

 

Well I hope it will work, then half of the problem is solved, - we will at least have decent apparel. But we are still left with the unnatural doll-figures. But half the problem is better than nothing. ( I always try to see the posetive!)


aRtBee ( ) posted Wed, 19 January 2011 at 2:31 AM

Quads are liked more for texturing / UV-map works and ZBrush-works. As BB said above, it does not effect Firefly rendering.

Quads and various tri's behave different in dynamic clothing. 

Read the http://www.renderosity.com/mod/forumpro/showthread.php?thread_id=2817155 thread about it, tip: start reading at the end.

In short:

Quads behave rubberish which is good for non-woven cloth like rubber, leather and fleece, and show a stable behaviour for mesh densities of 4cm or better. At lower densities they might freeze up.

Tri's behave like woven cloth, where pure diagonals are very jumpy and elastic (instable, you need lots of frames for a good result) and zigzags behave best.

Hex at last make the best knits (woolen sweaters, tricots).

All geometries respond differently to the shear/stretch/fold cloth room parameters. Increasing mesh density makes thinner cloth, not thicker.

Have fun.

- - - - - 

Usually I'm wrong. But to be effective and efficient, I don't need to be correct or accurate.

visit www.aRtBeeWeb.nl (works) or Missing Manuals (tutorials & reviews) - both need an update though


heddheld ( ) posted Wed, 19 January 2011 at 4:46 AM

Like a lot of peeps I had a good play with MD when it hit the forums

will admit to being impressed with its abillities and even made a couple of the dresses I made into conforming and dynamic  they worked fine (not better but no worse) the dynamics in MD is a LOT better then posers but then I guess its cloth sim an render are designed for the mesh it makes

I dont use autogrouping tools and doing the grouping by hand was a pain

things like shoulders/collars are very hard since the mesh isnt symetrical

never tried to add morphs but think the lack of any symetery would make that a lot more dificult

and lastly is the price, for me at least I would never spend so much on such a limited tool

the fact is it was made for fashion houses where one dress design might sell 1000's or 10000's of copies, IF they stripped out all the bits I dont need for poser an put it up at a sensable price I would consider it


vintorix ( ) posted Wed, 19 January 2011 at 4:55 AM

"I dont use autogrouping tools and doing the grouping by hand was a pain"

Marvelous Designer must be used with the OBJ2Cr2 tool, It is the only one that works.

Not using autogrouping tools at all seems like a no-brainer to me. Even if you know how to do everything you could use the tool as a head start and go from there.

I would be interested to see something you have done, do you have any links?

?

 

 


heddheld ( ) posted Wed, 19 January 2011 at 7:42 AM

well looked up this link its from last oct http://pfdlives.com/pfd/index.php?topic=838.msg9534#msg9534

 (dunno if you'll see it without signing in! but its not a bad group, ps I'm only a member there so cant change the rules ) got rid of MD an all files made with it around christmas time, was no chance of me ever buying it lol

if you cant see it guess I could rip the pics out and post them here

as for autogroup editor I'm sure its a usefull tool but I can group without it and my time is a lot more plentiful then my money lol

Obj2cr2 is good (like most of Phils stuff) but again I do it in a text editor with help from cr2 editor and morphman

Do plan on buying morphing clothes tho!! just tried adding them by hand and its soooooo tedious and not as easy as I thought it would be and it seems like I get some working an break others 

 

 

 


vintorix ( ) posted Wed, 19 January 2011 at 8:07 AM · edited Wed, 19 January 2011 at 8:10 AM

"Before you can login and start using the forum, your request will be reviewed and approved. When this happens, you will receive another email from this address."

I'll have to wait! :)

There are many different kinds of birds in the Lords garden, ups, I meant Poserworld. Not everyone have the ambition to paint like the old masters. Occasionally I forget that.

People work with so different things and from so many different viewpoints they could just as well be from Mars and Venus.


heddheld ( ) posted Wed, 19 January 2011 at 8:43 AM

"People work with so different things and from so many different viewpoints they could just as well be from Mars and Venus"  or maybe even a little farther away lol

Sorry about you having to join the forum (mind u might like it lol ), I have joined so many its a joke, then u find an interesting thread go to make a reply an notice the last post was like 3 yrs ago  opps !back out slowly lol

guess I still do it the old fashioned way but that was the only way when I started lol , did make a couple of things for v4(when she came out) but NO MORPHS couldnt get my head round them(then left it all alone for a while , heath issues, starting again with clothing, doin some bits for m4 but again the morphs are driving me nuts

if I was just making stuff for myself would build the cloth around the pose!! so much easier

 

Mus uni non fidit antro  ok ok so I looked up ur latin (not done it since school) 


moogal ( ) posted Thu, 20 January 2011 at 5:20 PM

Quote - I mean, do some people just insist on acting like it's the year 2000?

Yes, but I'm trying to stop doing that.

The model in my avatar (small, I know) was modeled in Caligari in 1994 and only has 3400 polys.  But if you really want to talk about people living in the year 2000, it's the people who still make excuses for the sorry state of cloth and soft-body dynamics in all four of the main figure posing programs.  Just sayin'.


moogal ( ) posted Thu, 20 January 2011 at 5:25 PM

Quote - never tried to add morphs but think the lack of any symetery would make that a lot more dificult

If you are talking about the morph brush, I don't think symmetry would be a problem.  It always seemed to me that turning symmetry on simply mirrored the previous morphing actions of the user, and wasn't dependant on the mesh itself.  I could be way wrong here...


MagnusGreel ( ) posted Thu, 20 January 2011 at 6:30 PM

..and thats when the fight started.

 

lets not go there ladies and gents!

Airport security is a burden we must all shoulder. Do your part, and please grope yourself in advance.


StaceyG ( ) posted Thu, 20 January 2011 at 7:21 PM

I had to delete a post in this thread that was highly inappropriate so just a reminder to everyone to please follow the TOS so this thread can continue in a productive way.

 

 

Thank you


corinthianscori ( ) posted Thu, 20 January 2011 at 7:38 PM · edited Thu, 20 January 2011 at 7:50 PM

file_464254.jpg

> Quote - No C4D for me, though I'm sure it's a delight.  I use Max.  I would like to be more into Zbrush but I feel that I'm just not ready for it. 

 

W, I use Max to make everything(rarely Zbrush). I can tell you from experience that Poser 6 doesn't seem to care about quads or tries when it renders.

On the subject of Poser's ability to handle those triangles or quads while working with dynamic clothes...

...Poser is better with triangles, generally. But PhilC is right: whichever result look best is the better option to go with. You can always run a quick cloth sim in Max to test your results in quads and then in tris. Thankfully, Max can just jump between both without a problem so do a quick cloth sim is easy and the results are comprable to Poser's.

Not to mention that Max does cloth sims MUCH faster than Poser(6, anyway) so get to testing!  Also if you make those radiating triangles/quads then you'll be MUCH better off in Poser's cloth room.

 

Below, here's a simple plane(imagine this is some outfit!! >..<). Just triangle the faces, and voila: instant fake quads that LOOK AND behave like triangles which would help cloth sims in Poser.


corinthianscori ( ) posted Thu, 20 January 2011 at 9:50 PM

I meant to mention that I used Turbo Smooth modifier on that middle plane to get Max to make those radiating quads you see in the third subdivided mesh.


heddheld ( ) posted Fri, 21 January 2011 at 2:17 AM

not sure why you call them fake quads !!

have tried the same thing in hexagon for use in posers cloth room

not a lot of testing but it seemed to make little dif to the cloth sim on items will a similar number of polys

only time I had probs with tris in poser is sometimes close to edges have had artifacts(blamed the tris I had left an removing them sorted the render ) and now an then with a 3ds model I had d/l'ed it had really long narrow polys and poser didnt like them


heddheld ( ) posted Fri, 21 January 2011 at 2:37 AM

Quote:

If you are talking about the morph brush, I don't think symmetry would be a problem.  It always seemed to me that turning symmetry on simply mirrored the previous morphing actions of the user, and wasn't dependant on the mesh itself.  I could be way wrong here... :unqoute

no I ment the mesh that MD gives out , it uses some fancy algorhythm (forgot its name) and is nothing like the nice neat meshes hex does, just made it hard to group using my current methods maybe things like autogroup would have helped


Schecterman ( ) posted Fri, 21 January 2011 at 12:08 PM

What Poser really needs is catmull clark subdivision surfaces that can be turned on or off and adjusted for subdivision level.

Then you use all quads or mostly quads, avoid "stars" or at least hide them in less noticeable areas, and you have no pinching, no non-planar polys, plus a better result with displacement maps.

...


Privacy Notice

This site uses cookies to deliver the best experience. Our own cookies make user accounts and other features possible. Third-party cookies are used to display relevant ads and to analyze how Renderosity is used. By using our site, you acknowledge that you have read and understood our Terms of Service, including our Cookie Policy and our Privacy Policy.