PDA

View Full Version : Cloth Topology


Stahlberg
05-15-2004, 07:36 AM
This shows what I aim for when modeling wrinkles and folds, except this one is idealized - it's all quads, and very symmetrical and homogenous. Normally I wouldn't bother with quadness, plus I try to make each fold a little different (the more different the better, since we all have an unhealthy tendency in the opposite direction, towards unnatural homogeneity).

http://www.androidblues.com/clothfoldtopol.jpg

I know derbyqsalano has a similar method, derby?

The recurring problem is, almost all cylindrical cloth objects (like sleeves, skirts, jackets and pants-legs) are better modeled with a topology grid slanted at 45 degrees to what we get on our standard primitive cylinders. It's very time-consuming to create such a 45-degree cylinder as this one:
http://www.androidblues.com/45degrtop.JPG

You could just save this cylinder when you're done, and reuse it - which is what I do - but if someone has an easier method to create this, please let me know. I tried the spiral-maker in Bonus-Tools, 2 curves going in opposing directions, then copy/rotate... but how would I turn the resulting NURBS curve network into a surface? I used to be able to do it with PowerAnimator, but it seems not in Maya. Maybe Byron of BPT could create another cool tool for us? [smile]

Another couple examples of the edges running at 45 degrees to the normal vertical/horizontal:
http://www.androidblues.com/jackettopol.JPG

Stahlberg
05-15-2004, 07:37 AM
And finally one that started from a vertical/horizontal layout, but then tweaked. It's possible to do it this way too, but I think it would have been cleaner and given me more freedom to get looser folds if it had been tilted 45 degrees.
http://www.androidblues.com/bootstopol.JPG

Everybody post your wrinkles!

flingster
05-15-2004, 01:50 PM
steven well i've been following the whole body topology thread for a while now with great interest and trying my best to apply it to c4d and its quadyness! heh heh...so i'm curious to see how this one goes...

one thing you mention is the topology of cylindrical surfaces like sleeves to be on an angle...personally i think you just hit the nail on the head so to speak with this comment....as if you go to a "good" tailor they will cut your cloth on the "bias" so it hangs better on the body when in suit form...it costs far more as a result as more material is used in the process..but the end result is more appealing. So this concept is actually used in clothing making....not high street manufacturing..but it does make sense when extended to this process also...
"bias" means = cutting obliquely across the warp.
hope this helps a little..so youre not just doing it for results..but its technically correct also...:beer:

animalunae
05-15-2004, 02:02 PM
I remember that sketch from somewhere Steven, hehe. Indeed my setup is based on the same theory, Although I generally start from a lowpoly mesh, then manually add the 45degree rotated quads by using Split Polygon Tool, only where I want them, when the quads have been rotated, I create a similar pattern as the one illustrated, but it basically consists of a pole (intersections between + and X ) and the double edges between them to create the hard edge for the wrinkle. In case anyone wants a very quick way to create a "twisted" cylinder; here's a how to:

First start off by creating a triangle (half of a quad), at a certain distance from the origin. Rotated 45 degrees and the long edge aiming according to the Y-axis and translated over the Z-axis (in this case) for about 3 units or so. Make the long edge a fixed unit long, like 2 units, then bring the other vertice (not from the long edge) and snap it to middle of the long edge. So now we got the long edge from (0,1,3) to (0,-1,3) and the other vertice at (0,0,3) (between them).
Now create a cluster of that middle vertice or use the manipulate poly tool. Rotate it 20 degrees or so along the Y-axis (360/20 = 18).

http://home.pi.be/~derbyq/Tutorials/CylTut/Triangle.jpg

Mirror the triangle to make a quad and delete the center edge (not deleted here though, depends on the purpose of the cylinder)

Now duplicate the quad but make sure the pivot is at 0,0,0.

Rotate 20 degrees along Y and move the quad to 0,1,0.

http://home.pi.be/~derbyq/Tutorials/CylTut/Create_3.jpg

This fits perfectly thanx to that subject in school we never liked, hehe.

Now we will duplicate the quad with these settings translateY=1 and rotateY=20. Duplicate it as high as you want your cylinder to be.

http://home.pi.be/~derbyq/Tutorials/CylTut/Create_4.jpg

Now group or combine the created faces and duplicate it, rotating it over 40 degrees till the cylinder is complete and tadaaaa... thanks to the math...

http://home.pi.be/~derbyq/Tutorials/CylTut/Create_5.jpg

animalunae
05-15-2004, 02:29 PM
Here's a screenshot I got from my cleric's coat which was the cause of you creating that sketch, right Steven?

It is not as dence as the sketch you made for it, because the mesh is heavy enough, I only have one edge for the crease here, and 3 edges along the horizontal edges on a cylinder to define the valley next to the ridge.

http://home.pi.be/~derbyq/Tutorials/CylTut/Cleric_Arm.jpg

animalunae
05-15-2004, 02:36 PM
And another post to post some wires... They are posted in the Body topology thread so I will just put in the links.

http://home.pi.be/~derbyq/WIP/Princess/OpenCoat_00.jpg

Pants Shaded (http://home.pi.be/~derbyq/WIP/Princess/OpenCoat_01.jpg)
Shirt Shaded (http://home.pi.be/~derbyq/WIP/Princess/OpenCoat_02.jpg)

Coat Wire (http://home.pi.be/~derbyq/WIP/Princess/Wires/Wire_CoatL.jpg)
Shirt Wire (http://home.pi.be/~derbyq/WIP/Princess/Wires/Wire_ShirtL.jpg)
Shirt Wire HighPoly (http://home.pi.be/~derbyq/WIP/Princess/Wires/Wire_ShirtH.jpg)
Pants Wire (http://home.pi.be/~derbyq/WIP/Princess/Wires/Wire_PantsL.jpg)
Pants Wire HighPoly (http://home.pi.be/~derbyq/WIP/Princess/Wires/Wire_PantsH.jpg)

Julez4001
05-16-2004, 03:17 AM
Very Final Fantasy:Spirits Within

Stahlberg
05-16-2004, 07:31 AM
Thanks flingster that was very interesting!
Thanks derby.

What about a topology based on bisected hexagons, or a randomized triangular one like you get when you turn a surface into a Maya Cloth surface? There's probably no real future in trying to model wrinkles, it's just what we have to do nowadays, for stiffer tighter clothing like t-shirts, jeans and boots etc. Soon I should imagine any kind of wrinkle or fold would be done procedurally, similar to how Syflex or Cloth works, only quicker and better, using some kind of dense randomized topology. I guess it all depends on how tight the clothing is.

Panupat
05-16-2004, 07:43 AM
very interesting theory. hmm actually derbyqsalano about the cylinder, this silly method just popped in my head. would it work? -_-?

make a plane, rotate it 45 degree
http://www.all-final.com/j-render/3d/45one.jpg

delete faces in top view....
http://www.all-final.com/j-render/3d/45two.jpg

non linear > bend
http://www.all-final.com/j-render/3d/45three.jpg

delete history, merge vertices.
http://www.all-final.com/j-render/3d/45four.jpg

this process took me ... 30 seconds.

animalunae
05-16-2004, 11:29 AM
My best friend just mailed me, he says, "hey I got this other way of making one of those cylinders" exactly the same way... even the screenshots are almost the same hehe... did you happen topick his brain hehe. Nice method though...

Julez4001: What is?

Quizboy
05-16-2004, 06:23 PM
Bodysuits! Anyone want to discuss bodysuits here made of lycra or whatever that substance is that they use for divers, neoprene i think. i've got to design a bodysuit made of stretchable kevlar fabric (copyright 2004 Quizboy) for my character.

And does the material of the outfit make any difference to the topology of the wrinkles? Besides smaller or longer, that's obvious...

Stahlberg
05-17-2004, 01:56 AM
Panupat good method, I tried that too earlier but couldn't get the nonlinear deformer to bend the way I wanted it to, so I ended up curling the plane by hand again, very tedious. Guess I should have read the manual better. :) This opens up the way for skewed topology as well - where the U and V are not perpendicular. Also, it becomes possible to vary the angle from top to bottom quite easily.

Quizboy - tight and thin body suits are fairly easy if you have a body topology you're okay with. Just sketch out where you want wrinkles and add them, merging them into the topology, similar process to how you built the body in the first place.
If the suit is thicker and more wrinkly it becomes a bigger job, but I think I'd still start off with a body as the source geometry.

edit: what I mean is, just to clarify, if the material is thin and stretchable and tight, the wrinkles will be very thin, closely spaced, small, and not so many. The only other change you need to do is perhaps a slight softening and lessening of surface detail, such as deleting the bellybutton. Look at latex-fashions for reference. :)

Viper
05-17-2004, 03:49 AM
Originally posted by derbyqsalano
Julez4001: What is?

Your shaded image looks a lot like the ones made by Francisco cortina...guess that's good eh? ;)

Ive been following the body topology thread reading everything and saving the pics that I thought would be a good reference and all...I must say, great work guys. It's very inspiring...I hope to start my own female mesh this week...gotta get back to the modelling madness ;) Maybe I'll be able to finally help on something ;)

BTW, it's no problem if I use MAX is it? ;)

Quizboy
05-17-2004, 10:13 AM
panupat that looks like a good way to do it, and now i have an excuse to use a tool i've never ever touched or heard of 'til now, non-linear bend!

thanks stahlberg!

If the suit is thicker and more wrinkly it becomes a bigger job, but I think I'd still start off with a body as the source geometry.

i think i'll go with thinner and less wrinkly then! hey it's my design, i'll take the easy way out...

Stahlberg
05-17-2004, 11:40 AM
You could also do the wrinkles with bump or displacements, more than one and animate fading from one to the other for joint-dependent wrinkling... but yeah you still have to smooth out the bellybutton and stuff like that.

MSB
05-17-2004, 02:48 PM
mmmmm,
I don't think that displacement map or bump will be good stuff for that.I think the wrinkles will depend on what do u have I mean do u have colth engine as syflex or maya cloth or u will use something like jiggle defromer.but I think if u model your cloth and make wires and u do your wrinkles it will be better even it time consuming.Or use Dynamics and soft body...


sorry for me bad english...

Stahlberg
05-18-2004, 04:52 AM
Small fine wrinkles on tight garments can be done quite nicely with bump or displacement. I saw this example once: a ballet-dancer's dress, don't know what you call the upper part, not the tutu... the bodice? Quite tight anyway. I think there were 4 different bump maps, one for each of bending left, right, back, forward, and they were driven by the joints. Looked okay, and saved a LOT of modeling. :)

animalunae
05-18-2004, 07:57 AM
Julez and Viper: Thanks a lot :)

peanut
05-21-2004, 12:30 AM
Bump-o

kryoboy
05-21-2004, 11:49 PM
you could just make a script to make it for you.
but i cant quite pin down how to approach the script, heres what i do know.

| Translate | Faces | Degree | Y Scale |
-------------------------------------------------
|----0.75-----|----4---|----90----|---0.5---|
|----0.75-----|----8---|----45----|---1.2---|
|----0.75-----|----16-|---22.5---|---2.5---|
|----0.75-----|----32-|---11.25-|----5.1--|


Yscale increments y * 2 + 0.1
Degree of rotation increments by d * 0.5
Faces must be (4) *2, (8) * 2, (16) *2, (32) *2 ...etc...
translate is the distance from point 0, this remains the same.
since the Yscale will move it outward to match the seams with the more faces it has.

so you can make a plane with an X inside

[x]

ans use those values in the duplication option box and apply the Y scale afterwards with all poly units slected. and make circle of x boxes, such as this but flattened out.

[x][x][x][x]||

then duplicate it on the Y axis as many as you like.


flat cylinder
[x][x][x][x]||
[x][x][x][x]||
[x][x][x][x]||
[x][x][x][x]||
[x][x][x][x]|| <--- seam of cylinder, since trying to explain in 2d
[x][x][x][x]||
[x][x][x][x]||
[x][x][x][x]||

combine polys, merge verts to 0.1 and delete the verticle and hoizontal spans.


XXXX
XXXX
XXXX
XXXX <--- magic a 45 degree fingertrap.
XXXX
XXXX
XXXX



kinda hard to follow, but i failed to make the script work as i liked because, i cannot find any relationships with the Number of faces and the degrees of rotations, unless i predifined each rotation depending on the number of the faces you like that is in the power2 with 4, 8, 16, 32, 64, 128, 256, 512.

would like to have a script to create the fingertrap by putting in values for hieght, and faces (faces would have to be in the 4, 8, 16, 32 ect) with my math, possible im sure but with numbers inbetween there is no visable relationship to the face degree rotation values.

flingster
05-28-2004, 12:09 PM
hey guys would like some advice or screenshots or shared experiences when it comes to modelling/topology for pocket areas?

eg front jeans pocket
back jean pocket
jacket pocket with flap

whatever you got would love to see topology.
do you guys make them real pockets? what about stitching and edges? what effect do you want on the rest of your mesh?

all to often people refer to wrinkles when they talk cloth...but as this thread has been pretty informative already at looking at things like flow its seems like a good progression to consider pockets....seams...stitching...variances between to types of material...dunno just a thought. thanks as always.:beer:

RichSuchy
05-28-2004, 02:56 PM
For pockets that I dont intend to use, I was thinking to paint weights on my colision object, or even modeling the masses there, to give the added thickness, however at least the top could be modeled. For back pockets, using the cloth as a wrap deformer for the pocket geometry might work. Its on my list of experiments to try. I prefer not to use garments but rather to use geometry though i've heard that that is less stable, my experiences have suggested the oposite is true.

flingster
05-28-2004, 03:37 PM
presumable the thickness part would be relevant on something like a pair of button fly jeans..or at least on the top parts of the pocket..i never thought about how the garment itself would move/deform in relation to the rest of the jean material in real life...seems theres actually quite a bit to consider.
:wip:

RichSuchy
05-28-2004, 04:22 PM
Originally posted by flingster
presumable the thickness part would be relevant on something like a pair of button fly jeans..or at least on the top parts of the pocket..i never thought about how the garment itself would move/deform in relation to the rest of the jean material in real life...seems theres actually quite a bit to consider.
:wip:

There are definately problems with the pocket idea I had, but after trying to model a bra and use it as part of my collision object for the shirt.... It was better to paint the bra into a colision object. Any interpenetration is bad juju for the simulation. It seems that when used on large production, a lot of tweaking by hand on a per scene basis is the order of the day.

With loose jeens, for instance a pocket may move around quite a bit, add a set of keys to it (in real life) and you have something dificult to simulate without keyframing fixes.

It may be better to just design into the rig keyframable colision objects to represent the below-cloth objects and hit those in a tertiary animation pass. (after main motion, and secondary motion)

flingster
05-28-2004, 04:32 PM
jeez cheers rich...like the idea of below surface collison object to represent things like keys etc. very cool...

what about the actual topology if youre going to model it anything i should watch out for..be careful about.etc...
thanks again...

:thumbsup:

RichSuchy
05-28-2004, 04:48 PM
Of course as stated before, avoid interpenitration like the plague.
To add to that, keep the angles in the facets of your colision object at less than 90 degree angles, the less the better, and keeping that in mind the cloth has to be dense enough to calculate well over troublesome areas. you dont want jaged edges in your colision object to poke through facets of your cloth mesh. I can't really boil that down to an exact rule of angles but try to considder it as you work.

flingster
05-28-2004, 04:52 PM
Originally posted by Rich Suchy
Of course as stated before, avoid interpenitration like the plague.
To add to that, keep the angles in the facets of your colision object at less than 90 degree angles, the less the better, and keeping that in mind the cloth has to be dense enough to calculate well over troublesome areas. you dont want jaged edges in your colision object to poke through facets of your cloth mesh. I can't really boil that down to an exact rule of angles but try to considder it as you work.

nice...thanks again bud...appreciate you taking time out to explain it to me...:thumbsup:

RichSuchy
05-28-2004, 04:53 PM
Originally posted by flingster
nice...thanks again bud...appreciate you taking time out to explain it to me...:thumbsup:

Your welcome.
:)

DamionShade
06-10-2004, 09:49 PM
panupat (or anyone else, for that matter), any chance on a little more clarification on the non-linear bend method?

I know you had easy to follow screen shots, but seems when I apply the bend to my plane, I end up with a weird burrito... low bound, high bound, curvature? I tried leaving everything at default with no luck, then seemed to be on the right track with changing curvature to 90, but no matter how much a shift the center point on the bend tool, it never comes out as a perfect cylinder.

any tips greatly appreciated!

Stahlberg
06-11-2004, 08:45 AM
Nice to hear someone else had the same problems I had... :)

ceql
06-11-2004, 11:42 AM
Oh oh!! Lemme play! :)

ok, This is the best way I could come up with...
It's very flexible and easy, but one step is a little CPU intensive for some reason :)


1. make any poly cylinder.
It can be any number of spans along the height or the axis, as you like...
(No caps though, please delete them!)

http://student.ci.qut.edu.au/~n4420454/3dmisc/cylinder/1.gif


2. select all faces

http://student.ci.qut.edu.au/~n4420454/3dmisc/cylinder/2.gif

ceql
06-11-2004, 11:42 AM
3. Edit Polys -> Poke faces
(Maya is a bit slow at this, even for a cylinder of this density. It should take a minute or two...)

http://student.ci.qut.edu.au/~n4420454/3dmisc/cylinder/3.gif

4. Polygons -> Quadrangulate

http://student.ci.qut.edu.au/~n4420454/3dmisc/cylinder/4.gif

hooray! hope it works for you :)

animalunae
06-11-2004, 12:02 PM
Ah hehe nice one

DamionShade
06-11-2004, 02:51 PM
ceql, you're my hero! thanks, your method worked like a charm.

Stahlberg
06-11-2004, 04:45 PM
Smoooth...

kryoboy
06-21-2004, 09:02 AM
doh, i should had thought of that lol.

animalunae
06-21-2004, 09:16 AM
I even have a faster one. Create a cylinder with a number of spans. Delete top faces, then:

Chamfer Vertex and set the width to 0.5... Done

I won :)

Luki
06-21-2004, 09:30 AM
so, here is my version of cloth, I didn't use the way like yours at all, all quad pretty simple way. It's still WIP, pants needs work, button is in a wrong place, shoes etc.

http://www.lelumpolelum.3d.pl/lepre_ogl.jpg
http://www.lelumpolelum.3d.pl/lepre_ogl_wire.jpg

yinako
06-21-2004, 01:38 PM
I remember reading maya cloth docs and it suggest randomly distributed triangles are ideal for collision calculated cloth look.

But I also remember on Syflex docs it suggest their patch polygons(grid) are better in there system fo desiered look.

I guess its a little contradiction to the two, and maybe it has to do with the collision engine.

Ziah
07-01-2004, 07:11 PM
you guys use this approach of the 45 degrees more out of calculation purposes or what?

I does seeem it would be easier for the topolgy of wrinkles etc but whats it got over just keeep it normal?



do you guys allways put cloth over the model or what??

Some people suggest that its a watse of time if the charters body is not needed or going to be seen

They just use a ref image for proportions and model just the clothes...faking every thing using methods like extending the flesh areas of the character eg female arm into the geometry of the cloth then adding a slight lip at the and of that cloth area eg shirt sleeve where the flesh and cloth meets

If the clothing is going to be baggy the just extend the gemetry in and terminate it. And use a smaller smaller low poly/triangle object as the colision ( eg smaller version of the cloth are itself )

I know some of the advantages could be shorter render time and calculations

whats the dis advantages of that?

animalunae
07-01-2004, 07:25 PM
I don't model anything I don't see

Ziah
07-01-2004, 07:27 PM
derby....... whut did you do with the rest of ur girl..the last one you were working on

animalunae
11-08-2004, 09:46 AM
I've been doing some animation tests, setup needs to be redone and texturing needs a revisit. Other than that I think I might be able to get this project a bit further...

Here's an animation I did a few days ago

Cloth Test Sorenson 3 (http://home.pi.be/~wimcoene/Cloth_Test_00.mov)

Julez4001
11-18-2004, 03:36 PM
nice ClothFX
care to xplain it a little more....

animalunae
11-18-2004, 07:29 PM
The lower part of the coat was simulated using a wrap from a syflex simulation, check the thread in the syflex section. I also wrote a tutorial on this, you can check it out at my site at www.ronin3D.net

CGTalk Moderation
01-18-2006, 05:00 AM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.