PDA

View Full Version : nCloth, Layered cloth explodes


HJ-panorama
08-29-2008, 07:11 AM
Hello


I have a character with multiple ( exactly 3) layer of cloth , the very first layer, which is underwear piece is fine when simulating..

but adding another cloth layer top of it makes both to explode, I checked the collision layer, not useful though..

I actually added another nucelos system to the second layer of cloth, it explodes strangely
I also increase substeps to 12 and collision iterations to 40 .. but no effect..

thank you in advance

john_homer
08-29-2008, 07:33 AM
hi,

Collision layer are essentially useless for this stuff, they create more problems than they solve because layer 0 does not leave room for layer 1, etc, etc for areas like the armpit or hip its a nightmare...

the 3 most common reasons I have seen for multiple cloths exploding is...
- substeps too low
- cloth thickness too low
- cloth thickness too low combined with trappedCheck on cloth objects (try to only use this on nRigids, self trappedCheck is ok though...)

the last one is most likely the case if a large chunk of a lower layer is suddenly on the other side of a top layer in 1 frame. would help to see what type of exploding you are getting.

for 3 layers of cloth you are going to need to go higher than 12 substeps (and pay the price in solve time) I would suggest substeps of 40-50 and max collision itters of 30-40.
after you get it working you can try dropping them down... but you wont get away with a lot less than 30 substeps...

also, check your relative masses, mass determines how cloths interact so if 1 has much less mass it will get shoved around by the others...

.j

HJ-panorama
08-29-2008, 08:35 AM
thank u

here it is, all I 'v done till now, please correct me if i did something wrong o.k?

1- for 3 layers of cloth, I have used a single Nucleus system the whole system is driven by one N system.

2- I boosted up the substeps to 40
3- and collision iterations to 40 as well as each nCloth shape node collision iterations
set to 20

4- I checked the Thicknesses, actually I doubled them with seftrapped check ON.

o.k, for now I start with lower 2 layers to be simulated , the very first layer is cached and fine...
the second layer suddenly explodes and never come back to its original position...

Castius
08-29-2008, 04:41 PM
Caching lower leyers does nothing for the simulation. nCloth will resimulate the lower lever anyway. But to confuse you it will still show you the cache. I so wish that was not the case but it is.

If you can't get them to play nice with layers and trap check. I pretty much only use trap check when useing layers like this.. Then duplicate your lower layer mesh and attach the cache created for the ncloth mesh. Then make it a passive. Then sim the next layer. If you go with that method you might want to use multiple necleus for each layer.

Castius
08-29-2008, 04:53 PM
This is the reason i don't use trap check on the same layer.

If you're useing trap check you have to make sure your pushout value is low enough to allow the mesh to escape that distance. If the mesh does not leave the pushout distance. Trap check will try to push it out again. But this time it will bring the lower mesh with it. Causing it to be "trapped" in a an endless loop. Useing col layers prevents trap check from influeneing lower layers. So that loop can not happen.

But as John said there or cases where you can't reply on layers.

john_homer
08-29-2008, 10:55 PM
why do you want to cache 1 layer them simulate the other?

This is what is giving you all the problems by the look of it..

everything in 1 layer, 1 nucleus should simulate this cloth fine.

Did you turn OFF trapped check?

.j

HJ-panorama
08-30-2008, 07:25 AM
why do you want to cache 1 layer them simulate the other?

This is what is giving you all the problems by the look of it..

everything in 1 layer, 1 nucleus should simulate this cloth fine.

Did you turn OFF trapped check?

.j

your right John, I deleted the cache for the first layer and it's fine now..

actually I cranked up every thing related to the quality, including substeps and the others..

but should I turn the trapped check box On or Off eventually??
I left it On for now and It seems to be O.k for the beginning of 500 heavy calculations ..
let's see what will happen..

john_homer
08-30-2008, 10:36 PM
but should I turn the trapped check box On or Off eventually??
I left it On for now and It seems to be O.k for the beginning of 500 heavy calculations ..
let's see what will happen..

best to have trapped check on only for nRigids, self trapped check can be used on cloth.

This is what Alias recommended to me.

however this does not mean trapped check will always make your cloth explode, but it was their recomendation...

.j

HJ-panorama
08-31-2008, 04:44 PM
sorry to bother again


I encountered a new problem and need to fix that..

the character's moving, so when his hand reaches the closest distance from his hip ( body) , the clothes which are trapped between the fingers and the body start to explode....

I've set the thicknesses low enough but It makes some kind of other problem like They are stuck to each other and then explode...

how can i fix this ?

thank you very much

HJ-panorama
08-31-2008, 07:12 PM
well, eventually i decided to make a exclude collide pairs constraint to exclude the fingers out the collision ....
so, by this way I avoided the cloth to explode at certain frames where hands are very close to the body... especially in walking periods...

thank u

HowardM
04-27-2009, 12:36 PM
Caching lower leyers does nothing for the simulation. nCloth will resimulate the lower lever anyway. But to confuse you it will still show you the cache. I so wish that was not the case but it is.

Do we know if this has changed in 2009?

Castius
04-27-2009, 04:23 PM
I only use 2008 in production. I have testing 2009 though. And I rememeber thinking the cached nCloth worked how i expected it to. On the other hand after working with nCloth longer i don't know if I care about that anymore. I'd probable just use a separate nucleus. and share the point cache. Same workflow less trouble.

CGTalk Moderation
04-27-2009, 04:23 PM
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.