PDA

View Full Version : nCloth air or wind scale bug


Castius
07-10-2008, 11:58 PM
We are having trouble using nCloth at a larger scale. We canít get a simple natural blowing cloth effect. And i think it's because the air model doesn't work well at large scales.

Test scene:
It has a small half cylinder with some wind
Then i have an identical cylinder setup but scaled up 100X.
Iím try to get the large cylinder to get even close to the behavior of the small cylinder.

Thanks
Scott

Duncan
07-11-2008, 01:06 AM
The 100X version will need the spaceScale divided by 100 and the windSpeed multiplied by 100.

Duncan

Castius
07-11-2008, 02:55 AM
Hi,

That's exacly what is not working.

small nucleus
scale 1
wind 10

large necleus
scale .01
wind 100

If you set wind to 1000 it way to much. And no amount of tweaking will make the the large object look like the small one.

I don't want to work at 100x but it's something we have to deal with. We don't want to scale all our work in order to work with nCloth.

TinuZ
07-11-2008, 08:30 AM
Hi,

I'm not sure if it will help you in this case but if you are using maya 2008 ext2 maybe you find some useful information here:
http://forums.cgsociety.org/showthread.php?f=86&t=649774

cheers

Castius
07-11-2008, 10:20 AM
Thanks, I have looked at that issue. I don't think they are related.

Our issue is with 8.5 sp1 not 2008. I have not testest this problem in 8.5. But it's possible that a change was introduced doring the update to 8.5 sp1. When they fixed a bug with large scale objects. That would crash Maya. One of our riggers did alot of look dev in 8.5. Where this blowing cloth was much eaier to acheave. I will find a way to get 8.0 installed a test machine tomarrow.

Duncan
07-11-2008, 06:01 PM
This problem seems to be due to lift. If you divide your lift on the larger object by 100 then the motion seems correct. (note that the windSpeed on the larger object also needs to be 1000, not 100). Thus to adjust for scaling you need to multiply the windSpeed and divide the spaceScale and lift by your scaling factor.

I think the lift may have a non-linear relation to wind speed, and thus it is not scaling in the desired proportion. (I've entered a bug on the spaceScale not compensating for this)

By the way... the windSpeed is currently in units/sec. We could have the spaceScale automatically adjust the windSpeed so that you would use the same numeric value in your example, although this might be more confusing, as then the speed would be then in the units defined by the space scale.

Duncan

Duncan
07-11-2008, 06:36 PM
Note that the problem I describe with needing to compensate for lift is also in Maya 2008.
There also seems to be another issue with bend resistance having problems at the larger scales. Some of the edges are flattening out. I've noticed this problem before, but have not tied it to scale. (this is a useful example file, thanks)
One thing that can help is to use vertex style self collisions and increase the selfCollideWidthScale so that the vertices overlap a little(thick self collisions are slow unless one uses the vertex selfCollideFlag). This creates a natural sort of bend resistance and can result in smoother meshes.

Duncan

Castius
07-11-2008, 06:56 PM
That does help a lot! Thank you for submitting a bug report.

I had tried an number of tests with changing lift. But never found the magic numbers. I was probable not keeping it simple enough. It doesn't help that more substeps are also required to keep the bend resistance from collapsing.

CGTalk Moderation
07-11-2008, 06:56 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.