I implemented collision detection not requiring a function based
environment in my mechanics simulation system.  It's based on intersection
tests in movement direction like in clothray and is of course much more
limited than the function based system but it seems quite usable in many
situations.  There are serious problems though with shapes not having a
defined interior since there is no way making sure the masses don't
penetrate such a surface. 

As said in another thread in this group i also tried simulating more
complex cloth objects sewed from rectangular patches.  The attached 3
pictures show a simple T-Shirt created with this method.  

First the elements of the shirt are created at appropriate positions
around the body.  This placement is critical because after the parts are
sewed in these positions using short length connections the cloth geometry
is relaxed using a gradient descent method.  If the patches are not placed
well parts of the body will stick out through the seams afterwards.

After that a normal simulation is calculated (about 10 minutes in this
case) Here the collar was the critical part since the stress due to
gravity extended the connections so much that the cloth vanished behind
the body shape between the nodes (although the calculations prevent the
nodes themselves to penetrate the environment).  Another difficulty are
the seams.  They don't have any bending stiffness right now so they tend
to fold when they should not.  Adding this is of course possible but will
involve more work.

Christoph

-- 
POV-Ray tutorials, IsoWood include,                 
TransSkin and more: http://www.tu-bs.de/~y0013390/  
Last updated 13 Aug. 2002 _____./\/^>_*_<^\/\.______

Post a reply to this message

Attachments:
Download 'cloth1_created.jpg' (17 KB) Download 'cloth1_relaxed.jpg' (18 KB) Download 'cloth1_final.jpg' (18 KB)

Preview of image 'cloth1_created.jpg'

Preview of image 'cloth1_relaxed.jpg'

Preview of image 'cloth1_final.jpg'

Now that's cool! Keep up the stupendous work, Christoph! I'm sure by the
time 4.0 comes out this will be part of it. :)

--
Anthony

Post a reply to this message

Good progress.

Can this be used to simulate a wind blown sail? The sail would also be
formed from stiched rectangles & would need to bend around ropes.

Tim

Post a reply to this message

Here is another one, this time on a human figure (obtained from
http://www.fantasticarts.com/3dmodels/3dfemale.htm )

Simulation is purely done with gradient descent method this time, real
physical simulation is difficult with formfitting clothes and does
probably hardly improve the results.

The cloth topology is the same as in the previous shirt, just longer and
tighter sleeves and a longer central part widened at the bottom.  The
shoulder section does not fit very well though, as you can see on the very
left the cloth is heavily stretched at the shoulders which leads to
problems.

A good design position for the cloth was harder to achieve this time
because of the more complex model.  I used functions to deform the cloth
patches for a good starting position, but i'm not sure if this would be a
practicable method in general.

Christoph

-- 
POV-Ray tutorials, IsoWood include,                 
TransSkin and more: http://www.tu-bs.de/~y0013390/  
Last updated 13 Aug. 2002 _____./\/^>_*_<^\/\.______

Post a reply to this message

Attachments:
Download 'cloth2.jpg' (28 KB)

Preview of image 'cloth2.jpg'

Tim Mcmurdo wrote:
> 
> Good progress.

Thanks.

> Can this be used to simulate a wind blown sail? The sail would also be
> formed from stiched rectangles & would need to bend around ropes.
> 

A sail would be easy, just a rectangular/triangular patch attached at
several points.  There is nothing implemented concerning wind yet, but
that won't be very difficult to add.

Christoph

-- 
POV-Ray tutorials, IsoWood include,                 
TransSkin and more: http://www.tu-bs.de/~y0013390/  
Last updated 13 Aug. 2002 _____./\/^>_*_<^\/\.______

Post a reply to this message