Some of my tests in the pursuit of non-intersecting, non-jittering, non-simulation-stopping cloth on a fast-moving character.

These are all V13 tests

Click to view attachment

I think most of the tiny jitters while the cloth is settling could be eliminated if it were possible to apply Channel Reduction to the cloth simulation results.

However, "Reduce" is only available if you have selected just one x, y, or z channel for one CP. Anything more and that option is greyed out.

Looks real good.

We tried Channel Reduction on Cloth... Unfortunately, the concept of doing it on each coordinate separately (x,y,z) is fundamentally unsound. For example, an "x" value would get reduced but "y" & "z" stay the same, however, the point was FINE to begin with BUT it's not after the reduction. (The results were jitters everywhere.)

I'm not insisting that doing it on each co-ordinate separately is the right thing to do. I'm noting that that is ALL you can do (in V13 anyway). you can only apply reduce to one separate channel at a time, one CP at a time. That's the only possiblity A:M presents me, which doesn't facilitate experimentation.

It seems to me a proper "reduce" in 3D ought to be taking X Y and Z collectively into account anyway.

Nice test Robert. That's some of the best cloth I've seen yet.

Those look great, robcat!

Looks pretty good to me ..cloth has always been trickiy

re-rendered one in Stereoscopic

Click to view attachment


how to view cross-eye Stereo

Very Nice! I'm looking to do some more cloth examples and tutorials soon.

Those are great examples Robert. Could you color the Deflector group(s) on Thom so we can see where the deflector(s) are? It looks like you have a fairly high collision tolerance, 2cm??

Thom is all deflector in those tests.


more like 8, but with Thom scaled to four times his normal size.

Large tolerance values seem to do better even if the character is scaled to make that distance appear smaller.

Great stuff, Rob. I would suggest making the cloth-mesh even finer(about 2wice as dense squares) I 've always had no problem as far as when cloth hits an animated body...but when the animated body hits the cloth- trouble...this test shows good promise for that.

I did try denser mesh at least once, but aside from quadrupling the simulation time, it didn't immediately get me more success with simulations so I didn't pursue that. I'll try again someday.

Another animator I know was trying to get a situation where a character was wearing a poncho to work. He had pretty much given up on simcloth, but i knew it had to be possible somehow. Rigging the poncho with bones was a pretty sad looking alternative.

So these tests are are just a little bit of further investigation of what i learned doing that.

BTW, those tests i show are the ones that worked. Many failures along the way with tangled, frantic looking cloth.

I love the stereo stuff. Did you create this with just A:M? If so, could you say how?

Thanks,
George

It's in the render options under "Stereo". If you are rendering to targas it will make separate files for right and left but if you render to quicktime it will render them side by side.

Check out the "Stereoscopic Rendering" tutorial in the Tutorials section