Forum: Poser - OFFICIAL


Subject: Dynamic cloth problems

Lully opened this issue on Jan 02, 2011 ยท 58 posts


aRtBee posted Mon, 03 January 2011 at 3:00 PM

Perhaps I may add a bit.

Just from the image - I don't have details - I infer the dress is about 50 vertices circumference and 50 height, making a 50x50 = 2500 vertex object of about 1 mtr circumference and 1,5 mtr height. This makes squareroot( 2500/1,5 ) = 40 vertices per meter, or about 25mm vertex density.

Vicky, inside the dress, is about 1,80 mtr height, and consists of 70.000 vertices. That is 200 vertices per meter = 5mm vertex density.

Hence the figure has a much higher vertex density than the dress, which is the reason to check the "figure vertex against cloth polygon" collision checkbox and the reason why the second one (poly to poly) will not add much value - it is for about equal densities.

Quad-based objects of about 25mm density are known for the mild rubberish behavior, which might be reduced a bit by lower Fold resistance and will increase by increasing density, and will especially be reduced by pumping up the stretch damping to 0,1 or above. Note that parameters like density and the resistances show different cloth behavior when altering them in orders of maginitude, like halving, doubling, tenfolding and the like. Personally I do not believe that much in the effects of minor teaks.

Please do not switch off air damping, it's the only force that can stop the dress from swaying sideways forever.

Quad based objects are known to show behavior that varies a lot with vertex density. Lower densities are less elastic and might even freeze up. This makes me worry about the sleeve ends, the chest portion and the skirt in the circumferential direction. It implies that portions of the dress will behave quite different from other parts.

Changing quads into tri-polies makes the cloth less rubberish and more towel / linen like, thinner and (far) more elastic along the diagonals. For the quality velvet-like dress you seem you create. my personal preference would still be quads.

Success

- - - - - 

Usually I'm wrong. But to be effective and efficient, I don't need to be correct or accurate.

visit www.aRtBeeWeb.nl (works) or Missing Manuals (tutorials & reviews) - both need an update though