hi, back again, was just updating my own site and doing some Cloth Room tests.

Laurie, you're doing fine, please continue. Stretch indeed is like an old iron spring, first you need some power to make it longer (resistance), and when you let it go it takes a while before it's back to its original length (damping). I'm going to find out what's behind those 50 and 0,01 numbers, in terms of units and real life. I intend to do Friction this evening. 

3Dave, well stated.
Your 1) Stage it! is so true, next to my findings tha animation shots should not last longer than 3 to 4 seconds anyway t make an intersting story. We live in the TV and internet age, shots are fast.
Your 2) Fake it! just mean: thins as a pro, as these guys don't have the time to get the full monty either. The difference it that they have a day to serve the customer, and I have only my spare time.

First, I'd like to note a bug (feature?). When clicking the Reset button in panel 4, my value for Stretch Damping is NOT reset to the default 0,01 value. I'm on P8SR3. can this be confirmed? Is it also the case on P7, other P8-SR's or PP2010?

Now, let's consider panel 1, Sim Settings, in more detail. In fact it covers two areas, 1) frame handling and 2) calculation handling.

1) frame handling.

It seems so simple, it takes the figures pose in frame 1, runs a number of drape frames to make a proper cloth fall over the figures body in a relaxed pose, then follows the figure when is animates from frame 1 to any end frame, and then inserts the result - excluding the drape frames - into the animation from frame 1 on.
And there is a button to do the drape only, without the time consuming sim. This helps you to determine a decent amount of drape frames. 

Now this is fun: just put some other value for Start Frame. Say your figure animates from frame 1 to 60, and now you simulate from frame 21 to 40. What will happen?

In my P8SR3, the drape still runs against the figures position in frame 1, then the simulation runs against the poses and positions of the figure in frames 1 to 20, and then the result is mapped into frames 21 to 40.
But since the sim determines the cloth vertex positions in world coordinates, the cloth in frame 21 will be positioned to fit the figure in frame 1 while the figure itself is moved away 20 frames. This is the nightgown chasing Vicky.

Can this be confirmed on other Poser versions? Can anyone think up a use for this? Because I think it's a bug (feature?). That is: it would be quite handy to run test sims on small ranges of frames before doing the whole one. But then, the sim should run against the figure positions in frame 21-40, in the example above. I wouldn't know about the drape.

2) calculation handling

The sim setting's offer three checkboxes and a steps-per-frame setting. The defaults are for hires tablecloth over a simple lowres prop (so the cloth can easily follow the props curves), and for minimum calculation times. By changing these settings you enhance the calculation mechanism, and the intention is to provide a more natural looking result.

For instance, when you've multiple cloth-objects (panel 2) colliding to each other (eg blouse over skirt), or the tablecloth has massive folds which collide to each other, it's well advised to tick the cloth-to-cloth checkbox.
And when you've got lowres cloth over a highres body, the cloth will face troubles to follow the curves of the figure, so it's advised to tick the object vertex against cloth poly.
With 2 steps per frame, it's like the algorithm looks not only at frame 15 and 16 positions, but at 15.5 as well. So when the figure is really moving you might get better results by increasing this amount mildly. It's not doing a complete subframe assessment I guess (I can't look inside) as calculation times do not go up linearly.

So I questioned: can these settings prevent the poke-through errors we were facing in the earlier posts? What does the enhanced quality result really mean?

Therefore I did a simple experiment: took a cone, scaled to 400% in Y-direction which turned it into a needle, and started to drop a piece of cloth right on to it. I noticed the following:

• at default setting, the cloth falls right through the needle

• ticking the checkboxes helps, for a few frames. The pokethough is not prevented but postponed

 - increasing cloth stiffness (fold / shear / stretch resistenace and stretch damping) made a difference. The stiffer, the more it seems to prevent pokethrough. Untill I prolonged the animation / simulation period. Then it appeared again that the problem was not solved, but postponed. On the other hand: this stiffened up kind of cloth becomes less and less realistic and elegant.

 - increasing collision offset (more on that in a next post) helps, as it thickens the cloth and puts it at some distance from the figure. At a firm thickness , the needle can't penetrate any more indeed, at any point in time. But again, then the results don't look realistic anymore. Real life favors silky night gowns over carpet-based ones.

So, what have we got (just thinking out loud).
Real cloth can be fibre weave based (leather, latex, fleeze are not. Sweaters can be knit). The fibres have thickness and characteristics (which differ per thickness) and the weave can be loose or tight. The fibre thickness and tight weave might prevent a needle to stick through.
Poser virtual cloth has a fibre-concept based simulation, as follows from the literature in earlier posts. But the fibres have no thickness (there is no parameters for this). It's only the fibre behaviour which is derived from the cloth parameters. Even leather and latex are treated as weaves. So, I guess, it wouldn't be that difficult for a needle to poke through. The algorithm can't stop that.

Hence, until the people who might look at the algorithms and vertices behaviour themselves come up with a different view, the best we can do is postponing the problem till a frame we're not interested in. Or, the other way around, we've got to chop our animation into pieces were each piece behaves properly. 

Which was exactly one of 3Dave's suggestions. And setting the calculation options might be of help, and 'solve' the issue in short animations.

- - - - - 

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