PDA

View Full Version : What's wrong with BULLET ????



prospector
02-27-2013, 09:57 AM
I was trying some hair dynamics (finally) and thought everything was finally working with hair dynamics...

But on this simple scene I figured out what was going wrong with my hair situation.
Bullet is screwing with LW.

Look at the heading rotations, with it off and then with it on.

Why is it doing this???

prospector
02-27-2013, 10:07 AM
Here is scene to check.
Just do a simulate

prospector
02-27-2013, 10:31 AM
It doesn't look as tho actual object is being effected, but Bullet is using the values to do calculations, which is just screwing with hair dynamics and making big knots in hair.
pay particular attention to frames 76 and 77 where values go from 167.62 to -157.70.

Kaptive
02-27-2013, 11:48 AM
Hello,

I just gave it a try. I'm not sure what effect it is you were going for, but I think what it might be is that bullet requires (for decent results) the polygons to have some level of thickness. Try the attached scene.

I've slightly altered the rotation of the object to a TCB curve, so the start and stop are smooth... I thought this might help the calculation so it isn't such a jolt. But having given the circular object some depth, it seems to work better. Anyway, have a look and see what you think!

Kris

112009

prospector
02-27-2013, 12:17 PM
Loaded and tested..
strange tho

with the exact numbers in each field (loaded mine in another layout to match side by side), yours gets no centrifugal force on hair and it just twists under object.
That wouldn't work for fast head turns like skater or balarena type movements.

But strange that such a difference as thickness will make a big difference in results.

and yours also changes numbers from pos to neg.


I'm not sure what effect it is you were going for,
realistic hair dynamics really.

Also upon furthur investigation...all numbers change..rotation, translation, and the object is really rotating by 359 degrees, and translation jumps from values of .0007 um to over 2/10 ths inch in some frames tho graph editor does not show any changes at anytime.

prospector
02-27-2013, 12:25 PM
This is best I have gotten so far, but as seen, there are strands above shoulder and some in bangs (seen on the tilt forward section) that are not being touched by calculations tho they have the same properties as rest of hair strands. And at the end some sections just kinda take their time sliding across face.

LightFreeze
02-27-2013, 12:47 PM
I think the docs recommend you do it this way

prospector
02-27-2013, 01:09 PM
downloaded and tried.
Again, no centrifigul forces, which there should be, like in the little vid.
Played with the gradient values and it goes from just twisting under object to holding the shape so much that there is no hair flowing naturally.
But will try other number changes to see if there is another combo that gives good results with a gradient applied (as per docs)

LightFreeze
02-27-2013, 02:18 PM
your demo disc is a bad model for developing centrifugal forces as it spins on its own axis, the strands near the centre are only being twisted when the disc spins, a sphere would be a better test model. I have offset the disc from centre to test it that way

LightFreeze
02-27-2013, 02:47 PM
with a test sphere this time