![](/i/fill.gif) |
![](/i/fill.gif) |
|
![](/i/fill.gif) |
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
All Hail the Heun!
Post a reply to this message
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
sweet - can hardly wait for the next one :)
--
Rick
Kitty5 WebDesign - http://Kitty5.com
POV-Ray News & Resources - http://Povray.co.uk
TEL : +44 (01270) 501101 - FAX : +44 (01270) 251105 - ICQ : 15776037
PGP Public Key
http://pgpkeys.mit.edu:11371/pks/lookup?op=get&search=0x231E1CEA
Post a reply to this message
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
I dunno if there is a name for the method (Runga-Ketta appears to work
differently), but I found that the third and higher degrees of motion
can be approximated from the two next lower degrees of motion, in the
same way that the second order is derived from the location and the
first order. It greatly reduced the need for dampening to keep things
stable.
Regards,
John
--
ICQ: 46085459
Post a reply to this message
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
Nice work. If you want even better accuracy I reccomend the 4th order
RK scheme. Paired with an adaptive timestepping routine it should be
both faster and more stable than Heun.
sig
--
ICQ 74734588
Post a reply to this message
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
"Rune" <run### [at] mobilixnet dk> wrote in message
news:3c6d9914@news.povray.org...
> Wo-o-ow!
>
> Just like looking at the real thing! :D
>
> I just can't wait to see the more complex animation!
>
> Rune
Thanks!!!
Well... I left the computer running with very fine settings (much more
precision and a much finer mesh) but I forgot to turn on my primitive
cloth-cloth collision detection... However I think I put it TOO much
precision as it had only finished 336 frames in 48 hours!!!! I'll put the
computer to render the frames this night and I'll post it tomorrow in the
morning, I hope it will look good.
Thanks again for your nice comments!
Fernando.
Post a reply to this message
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
"Tony[B]" <ben### [at] catholic org> wrote in message
news:3c6dbfe7@news.povray.org...
> All Hail the Heun!
Thanks!
Fernando.
Post a reply to this message
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
"Rick [Kitty5]" <ric### [at] kitty5 com> wrote in message
news:3c6e337e$1@news.povray.org...
> sweet - can hardly wait for the next one :)
Thanks, I'm glad you liked it!
I'll post the next one in a few more hours!
Fernando.
Post a reply to this message
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
Your comments remind me that not very long ago I read about a very
interesting phenomenon: In cloth simulation sometimes it is "better" to use
Forward Euler than Runge-Kutta, because the inaccuracy of F.E. sometimes
acts as "natural dampener" and is not necessary to include dampening
explicitly in the model, whereas R-K looks very bad if it does not have this
dampener.
At first I thought this was very strange, but later I observed a similar
behaviour when I first tried the Heun method: I saw that the cloth looked
much more "rubbery" than before. That's why I had to implement a better
dampening routine to avoid very "violent" vibrations in the cloth. However I
think that Heun + dampening is much faster and stable than F.E. alone.
Thanks for your comments,
Fernando.
Post a reply to this message
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
"Sigmund Kyrre Aas" <as### [at] stud ntnu no> wrote in message
news:urps6uo8pk9fjqhqe0th0jtmoi68c8ih78@4ax.com...
> Nice work. If you want even better accuracy I reccomend the 4th order
> RK scheme. Paired with an adaptive timestepping routine it should be
> both faster and more stable than Heun.
Thanks Sigmund!!!
I think a higher order method would not help *that* much... The main problem
is not supposed to be about accuracy, but stability. I have read that the
derivatives of the forces of the cloth can be quite high, and this can cause
unstability problems with EXPLICIT methods (Forward Euler, Heun,
Runge-Kutta), but stability problems can be fixed with IMPLICT methods (I
only know Backward Euler).
Nevertheless, I've found that the Heun method works very well with
not-very-small timesteps and haven't had (fortunately) stability problems.
I think I'll play just a little bit more with cloths and then I'll be back
into something else... maybe fluids?
Anyway, I'm really happy that you liked it! Thanks again!
Fernando.
Post a reply to this message
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
Very nice. :) *big drooling grin*
I don't really think there's a need for any higher order method than Heun's
method.
Looking forward to see further developments.
regards,
Simen.
>
> * Goodbye Forward Euler! Now I'm using Heun's method (also known as
Improved
> Euler), it is a second-order explicit method, similar to Runge-Kutta, but
RK
> is of 4th order. It seems that I can now take much larger steps without
> worrying too much. I'm still using a variable-timestep scheme.
> * New damping process to avoid in-cloth oscillations.
>
Post a reply to this message
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |
| ![](/i/fill.gif) |
|
![](/i/fill.gif) |
|
![](/i/fill.gif) |
| ![](/i/fill.gif) |