|
 |
This is in MPEG1 format made by TMPGEnc, I hope it works for everyone.
Updates since last version:
- Rewritten in C++
- Extra springs added to stop material "folding" too much
- More detailed mesh
- Smaller time step in simulation
This one took about 5 minutes to generate 200 frames, then another 30 mins
or so for rendering.
Still on the list:
- Use RK4 to integrate
- Adaptive simulation time step size
- Add drag force based on surface area perpendicular to velocity
Post a reply to this message
Attachments:
Download 'scene000.m1v.mpg' (627 KB)
|
|
|
|
Odd, IrfanView 3.85 couldn't read it - it's best to chose the 'system'
option in TMP, that makes a 'real' MPG file (extension mpg instead of m1v).
Bottom right hand option. TMPGENC will read it, however. I used TMPGENC to
produce a system file, then I could view it.
What are the advantages of this over the CLOTHSIM in POV/MegaPOV?
It looks like there needs to be some damping - it's awfully 'twitchy' after
it mostly settles.
Are you making one step per frame or more?
I've been trying for a while now to get a decent 'metal plate' simulation (I
have a robot tearing through something), so I've been looking at cloth
sim's. So far for most stuff I have to admit I use Poser's cloth sim, it
seems fairly fast and flexible; of course, it isn't free :(
== John ==
"scott" <sco### [at] spam com> wrote in message news:4225a2fa@news.povray.org...
> This is in MPEG1 format made by TMPGEnc, I hope it works for everyone.
>
> Updates since last version:
> - Rewritten in C++
> - Extra springs added to stop material "folding" too much
> - More detailed mesh
> - Smaller time step in simulation
>
> This one took about 5 minutes to generate 200 frames, then another 30 mins
> or so for rendering.
>
> Still on the list:
> - Use RK4 to integrate
> - Adaptive simulation time step size
> - Add drag force based on surface area perpendicular to velocity
>
>
>
Post a reply to this message
|
|
|
|
"John D. Gwinner" <joh### [at] cornelledu> wrote in message
news:4227540d$1@news.povray.org
> Odd, IrfanView 3.85 couldn't read it - it's best to chose the
> 'system' option in TMP, that makes a 'real' MPG file (extension mpg
> instead of m1v). Bottom right hand option. TMPGENC will read it,
> however. I used TMPGENC to produce a system file, then I could
> view it.
OK I'll remember for next time.
> What are the advantages of this over the CLOTHSIM in POV/MegaPOV?
Probably nothing at the moment, I was mainly using it as a programming
exercise for myself. I hope to expand it in the future, and I assume it's
more flexible than CLOTHSIM because I've written it and it's in C++ :-)
> It looks like there needs to be some damping - it's awfully
> 'twitchy' after it mostly settles.
Yeh, one of the things on my list is to add some damping based on the patch
velocity.
> Are you making one step per frame or more?
More than one, quite a lot more than one. BUt this is because it's still
using the standard quick'n'dirty integration method. I plan to use RK4 and
use adaptive time-step, this should make the sim a fair bit quicker and
perfectly stable.
> I've been trying for a while now to get a decent 'metal plate'
> simulation (I have a robot tearing through something), so I've been
> looking at cloth sim's.
Hmm that's a tricky one, I guess you just need a *really* stiff cloth, and
one where the nodes will split if the forces get above a certain amount.
THis is something I plan to do with mine, it should be quite easy to just
spawn the node into 2 if it is being "stretched" above a certain amount.
This should (hopefully) cause some realistic tearing.
I've also been thinking a bit about adaptive patch sizes, could be quite
interesting...
OK, enough talking, back to some "doing".
> So far for most stuff I have to admit I
> use Poser's cloth sim, it seems fairly fast and flexible; of
> course, it isn't free :(
> == John ==
>
> "scott" <sco### [at] spamcom> wrote in message
> news:4225a2fa@news.povray.org...
>> This is in MPEG1 format made by TMPGEnc, I hope it works for
>> everyone. Updates since last version:
>> - Rewritten in C++
>> - Extra springs added to stop material "folding" too much
>> - More detailed mesh
>> - Smaller time step in simulation
>>
>> This one took about 5 minutes to generate 200 frames, then another
>> 30 mins or so for rendering.
>>
>> Still on the list:
>> - Use RK4 to integrate
>> - Adaptive simulation time step size
>> - Add drag force based on surface area perpendicular to velocity
Post a reply to this message
|
|
