PDA

View Full Version : Physics and forces in V8



Turner
01-24-2004, 09:00 AM
Well, I've been reading and looking and can't seem to find an exact answer to some questions...

Basically, as rudimentary as it is, Carrara 3, which I've been using, has some fairly elegant physics and force systems -

For example, I have a bunch of objects resting on a surface, and I pass a point force over them, and they pull towards it while still bouncing off of each other. As the point force moves through space, the objects convincingly "swarm" around it while still colliding with each other:

http://www.turnerdesign.net/force/

You can also put whatever stationery or moving point forces in a scene, and "drop" objects into it which then are pulled around by the point forces. Very nice stuff.

It isn't fast, but it works and is VERY easy to implement.

Naturally, you can also drop objects which bounce, and slide, and so forth...

I'm trying to figure out if this is what the hard body dynamics in V8 is all about, and if the same type of forces that can be applied on particles in 7.5 will be able to be applied to objects.

Anyone know if it's coming in 8? I realize there are add-ons for this but I can't shell out the cash just to goof around, I'm just learning this stuff and will be for some time!

Thanks,
Andrew

Dodgy
01-24-2004, 11:05 AM
Have a look through the animations. Lw's new dynamics look like they'll be able to do anything you care to throw at them :)

Turner
01-24-2004, 12:15 PM
Yeah, I watched them - still, I'd like to read a list or something, as the vids didn't quite cover what I'm looking for.

The "breaking the robot dog" routine is nice... I want to see more!

Andrew


Originally posted by Dodgy
Have a look through the animations. Lw's new dynamics look like they'll be able to do anything you care to throw at them :)

Ramon
01-24-2004, 02:11 PM
Originally posted by Dodgy
Have a look through the animations. Lw's new dynamics look like they'll be able to do anything you care to throw at them :)
I hope so, I've been planning on testing it out with a mond of gold pieces. I would make the clylindrical pieces as poly effecient as possible given the camera distance. I will also make sure that the pieces do not comprise the whole mound - only the top but, I think that it still might choke the L8 dynamics. I hope not though.
What do you guys think?

Meaty
01-25-2004, 10:11 AM
Originally posted by Ramon
I hope so, I've been planning on testing it out with a mond of gold pieces. I would make the clylindrical pieces as poly effecient as possible given the camera distance. I will also make sure that the pieces do not comprise the whole mound - only the top but, I think that it still might choke the L8 dynamics. I hope not though.
What do you guys think?

I guess it remains to be seen just how efficient the physics engine is, more likely than not, you will probably have to use lower poly proxy items to control the motion, and substitue the higher poly model for rendering. Although, i do not see how this could work if the model is set to 'Pieces' like the dog in the video. I am very interested to see how well it works! I am tempted to build an Impact 3 - like mouse trap demo for when 8 comes out.

Ramon
01-25-2004, 12:39 PM
Meaty, cool I'd love to see it. Also, I wonder how well a poly proxy would work. If the dynamics/collisions are based onthe poly surfaces contacting each other, then a low res proxy would not be describing the shape of the final rendered object correctly (due to the proxy objects having much less polys) and thus, I would imagine that there would be pieces that would be intersecting each other.
Is that the way you think it might work? Good question for Proton eh?

Dodgy
01-25-2004, 02:56 PM
For a cylinder, an 8 sided object would probably be sufficient, in an animation you'd probably never be able to tell, you go back and watch some heavy dynamics effects movies frame by frame, you'll probably see all kinds of dodgy stuff going on :)

Ramon
01-25-2004, 04:19 PM
Yeah, especially at the distance from camera in which I am planning on rendering them. I was thinking of using animated UVs (when 8 comes out) to map an AE animated sequence on the gold mound - of gold pieces sliding over one another which will be just under the "layer" of dynamically animated gold pieces.

In theory, do you think this might work to give the illusion of depth - realistically? This would cut down the need for exorbant calculations.:confused:

Dodgy
01-26-2004, 03:16 AM
Yup. You could probably have a pile of static coins and a simple collision box or sphere under it all to speed things up.

Ramon
01-26-2004, 01:29 PM
Yeah.
Thanks for the reply and advice.

Crocodilian
01-28-2004, 10:10 AM
being able to do good looking work with simulation requires being skillful in use of proxies. Ramon is correct, poly proxies don't precisely describe the shape of the parent, but that's where skill comes in.

If you say "I want to run my simulation only on the hi-res real thing", be prepared to come to grief. Bear in mind that there are plenty of other things, like displacement maps, which will also affect the final geometry so that it differs from what was tested.

The aim of a dynamics simulation is not to produce a completely accurate physics model --although you can do that-- its to produce something that "looks right"

My experience (3D studio's Reactor plugin--powerful, but maddeningly hard to use; I'm soooo looking forward to LW), has been that time spent skillfully setting up proxies is always well spent. . .full res simulations can take hours and hours.

And remember, because you're not controlling this the way you do with keyframe, you need interactivity-- the only way to get the wall to break "just so" is to tweak the parameters. . .

Turner
01-28-2004, 10:20 AM
I posted my original question after watching the Maxon Cinema 4d rigid dynamics simulations on their site.

Good stuff, I hope LW8 will include these type of dynamics.

Ramon
01-28-2004, 11:05 AM
Sorry Turner, I really didn't mean to hijack your thread. I was just curious as to answers with respects to 8's physics.

Turner
01-28-2004, 11:14 AM
No problem at all!

Your question is very interesting. I've thought about it myself, though I'm an amateur in Lightwave. I'm still thinking about it - it's like looking for a way to fake something that looks/acts real with the greatest efficiency, and deciding where the tradeoff between realism and efficiency occurs.. then trying to find a way to increase realism without decreasing efficiency!

I guess this is why big studios write their own software...

As far as Maxon goes, I just wanted to make note of the point of reference I used before writing this thread. Their examples are fairly brilliant.

Cheers!

Andrew


Originally posted by Ramon
Sorry Turner, I really didn't mean to hijack your thread. I was just curious as to answers with respects to 8's physics.

Ramon
01-28-2004, 12:35 PM
Originally posted by Turner
No problem at all!

Your question is very interesting. I've thought about it myself, though I'm an amateur in Lightwave. I'm still thinking about it - it's like looking for a way to fake something that looks/acts real with the greatest efficiency, and deciding where the tradeoff between realism and efficiency occurs.. then trying to find a way to increase realism without decreasing efficiency!

Hey Andrew, that's exactly it! "fake something that looks/acts real with the greatest efficiency".
In the end, the audiance will only be able to see (in a majority of cases) limited areas of the scene and a distance to boot.