View Full Version : FXWars! Roller Coaster!: Laluneverte

 laluneverte10 October 2004, 01:28 AMHey. We join this contest as a team. We are> laluneverte (helge), NTMonkey (lu), deepGreenJungle (aa.) We are using XSI 4.0, and did our own custom build simulation setup. This is the first simulation test whatsoever. It is NOT supposed to be a real rollercoaster yet, just a test for the dynamics. No keyframes used at all. http://217.160.138.15/helge/rollercoaster/sim_test_01.avi (2 megs) cheerio, theTeam
DeepGreenJungle
10 October 2004, 07:44 AM

ACantarel
10 October 2004, 08:16 AM
Looks very promising!

Are the waggons banking in the curves?

André

laluneverte
10 October 2004, 08:59 AM
Yes exactly.

We use a nurbs extrude using an additional upvector curve.
+ duplication along path.

Test setup.... (poles not connected to rails yet...)

http://217.160.138.15/helge/rollercoaster/roller_01.jpg

laluneverte
10 October 2004, 09:28 AM
the next simulation test showing banking>

http://217.160.138.15/helge/rollercoaster/sim_test_02.avi

http://217.160.138.15/helge/rollercoaster/sim_test_03.avi

rdj
10 October 2004, 02:30 PM
thats last avi had a very realistic simulation, cant wait to see some rendered stuff with the models!

colt
10 October 2004, 05:56 PM
Really nice. How do you drive the rc along your curve? Is it constrained to the curve and still reacting to forces?

laluneverte
10 October 2004, 06:49 PM
yeah.

basically we reimplemented a rigid body constraint, which is some sort of a path constraint including gravity simulation.

we have a speed value which changes over time, which is an interpolation of the following equasion per car:

if the angle between the curves tangent at the cars position and the gravity vector (which is negative Y) is more than 90 degrees, decrease the speed, if it is less, increase the speed (multiplied by the angle).

additionally, the constraint has a second curve, which we use as an upvector, so that we can control the banking on the track.

colt
10 October 2004, 07:39 PM
I'm doing something pretty similar in maya, since its rigid bodies let me down. It's not to practical, since you're limited to one path and it's hard to influence through other forces.
But it works.
I feed it into the uValue of a motion path and let maya do the banking.
Have the other cars their own simulation or just a offset to the first one?

laluneverte
10 October 2004, 08:14 PM
the big drawback of a softwarepackage "doing" the banking is that you don't have control over it... we want to include some twists in our track.... crazy stuff... ;-)

all of the cars are simulated and fixed to each other.

Sir_Vuclan
10 October 2004, 10:49 PM
wow, thats looking really nice, I like the design in the first pic.

ntmonkey
10 October 2004, 01:50 PM
Helge,

I'm glad I didn't insist on using the software's dynamics system. I think we would have been fighting with it for a long time. Good call on writing our own.

Now we just need the designs for what you call "wagons".

And yeah, I'm with DeepGreenJungle. Next time, when I have another "Hey! Let's do this!" idea, just stop me.

peace,

Lu

laluneverte
10 October 2004, 05:51 PM
sure.

I know how to handle ya.

new test> little fast though...

http://217.160.138.15/helge/rollercoaster/sim_test_04.avi (5 megs)

DeepGreenJungle
10 October 2004, 07:51 AM
Hey here's a screenshot of a little thing I've been working on for this: procedural generation of support pylons based on the coaster's curve. Right now, I'm just generating curves that will referenced using our track generation/instancing system... so it ain't much to look at yet... but an idea of where we're coming from. The idea is to be able to open a blank scene, draw a couple of curves and a ground plane, click go and get a working coaster. The next step is maintaining the dynamic links from the curves to the geometry, so that the curves can be adjusted in the scene, and the track will updated to compensate. The final (and oh-so-delightfully over-engineered) step is to write another script that will generate the curves automatically, so you can just open a scene, click a button and get a working roller-coaster. Soooooo unnecessary. Soooooooo much fun.

Ok, I'm done with buzzwords that Helge feeds me, have a good night everybody.

http://217.160.138.15/helge/rollercoaster/pylonScreenShot.JPG

ACantarel
10 October 2004, 10:03 PM
Wow! This looks very promising!

"procedural generation of support pylons based on the coaster's curve"
Is this something self-written or is it something like a spacing-tool implemented in the XSI package?
So on each point of the track-curve a pylon will be generated with the pivot-point aligned to the point of the curve? How do you orientate them? Is XSI giving you the normal orientation of each spline point? Hmm... must be. Or how did you rotate each pylon?

Hehe, maybe you build a small database of loopings, curves at different angles, straight lines...
-> Select them into a list like you want, align and position them in the order of the list, evaluate -> track finished :)

André

Edwin Braun
10 October 2004, 03:09 PM
yeah.
..........we have a speed value which changes over time, which is an interpolation of the following equasion per car:
if the angle between the curves tangent at the cars position and the gravity vector (which is negative Y) is more than 90 degrees, decrease the speed, if it is less, increase the speed (multiplied by the angle).
.........

Do you also care about the mass of the car also? It is an important part if you have more than one car and steep curves! The dynamic behavior is different especially if there are no other forces than gravity. Can you put a heavy passenger in the back of your car and see how the whle system behaves with that ?

Or lets say a heavy car in the middle or in the front ? If it is a gravity based simulation you are doing such effects are also important :)

I guess friction is also part of your simulation? Or is it a contact less magnet based car (it has also some friction) :)

Anyhow keep up your simulation work I wonder how this turns out. What kind of programming tools do you use?
edwin

laluneverte
10 October 2004, 07:41 PM
Hi Edwin,

our simulation actually doesn't use mass, but it will, I guess... ;-) thanks for the suggestion. We use friction, which I will update so that it is also dependent on the mass of each car.

We use javascript only to do all of the stuff.

ACantarel: It is our own written setup, and yes, XSI offers you all of the curve data you need to do this kinda things like tangents, normals, binormals and so on....

Helge

DeepGreenJungle
10 October 2004, 08:37 PM
<QUOTE>Wow! This looks very promising!

"procedural generation of support pylons based on the coaster's curve"
Is this something self-written or is it something like a spacing-tool implemented in the XSI package?
So on each point of the track-curve a pylon will be generated with the pivot-point aligned to the point of the curve? How do you orientate them? Is XSI giving you the normal orientation of each spline point? Hmm... must be. Or how did you rotate each pylon?

Hehe, maybe you build a small database of loopings, curves at different angles, straight lines...
-> Select them into a list like you want, align and position them in the order of the list, evaluate -> track finished :)

</QUOTE>

Yeah, we wrote our own stuff using Jscript and are using a few compiled tools that Helge had laying around. There's good stuff in the XSI object library for getting information about curves and patches (for the ground plane). As for orientation there's a second curve (hidden in the screenshot) that defines the up vector for placing objects along the track. The pylon thing still needs a bit of work to make sure they won't intersect with the track or other poles or things, but all in good time.

laluneverte
11 November 2004, 02:49 AM
just a quick test using the revamped simulation.

I added friction based on mass as well as friction based on overall angle.
So now the friction also depends on the "bend" of the track.

http://217.160.138.15/helge/rollercoaster/sim_test_05.mov (6 mb)

Sir_Vuclan
11 November 2004, 03:17 AM
The video is really wide. It plays in my browser window and I couldnt get to the play buttons and my screen is set at 1280x1024. Anyways, nice update, looks like its working great. I like the look of the track, with the little blue things on it, looks neat.

laluneverte
11 November 2004, 04:59 AM
upps. sorry.

I always forget about these things as I am using a dual monitor setup.

*hitmeintheface*... I will do better next time.

ACantarel
11 November 2004, 08:16 AM
Very nice! The dynamic looks authentic!

André

FrankFirsching
11 November 2004, 10:21 AM
Great results! The design is also very nice. How is it going with your automatic generation of support pylons?

Frank

Yodimus
11 November 2004, 05:56 PM
DeepgreenJungle - YO! It figures I'd find you doing something pointlessly insane.

- Eugene

Odrakir
11 November 2004, 01:35 PM
wow!:eek:

You have to show us how to make that!

beccabaldwin
11 November 2004, 08:25 PM
lookin' sweet guys! that last video is mind-blowing.

Yodimus
12 December 2004, 02:59 PM
...And then, in a cloud of smoke, they disappear. Weird.

CGTalk Moderation
01 January 2006, 05:00 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.

1