View Full Version : FXWars! RollerCoaster!: Sir_Vulcan
10 October 2004, 11:38 PM
Ok, This is my first try at any competition here, hope I dont come in last. Anyways here's my first tests. It doesnt look like much yet, but I wanted to get the Physics setup first, then I'll work on the design and modeling. I am using Max6 and Reactor 2.
I only have two cars so far, to setup the main wheels and links between cars. Ill add more later including the under and (maybe) inner wheels.
LINK TO FINISHED COASTER PAGE
10 October 2004, 11:42 PM
10 October 2004, 11:43 PM
Hey, looking nice. But what about turns? :p
10 October 2004, 01:55 AM
Got an update, I finally got it to go over the loop, theres to much drag some where. Hopefully my new version will work better.
Anyways here's one of my tests .mov
Tell me what you think...
10 October 2004, 09:56 PM
Ok, im having a lot of trouble with the banking around turns and loops and such. I have tried to use splines, NURBS, and bones.
The spline won't let me control the banking and often banks the wrong way or just makes a complete flip.
The NURBS is great for controling banking with the 2-rail extrude, but then it scales the track to fit between the lines, so my rollercoaster can't go down it because the track changes shape.
the bone idea that Ironlips is using is really nice, but he's using Lightwave and I cant seem to find such a nice feature like that in 3Ds max6. I know it has the spline IK chain, but I have to make each and every bone, and then go back and change the height so there all the same.
I have also thought of just using a mesh and then extruding and moving, extruding and moving, but that would take forever to do, and no easy way to edit the track once is done.
If anyone could give me a tip as to how to control the banking, that would be very nice.
10 October 2004, 03:01 AM
One way to do the banking could be using the loft tool. Create the spline for the path and the shapes for the shape to be extruded along the spline and then after lofting you can edit the "twist" graph under deformations or something like that, so that you can get the shape of the track right.
10 October 2004, 03:10 AM
ya, thats what I ended up doing earlier today, I had tried lofting before but I was doing it backwords. I was picking the shape and lofting it along a line, instead of picking the line and lofting the shape. I think I have it all figured out now. Ill be posting another testing clip soon. Then I think I am ready to make the complete track and work on the visual part.
Thanks for the tip
10 October 2004, 03:50 PM
Ok, here's an update. On the fourth version of the cars, and I got the loft and banking to work right. http://www.uni-weimar.de/~rehberg/vulcan/CarPhysicsTest04.mov
Time to make the entire track and start adding details.
10 October 2004, 01:50 AM
Nice setup! Seems to work very well!
10 October 2004, 01:54 AM
the thing usually doesnt slow donw on loops tho
10 October 2004, 01:58 AM
Ok, here is my track design. I might have to change a few things depending on the physics, but the shape is there. Ill post a picture of the cars soon too, so you can see how they are setup.
So, what do you think?
10 October 2004, 02:14 AM
here's the car setup I have right now, I was originally planning on four rows of people, but I think the cars might be too long to bend properly around the track. So I might make it into two rows per car, and six cars per train.
11 November 2004, 03:05 AM
Update on the track. It's a real pain to get this thing to bank right, expecially on the looping parts where the track flips over. I have to go through foot by foot and edit the twist curve. And on the loops the angles on twisting can go from +125 degrees to -154 degrees in just a fraction of an inch. Theres got to be an easier way.:shrug: oh well. it works.
Note: I've only gotten to the first corkscrew with the banking.
11 November 2004, 05:06 AM
Do you use two curves, one for path and one for twist?
If so... make the second one a clone of the first one and move each
control point a lil bit along the normal of the of the first one as the
start... makes it easier to tweak....
11 November 2004, 08:23 AM
"It's a real pain to get this thing to bank right"
Yeah, welcome to the club. This is really a pain. The funny thing is when you try to spread the mountings of the main tubes along the track. The "banking" of the tubes was done with the twist curve. For the mountings you use the spline (with the pacing tool). There´s no way to get the twisting information into the spline so the mountings have to be rotated each by each. Or did I miss something?
11 November 2004, 09:14 PM
laluneverte: No, I am only using one spline (and then editing the twist curve in the lofts Deformaiton properties) I don't know how to use another spline for twisting the track. :shrug: I am using max 6, so if you have any suggestion feel free to share them.
ACantarel: lucky for me my track is solid, so I dont have to worry about the supports between the rails. I will have to make the track supports (by hand) but that shouldn't be hard.
11 November 2004, 05:17 PM
ACantarel: I had the same problem when trying to attach the ties. But it seemed that once I messed with the axis orientation and stuff, it actually did follow the twist using the loft. The screen that is in my topic shows those ties, I made the ties, then used the spacing tool along the original spline, and it actually did bank itself to the loft. It was strange, and I am still trying to figure out how it did that. I can get it to work on like straight hills, but I will try some banked curves in a few hours and see if I can get it work....
One question for Sir_Vuclan though, how do you plan on doing supports?
I haven't tried it yet, but only ideas I have are to either model every support, or use splines to connect the track and footers, then loft a shape onto those splines.... but I would like to hear your input....
11 November 2004, 02:48 PM
I think I am just planning on using splines and lofting the supports or in some cases just make the spline renderable. I haven't thought about it to much, Working on getting the track built so I can test to make sure the cars make it all the way around. Then after it all works, Ill add everything else.
11 November 2004, 07:14 PM
Ok, new version of...almost everything.
Cars (made them shorter to bend around track better)
added more cars
tweaked Reactor more
Its getting there...
11 November 2004, 04:49 AM
I have the coaster working around most of the track now. This clip doesnt show all of it, but you get the idea. I have a lifting device (box following a spline) to get it up the hill.
I have also started to work on the enviroment around the coaster.
Link to the newest Movie:
11 November 2004, 11:42 PM
Ok, I finished my coaster. Now I just have to let it render for about a day. All the physics ended up working great. There are some things I would do differn't if I have to to this again. But for my first FX-Wars Challenge, I think I did ok.
It may not look the best, spent very little time on the eviroment. But thats not the point, the simulation works. I would like to add more, but not enough time to work on it.:sad:
Final Specs for the coaster:
Length- 1,983' 11" (604.722m)
Height- 112' (from lowest point to highest point)
Time- about 1 minute from start to finish
Used Max6 with Reactor 2. (and one stressed out computer)
Each train car was bound to the track by 6 wheels, Two on top, Two inside, Two on bottom. All the wheels are constrained to their car with wheel constraints. Each car was coupled to the car before it using a point constraint. The lead car uses a hinge constraint to allow it to bend around the track.
Simulation time- about 2 hours for 1550 frames, with 15 sub-steps per frame.
I am planning on rendering it 3 times.
- On the coaster with enviroment
- On the coaster with no enviroment(wireframe)
- Stationary sky cam.
Ill be posting pictures and movies the next few days as I get things rendered, so Check back and LEAVE A COMMENT (whats good or what needs to be fixed), so I can do better next time.
It was a lot of fun and a GREAT learning experience. (I am self-taught, about 2 years worth). Good luck to everyone else :thumbsup:
11 November 2004, 11:59 PM
Looking really good!
I am very impressed! using reactor for that whole thing must have been a super pain!
11 November 2004, 12:00 AM
11 November 2004, 12:11 AM
Car setup (What I used with Reactor for simulation) I know the car frame is complicated, but reactor doesnt really care because it used for weight and as a parent for the wheels. I didnt use it for collisions.
11 November 2004, 12:33 AM
Some on board cam shots
11 November 2004, 03:53 PM
Ok, here is one of the Final Movies. This one is On board the train with the enviroment. Ill post the wireframe one soon(just the coater and the track)
It took me many hours to make and many more waiting for things to calculate. I started it a few days after the competition started and worked on it a few hours a day until yesterday.
This one took 23 hours to render on my poor computer and the compression pretty much wasted the texturing. But it's still cool to watch.
Its 640x480 and filesize is 14.8 Mb
12 December 2004, 01:12 PM
If you like seeing the wireframe part of an animation, then this one is for you. It only show the peices I used in reactor for the simulation and theres no mesh smoothing or anything like that. Its exactly what was used for calculations.
Heres the link:
For those that are interested, I'll post the number of polys for each piece
Each train car - - 2280 faces
Each Wheel - - 2090 faces
Total for the Track - 167264 faces
12 December 2004, 01:51 PM
Wow! that is really cool!
12 December 2004, 01:56 PM
looks great, I would never have thought about using reactor for such an animation...
btw. nice to see some folks from weimar here :bounce:
12 December 2004, 04:12 PM
Hey Vulcan, I really like your setup. As I said before it somehow "feels" right. Maybe just a little too smooth in some sections, but very convincing anyway. I'm pretty impressed that reactor can simulate collisions of so many wheels with so many polys without errors and in a "reasonable" time. (But I guess tweaking it was very time-consuming ... )
You really rebuild the workings of a coaster, without simplifying or faking things. Did you also use correct weights and sizes? How do you get it up the first hill?
The interpenetration of the cars spoils the look a little, but I guess those curves are just to hard. I'd expect to see some broken necks in a real coaster in those. The ride through the waterfall is nice, but that bend afterwards is pretty sick. :argh:
Anyway, congratulations. :applause:
12 December 2004, 05:49 PM
Hi, Thanks for the comments
ernest - Reactor worked great, I was really impress with it. I have used it before but not to this extent.
colt - Tweaking reactor wasn't to hard once I figured out what to tweak. It handles the polys pretty well. There were too many polys for the real-time previewer, but I didnt use that a lot. The key to getting the simulation time down is setting the collision pairs. If I didnt set it up and left it at default where everything collides with everything, it would take many more hours to calculate.
ya, the cars do interpenetrate each other in a few places, but thats because I added them after I simulated the train using the frame part. Guess I'll have to remember to think that through next time and build the cars before.
The weight of each car is about what it would be in real life.
The size is close to the real things too. The cars are a little shorter(height) then should be, it made the coaster look more streamline and faster.
You can see the two places where I pulled the coaster up hill in the wireframe movie. There are two boxes that grab a hook on the bottom of the front car and pull it up. Those boxes just follow a spline. I tried a motor and pulley system with reactor but it was giving me trouble.
The bend right before you go back into the waterfall is really tight, I was going to fix it but forgot to. The coaster barley stays on around that bend.
12 December 2004, 02:00 PM
Yay, we started voting, go vote for me :thumbsup:
12 December 2004, 11:00 AM
wooooooooooooaawwwwwwwww!! incredible work !!
so u really do that with reactor Wheel??? I m interested in your wheel settings because discreet give a poor ridiculous example and setting the Wheel is not really easy. So I never try because of the lack of references.
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.