View Full Version : FXWars! Catapult! (Trebuchet) : Westiemad

07 July 2004, 06:59 PM
This is the small humble area when my work will hopefully take place, although this cloth and hair trigger simulation does have me a bit jumpy.

07 July 2004, 07:06 PM
Mental notepad. (brain storming may not make sense to any readers)

Things to condiser during the simulation process.

Resolution of objects, naming, constraints, extensive testing on simple models to prove concepts.

Model, mostly done in its simple form, not that i'm a master modeler by any ways or means.

Textures, planar projections.

Dynamics, fields, hair or weighted CVs, possible dust, and maybe fluid effects (cause they look nice).

Main things to look into, unhitching of the rope, could be tricky. Cloth holding sack, either cloth, or softbody with a wrap deformer and some low res version for collision detection.

Software: maya 6

07 July 2004, 07:20 PM
Note, sort out the crappy facets on render, they are new since i broke mental ray playing with it last night. I know its not a good render before you tut aswell.

07 July 2004, 11:39 PM
ok, originality is not my strong point. Its a classic boring one.

To do - rig up arm so it swings


- simulate rope

- come up with some funky ideas

- finish the actual model before doing any of the above.

07 July 2004, 01:53 PM
ok i did a quick test to try and get the motion of the main arm right, so feedback on this would be nice. Next simulate ropes and pullies. I haven't played with dynamics in years.

To think of, way of winding down arm and attaching hook, (for look only). (1.12mb playblast)

07 July 2004, 02:22 PM
that is freaking unbelievable.

Very VERY nice job.

I wish I could make mine like that.

07 July 2004, 02:25 PM
I can help you make one if you want? It wasn't as hard as I'd thought, although I tend to bark up the wrong tree for a while. If you have any questions m8 i'll help as much as I can.

07 July 2004, 05:21 PM
ok, have made something to throw, it was rope, but they also made brackets on a ball obect, so i've done that, its no simpler just different. Here is the render for throwing the object (it doesn't go as far as it looks, e.g. its not an ICBM)If this was a real catapult I'd have smashed it to bits twice and killed the blokes loading it with a few miss fires, still as long as i didn't get hurt...

Still to do:
Ropes, particles, enviroment, object to hit. "Tart up catapult".

07 July 2004, 05:29 PM
ok, have made something to throw, it was rope, but they also made brackets on a ball obect, so i've done that, its no simpler just different. Here is the render for throwing the object (it doesn't go as far as it looks, e.g. its not an ICBM)If this was a real catapult I'd have smashed it to bits twice and killed the blokes loading it with a few miss fires, still as long as i didn't get hurt...

Still to do:
Ropes, particles, enviroment, object to hit. "Tart up catapult".

Looking Good :)

I want to see it knock something big down!

BTW... I added you to my buddy list on Yahoo Messenger.

07 July 2004, 06:01 PM

That ball strikes me as falling very slowly? Eh, not that I got my dynamics all proper either so I'm not one to talk. I just.. ah.. think maybe it falls too slowly? :) Or the scene scale is bigger than I think, that could be it.

Other than that, looks good. I'd suggest concentrating on the physics first, then the model later, because just getting the dynamics all nice and solid will take work.

07 July 2004, 09:44 PM
i think it might be the angle of the shot and the way its thrown, I started to make a sort of tower thing, not finished but u get the general idea.

07 July 2004, 10:22 PM
Hey man, put a lot of thought and R&D into your wall destruction before building something too elaborate there. If you try to run a simulation with a ton of bricks and want them all to be rigid-bodies, you will end up calculating for a year! I ran a simple test with about 300 bricks and a simple ball and I couldn't even get past frame 2 (I waited about 15 seconds and decided it wasn't practical). Your best bet will be to have a number of proxy objects that are collision objects, maybe 50 blocks tops. Have a collision event that says when they are hit, they bust up into smaller, instanced rocks that go flying. This will likely be the method I'm going to end up using but I figure before you put much time into a big elaborate tower, I'd warn you about the calc time.

07 July 2004, 10:30 PM
they won't all be on the same collision layer, but I understand totally what your saying. I blew up a columb into bits and it took a night to render the 10 second shot. That is a complex as I think I'll make it, and not all the blocks will be seperate, depending where i get the ball to hit it, i may combine a lot into 1 object. I chose the tower to be semi smashed to bits already to lessen the amount of work needed to do. I'm sure when i actually get round to having to make the simulation I'll PM you with a "you were right".

Thanks for the feedback tho, and await that PM.

07 July 2004, 11:58 AM
ok 10 mins this morning on the projectile. Not finished, but its a different. Can you guess what it will be when its finished?

07 July 2004, 03:57 AM
I can't remember the name of them, but when it is finished that thing looks like it will contain a (clay?) pot with oil and a wick, so it will explode on impact. If I'm right, then you will know what I am talking about ;)

And looks good!

07 July 2004, 09:36 AM
pretty much, its gonna contain a fire, which can then happily streak across the sky. I might change it slightly tho.

07 July 2004, 03:10 PM
trying to be a bit more adventurous (sp) this time, I've made a new catapult this time its on wheels, apparently wheeled ones through a third further? I will teather it to the ground using ropes, but will allow it a bit of movement.

07 July 2004, 05:23 PM
Awesome man, wheels is the way to go! I am going to commence building my final model tonight and going to put wheels on mine as well. I don't know if the dynamics sims will accurately transfer inertia all the way down through the model (the entire model will have to be a rigid body or multiple rigid bodies) or not but it should be cool to see how it works out. Post some more details when you get to the dynamic part.

07 July 2004, 07:19 AM
Your catapult looks good and i cant wait to see the fireball across the sky.
The reason the wheel thing works is that instead of waisting energy on the ground you use the hevyer weight at the "front" as a propelat for the catapult (PS: it will get back, but then the ball is fired and wont be affected by it) and therefor gives the ball extra speed. This will not work if the ball is to heavy (I think) because then the force will be equal.
Good luck!

07 July 2004, 08:25 AM
Looks good! The wooden boards are awesome.

07 July 2004, 10:57 AM
i'm still endevouring to make it as dynamic as possible, it now rolls and moves like a real one (sort of ;) ) and I playing with the release mechanism, so its dynamic also, going for the sack on a piece of rope, i fear I might have bitten off more than I can chew, although I am like every other typical male, who likes cars, and blowing stuff up.

07 July 2004, 04:47 PM
That's a nice model man! I love the wooden planks and the projectile just looks great! How about throwing a cage with a man inside? Aparently that was a popular thing to do.. :)

07 July 2004, 05:14 PM
Thanks echo and flipped :) .

At the moment I'm having problems getting the cloth looking sack to work, its very twitchy, and the normals are either one way or the other so it won't collide, or it won't grab the ball. Any clues would be handy.

Here is a shot of it wheeling. (

07 July 2004, 05:32 PM
Looks pretty good, but the wheels don't rotate quite right. The front left one slides backwards at the start, for instance. And looking around the forum, it seems that getting the rolling to work is problematic for more than just you.

07 July 2004, 05:37 PM
yeah i think it might tilt the frame slightly, i'll have to have a look, i tried a few values.

07 July 2004, 05:39 PM
Could you post a shot of the rig? Maybe just the constraints visible, or explain how it's set up or whatever you want. What kind of constraints are holding the whole thing together and such.

Not just to satisfy my curiosity :), I was thinking someone here could maybe come up with a solution for those wheels once the details are known.

07 July 2004, 06:00 PM

The Rig:
The major parts are all grouped together, so u have each wheel, the frame and the arm all in seperate groups. These are then all made in active rigid bodies. Hinge constaints are then made between the arm and the frame, then each wheel and the frame. The hinges can then be moved into position (of where u want it to pivot). I had a problem with whated out hinge constraints, so i created pins, then moved them into position, changed them to directionalhinge, and rotated them, and then changed those to hinge. You can just create hinge tho. All parts have gravity. I have changed the collision layers for the objects. Basically, if you have a number of objects and they are all on different collision layers, they will never hit each other. This makes it easier on the solver to deal with everything. The frame is on layer 0, arm is on layer 1, and the wheels on layer 2. I have used stand ins for the wheels. A stand in is exactly that, it stands in to the object, but is perfect in tesselation, so the wheels are spheres. To make the arm swing I have used a spring constraint. One end is on the arm, (which has its centre of mass out towards the weights), and the other end is parented to the frame (not constrained to it, eg i didn't select the arm and the frame and make a hinge, as this causes the 2 to jump together). Its parented to the frame so that as the catapult moves, the arm can still return to the centre of the swing, and not try and go back to its original position somewhere in the distance.

Things to check> normals, bit me again today, mass, things act very differently on the mass attribute, collision layers.

I'm having probs with the sack, still its all good fun, i don't think this is gonna be as impressive as Derbys tho (by a long way).

07 July 2004, 02:42 PM
i made my catapult completely dynamic, it releases the stone exactly like one would in the real world, and it plated back in the viewport. The downside was that after much tweaking, I couldn't get the stone to fly more than 50ft. So I have taken that completely dynamic idea away, and have stuck to a rope and a large fireball. Not as complete as Derby's, but hey ho.

07 July 2004, 01:13 PM
*mental note* large fluid containers take far to long to render, although the effects look nice. 191 640*480 frames took a night to render, way to slow for me. must finish destory and cache all.

07 July 2004, 06:03 PM
I'm not dead. Or have i finished, I have to finish writing a script and then rendering it out (after baking it), here is a render of the catapult, not in its final surroundings mind, but i hope its dynamic enough for u, no keyframes at all.

To do:
Landscape - model the ground, and maybe a tree (only 1 tho, jk)
Rigid body expression - clever bit to make it play in the viewport
Bake simulations - make everything predictable
Render - yes you will have to perfom a crappy lighting job on it all, and not make everyone sick with your camera cuts.

TEST ANIMATION: ( (1.36mb)

07 July 2004, 02:24 PM
yay the coding is complete, after spending the morning trying debug something. Hopefully u guys will give me some feedback on the animation on the above post, that fireball will smash into the tower on the first or second page. Can get on with that bit now.

07 July 2004, 08:04 PM
Hey Westie atlast we got to see your fireball it looks really good. The particles may be to big and to few, when its sitting still its buring the frame to much (maybe lowering the wind) and it looks too systematic. Ofcourse i can understand that adding more particles could make the rendering time unbearable.

Have fun

07 July 2004, 08:24 PM
many thanks squire, I shall make the changes, I'm sure if i ramp the rate up a few times it won't make too much difference.

Thanks for the feedback.

08 August 2004, 02:50 AM
Good job so far! Could you please explain how you did the ropes?

08 August 2004, 08:00 AM
Looks awesome :) although, maybe after it rolls forward,and goes back, make the front ropes go taught and have a slight bit of roll the other direction agtain, just to show this is an incredible force.

08 August 2004, 09:02 AM

Make a cylinder and scale it out.

Increase the number of divisions along its length (i think mine are 10)

Select all the points at each division and make a cluster (you'll end up with 10 clusters i think)

next create a poly cube (i scaled my cubes to 0.5 in all axis)

point snap the cube to a cluster, then duplicate the cube and point snap it to the next one and so on, until you have a cube at the position of each cluster.

freeze transform all the cube values

select the cubes and make them active rigid bodies

with the still selected add them to gravity

press play to check that the all fall, then rewind

create a pin constraint between all the cubes, first and second, second and thrid, thrid and fourth etc.

press play and they will all fall to earth connected

you can then pin or spring constrain one of the end cubes to a passive rigid body so it doesn't fall forever.

Thanks for the feedback guys, I will try and get it all sorted. Much appriciated

08 August 2004, 01:55 PM
damn i forgot to mention. once the cubes all fall how you want, point constain the coresponding cluster to its cube this way it will move the cylinder, don't orient constrain it tho as the cubes will spin.

08 August 2004, 10:42 PM
hey Westie

tx for the info on the ropes...
im begining to work my dynamics,, and was craking my head on that :)
cool work M8

08 August 2004, 04:27 PM
glad i could help you out m8, any questions just post them.

08 August 2004, 11:19 PM
This post is the same in my WIP thread as it is in the entry thread.

Well I'm done, I won't have any more time to do anything on it, there are 3 files avaliable, a crappy encoded one that is about 12.51mb, and a nice shiney one (thats the one you should get) thats about 46.23mb. There is also the required simple side view with all the bells and whistles turned off (6.5mb).

About the entry.
There are no keyframes on any of the objects, except a camera which are stepped keyframes.
There are a few expressions to drive the birds, the arm of the catapult (to ensure it evaluates corretly, which is more of a debugging command), one on the walls of the castle to get them to react to gravity.
Particles are cloud type, software rendered, 2 particle emitters, shaded using particle cloud, with ramp opactiyPP and radiusPP.
900 frames long.
Envsky for the sky texture.
Pin and hinge constraints used only.

I have learnt a fair amount in this little project, I've started to flex the dependancy graph in maya in a way that it shouldn't be used, and have got a couple of new tricks up my sleeve, as well as a new found confidence in trying out new stuff. I'm not a texture artist or a lighter but dynamically its pretty sound. Hope you like it as much as I enjoyed making it.

If anyone could help me host them I would be very very appricative. ( (NOW UP!!) (



08 August 2004, 03:49 PM
There are 3 stills for the 3 parts of the video.

08 August 2004, 07:36 PM
Hey westiemad great job :thumbsup:

The catapult motion looks just right, you really get the feeling its a huge solid object slung around and out.

My only crit is the impact it looks like a slow-mo but im guessing thats what you were after, Personally id love to see it at full speed blowing that wall apart! :)

anyway looks great!


08 August 2004, 06:09 AM
thanks :) , it does go a bit slow mo at the end, but it all about the catapult animation, e.g. how it behaves dynamically.

08 August 2004, 09:10 PM
new mirror added to help the file downloads, as reports of problems on the lycos ones.

CGTalk Moderation
01 January 2006, 06:00 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.