tank tread/chain saw chain rig

[fjg]

I'm trying to make a rig for a tank-tread/chainsaw type mech thing. It doesn't have to be a perfectly
real sim. I'm not really concerned with the treads matching up with the ground, actually a more exaggerated
motion is the goal - imagine a chainsaw.

See video.

The idea is to use a strip of geo as a softbody (Momentum) and adjust the shape of the tread/chain 'band'
to make it look like its spinning (rounded) or relaxed (draped). The 'pressure' and 'recover shape' parameters
of Momentum are great for this. So far, this is looking good... However, The strip isn't actually rotating
of course, and that is where one problem is.

tread band, no 'teeth'

I tried a few things:
Shrink-wrapping another geo to the tread band(and rotating that) was not successfull because of the oblong shape, but
maybe I didn't set it up quite right.

I thought raycasting a bunch of points from a null in the center of the band (and rotating the null) might work, but
because the tread geo changes shape fairly dramatically, it would cause the instances to pop and not be evenly
spaced when the tread is relaxed.

I think the best way is going to be some sort of uv to geo animation, but I can't figure out how to get that many
instances working, This is a great script for one object, but I'm not sure about looping and motion blur etc...

The other issue is that the Momentum softbody doesn't seem to like being enveloped/parented to null or whatever to animate
it (like a syflex softbody can).

Any ideas on either of these things? maybe animating along a curve deformer attached to the tread-band?

[azurecgi]

I used this tutorial as a starting point - http://m.youtube.com/watch?v=4yQua6SGf1A

I used nulls instead of cubes and then enveloped the tread mesh to them!

[fjg]

thanks - for the reply. For some reason, tho, I can't get the link-with part to work (it wont cycle correctly). < -- EDIT: works now, needed an open curve.
This may work if I can transfer the 'roll' created by the ribbon to the constraint.

[fjg]

So here's an update test using some temp geo.

It's working for the most part, thanks to the UV slide compounds linked above.
It didn't seem to like the uv unfold op, not sure why - using a cylindrical projection works fine.
Anyhow, as I populated the tread ribbon with 'teeth', by duplicating and using an offset in the
ice graph, I've run into a little problem with cycling. Because of the offset, the teeth reach
the end of the ribbon geo before the null on the flattened mesh gets a chance to start the cycle
over. This causes it to pop back to the flattened mesh until the null starts the cycle over
(you can see at ~2.5 seconds). The effect will only get worse with more offset.

any suggestions?

[Mathaeus]

it seems to be possible to "foolish" the "UV deform" by modulo. Take look at ICE tree on "thread move", try to move "MOVE null" along world x.
Anyway it doesn't work when "move null" goes into negative world X, I guess ICE don't like negative value as input for modulo - shouldn't be a big problem, you
can add some huge positive value in advance, modulo already keeps everything into same range.

take it here

About UV unfold, I guess it adds some different name of UV projection, instead of default one, referenced in 'UV deform on catcher' compound.

[fjg]

Thanks for response. This looks great, but there is one hitch (I was unclear earlier). The compounds
I ended up using are from the file am-UV2Position-Object.emdl for sliding an object on surface (the one
with the rocket). So all the teeth are separate objects, not one mesh. I'll try to hack this in to that ice
tree, but I may come back for help!

[Mathaeus]

Thanks for response. This looks great, but there is one hitch (I was unclear earlier). The compounds
I ended up using are from the file am-UV2Position-Object.emdl for sliding an object on surface (the one
with the rocket). So all the teeth are separate objects, not one mesh.

well I didn't played with this compound in about last two years... I think there's 'set average' of point position or something. If so, you'll want same pattern with modulo, to be sure that all points are jumping back at once. But, maybe it's good time to re-consider this setup, probably to try point cloud with instanced teeth, if you already want teeth to be rigid. I guess, now you have a lot of ICE trees all around, while with instanced point cloud you;ll have just one ICE tree.

[fjg]

I think you are right, I'll probably end up with a pointcloud. I just have to figure a way to make them traverse the
geometry.

Thanks for your help!

[3DMastermind]

I'm sure this is a great learning opportunity for you doing this in ICE but since you don't need simulation, you are going about it all wrong.
This should take 10 mins to accomplish:)
As you said follow my tank tread tut as a base:
http://youtu.be/4yQua6SGf1A
but use a long subdivided cube for the belt instead and deform by curve
In the deform params make the belt spin continuously
and if you want to show that expanding contracting effect on the belt...
Shape animate the curve and animate the shape params.
You can also simulate that motorized effect on the belt by playing with a separate shape parameter and going back and fourth according to the speed of the motor (perhaps with an expression on the animation param).

ciao

Newcomer (<20 posts) alert: please use the URL tags - HB