tanks, but no tanks

[robcat2075]

What would be the least obvious way to make tank treads?

 

Tank treads made with, what else... procedural materials!

 

The top is the final render, the bottom is the wireframe.

 

treads9CompH264900.mov

[Darkwing]

and Rob has done it again. Next thing we know Rob will have built a model as detailed as Al's Discovery or something, except completely with procedural materials!

[jason1025]

That is so impressive. Tutorial!!!!!

[TheSpleen]

Looked perfect!

[largento]

That's extraordinary!

[Paul Forwood]

Amazing!

[williamgaylord]

Brilliant! You're a genius!

 

I bet you could use this idea to get a "baloon" tire to bulge at the bottom, too! Can you sum two such effects--one for the tread and one for the bulge?

[robcat2075]

Here's a slightly more detailed treatment of the tread. (I have no idea what is on the inside of tank treads.)

 

 

Another day lost to A:M!

 

I started this just before lunch and got the basic tread concept working in an hour, but then i couldn't leave it alone. Now my stomach is grumbling.

 

 

I bet you could use this idea to get a "baloon" tire to bulge at the bottom, too! Can you sum two such effects--one for the tread and one for the bulge?

 

I think you're limited to one displacement material on a surface.

 

But you can model the bulge in the tire and spin the tread on the model, much as I'm doing here.

[Darkwing]

I'm assuming this must be quicker to render than a model(?)

[robcat2075]

I'm assuming this must be quicker to render than a model(?)

 

Probably not, displacement is tedious to render. In this case 16 passes take 2.6 minutes per frame.

[mtpeak2]

Pretty cool Robert.

 

I'm guessing the geometry is circular, then posed to the tank tread shape?

[itsjustme]

Extremely cool, Robert!

[johnl3d]

Great Idea worth the lunch delay

[Xtaz]

you are sick !!! bravo !!

[brainmuffin]

That's awesome! I'll take the render hit over having to rig treads any day! Is this going in the book?

[Bertrand]

What would be the least obvious way to make tank treads?

 

Tank treads made with, what else... procedural materials!

 

The top is the final render, the bottom is the wireframe.

 

treads9CompH264900.mov

Great job for the movement we can feel the elasticity of the rolling chain.

[steve392]

Thats a clever trick Robert ,I hope this aproach is going into your book.I know nothing of how this stuff works so would be a new thing for me to tackle

[robcat2075]

Thanks, everyone, for your positive comments!

 

I'm guessing the geometry is circular, then posed to the tank tread shape?

 

That's exactly right. It's all "Extended GridTurb" combiner.

 

I hope this aproach is going into your book.I know nothing of how this stuff works so would be a new thing for me to tackle

 

My problem is figuring out how to teach it beyond just giving an exact series of steps. As it is, the process is rather un-intuitive. The tree you make in the PWS gives little hint of what the result will be.

 

If we could drag and copy nodes in materials around that would be a big help, but we can't.

 

Also, these displacement things can behave very oddly in animation. How to explain that it's not going to be perfect?

[steve392]

Maybe a step by step for a simple as you can example would be good and then explain that it has to be played with to get more of the not so simple stuff.But I can see how this would be a lot of text and take a lot of room up ,maybe you could referance a tut on your site as a way forward Rob

[John Bigboote]

Color me impressed! MOST impressive about this is the highly accurate way it is animated, with the underlying mesh static. It begs to question, are the two treads independant of each other? Meaning, if you were to animate your tank going around a corner, could one tread spin faster than the other? Great work Robcat, you've opened many an eye to the power of procedural materials, Brian Prince would be proud.

[Darkwing]

speaking of procedurals, I found out the other day that one of the turbulences (perlin I think) was created by the special effects artist for the original Tron film and the purpose of the turb was for Tron. Nifty little facts you find on wikipedia

[robcat2075]

Color me impressed! MOST impressive about this is the highly accurate way it is animated, with the underlying mesh static.

 

There's another hard thing to teach. Rotating the materials (there's four that have to move exactly together) just the right amount so the tread appears in synch with the ground is tricky. It's not quite right in my demo.

 

 

It begs to question, are the two treads independant of each other? Meaning, if you were to animate your tank going around a corner, could one tread spin faster than the other?

Yes. You'd animate separate instances of materials on each tread.

 

 

speaking of procedurals, I found out the other day that one of the turbulences (perlin I think) was created by the special effects artist for the original Tron film and the purpose of the turb was for Tron.

 

About 10 years ago I went to a talk by one of TRON guys, Chris Wedge. He said there was no interface in their software at all. Everything was done by coding on punch cards to control the renderer. Ouch.

[3DArtZ]

Robcat, I for one am sick of you lies!!!! you obviously faked the tank treads! lol.

No but seriously great work. I wonder you probably could do airplane propellers like this too?

Or maybe bullets feeding into a machine gun?

very very cool stuff you've been posting.

Mike Fitz

[agep]

Very cool!

[mtpeak2]

You could use expressions to animate the rotation of the materials automaticly, just like the wheels in my suspension rig.

[robcat2075]

You could use expressions to animate the rotation of the materials automaticly, just like the wheels in my suspension rig.

 

Does that handle corner-turning?

[robcat2075]

I wonder you probably could do airplane propellers like this too?

Possibly, although a propeller is so much simpler than tank treads to rig and animate that doing it conventionally is probably better.

 

 

Or maybe bullets feeding into a machine gun?

 

Yes, that would be very much like tank treads. I'm not sure we have a combiner that would make a nice cylinder shape for the bullet casing. Bullets move thru their positions so fast that strobing might be a problem.

 

This tank tread thing is best for repeating shapes that move quite slowly.

[mtpeak2]

It could with the right setup.

 

Here's a test using a bones translation to drive the rotation of the material. (note: the material is not as good as Robert's )

mat_test.mov

[mtpeak2]

Here's a new test with turning.

mat_test.mov

[itsjustme]

Here's a new test with turning.

 

Very cool, Mark!

[robcat2075]

Here's a brief clip showing some of the oddness complex displacement patterns have when animated.

 

treads11_LoHiComparison_H264.mov

 

Watch the top version around the edges of the tread and you'll see strange flickering and rippling.

 

I think it has something to do with displacement texture being seen at a very shallow angle. The ripples seem to be about one pixel in size rather than all different sizes, which makes me think that somewhere between the material that generates the grayscale pattern and the actual displacement process, a value is getting passed thru a lower precision than it ought to be.

 

Just a guess, I really don't know much about how it all works.

 

The bottom is the same frames, but rendered at double res then shrunk down. Less flickering and rippling, but not completely gone either. Its ripples were about one pixel on the original large frames also. Haven't tried quadruple size.

[pixelplucker]

That is pretty dang slick, does it render faster than pure geometry based treads?

 

Many years ago I had to do an animation that had a conveyor belt, I used deformations to conform the objects to a belt shape which caused some distortion on the height. Back then displacement mapping wasn't all that great in EI and required high Dicer values to smooth things out.

[robcat2075]

Here's a comparison of single, double and triple resolution renders of the problem spot. Flickers and ripples are gone in the triple res render.

 

treads11SingleDoubleTripleMP4.mov

 

That is pretty dang slick, does it render faster than pure geometry based treads?

I'm sure it's slower. Also, Displacement only shades properly in multipass so you can't take advantage of faster regular renders.

 

Many years ago I had to do an animation that had a conveyor belt, I used deformations to conform the objects to a belt shape which caused some distortion on the height. Back then displacement mapping wasn't all that great in EI and required high Dicer values to smooth things out.

I'm finding that over-rendering helps in A:M too.

[NancyGormezano]

I'm finding that over-rendering helps in A:M too.

 

A slight blur post effect might also help in anti-aliasing instead of oversampling? Reduction in resolution of the original render averages the pixels in some manner which is a form of anti-aliasing. Same with blur.

 

It looks to me like displacement runs into problems when it resolves to less than 1 pixel. That's what anti-aliasing is for.

[robcat2075]

I'm finding that over-rendering helps in A:M too.

 

A slight blur post effect might also help in anti-aliasing instead of oversampling? Reduction in resolution of the original render averages the pixels in some manner which is a form of anti-aliasing. Same with blur.

 

It looks to me like displacement runs into problems when it resolves to less than 1 pixel. That's what anti-aliasing is for.

 

Well, let's try that.

 

Here's the same three panels, but the top {original] render at the top is blurred 2 pixels, which is a lot of blur.

 

treads11Single_blurredMP4.mov

 

The flicker is still there, blurrier, but still obvious. Downsizing a larger render is the only way to get more actual anti-aliasing.

 

But I'm glad even that works. I thought it was hopeless.

[NancyGormezano]

Here's the same three panels, but the top {original] render at the top is blurred 2 pixels, which is a lot of blur.

 

The flicker is still there, blurrier, but still obvious. Downsizing a larger render is the only way to get more actual anti-aliasing.

 

But I'm glad even that works. I thought it was hopeless.

 

Yup - too blurry. Don't think a smart anti-aliasing filter would help either as in the smaller images, the displacement resolves to too few pixels (stairstepping/edge crawling effect is due to contrast with adjacent pixels). Need the extra pixels in larger images so that they can get "sampled/averaged" to more graduated, less contrast tones. No different than noisy, busy textures, or thin hair rendered at low resolutions flickering. I'm guessing.

[3DArtZ]

I find most of your results to be acceptable for use. I mean, the part that is strobing a little bit on the top back of the tank tread... most tanks today have that part coverd.

I would love to see what a render might look like with a basic tank shape body attached rolling over some grass or whatever.

NIce stuff.

Mike Fitz

[robcat2075]

I wonder you probably could do airplane propellers like this too?

 

You made me do it...

 

A procedural material that makes... a propeller!

 

propShadedFinal.mov

 

It's quite slow to render, so I recommend everyone continue modeling your propellers as usual, but for the curious here's a PRJ with the the material

 

Propeller05c.prj

[brainmuffin]

I was actually thinking about propellers the other day. I thought it would be interesting to try rendering a modeled prop spinning with motion blur, and then compositing several iterations of it in AE, then using that as a material on a disk. So when the engine starts up, the modeled prop spins up to speed, then gets replaced with the disk for full speed spinning.

 

I'll see if I can run a quick test off later today...

[NancyGormezano]

you are sick !!! bravo !!

 

Yup - that propell-lah thang has me totally convinced..have to agree with Marcos.

 

Quite Amazing! (as well as wonderfully ridiculous)

[itsjustme]

Another cool demonstration, Robert!

[3DArtZ]

RobCat you should see a doctor... you're one sick mofo!!!!

the propeller looks fantastic!

 

Mike Fitz

[3DArtZ]

Hey Robert, you're doing all this stuff with just the a:m materials? You could probably do similar things with using texture maps?

Mike Fitz

[brainmuffin]

Robcat, when you're rendering the treads, are you using fielded rendering? If not, is the crawling better or worse with fielded, I wonder?

[NancyGormezano]

Hey Robert, you're doing all this stuff with just the a:m materials? You could probably do similar things with using texture maps?

 

Noooooooooo....!!!

 

If Robert did it a lazy way, then we wouldn't be able to think him crazy...I kinda enjoy thinking of him as insane.

 

This experimenting, besides having great entertainment value, has the possibility of providing a spark for something more useful ...

 

It's known that complicated materials have an impact on render times (sometimes quite severely) - decals are usually, almost always faster, and easier to create, but of course have other drawbacks.

[robcat2075]

Hey Robert, you're doing all this stuff with just the a:m materials? You could probably do similar things with using texture maps?

 

You certainly can do similar things. The reason i go into materials for this was that I noticed displacement maps get a very grainy and/or stepped look at high displacement settings unless you used very high-res maps.

 

Materials, on the other hand, have infinite resolution which makes them well-suited for creating the grey-scales that drive displacement.

 

Displacement materials seem to have fewer of the blips that displacement maps have.

 

Also, I like the stunt value of doing something that no one would have thought possible.

[robcat2075]

Robcat, when you're rendering the treads, are you using fielded rendering? If not, is the crawling better or worse with fielded, I wonder?

 

No, haven't tried it since there's no case in which I'd want field rendered footage these days. In most CG field rendering makes things crawl more rather than less.

 

But I haven't tried it. Also i don't even have a way to properly play field rendered footage. I suppose I'd have to burn a DVD and play it on an NTSC set.

[robcat2075]

Slight further refinement. After a bit of research I've found that tank treads usually have some sort of tooth on the inside so they can be engaged by the drive wheel.

 

 

 

And you can see this animated (with no flickering) at my A:M tests gallery

[dinocarl]

This is a great thread, Robert. Wonderfully innovative. Are you now using mtpeak2's expressions to drive the movement now, or are you still animating the material?

[Xtaz]

GORGEOUS !!!!!!