Animation Concept Question

[brucelay]

HI there,

Does anyone know what Layer animation means? Recently, I heard that Pixar is using this "Layer aimation" concept for for many years for their productions but I read through the whole Rickard Williams animation survival kit but I still could not find the meaning of Layer animation. Could anyone refer me to a link or book to me? (about layer animation)

thanks!

[lisux]

HI there,

Does anyone know what Layer animation means? Recently, I heard that Pixar is using this "Layer aimation" concept for for many years for their productions but I read through the whole Rickard Williams animation survival kit but I still could not find the meaning of Layer animation. Could anyone refer me to a link or book to me? (about layer animation)

thanks!

At least In my knowledge Layer animation means to put one animation over another.

For example you animate the overall movement of the arms of a character, and later of it in a diifferent layer, so you are not affecting the previous animation you refine it.

One of the goals of takes in Houdini is to do this. Create one layer and then animate anything there. Craete a child take and refine te animation, you are modifying the previous one but not loosing ot,, if you like the chamges then merge your child animation to the parent and you get the final movement.

Hope this helps ..

[thekenny]

Essentially it is animation on top of animation- think of it as non-destructive animation. Which is what CHOPs was intended to do, but it isn't as simple as that.

XSI has layer channels manipulation.IMHO it is pretty cool (but I haven't used it myself in production, but in theory it is what you would want).

This concept is really powerful, and I'll give SESI props for TAKEs, but really it isn't the same thing. TAKEs is a just animaiton, either you have this or you have that. Some may argue that you can use it to 'layer' animation to provide different results, but in the end it is a TAKE and as a result it is finite. I'm not looking to start a TAKES war but the real power to do this would be to allow CHOPS to be a subnet of a channel in the channel editor where you would manipulate the incoming data with other nodes and channels and essentially redirect the output to the same channel. That is a bit abstract but anyone who has used CHOPs to manipulate data even spare channels you will know what I mean.

-k

[andrewlowell]

Well, the problem with CHOPs and layered animation is that there are too many options to make it something you can "just animate" with IMO. But, I think with DA's and a decent animation-baking workflow CHOPs could still be an excellent tool for animation layers. Of course it goes without saying that all the other apps this is a completely non-procedural process, CHOPs is 100% procedural if need be.

I think of those "animation chunks" , and video editing style interface when I think of this method. Maybe it would be nice to have a tool similar to the channel editor to do layer/block editing, Actually, if Houdini does ever adopt the typical non-procedural layered approach found in other apps I think it would be nice to be able to load in COPs for video editing and CHOPs for sound or animation editing, just a suggestion.

Edited January 2, 2008 by andrewlowell

[QuantumChaos]

I think "layer animation" might also refer to "layering" as opposed to "pose-to pose" animation. With the pose-to-pose method, the 3D animator works in a manner similar to 2D animators, posing everything in a keyframe right down to the fingers and toes and then moving on to the next keyframe. With the layering approach, the animator gets the movement of the hips and the weight right first and then moves on down to the legs and feet. Once the legs and feet are completed, the animator will then work on the upper body and finally on the fingers and expressions.

John Lasseter talks about the two methods in his paper "Tricks to Animating Characters with a Computer" (Sigraph '94)

which can be found at http://www.siggraph.org/education/material...4.htm#keyframes, although he does not use the term layering in this paper.

Edited to add:

In their book "Thinking Animation", Angie Jones and Jamie Oliff use the the term "layering" to describe this method of working, and I think I have seen it described as layering elsewhere by John Lasseter, although I can't find any references.

Edited January 13, 2008 by QuantumChaos

[gpa]

At least In my knowledge Layer animation means to put one animation over another.

For example you animate the overall movement of the arms of a character, and later of it in a diifferent layer, so you are not affecting the previous animation you refine it.

One of the goals of takes in Houdini is to do this. Create one layer and then animate anything there. Craete a child take and refine te animation, you are modifying the previous one but not loosing ot,, if you like the chamges then merge your child animation to the parent and you get the final movement.

Hope this helps ..

Could you give me the links to the definite tutors on how Layer animation works in Houdini?

Thanks

[v_m]

found this post accidentally... so just in case it's helping anyone, one easy/cheap way to layer animation in pretty much any 3d application:

you have a controller, you duplicate it (or add a null at the same position), if the controller already has animation on it you delete it, making sure you don't offset its location, and this controller can be a hierarchical parent OR child of the original controller, but normally you want it to be a child - if there already is some animation on its parent, it will translate/rotate along, and not be left behind. this way you can add an entire hierarchy of controllers, even during animation, if you're allowed to modify the rig I prefer doing it this way, so I don't build unnecessary layers, but for a cleaner interface, the otl should have those hierarchies of controllers already made easy to hide/unhide, that just means the rig will have to have something like 2, 3, etc layers of controllers.

on the other hand, it would be great if houdini would add native support for animation layers, like the ones you find in motionbuilder.

[KrisStaber]

the layer animation you mean (pixar talk) has definately nothing to do with technically layering animation curves on top of each other.

like quantum chaos already explained

its about animating sperate bodyparts independent from each other well not independent, but seperately

so you usually begin with hiding everything but the hip (hide arms legs head upper body)

then animate the hip movement till the timing and weight and movement looks good

then unhide the upper body and do the animation on that so that it works with the hip movement

and so on

and so on

its a very 3D ish animation approach

Pixar at the beginning only worked like that and taught all the animators that approach as the 3D way of doing it

nowadays most people use more of a 2D ish approach where they pose the whole character - so thinking about the key poses first and not worrying about timing at the beginning

some animators start off with 10 key poses for a 400 frames shot and having them all on the first 10 frames

then adding breakdowns (first 20 frames) and then moving then out on the timeline thinking about when what pose shoud be

the third technique is straight ahead where you just go off and animate from frame 1 to 400 (often used for action shots/moves)

from what i heard animators at pixar roughly work 1/3 layered, 1/3 pose2pose, 1/3 straight ahead depending on personal preferences

I dont know about other studios (other major feature film studios might have some animators animating layered too)

richard williams didn't write about the layered approach because he probably didn't know about it - and maybe still doesn't know since its a very 3D ish approach.

Edited July 19, 2008 by KrisStaber