PDA

View Full Version : DIFFERENT ANIMATION SOLUTIONS: faster & easier !



robertoortiz
10-06-2012, 10:06 PM
Sorry for the double post, I posted thin on the wrong post.


Repost from CGSociety
DIFFERENT TOOLS FOR CHARACTER ANIMATION: faster & easier !
Hi everybody,

Coming from CFT-Gobelins, i worked during 12 years in video games development mostly as Lead Animator. But i'm crazy about ergonomy and designing new ways to animate with unique purpose: really easier, really faster and with the best comfort of work.

About 9 years ago, i designed the complete animation/mesh preparation pipeline of Eden Games intern software.

Please take a look here (tech videos): http://vimeo.com/acharrier

Coming from traditional animation, standard softwares always seemed to me complicated and laborious : i mean, my job is to animate, so, making posings and edit timings, and not to spend time to "handle" the software.
From their beginnings, 3D softwares are generalists, using about same windows and manipulators for all needs. They were all built on about the same basis and today they all look the same.
My idea was to going back to foundations:
- Re-work the joint to finally use it as it deserves. What is the interest of curve controller when you can select easily your joint and reset all its channels ? no more (at least for the body)!
- Propose a keyframes manager really designed to handle keyframes: with all keys easy to locate, easy to grab, easy to edit (with special rules).
- Find an IK style solver system with all the advantages of standard IK but without its disadvantages. Something you can put in 2 clicks, which is ready to work and easy to handle.
- And lots of other features, layouts and rules that make you go fast ! really fast !
And of course, i wanted that all of this be fully generic ! no specific dev per characters or else, all of this work for everything !

Please share your opinions with me !

This software represents the past, and i've got lots of ideas for future.
Today i'm no more at Eden Games, i would like to find a role in R&D as User Experience Designer in Animation software. I really believe anim workflows need to be deeply improved. I think lots of users would like fresh features and most of all, easier.
There are really things to do !

Please find as attached piece a hierarchy comparison illustrating well my vision of pipeline

Thank you

Cordially

Aurelien108349

djwaterman
10-07-2012, 04:40 AM
Something that bothers me about most CG animation I see, even top end stuff like Avatar, is that joints still seem to have a puppet look to them, elbows and wrists bend and it just looks wrong for some reason. Is it possiblee that the single pivot rotation point of a joint is not really how an organic joint of bone and muscle rotates. This is just an observation I have not backed up by any data.

bobakabob
10-07-2012, 05:02 AM
Something that bothers me about most CG animation I see, even top end stuff like Avatar, is that joints still seem to have a puppet look to them, elbows and wrists bend and it just looks wrong for some reason. Is it possiblee that the single pivot rotation point of a joint is not really how an organic joint of bone and muscle rotates. This is just an observation I have not backed up by any data.

Agree, also in many top end CG animation films, despite the sophisticated movements characters have an eerie weighlessness about them, there's no sense of mass e.g. Shrek. Not sure if it's being addressed.

kmacphail
10-07-2012, 06:23 AM
Is it possiblee that the single pivot rotation point of a joint is not really how an organic joint of bone and muscle rotates. This is just an observation I have not backed up by any data.

This is absolutely the case. An elbow's center of rotation is not fixed as it pivots, and I expect this is true for most joints.

I've seen an accurate elbow rotation modeled with a simple four-bar linkage, it should be fairly straight forward for rigging pros to replicate similar results.

Cheers,

-Kevin

Ryan Roye
10-07-2012, 08:11 AM
I'm fine with today's animation tools personally, so long as I always have access to a tool that doesn't require you to SELECT an item before manipulating it (IE: IKBooster). Having to select crap is such a huge waste of time and I don't understand why even top end software forces you to do this (unless I'm missing something)

It is the deform side of things that is the real problem; getting decent results means jumping over too many hurdles with today's software as far as I can tell. Multiple bones... relative movements... bone types... I'd love to just have a one-item solution for limb joints. I get that things get complicated around the shoulder especially for muscular characters... but I'd personally be happy with a simpler solution to getting good lower-limb-joint deforms. Sure, one could create a dense rig that takes care of these things... but that just means it makes things much slower if more than one character is being animated in a scene (this of course is a problem when 2 characters need to interact).

Oh, and one thing I'd like to add is the idea of physics-based animation (not running simulations). Making software aware of boundaries and really being able to push your character against a surface without hand-keying everything would be nice... I'm not talking about IK. IE: Make it so that you can make an IK control and/or character's skeleton so that it cannot pass through defined walls/surfaces... and it will automatically nudge them as needed. Make it so that you don't have to "pretend" the object a character is interacting with has some kind of density to it.

jeric_synergy
10-07-2012, 08:20 AM
Is the skeleton deformation issue because the bones are in a mathematical hierarchy rather than a physical model simulation?

Advanced rigging w/constraints could address that, but you'd want good management tools so you'd only have to do it once, then just plop it into new characters.

ShadowMystic
10-07-2012, 01:42 PM
I'm fine with today's animation tools personally, so long as I always have access to a tool that doesn't require you to SELECT an item before manipulating it (IE: IKBooster). Having to select crap is such a huge waste of time and I don't understand why even top end software forces you to do this (unless I'm missing something)

It is the deform side of things that is the real problem; getting decent results means jumping over too many hurdles with today's software as far as I can tell. Multiple bones... relative movements... bone types... I'd love to just have a one-item solution for limb joints. I get that things get complicated around the shoulder especially for muscular characters... but I'd personally be happy with a simpler solution to getting good lower-limb-joint deforms. Sure, one could create a dense rig that takes care of these things... but that just means it makes things much slower if more than one character is being animated in a scene (this of course is a problem when 2 characters need to interact).

Oh, and one thing I'd like to add is the idea of physics-based animation (not running simulations). Making software aware of boundaries and really being able to push your character against a surface without hand-keying everything would be nice... I'm not talking about IK. IE: Make it so that you can make an IK control and/or character's skeleton so that it cannot pass through defined walls/surfaces... and it will automatically nudge them as needed. Make it so that you don't have to "pretend" the object a character is interacting with has some kind of density to it.

That's what we do as animators. If it was that easy anyone could do it.

AbstractTech3D
10-07-2012, 03:09 PM
Technically, boned joints articulate. So in the knee the tibia traverses the contour of the femoral condyles (the double large bumps at the end of the bone) as the leg bends. So I think the axis of rotation is not a constant.

GregMalick
10-07-2012, 09:37 PM
I'm having physical therapy for a frozen shoulder - and I gotta say that the shoulder is one of the most complicated joints in the body. It kinda floats around in this capsule of muscle and cartilage and has quite a dynamic range of motion (when it's working right). There is no way that a simple rotational joint can mimic all the movement one's shoulder actually does.

Just something to think about. I'm not a doctor - this is just what I've picked up during therapy.

geo_n
10-08-2012, 01:17 AM
Looks cool! The concept of character manipulation is similar to Eric Smiths maestro plugin for lw of active multipurpose hotspots.

http://www.stillwaterpictures.com/Maestrodemos/animate.mov
I guess they both don't like to select nulls and controls on the viewport as both mentioned. I still use maestro because I keep clicking other stuff in lw.:D This maya system is even more streamlined with that small panel.
Then he shows something that Genoma might have. Automatic creation of Ik/Fk and linking. Really looks like you can rig any character in minutes looking at his demo reel.
The guy doing the vid though is a master at skinning and mesh preparation. Video two shows a very typical method of skinning in AD software but his experience probably makes his characters deform better using only one skeleton rig. He doesn't use a deform rig at all that slows down viewport speed.
So with lw we have almost similar concepts, maestro and genoma. Its up to devs to take it further.

lwanmtr
10-08-2012, 02:49 AM
Setting up shoulders, elbows and knees to 'float' is possible. I toyed a little with it, when I was in school, but it gets too complicated (at least for me) to actually be usable quickly...had too many controllers in just the shoulder to upper arm joint.... and forget adding mocap data.

So for now, it works ok to use the traditional methods.

As for Bobakabob's comment about mass, thats more of an animation issue and not rigging. It seems they try to have a smooth transition from key to key and forget that people do have sharp movements also.

- - - Updated - - -

Setting up shoulders, elbows and knees to 'float' is possible. I toyed a little with it, when I was in school, but it gets too complicated (at least for me) to actually be usable quickly...had too many controllers in just the shoulder to upper arm joint.... and forget adding mocap data.

So for now, it works ok to use the traditional methods.

As for Bobakabob's comment about mass, thats more of an animation issue and not rigging. It seems they try to have a smooth transition from key to key and forget that people do have sharp movements also.

Aurel77
10-08-2012, 09:17 AM
Hi Everybody,

I'm Aurelien Charrier, the guy who made theses videos.
Don't hesitate to participate on the CGTalk initial discussion here : http://forums.cgsociety.org/showthread.php?f=2&t=1073828&page=1&pp=15
Your questions and opinions are welcome !

Thanks

Cordially

Aurelien

geo_n
10-08-2012, 10:22 AM
Get in contact with newtek. Some devs are lurking in the forums. They surely need fresh new ideas and innovative features in lw.
First though I would patent all ideas before revealing more. :D

toby
10-08-2012, 04:21 PM
Agree, also in many top end CG animation films, despite the sophisticated movements characters have an eerie weighlessness about them, there's no sense of mass e.g. Shrek. Not sure if it's being addressed.
Yep, Shrek is what I bring up when I see most of today's animation, like the recent Spiderman. Animation seems to be the only discipline that's not improving over time. What I think really needs to be addressed is the movie studios. Decent animators are quite capable of adding weight if the studios are willing to give them the time/budget, just look at the cg tiger in Life of Pi, it's incredibly real, and tougher to animate than humans.
http://www.youtube.com/watch?v=mZEZ35Fhvuc

I don't think the fact that most animators are trained in the Disney, cartoony/"better than real" style, helps much.

Titus
10-08-2012, 04:34 PM
I remember the time (late 90s) when Disney and other studios created custom software for traditional animators moving into 3D. For Chicken Little they had a couple of interesting tools:

"A tool called Shelf Control-akin to a Fisher-Price version of Maya’s Hypergraph-simplified the selection of character geometry and rigging. Instead of a node tree, the animators would see a flattened-out character, and by selecting the head, fingers, or other body parts, they could choose the corresponding joint on the model. Meanwhile, Disney’s Timing Chart replicated the charts that 2D animators are accustomed to using for their keyframes.

As a matter of fact, capturing the snappy, energetic feel of traditional 2D animation demanded a lot of keyframing. “There was almost a key on every frame; very little was left to the computer,” notes Ryan. “In contrast, on Dinosaur, we’d have a key at Frame 1 and a key at Frame 9, and let the spline do the rest.” So that the animators could define a character’s silhouette as precisely as they would in 2D, the artists applied numerous influence objects (which were more predictable) and added Maya’s sculpt deformers to the Chicken Wire deformers."

The animators had a tool to draw the silohuete with a Cintiq and the character aquired the corresponding pose, something I'm not seeing anymore.

http://www.cgw.com/Publications/CGW/2005/Volume-28-Issue-11-November-2005-/The-Skys-the-Limit.aspx

Aurel77
10-09-2012, 01:11 AM
Hi,

Those development directions are interesting. But, as on Titus post , i think we can't just exactly apply a workflow or a way to work without adapt it. If the animators still put a key on every frame, there are no interest.
The direction i try to work is a better blend of both worlds, a good fusion of the both ways of work, because both domains can bring something to the other.

tudor
10-09-2012, 02:47 AM
for adding weight etc I would like layered curves.
Do your main movement. then have a second curve that you can tweak without messing with the first one. Like an adjustment layer.
there are ways of setting this up now, but having it built in would be nice.

jeric_synergy
10-09-2012, 07:18 AM
for adding weight etc I would like layered curves.
Do your main movement. then have a second curve that you can tweak without messing with the first one. Like an adjustment layer.
there are ways of setting this up now, but having it built in would be nice.
I saw a MAYA demo that had exactly that. It looked extremely useful.

Built-in would be nice, but I'm almost positive we could accomplish the same thing (although w/o a nice UI) using DPKit nodes.

Aurel77
10-10-2012, 10:42 AM
The original software (the one shown in videos) did not work well with curves editing. I mean, it could have worked, but at that time we did not have time and people to make it work. So i had to make a choice and i choosed to keep the special keyframes manager and not the traditional curves editor. I don't at all regret my choice. Even sometimes it could have been useful, it proved we can animate without curves (i mean it's totally possible).
For the future, i would like both ;-)
And layered curves too.

jeric_synergy
10-10-2012, 10:27 PM
for adding weight etc I would like layered curves.
Do your main movement. then have a second curve that you can tweak without messing with the first one. Like an adjustment layer.
there are ways of setting this up now, but having it built in would be nice.
IIRC, the Maya way sorta (& I'm prolly totally wrong) duplicated the original curve, versus making a new one. Then the animator tweaked this second curve, which conceptually made it easy to have them diverge slightly, and maybe more importantly, converge after whatever got tweaked.. Since the whole point is TWEAKING, this makes perfect sense, PLUS makes it trivial to delete the secondary curves w/o starting over totally.

Then, no doubt, you could specify another curve which would modulate how much the secondary curve(s) affected the primary curve. Like a "balance" knob on an amplifier. Either 100% Primary or 100% secondary, or any mixture between.

Compared to SI and Maya, LW doesn't seem to do much with its curves. :devil:

jeric_synergy
10-10-2012, 10:32 PM
Animation seems to be the only discipline that's not improving over time.
All they have to do is quit putting animation houses out of business, and the discipline will accumulate.

tudor
10-11-2012, 02:57 AM
Animation does improve. It is however very time consuming and 'never' done.
You can always tweak more and more and get it better and better. Add more weight, secondary motion etc.

I need curves to animate. As animation is time consuming I have some tricks that I use when I need to get something decent out fast.
Basically the fastest way from pose to pose to final animation; offsetting curves.
Look at what feel you want. Drag or lead. Move curves for spine, head etc to start earlier/later than the rest of the items. It takes the blockiness out and gives you a more fluid pose to pose fast.