Animation Personal Trainer: Session 3!!

It’s that time again- time to gear up for another APT session. We recently wrapped up APT session #2 and it was another great session. The students did some amazing work and made some great progress. I’ll be posting a session 2 reel sometime next week so keep an eye out for that. Some of the stuff the students did was really impressive.

I wanted to get a jump on things and announce that I have scheduled APT Session #3 for July & August of 2007. Class will begin Monday July 9th and will run through the end of the first week of September. Unlike previous sessions I will open registration immediately rather than wait for a specific date to take enrollment. So if you’re interested in taking part in the next APT class, be sure to check out the updated APT FAQ and head on over to the kLango Online Store to register today.

And I have a nifty deal going, too! If you register before June 1st 2007 you’ll get a 10% early bird discount on the cost of the course (that’s almost $200 savings. Who says I’m not generous? heh).

Space- as always- is limited, so head on over to sign up for your turn at proven, personalized animation training the kLango way!

Wow, that was cheesy. heheh.

Some Misc. Eye Rig R&D

Awhile back someone emailed me about how I rigged the eyes for this character. (click to watch a movie of him being posed)
Anyhoo… I wrote back explaining a few things. Earlier this week I was going through my sent email looking for something else and I found my reply. I thought- Hmm. This looks interesting. Might make a nice little blog post. So here goes…
I’m just using “clam shell” eye lids. The lids aren’t modeled into the over face mesh but are just two half globe shells that rotate around the eyeball. This makes them easy to set up for a system like this. Plus I’m not a gifted modeler and it’s always hard for me to get the built-in lids to look/work well. For the rendering style I hope to use for this guy it doesn’t end up being a problem with the clam shell approach. The actual shells are nurbs revolves.

Each lid is then deformed by a series of clusters that work together like a folding fan along a hull of the NURB. Each cluster is centered at the eyeball center and each one is driven a certain percentage to create a smooth, fanned out deformation to shape the lid.


Then I put an FFD (free form defromer- aka: Lattice) around the whole eyeball/pupil/lid system. The idea here is that when the lattice is deformed the whole eye system shares that deformation. This way I can layer the clusters that shape the lids underneath the larger lattice that shapes the whole eyeball/lid complex.

Then the lattice is deformed with a wrap deformer. A wrap deformer in Maya is like using any geometry as a kind of lattice itself. In this case I’m using the face rig mesh to deform the lattice. So if the brows or cheek are deformed using other controllers this deformation then affects the eyeball/eyelid complex at the same time. So with one controller I can shape the brow and shape the eyeball shape as well and still have the lids rotate around the eyeball and be deformed themselves. This is how you get the eye to deform with the brows all at once.
Here you see the grey mesh has been deformed. This grey mesh then drives the grid lattice around the eyeball system which then deforms the eyeball and the lids all at once.
One caveat to a system like this is that it doesn’t play nice when it moves through space. To solve this I rig my facial system on a seperate copy of the rig that only works for the face. It’s still in the main rig file- just hidden from the animator. The animator controls the face with controllers on the main rig, but the output of the controllers affects the face rig. Then the deformation result of the face rig is piped back onto the main rig using a morph target blendShape. So animator controls main rig A for all body & face animation, but any facial controllers really affect face rig B and the result is then piped back onto main rig A. This is a fairly common approach in rigging these days- to use a seperate face rig that doesn’t move through space. It’s hard to do complex facial set ups right on the body deforming mesh. The key to having the facial output and the body rig play nice is to keep the face blendShape input come lower in the input list for the body mesh than the body deformation stuff.
OK, that was real geeky. Rigging is one of those black arts that makes CG so much more of a pain in the butt to do than other forms of animation. It’s a bit of a puzzle solver’s game that requires a person to be extremely creative but not necessarily highly artistic- if that makes any sense.
But for those who are interested in this kind of stuff, there you go.

Worshipping at the altar of innefficiency: part 2

Read the previous installment to catch up. Some good comment action going there as well.
I think there are other contributing factors that attribute to why current day animators are significantly slower (and arguably make poorer animation) than those of a bygone era.

First- The tools define how we do the job. In the infancy of Cg animation programmers invented new ways of doing things that weren’t even close to being based on the way animation had been done for decades. Language was borrowed and bastardized to mean something else. Techniques and tools had some loose association with old established methods, but in practice were completely different. As a result for a long time CG animators couldn’t see any parallels in how animation was made “back then” and now. To this day a huge number of CG animators can’t read a timing chart. It’s so foreign to them as to be unusable and irrelevant. But if you can read a timing chart then you have the ability to understand timing in a very practical and useful way- something which is of great value to all animators in any era regardless of medium. And that’s just one example. The result is that we have lost out on generations of “back then” know how. It kinda happened by default, but it was like there was this new tool and all the old ways don’t even apply. We could read the right books, but the practical “how to” parts seemed like a long lost craft- like wooden ship building or clovis point arrowhead making.

Some folks in the comments for the previous post mentioned rigs as being a problem. I used to think rigs were a major problem- and I do think the general craft of rigging is in need of fresh ideas and new solutions. But I don’t know just how much a rig has to do with animator speed. An experienced animator can make even a slow rig work for them. Rig limitations for obtaining a certain shape is one thing- but that’s more of a qualitative complaint, and a valid one at that. It does have a bearing on getting things qualitatively and that reflects itself quantitatively as well. But in my eyes the primary way a bad rig can slow down productivity is if it is too cumbersome in how it utilizes the computing power. Meaning- if the animator has to wait 10 or 15 seconds to see the puppet being moved into place on screen, then we have a valid speed issue. Thankfully most rigs aren’t quite that slow (but I’ve worked with some that were. Pure pain.). So unless your rigging folks are making a rig that is agonizingly slow to update, redraw and deform in the given buffer memory space for the frame you’re on I don’t think it has a lot of bearing. Now if you’re a timeline scrub monkey, well then the problem isn’t the rig. The problem is you and the way you animate. After all pencil animators can’t scrub in realtime either and they’ve been animating for a long time with good result. So I don’t buy timeline scrub speed of a rig as a major problem because I think a workflow that relies heavily on timeline scrubbing is a flawed workflow. Allegorically I liken it to a cabinet maker who instead of measuring twice and cutting once instead puts his saw away and decides to sand the piece of wood down to the proper length, constantly stopping to see if it fits yet.
The visual style of CG is generally pedantic. All those details, all that versimilitude to reality pushes our efforts into the little things that used to be exaggerated out. What old masters used to distill down into an illusion of life we spend lionizing in replicating life. This is true from design down to motion and then on to rendering. Smooth trumps strong nearly every time. The realism of Cg has always been a siren song and all that fine twiddling makes the entire process achingly slow.

It doesn’t help either that we fill out the ranks of our studios with younger animators who have never benefited from apprenticing under an experienced animator as an assistant. Economics dictates that we need to fill out a good chunk of the staff with young animators right out of school or with less than 2 years working experience. The quick rise to full film animator status adds a kind of legitimacy to the workflow of the young animator, thus cutting short many advances in their development. After all, why should we bother learning to do things a better way? We’ve got our job at (insert big name studio here) already- seems good enough for us. But there’s a danger in this- the stagnation of learning only cements bad methodologies and as time passes these weaker methodologies become the new defacto “right way”. Youth by nature has a dismissive hubris toward age and experience. Sometimes this is good. Often times it is bad. Old knowledge is devalued or lost. I am convinced that if I had spent 2 or 3 years apprenticing under an experienced traditional animator that I’d be much further along in my skill than I am today. But these opportunities don’t even exist in CG and they are increasingly scarce anywhere in the world now.

Then there’s the endless noodling of CG animation. The “easiness” of change in CG is a trap. Animators noodle. Supervisors noodle. Directors noodle. Producers noodle. Executives noodle. In an effort for everybody to take part in the decision process nobody is decisive- so the work constantly gets reworked. There does come a point where continual tweaking makes a scene worse- not better. Sometimes a thing really is better the first or second time instead of the 8th, 9th or 20th time. Of course for that to be true you do kinda have to know what you’re doing. And just as often after a scene is noodled it’s not really any better- it’s just different. When I was a supervisor I always asked myself The note that I want to give to this animator about this scene- will it really make the scene better/clearer/funnier/stronger? Or will it just make it different in a way that is more like the way I would have done it? The former is a valid supervisory note. The latter is ego stroking and frustrating to the animator. We have far too much of the latter in the business these days.

Whatever the cause, the reality is that it takes us a lot longer to get the job done than it used to take this other bygone generation of animators. But I don’t think the task is beyond us if the current marketplace was different. I think experienced professional animators of this CG generation could match the old 1938 numbers if we cut out the noodling and decided to use a more exaggerated style of motion. But even then fixing those two conditions isn’t enough if you really don’t know how to animate and you have to rely on brute force to get things looking good. Brute force doesn’t scale well.

Worshipping at the altar of innefficiency

Have you ever had your entire concept of the craft of animation shattered? About 3 or 4 years ago I did and it blew my mind. I’d been talking with some older hand drawn animators, listening to their stories, as well as reading some interviews and such. When I found out about how much footage old school Disney animators and Warner Bros animators used to do back in the golden age of animation and I just couldn’t believe what I heard. They were absolutely smoking what we do now- and they were making work that was magnitudes better looking than what we do now. This so ruptured my concept of what it means to do quality animation that I then went on to try and understand what it is that I had gotten so wrong all these years.

I got to thinking about this again when Hans Perk posted a 1938 Disney memo on his blog the other day. The memo details the footage accomplishments of various animators in the Disney studio at that time. The numbers are mind blowing. The average animator at the Disney studio in 1938 was producing 3.75 feet of animation per day, 18.75 feet per week or roughly 14 seconds per week.

I’ll let that settle in for a moment.

For many years I labored in CG animation under the illusion that if only I had a lower footage quota- then I could do top notch work. I recall wistful conversations with younger animators in lower tier studios about the low footage quotas of 4 seconds per week at places like Pixar, Blue Sky, Dreamworks, etc. It was like some holy grail for those of us laboring away in the 12-18 seconds per week world of direct to video. If only we didn’t have such cheap bosses then we too could make really high quality animation. Today I’m not convinced that this is necessarily true.

Of course footage quotas are always sliding depending on shot complexity, etc. But in general your typical feature film CG animation studio has an average per animator footage quota of roughly 4-6 seconds per week. 4 in the beginning of production, 6 towards the end. Some studios do more, some less, but these are the typical numbers for your “high end” feature shops. I was under the misguided notion for many years that if only I could have such luxury then I’d be a good animator.

I’m of the opinion now that I was wrong. How else can we explain that Ollie Johnson, considered a slow animator at Disney in 1938, did almost TWICE that? Are we making animation that is 2x better than Ollie Johnson did on Snow White or Pinocchio? I’m not convinced of of that. And what about guys like Norm Ferguson who kicked out a staggering 23 seconds per week? I read recently on Michael Sporn’s blog that during the 70’s on the Doonesbury film that animator Tissa David was doing 100 feet per week of nearly full animation. That’s over a minute per week!! What did they have in the water back then? Clearly there must be something else - or a combination of somethings- at work here.

I used to have a number of arguments to excuse away my own inefficiency and poor results. “Yeah, but those guys didn’t have to clean up their work. They had assistants do that.” True. They usually had just one assistant. So even if you cut the footage in half the numbers are still impressive by modern film standards. That means Norm Ferguson and his assistant combined were doing 12 seconds per week of final polished animation on Snow White. Frank Thomas and his assistant were doing 10 seconds per week of final polished animation combined. Milt Kahl & assistant over 7 seconds per week. Same for Art Babbit, Grim Natwick and Ward Kimball. But I think perhaps even this is too simplistic a point. The assistants drew everything based on the keys, breakdowns, extreme drawings and timing charts invented by these animators. Meaning the animators defined every single creative aspect of the performance and the assistant drew the inbetweens and did clean ups based on those decisions. Every single creative performance decision - all created, drawn, noted and communicated to another person- without the benefit of instant video feedback at the push of a button! Think of it. How many of us could do 10 seconds per week of final animation at a level that Frank Thomas did in 1938 without making 20 playblast previews a day?

“Yeah, but these guys are legends. They’re not like us. Give us 40 years in the business and I bet we could do the same.”. Not quite. The footage numbers on Hans’ blog are from 1938. Long, long, long before these guys were the 9 Old Men or whatever. They were young guys. Probably not a gray hair in the entire crew except for maybe Grim Natwick. They even had the added burden of inventing the entire medium while hitting these numbers.

“Well, they were animating on 2’s. They didn’t need to clean up 24 drawings per second like we do.” Kinda. Often they worked on 2’s, but in full feature animation back then more often than not they worked in a mixed 1’s and 2’s format. A lot of CG animators wouldn’t even begin to know how to handle that. I know I didn’t until I took the time to learn how to do it- from an old school animator, of course. Mixed timing in many respects is more complex than fixed timing on 1’s.

One by one I learned that my excuses were just that. Sooner or later I had to face the fact that the reason I was slower and made poorer results was that I really did not know how to animate. I could make fine enough animation results, get work. I had an eye and a talent for motion and performance, but that’s a different thing altogether than knowing how to animate. It’s the difference between being a food critic and being Emeril Legasse. The critic knows food and definitely knows what he’s talking about- and if given the opportunity probably could plunk about in the kitchen and whip up a nice meal from a recipe in an hour or so. But the chef is the one who could make the dish from scratch in under 20 minutes and give you 10 variations on the theme. Both know food- but only one really knows how to cook. In the same way I had to admit that I knew animation, but I didn’t know how to animate.
More to come…

Why do you seek the living among the dead?

Today marks the great celebration of life over death. Today marks the day that Jesus Christ- the Son of God- walked out of a tomb after his body lay cold, lifeless and without a pulse for 3 days. Today marks the day when a way for us to get back to God was sealed because the seal on a gravestone was broken. For two millenia now men and women the world over have gone to their graves full of hope- often with great pain and sacrifice- because they believed the same thing. For two millenia now unbelieving men and women have mocked and scoffed at those who believe the resurrection of Christ is true. For centuries men have sought the body of Christ in vain. Technology has not made this search any more fruitful. It will never do so. The reason why is simple: You cannot find a living body among the dead.

So to everyone who believes- and even those who think all of this is foolishness- I wish you a Happy Easter!!

Metapost: VTS20 now available in Back Issue Store

I don’t hype my back issue store here a lot. I dunno why- guess I’m not the world’s smartest self promoter. Anyhooo, I do have an online store where non subscribers to my Video Tutorial Service can download individual videos a la carte. One hitch is that non subscribers have to wait 6 months longer than subscribers to get access to videos. It’s kinda an incentive to make my awesome subscribers feel like they’re special- which they are!

Anyhow, at the beginning of each month I make a new back issue VTS video available for public purchase. This month I’m releasing VTS20 (subscribers are currently on VTS26). The topic is Blocking & Breakdowns Part III. I think the whole Blocking & breakdowns discussion is some of my better work in the VTS if I may say so myself. You can check it out in the back issue section of my online store.

Toon Snap Tutorial now in Korean

Special thanks to Choi Kyu-bo for his efforts in translating my older Toon Snap quick tip tutorial in Korean. So for you folks who like to read animation tutorials in Korean- and we know there are many of you - then enjoy this new translation.