A number of years ago I wrote my Pose to Pose- Organized Keyframe tutorial. In it I advocate a method of acheiving overlap and drag on objects that- at the time- I thought yielded decent results in a quick and easy fashion. In the ensuing years I have come to change my opinion about the merits of such a workflow. In this follow up article I hope to point out the problems with my earlier ideas and propose a better solution to the challenge of creating great overlapping action and drag in your animation. So think of this as a self repudiation of my earlier ideas. As I grow as an animator I feel it's my responsibility to correct my own mistakes. So follow along as I trash my own theories!


The Problem:
In my earlier tutorial I wrote....

So for offsets I'll shuffle my head from my spine and my spine from my hips to have the torso of the body sorta flow into a pose. Depending if I want the motion to lead with the hips or the head will determine which way I'll offset my keys. Sometimes I'll shuffle some keys and things will look awkward for a certain pose or transition. That's OK, I can shuffle them back for that spot. It's a cheap way to get rid of that robotic feel of pose to pose. I'll offset the hand's rotation to occur a frame after the hand hits in place. ...

The article has served many people well, and many working animators today tell me they are thankful that I wrote it at a time when they needed to know how to get things done. So for this I am grateful. However the original article was not meant to be the end of the discussion, and as such I have come to understand that there's a problem with this approach. The problem is that this method of creating overlap and drag introduces too many motion problems to be used very effectively without requiring significantly more time to clean up later. It was originally posited as a "quicker" way to do offsets, but time and exprience has taught me that in fact the opposite is quite true. Namely, working this way requires more time to clean up, not less. Let's explore why.


Example 1: Shuffle key method
Let's start with a simple bit of blocked animation. A character has two key poses with two breakdown keys in between to help define the arc of the motion and some of the favoring. Here is the dopesheet representing the motion....


And here is the motion (for the sake of clarity we are leaving the curve tangents in linear):

no overlap (no overlaps)

Indeed, it's mechanical and in need of some help. It has some nice breakdown ideas, but you'll notcie that all the body parts leave and arrive at the same frame. We want to build some more overlap in here to lossen things up a bit. Using the method I outlined above I would shuffle the various keys around to create overlap. Here's what the dopesheet would look like.

shuffle dope

So far we're right on course with my earlier tutorial. But let's look at the motion...

shuffled movie (overlaps done via key shuffling)

Uhh...Kinda junky. I mean, it's hard to say that it's any better than the flat linear version before. Some things are indeed offset, but there are problems with the motion. It needs more work to smooth out the little hitches and pops and junky bits of noise in the movement. What went wrong? In the old days I'd just dive into the f-curve editor and start scrubbing the timeline and the tweaking the curves to try and smooth this stuff out. Eventually I'd get things working OK. However after a few years of this I realized that I was spinning my wheels a bit. My work took a lot of clean up and the results weren't as strong as I wanted them to be. That and my wrists hurt from all the scrubbing and tangent handle tweaking in the curve editor. The low, dark stink of fear crept in as I realized that I was well on my way to carpal tunnel syndrome if I didn't fix things. So like other times in my career when I came across a problem I decided that there HAD to be a better wasy to do this (that's the impetus behind all the tutorials on my site in the first place. I am driven to find a better way to do this stuff- and then share it. :o)

Anyhow, like other times before, when I find myself struggling with a methodology in CG I stop and ask the simple question that I always ask:

How did those old school traditional hand-drawn animators handle this?

They didn't have little magic widgets to push around on a computer. So how did they manage this? How indeed.


Old Becomes New Again

As you may have guessed, hand drawn animation has no magical little black squares to slide around on a computer screen to get the feeling of drag or overlap when their drawings are projected at 24fps on a big screen. Hand drawn animation didn't have timelines. No f-curves to tweak. No free inbetweens to look at. All they had was their brains, some paper and a pencil. So how did they get overlap into their work?

The same way they got everything into their animation. They drew it.

Yes, absurd as it sounds, I propose that you can solve this problem by making better drawings. Ahh but the cries rise up: "This is CG, you idiot! There aren't any drawings!"

So says you, my quick tempered friend. Yet look at your computer screen when you're working in Maya or XSI or whatever it is you use. What do you see? Right there, sitting in front of your eyes? It's a picture, no? A picture of a single moment in time. A single frame. A... drawing? Yes, a drawing. Not drawn by hand, but drawn by the computer. Yes there's lots of wizardry and magic behind all those glorious pixels, but the basic essence of what lies before you on screen is this: the computer is representing a drawing of some mathematical construct we call a "character". It will faithfully draw this image according to your exact specifications. Meaning this: if you want the result to look different than it does right now, that's YOUR responsibility. It's just a dumb computer. Ultimately if you are a CG animator you are manipulating a clumsy, mechanical pencil to create drawings. These drawings then are projected at 24 frames per second onto a screen to give the illusion of life. Hey, just like hand drawn animation!

So, what does this mean for us now? I think we threw out the baby with the peg bars. We need to think of these things as drawings, not magical pixel fairies that can be mushed about with spaghetti strings of math.Ok, back to our original discussion: How do we create good overlaps and drags in our animation if we don't use the above mentioned "keyframe shuffle" approach? You know the answer already. By making better drawings- drawings that put the overlap right into the image. Let's explore...


An Example: Drawn Overlap
OK, let's take the same scene we have above. Two poses, two breakdowns, no offsets. To remind us, here's the dopesheet image again...


Good. Everything lined up like ducks in a row. And again as a reminder, here's what that looks like moving, sans overlap in the drawings....

no OL (no overlap in motion)

So we've re-established that this isn't so good looking. Now, let's get down to business.
First, we need to THINK. Think, what do we want to lead? What do we want to stay behind? Once you figure that out based on your mad animashun sKiLLz you're ready to start editing your drawings. First, let's look at pose 1...

pose 1

Nothing special, but that's OK. Now, before I get too far into this, let me say here that this method will most likely require you to create additional "drawings". That is, you're gonna have to do a little bit more work here. "More work?! I thought you said this was better?! You suck, Lango!". Ahh yes. At first that sounds like a total raw deal. But remember the previous method? Look at that shuffled offset dopesheet again. Go ahead, I'll wait. You see all those offset and shuffled keyframes? You have to clean all those, and you have to get the motion all smoothed out, too. So it's not like there's no work in the other way. The difference, and this is key, is where you spend the time.
Think about it: How long does it take you to tweak the position of a character's body in the viewport? The feedback is pretty immediate. No need to make a preview. Just grab the controller and look at what you're doing. Fast, simple, direct. Kinda like drawing, huh? OK, now how long does it take to clean up a bunch of abstract offset shuffles that make goofy funky hitchy motion? Can you quickly and easily tweak something and just look at it to know that it's right? What about tweaking curve tangents? At the very least you're looking at repeated preview renders and lots of twiddling of things in the Spaghetti Box TM (that's my new name for any curve editor in CG). So yes, you will make a few more drawings this way (keyframes in CG speak), but in the end you'll spend a fraction of the time to clean up the motion later. Don't believe me? Follow along and you'll see why.

Now let's look at the scene again. Here's the first breakdown (this was in the original "blocking" that we started with)...

11st breakdown

This was achieved just by editing the one that was already there. No new keys were added.

OK, now the second breakdown drawing.

2nd breakdown

Again this was in the original, all we're doing is changing the position of the character's body parts here. No new keys being added.

Do you see what I'm doing there? But now we want to create some extra drag at the front of the move and at the end, but I also want to build some ease while I'm at it. Let's kill two birds with one stone. So I decide that I need a new drawing at the begining fo the move before the previous breakdown drawing we originally had. Just a little something to help define the ease and the overlaps a bit. Here's what I do for this new breakdown drawing....

first ease BD

And much like the beginning of the move, I want to better define things at the end of the move. More ease, more overlap solutions. So here's what I did for the last breakdown drawing.

last ease

Like the first this is a new one that I decided to add to again help sell the overlap and the eases. And again, the last pose, which I didn't touch at all.

pose 2
Nothing too complex. Have a look at our dopsheet to see what this looks like under the hood.

drawing dope

Simple, clean. Nothing crazy here. It has two extra keys, but I showed why I wanted them. Yes, the extra drawings were more work, but they're the right kind of work. I tried to explain all the thinking in each JPEG so I hope you read the words and didn't just look at the pretty pictures.

So, now, here comes the test. How does it look when we let this thing play? Well, look for yourself...

drawn OL

Hmm. Not bad. But is it better? OK, side by side.


Alright, now you gotta admit that the "drawn" approach has given us better results than the shuffled approach. we have good offset, we got some eases out of it as well. You can just tell by looking at it that it will be much easier to polish out than the other one. See how the right hand pops out and in again in a very awkward motion path in the shuffled one? See how the head has a nasty hitch in its arc coming up? And notice how stiff the hips feel? The left hand is also a bit of a mess. I mean, it's better than not having any overlaps or drag at all, but this thing clearly needs more work. And I didn't stack the deck here. I did the shuffled method the way I outline it in my old tutorial. Which one of these two versions of the motion would you rather clean up? Which one do you think will be easier to get nailed down smooth and groovy? Yeah, I thought so.



One last caveat. This way of working does require a little something extra. Namely, it requires that we actually know what we're doing. I can already hear the cries of lament: "Hey, no fair! You suck, Lango!" Yes, sad as it may be, we have to grow up and learn how to animate if we want to work this way. But that's a good thing. I mean, think about it. This is the only way you could do this for over 80 years. The only way. Generations of animators have come before us. They learned this. They eventually learned to animate. Let's not be content to be merely talented discoverers of happy accidents, but really, truly know how to animate. And hey, if you disgaree with me, that's cool. But we are a new generation of animators. Are we serious about knowing what we're doing or are we going to be prisoners to a technological crutch? If the electricity went out, would we still know how to animate? It's a fair and honest question for this generation of animators. If we're smart we'll pay attention to those who have gone before, even if they went with only a pencil and a stack of paper.