2011-08-18

Spacing analysis: a calligraphic path of action

I recently took the time to purchase a copy of Cats Don't Dance online. It's a bit hard to come by, a lovely yet all-too-obscure work.
The animation in this movie is excellent and well worth studying. The storyline is fairly simple (though pleasant, with a nice soundtrack), but the animation artistry is great; it is easy to understand how it managed to win Best Animated Feature Annie in 1997.
So here's another spacing analysis, my third so far.
First the clip, a shot about 100 frames long:
Then my Photoshop magic, to study the spacing:
It's tricky to find a good clip to study this way, though it's definitely one of those "when you see it you know it" kinds of things. This one was quite interesting. I only managed to trace the head and the tail; the hips are doing their own thing inbetween and the feet are well worth a flipthrough to take a look at on their own. Each of these has its own path of action happening that, when traced out, is calligraphic in compexity, seemingly all over the place yet artistically unified. Indeed, the thing that happens at the top of the ladder could very well be taken as an easter egg, someone's initials; but that's pure speculation.
It's all highly syncopated timing-spacing-wise, which fits to the syncopated accompanying sound track, as well as to a general overall animation timing ideal that closely correlates with the same musical ideal. Perceptually, pleasing and interesting. In spacing, note the corners in path as well as the loops; similarly, in timing, there are offsets of holds and the opposite extreme of smear frames. As I said, calligraphic is an apt description. I've wondered about possibly setting animation to the path of a calligraphic pen before, copperpoint style; this is an excellent correlation to that concept, the best I've seen to-date.
This sort of thing is not easy to do.
I thought I'd also include an examination of some of the more interesting smear frames in the sequence:
The first, on the left, is an "impossible" step-up-the-ladder that, in a single step, gets Danny to the fifth rung up the ladder in as many frames, while facing backwards and looking nonchalant. A lovely bit of nonsense that somehow seems perfectly natural in the context of the shot.
The two most extreme, on the right, flip his body through a pretzel-path.
As befitting of a smooth operating-cat.
Smear frames remain one of the greatest points of contention for me with regards to 3D animation. A good rig will be able to handle a degree of the distortions served up by an extreme cartoon style by permitting some stretching or even bending of parts and extremities, but one still tends to run into a brick wall at some point, and I'm not sure whether rendered motion blur could even be capable of pulling off the pretzel example hilighted above (one would need presumably subframe sampling and subframe keys to do it: i.e. a lot of extra work combined with very slow rendering: not an efficient solution). A good smear frame will eliminate strobing and make extreme fast action seem plausible (explicable in part as the mimicking persistence of vision and in part as the emulation of camera motion blur).


If I can manage to identify the animator who did this shot, I'll be sure to update this posting to give the name. (If you happen know, please add a comment with a link for verification) Excellent work!

2011-08-04

Target-Based Animation

(Discussing the integration of pencil testing in the 3D animation workflow)

I've been including pencil testing in my workflow for the first time during this class, and have been experimenting with it and thinking a lot about the utility of including such methods in an animation workflow.
Developing an efficient workflow is a big challenge in animation. Animation is arguably the least efficient artform (or, conversely, the most involved artform), so finding ways of accomplishing each step of the process in an efficient manner is an important aspect of making animation doable in the first place.
So it would seem that >adding< a step to the process is kind of like shooting yourself in the foot. There are only so many hours to the day already.
A revolutionary feat of 3d animation is that the computer does all the drawing for you. So you don't have to do any drawing yourself, right? Well, yes and no. The thing is, drawing still remains the >most efficient< way to develop and capture visual ideas, and it can also efficiently develop and capture many abstract and even literary ideas as well. So it's important to never be too far removed from drawing when developing ideas (of any sort). Throwing drawing out the window would be a very grave mistake.
I've been pondering a theme in this context: Target-Based Animation.
A lot of animation is "already" target-based in the sense of being built upon reference video that will be shot or sourced to guide the process. In this case, the "idea" is already well-into-play: you'd usually want to build from video of similar angle, similar timing, posing, overall feel and so forth.
One can most certainly work directly from this material. It can be a primary guide.
However, you see the effect of this. If no additional thought is put into the matter, you get an animation sequence that looks like reality. Which, for animation, is often problematic: because reality is "soft". When animating on the basis of real-world timing, spacing and posing the results tend to look slow and will also typically not be very interesting to watch.
So you want to push beyond reality to get better development and refinement of your visual and motion ideas.
We can push the ideas farther in our poses, this is true. But THIS is where it gets a bit fishy. Why? Well, "pushing your poses" amounts to >dabbling<. What you get into is iterating. You try this, you test it. You try something else, you test it. You iterate, and that's how you get somewhere with "pushing your poses" and "pushing your timing". There are a few concerns, three KEY problems that I can readily identify with this sort of a workflow:
1) it's time consuming.
2) you never know how or when you're going to arrive at your creative destination. This is an issue in terms of workflow
3) probably worst of all: you may >never< reach a meaningful destination.
Why not? Because you don't have a TARGET (whether clearly or even roughly) that you're working toward. The only clear target you have is your reference video. It's quite likely you'll end up having something that retains a strong tie to that reference. As for the "pushing" that you've added to that: that's a wildcard. So of course it may work (I'm not at all saying this can't work, I'm just saying that it can be problematic and is less than ideal), depending on your sensibilities. But it may also look distinctly like "ref video plus some aimless dabbling", especially if you're tired or having a bad day.
Right. So here's what I'm getting at: drawing >IS< the most efficient way to develop and capture creative ideas. So if you draw through your sequence, you'll be doing this. THIS is the TARGET that I'm referring to here. The ref video takes you halfway, and dabbling around can take you further, but if you have this drawn-out target, the in-cg work is less about "dabbling" and more about working toward something you have already established. This is significant.
So, in short: including pencil testing in your worflow introduces the >setting of visual targets< into the planning phases of the workflow. Planning is where all the most significant decisions are made, so targets are then set before anything starts getting "locked down" in the "actual" scene.
Pencil testing does take more time. It still may not be necessary for a lot of animation situations, particularly as one's experience builds. But certainly when working through any sort of difficult segment, it's a tool of first choice that I know I will continue to turn to.
So, here are some examples to illustrate my point:

Here is a segment of a body mechanics shot I'm working on, showing pencil test planning first, followed by a playblast of the work in progress.
It's worth pointing out that pencil testing can be done at ANY stage. It isn't at all limited to planning phases. Our AM mentor feedback often includes drawovers on our animation. Any half-decent 2d animation application will handle importing an existing playblast / sequence / ref video / whatever and allow drawover work to further develop ideas. So a pencil test can be brought into the loop at any time it's needed. In turn, you'd want to rotoscope the pencil test back into the cg environment, lining things up exactly with your existing camera angle. This is where the method has an obvious advantage over simply drawing on paper: in a pencil test you're developing >actual< timing and spacing for your shot - something you can't easily do on paper. You like to do your blocking first? Go right ahead. Block it out, THEN do a pencil test over that. That can also work. Though I would point out that this can potentially impede the pencil test process if you allow the less authoritative blocking to overly influence the potentially more authoritative drawing (the latter being nearer the realm of ideas and ideals, more immediate, more guestural, and unhampered by all the complicated limitations of what a rig is readily capable of doing.)

It's a challenging shot. The tennis player is to be sleek, quick, an expert at the game. If I were to film myself as ref for this kind of thing the shot obviously wouldn't be "headed" in that direction, so I have sourced a bunch of pro tennis reference to look at and work with: one typically does well to work from the best available authorities. But this material only goes so far. So resolving difficult spots via pencil testing was most certainly useful. I was able to nail down timing, path of action, and the extent to which I would be distorting the body in advance. You can see that I have STILL done things very differently in the actual shot. However, the pencil test has played a big role toward getting it right.
The side-by-side image is to also illustrate: if I were to have built my poses directly on reference video, without this kind of planning, I seriously doubt I would ever have arrived at such extremes of posing as you see here. I would rather have been likely to treat the rig realistically as per the realistic target in use.
It's also important to note that this kind of distortion is only relevant in certain situations, i.e. extreme action. I "went all rubbery" with this shot, and the shot I'm working on after it is a complete contrast - so I have to also be careful to give the upcoming "Nervous Beginnings" shot a completely different treatment, accordingly.