A recent project has had me exploring the craft of rotoscoping, and I thought I might share my thoughts on this oft maligned technique.
Some of these sequences work better than others, but on the whole, for myself, I find them pleasant to look at. Yet this alone is a compliment rarely made towards rotoscope, which in animation circles is considered a fancy word for tracing, or cheating, and synonymous with ugliness.
Rotoscoping , according to Wikipedia, “is an animation technique in which animators trace over live-action film movement, frame by frame, for use in animated films” – which highlights the problem most people have with traditional rotoscope right there – “frame by frame”. What is a length of film or video? A moment captured in time? Nothing of the sort. A moment exists infinitely. Technology might be used to record that moment at 25 frames per second, or a thousand frames per second, but life itself has no frame rate. So a length of film in that sense, is but a sequence or arbitrary, semi-regular instances, captured over a finite time.
In animation, we capture a moment more selectively. The gesture of an arm, say, is portrayed by the artist selecting what they see as the most important or effective instances, needed to deliver the meaning or focus of that gesture. This act of artistic interpretation, impressed upon the canvass of time, is what allows us to make judgments about animation. We can have favorite arm gestures, arm gestures that pass without meaning, or gestures that throw us out of a films reality, precisely thanks to that process of selection. This is why we dislike tracing.
Tracing, or rotoscoping, binds us to the artless selection process of the camera shutter, a selection which includes not only the desired focal action, but the inconsequential, involuntary movements too. This is not a problem for live action. In live action, if the focus of a shot is supposed to be the kick of a leg, then the behavior of the opposing leg, or of the neck or of the nose, are all part of the flow. The problem for the rotoscoper is that these involuntary motions are not captured in their entirety, but again in random static instances. A blink, or a baring of teeth might exist for but one frame, and the rotoscopers choice is to either follow this ‘reality’ dogmatically or selectively ignore it in pursuit of a more streamlined animation. To put it in laymen terms, think of the last nip-slip you scrutinized, or 9-11 video you watched frame by frame. We all know what the Zapruder film looks like – as a sequence of images it’s completely abstract.
This is what forced me to my own conclusions about the ‘correct’ way to rotoscope. The animations I’ve created are deliberately free from the ‘true’ line of the source material, and are deliberately low in frame count. I’m fortunate that the project allowed for this and that I wasn’t forced to clean them up further, since to do so would have been to murder the art. Most of these sequences were done on threes (holding each image for three frames, running at 25 per second) which I honestly feel is the optimum frame rate for rotoscope, in any situation.
An inoccuous sounding statement, but one that opens a can of worms. Working on ones and twos (changing every frame, or every other frame respectively) is what we call ‘full animation’. It is the stock in trade of the Occidental animator and tied almost inextricably to how our culture has decided animation ‘should be’. Taking our lead from America, the west as a whole has succumbed to the dogma that persistence of vision, is integral to accepting sequential drawn imagery as ‘real’ – that to drop to a frame rate where we can register each image, is to dislocate the audience. Yet ‘full animated’ rotoscope is broadly accepted as dislocating the audience in precisely this way. Are we to conclude that certain illustration styles are simply unsuited to animation? That seems terribly limiting.
By proposing that rotoscope works best on threes, I place it within the remit of ‘full-limited’, a wholly other approach to animation favored by the Japanese. Full-limited proposes that animation can still thrive within the very limits of persistence of vision, by engaging the viewer with the chance to enjoy each image. That to see the drawing for what it is and to understand it as conveying a reality in the same instance, is our natural way of seeing. In short, they credit us with the ability to multitask.
Full-limited allows animation to be what it should be – drawings that move, and allows that description be applied to any art style. This was good for my purposes, since my other main conclusion through this, is that where rotoscope is concerned, there’s a point at which you should just stop drawing. I entirely believe that very detailed drawing styles are appropriate to animation – just not rotoscope animation. It’s the combination of unselective motion and unselective design that makes so much roto so ugly. For my own part, I made efforts to leave out anything that wasn’t servicing the movement. Knowing that each animations relationship to its video source, was going to be disarmingly close no matter how loose I played it, I purposely left off noses or reduced whole bodies to silhouettes where the focus was supposed to be on the face.
The other golden rule I found, was that the trick is to illustrate over the image, rather than trace it. The distinction being, that using the video as a guide to where a hand aught be positioned in space, I would then draw a hand in that place, rather than slavishly scrawl where the fleshy pixels ended. I also found it helpful to watch each clip and then decide on an appropriate character design to depict it. In many cases I would use the real face as no more than a guide for rotation, like the old cross-hairs method, and apply an original animation to that, perhaps only using the video for key frames. Emotion tracking, if you will.
Anyway, that’s what I learnt this week. Hope you enjoy my little experiments, and that I managed to get all the way through that, without mentioning A-Ha once.