270 Frames per Week

I've been a fan of rotoscoping ever since I walked into the Cafe at CMC six years ago and saw Waking Life playing on the big screen. For those of you not familiar with this or Richard Linklater's other roto'd film A Scanner Darkly, rotoscoping is a technique where each individual frame of a film is traced over by an artist. The result is an animation style that reflects the Screen shot 2013-01-26 at 1.25.05 PMmacro-movement appearance of real life but maintains a level of "kinetic chaos" on a micro level. An example from my favorite scene in Waking Life can be seen here. The original rotoscope was a film projector that projected an image onto a translucent drafting board. Today this is done digitally.

About a week ago I stumbled upon an iPad app called Brushes 3 which is by far the best freehand drawing app I've ever found for iOS. What made this app truly excellent was its ability to work with multiple transparent layers, import images directly from DropBox and export them as PNG files. What this means is that I can take a video clip, export it as a JPG sequence using Adobe Media Encoder, import all those images directly into the app, trace them and then export them back to DropBox (which syncs automatically to my computer). The app itself also has a massive range of brush options that can be saved as presets between which the user can change rapidly. All this to say that the way in which this app manages files and the artistic process allows the time spent dealing with each individual frame to be cut down dramatically. This is important since just a few seconds of film represents hundreds of individual frames.

So with this new editing method, I decided to try my hand at a rotoscoping project of my own. I shot some test footage and cut it together into a 22-second clip. Based on previous experience, I knew it would be important to include four specific elements in this test:

  • Hands
  • Faces
  • Straight lines
  • Movement synchronized with sound

Hands are incredibly difficult to draw because they are asymmetrically segmented and bend in strange ways when moving. The holy grail of an animated face would be to actually capture the subtle details that make a person recognizable so that would be a solid Screen shot 2013-01-26 at 2.24.09 PMchallenge. While rotoscoping creates a chaotic style of movement, I wanted to know the extent to which the eye would be distracted by curved or moving lines that it expected to be straight. Finally, I wanted to see the extent to which precise movements that the brain expects to be synched with specific sounds could be replicated in hand-drawn animation. With the test footage assembled I rendered out the footage as a JPG sequence and started to work.

Rather than take on a task that was too big too soon I decided that the final video would play back at 12 frames per second instead of the original 24 which cut in half the total number of frames I had to trace. I was still left with 270 images but that turned out to be a good introduction to the process. It took about five days with varying amounts of dedication to work my way through and in the end I was extremely satisfied both with my first attempt as well as the potential for utScreen shot 2013-01-26 at 2.27.07 PMilizing this technique in future projects.

In each scene of the test video I decided to pick certain elements that I would trace precisely and other that for the sake of time I would allow to be sloppy. The combination of varying levels of attention to detail fit together nicely since even my most dedicated efforts to trace the details of eyes, noses, mouths and ears resulted in what I call "rotoscoping drift" from frame to frame. Tracing hands exactly as they appear in an image also looks really unnatural. Without the fine details of skin shading, fingers look fat and misshapen unless you intentionally scoop out the spaces between knuckles and slightly exaggerate the sharpness of the joints. Finally, finger nails are a pain in the ass. I never figured them out and largely chose to neglect them by the end.

All in all this project was a ton of fun and the patience definitely paid off. When I imagine taking of the same task for, say, a 3-minute music video with upwards of two or three thousand frames, I wonder at what point the novelty will wear off?

Enjoy the video.

[youtube http://www.youtube.com/watch?v=_o-BEujtt7w?rel=0]