While working on my documentary segments, I have explored a technique of rotoscoping that allows me mix unrelated pieces of footage together into a new, cohesive scene, and can also let me single-handedly build a shot that starts in orbit and ends on the sidewalk in Washington D.C.
You can see what I mean in my demo reel to the right.
The first example with the map was created by taking screenshots in google earth, essentially using the entire planet as a virtual stop-motion prop. Take a screenshot, turn the Earth slightly, take a screenshot, grab, turn the planet, screenshot, grab, drag, etc.
Once I have enough frames to move across the continent at a slow enough pace, I import them all into flash onto the timeline, and trace them over one by one, adding the graphics needed for the shot.
I could have used a single map image and just moved around that, but tracing over a few hundred frames provides a handful of benefits. Firstly, it conveys the slight curvature of the Earth. Secondly, there’s a pleasing squiggle effect. Everyone loves squiggly rotoscopes. Thirdly, all that tracing gives me some needed time drawing on the tablet. I’m still haven’t quite gotten the hang of drawing from a tablet, having my hand and my line separated by such distance of cords and programs. I still do my character animation on paper and scan the drawings to trace in photoshop. Flash is great for rotoscoping, though.
The final benefit is being able to seamlessly transition to a crazy falling from orbit shot.
Tracing over Google Earth basically lets me do any aerial or landscape/cityscape shot I could think of, on location.
Tracing over Youtube lets me take two people doing two separate things and make them part of the same scene.
Here’s a shot I’m working on involving a guy sitting at an interview and telling a story about being harassed at a baseball game for not standing during “God Bless America” out of protest.
What I can do with rotoscoping is take the footage of him sitting in his interview chair talking, and move him to a seat in a stadium.
Next, I looked on youtube for videos of drunk, belligerent fans at baseball games. By tracing over them, I can have the interviewee recount the story as it happens around him. Here’s a mockup of what I was aiming for when I started out.
The best part of this technique is that I can modify existing footage to take it out of context. I found this clip of a cop chewing out some skater kids.
(fun tip: use the downloadhelper extension for firefox to download youtube videos: http://www.downloadhelper.net/ You can convert them to anything, but I’m importing to flash so the raw flv file is actually what I need)
Putting it together, I remembered how it was working on the rotoscopes of the people dancing in my reel. Those shots were also lifted from youtube videos. When you zoom into a youtube to try and trace a person, this is what you get.
With the HD footage of the interview, I was able to get really clean tracings to capture his likeness and body language, but the youtubes from cell phone cameras are much more difficult. The trick is to treat the youtube video like it was a rough animation drawing, and I’m cleaning it up, getting it “on model”.
Notice how I get rid of the police uniform to turn him into an angry fan. The vague form, the gesture, becomes the main point of the youtube. What I’m doing is like halfway between animating and rotoscoping. I’m always going back and forth between my drawings to make sure they’re matching up. I’m much less concerned about tracing whats in the video. It becomes a whole new performance.
Now I can take these frames and copy them over to the scene with my rotoscope of the interview. I haven’t colored it yet, so here you see it without the seats in the background.
So what I’ve been able to do is take a guy sitting in a chair telling a story, and a video of a cop yelling at some kids, and build a recounted scene at a baseball stadium using only youtube and flash.
I’ll post the footage here once it’s complete.