It’s hard to believe that over a month has gone by since I last posted. But a lot has been happening. For one thing, I finished the dog model, using balsa wood for the head. That’s really yesterday’s news though, because, as I’ll explain shortly, plans have changed. As much as I enjoyed building him, Fido here isn’t going to have a starring role in the movie anymore.
I’ve also created a tentative 12-day shooting schedule, and I’ve been talking to a professional director of photography who’s very interested in helping me on the project. Ivan’s got an impressive list of camera man credits for clients ranging from Coca-Cola to the Chemical Brothers, and he’s eager to work on Electric Town. The camera he shoots on is a RED Epic, which, in addition to being awesome in itself, will also give us a chance to fully explore the RED/Blender workflow. Didn’t know there was a RED/Blender workflow? There will be by the time Mango is finished!
As I mentioned, there has been a change in plans in terms of how this movie is going to be produced. In short, I’m going to be largely ditching the stop-motion aspect, at least as the primary means of special effects. Don’t get me wrong, I love stop motion. I love the physicality of building things for stop motion and the fact that you don’t have to sit at a computer the whole time to do it. And the results have a feeling to them that isn’t the same with computer-generated animation. But for this project, the goal is to make the movie and tell the story the best way I can. Given the variety of challenges I’d have to deal with to do a good job with a combination of stop-motion, puppetry, CG, and live-action compositing, and considering the absolutely amazing new features now integrated into Blender, I’ve decided to go all-CG for the special effects.
The biggest developments in Blender that has made it possible to consider this are camera and object tracking, and Cycles, the new unbiased renderer. These features already work incredibly well, and after the Mango project finishes off, they will be even more refined. I now have a clear idea of how all the effects necessary for Electric Town can be accomplished in Blender.
This simplifies things a lot. Matching lighting, for example, would be a nightmare if I had to use real lights in a (non-professional) stop-motion studio to composite into a natural-lighting live-action plate. On the other hand, using HDR image-based lighting in Cycles, it will be possible to get very good lighting matches in CG. Going with CG also gives me a lot more leeway in terms of character an prop design. I’m no longer stuck with shooting only what I can build in my basement.
This doesn’t mean that building the dog model was a waste. First of all, it was fun in itself, and now I’ve got a neat poseable robot dog model to use as a door stop. One of these days I might even animate him, just for kicks. For Electric Town, I think the model might actually prove to be very useful on set for blocking, lighting, and eyeline purposes, so it’s not a bad thing to have around. Plus, after all the work I did on the dog model itself, it was also trivial for me to build a nice dummy object for object tracking. When I bring this back to Japan I will paint it black and adorn it with brightly colored beads to make it easy to track. The actor will hold it (or in one case throw it) and the CG robot dog will be added in its place in post-production. The clip on the front is for a shot where the robot dog is grabbing the character’s clothing in its mouth.
Making the decision to stick with CG for the effects lets me focus my energy on the skills and techniques I’ll need to master to get the results I want. There’s still a lot I need to learn about HDR image-based lighting, motion tracking, and compositing, but at least I know what it is I need to do.
Of course, it’s also cheaper. All those bolts and washers were starting to add up. And the various decorative elements I wanted to add to the model would have pushed it into fairly serious money. With a model made of verts and polygons, the only cost is in render time.