Tuesday, July 24, 2007

The Fool’s Errand, Part 2

Wow, 10 days since the last post. As you might have guessed I’ve been super busy, but I didn’t want to let this topic die just yet. In my first post on The Fool’s Errand topic I noted how it’s not all that feasible a goal to try and imitate the bigscreen results in your personal short film (especially if CG is your medium of choice). There’s been some great discussion in the comments, so let’s hope we can keep the conversation rolling. In this installment I put my cards on the table to demonstrate why I believe what I do about all this.


This is a test render from a short film that I had been working on for about 4 years in my spare time. This was an early “look of film” concept test, done sometime in the summer of 2004.


This is a rendered push in on the character in the set. Caveats: It’s rough, the animation is totally scratch (as in Definitely Not Final) and there are some render glitches (flickering shadow maps, etc.). All par for the course for an early test.


And from way back in early 2004 is a walk cycle test with the original version of the character design and rig. The reason it plays all stuttery is because it’s 12 fps. My early tests are often on 2’s to keep things light and simple.

Here are some other test renders as well as some boards from a sequence from early on in the short… (click on each to see bigger)

caveInterior.jpg controlRoom_ambOcc.jpg hallway_RedWhite.jpg

murrayRoom_Blue.jpg murrayRoomShelf.jpg contactSheet.jpg donut.jpg stalagmo.jpg iDot.jpg rocket.jpg

boards_01_sm.jpg boards_02_sm.jpg boards_03_sm.jpg

As you can plainly see this stuff nowhere near as detailed or lush as the Ratatouille scenes mentioned in my previous post. There are all kinds of style and design deicisions in here meant to keep the workload as manageable as possible. I consciously chose to work from simple sit-com type stage sets. Most sets don’t have a fully built 4th wall. In addition I was careful to not overbuild things that won’t be seen up close. I had no plans on building a set so completely detailed that I had total camera freedom without fear of showing the technical uglinesses that come for free in CG (model faceting, low texture resolutions, crawling shadow maps- all computationally and laboriously expensive to fix). The shots were already in mind when the sets were built. Additionally I consciously chose color and lighting schemes that were roughly monochromatic, thus reducing the demand for highly detailed textures to give objects their sense of meaning and presence.

On the technical side I borrowed some techniques I’d learned in the film biz. I developed lighting rigs and production tools that would allow me to quickly block in my lighting - all while automatically taking into consideration which set or location the character was in. That would speed up baseline lighting and allow me to focus on finer details. Render scripts were developed, again to automate a lot of the gruntwork of getting images out. Compositing scripts and recipes were also created, as were color correction recipes and widgets. Animation wise I’ve got my library of various tools that built up over the years to help me move quickly. A pose library tool was adapted to allow me to quickly block in broadstroke foundational poses for detail areas like fingers and faces, again allowing me to focus my time on working the details and finer subtleties rather than build everything from scratch.

In short, I used every trick or technique I’d ever learned in the biz about how to streamline production without sacrificing quality. I just built smaller, home versions for my own use. There wasn’t a single area of production where I was not leveraging my many years and many different kinds of studio experiences.

Yet even with all of these design and production shortcuts I was doomed. The amount of work was too much. I bogged down. Animation tests revealed that the character rig wasn’t going to hold up. Skinning fat characters that maintain good volume but are still flexibile and have the ability to be pushed to some extreme poses is perhaps one of the hardest rigging tasks in the business. This rig had 60+ pose deformers just on his hips and belly to maintain proper volume and it still was not enough. It was looking like a bag of grape jelly in even moderately extreme poses and motions. Seeing as feature film quality was the standard and The Incredibles had totally redefined what the bar was for this kind of stuff I had to completely rethink my approach. I started to see that a muscle & fat simulation solution was going to be the only thing that would make him look right. After some messing with Michael Comet’s cMuscle system (I still had my old beta testing copy lying around) I saw that I could do it, but it’d take me a year probably to get it right. It’s at that moment that I said Enough. I’d already rebuilt the guy 3 times over the previous two years. It was too much. And it’s not like I could throw him away and go with a skinny guy because so much of the short’s story was built around his bumbling, fat, clumsy and slovenly ways.

Aside from the rig problems, there was the sheer workload of the animation. The animation was going to be highly polished film level Cg work- something I have experience doing and can do pretty well. But that takes time. Even though I work fast the best I could hope to accomplish for my own film would be 2 or 3 seconds of final, full, CG feature film quality animation per week. I lacked sufficient man-days to do any more. To do a minute would take me 6 months of animation. To do the full 8 minutes for the story would take three and a half, maybe four years. Pre-production had already cost me almost four years. Re-working the rig for the fourth time would take another 8 months or more. I haven’t even mentioned lighting, FX, rendering, compositing, sound engineering, music, etc. In effect I had chosen a style of animation and design that required far too much labor.

I fell into the trap and the technique was the goal. I was trying to match the visual tones of a larger film production all on my own. And I wasn’t even trying to match that level of detail exactly. I did make plenty of compromises along the way. Yet even with all that I was still left with a project that was easily 10+ years from completion.

I know, I know. Shame on me. But understand- I could do the job. I could do it all well enough to make it work. I wasn’t on track to embarrass myself here. Yet I learned a very hard lesson in my youthful exuberance: Just because you can do a thing doesn’t mean you should.

In another post I’ll explain why I made this doomed choice in the first place.

No comments: