Johnny’s Workshop

June 15, 2010

I ain’t got time to render.

This little experiment left me convinced that pure CG was not the way to go; not when you’re a few hundred CPUs short of a render farm, anyway. I guess that line could be used as a clever insult in the VFX industry. “I’m afraid Billy’s a few CPUs short of a render farm.”

Of course, in the games industry, if you worked as a level designer and you weren’t very good at your job, your co-workers would say, “That guy’s a few destructible wooden crates short of a level.”

Pretty clever, huh?

I began this experiment toward the end of March last year thinking I could do with some practice before having a proper stab at that “UFOs at the You Yangs” project. The latter was going to be entirely CG and involve a detailed ecosystem, clever lighting, and some kind of fancy atmosphere with clouds, haze, etc., so it made sense to try a simpler version beforehand. The following elements I considered essential:

  • A rural setting.
  • Stereotypical flying saucer.
  • Realistic lighting.
  • Handheld camera movement.

The other requirement was that I use Vue 7 xStream and nothing else. Ironically, I went into the project hoping to get a lot better with Vue and emerged from it convinced that studying Vue was a waste of my precious time (I think that qualifies as irony). Not because Vue is bad software—it’s really not; it’s awesome … awesome to the max—but because the render times are usually horrific. I actually spent a couple of weeks studying and experimenting with all the render settings. Achieving a high-quality result was easy enough; it just meant you had to wait forever.

Take this 7-second video, for example:

There’s only one model in the entire scene, yet it took ELEVEN WEEKS to render. Okay, that’s not true; it actually took 8 hours, 15 minutes but that’s still way too long. I mean, it’s not even a good video; it’s just an F35B from Battlefield 2 with some dark, menacing clouds in the background. Oh, and the resolution was 800×600. Imagine trying to render a high definition (1280 x 720 or 1920 x 1080) shot of a forest with a running stream or something. You’d die of old age before you had the first two seconds. Even if you lived long enough to see those seconds, you’d probably discover they were riddled with inexplicable artifacts.

Like these:

Took a quadcore CPU just over ten minutes to render that one frame.

“Uh-huh, and what are those stupid watery columns?” you ask. Oh, you noticed those? I have no idea where those came from; ask the propeller heads who engineered Vue’s built-in renderer.

So, to summarise: No matter how much tweaking / optimising you do, relying on Vue for complex, animated scenes really is asking for trouble. For single frames, it’s brilliant—just look at the e-on software gallery—but for pure CG video, you really need a render farm. Or else a whole lot of patience. I’m prepared to wait years for YouTube to produce a celebrity who isn’t a fucking obnoxious douchebag but you’ve got to draw the line somewhere.

Of the scene I’d originally envisioned, I produced only one clear render:

Also, besides a one-off, low-resolution test—with almost all the render settings set to ‘Low’—I didn’t even bother with the full animation:

(Some of the trees in Vue don’t look so flash when they’re subjected to wind.)

Eventually, in a bid to salvage the scene, I removed the flying saucer and rendered yet another clip featuring the Strider from Half-Life 2:

Strider at Farm

Click to enlarge.

After testing a couple of atmospheres and their impact on render times …

… I went ahead and rendered a very basic version of the entire animation:

I might have experienced a sense of relief when it finished rendering but my hard drive died a few hours later.

So, why all this Strider crap? Well, back then, I was still hell-bent on creating a Half-Life 2 short film. My ingenious plan was to take advantage of the massive, pre-existing fanbase of Half-Life players and thus have a guaranteed audience of millions. How’s this for vindication of that theory? The Purchase Brothers actually did make a Half-Life 2 short film, racked up over 1.5 million views within a week (on YouTube alone) and were then signed up by CAA, one of the biggest talent agencies in Hollywood.

Man, I’m glad that didn’t happen to me!

After conceding that I’d been beaten to the punch, I said to myself, “What the hell was I thinking, anyway? I should be using my imagination to create something entirely new; not exploiting a pre-existing franchise for personal gain. I’m not Uwe Boll.”

I went on to write just under twenty stories and from that batch selected one to turn into my first short film. So, from now on, my every waking moment will be dedicated to that project.

Hopefully, it won’t suck.

Blog at WordPress.com.