Speaking With the Blizzard Cinematics Team

TheFrunj writes "Blizzard's games, such as World of Warcraft and StarCraft, are perhaps known for their breathtaking cinematics as much as for their gameplay. The process behind these cinematics is complex, involving an entire modeling and coding team — they even needed to make their own physics engine! AtomicMPC has the full story behind these incredible videos. Quoting: 'Storyboards become animated storyboards (thanks to a storyboard team who are also trained animators) complete with temporary music to help set the tone and get the pacing right. Then the animators and modeling team get involved, creating a rough 3D version of the final animation. While the modeling team works on characters and environments, the animators work with very low polygon characters and as the models are refined and updated the animations are refined. This is also where production technicians get involved, supporting the more technical developments like hair and skeletal systems for the characters. And then the effects team kicks into gear. '"

Comments

Re:we want more

[Anonymous Coward]

It's amazing how the high quality of Blizzard games is instantly noticed. I remember back in 2001 when I was amazed by the cinematics of Diablo 2. When I was at my friends house watching the Diablo 2 intro, I said "that's high quality texture!" "What?" he said. That was in Syria, right next to Iraq. I wonder how many Iraqi's have played Blizzard games before.

"...the first step to controlling your world is to control your culture. To model and demonstrate the kind of world you demand to live in. To write the books. Make the music. Shoot the films. Paint the art."
Chuck Palahniuk

"Oops, no time for discussing compositing"

[shidarin'ou]

As a compositor, it's probably just me, but it seems like every article on post production effects (for movies, games, commercials, etc) spends so much time on modeling, effects and lighting that by the time the writer has hit the compositing part of the pipeline they've- oops! run out of time.

There are fantastic modelers, trackers, texture artists, lighters etc in the pipeline, but here's the thing- all of us share the burden of producing a good looking shot equally.

Without a good model, no amount of work by the texture artist lighter and compositor can make the shot look good. Without good texture art, the lighter will never get good light play off the model. Without good lighting, the compositor's workload increases 100 times as he now has to "fake" lighting and shadows in 2d. Without good compositing, excellent lighting, modeling and texturing will still scream CG to any audience.

Personally, I find it very cool to be at the end of the pipeline- things come to me and I get to throw them into the pot; stir and make the entire shot come together. I feel bad for the modelers and animators who don't get to see their final product- what I send to the client is what I see on the big screen in another month.

Of course, being at the end of the pipeline would be cooler if all the magazine editors didn't run out of space at the end of the article :(

Re:"Oops, no time for discussing compositing"

[Animaether]

A compositor usually does more than a little touch-up. These days, a compositor may very well be expected to add rim lighting to an output render using nothing but the original output render with its auxiliary data (normals, z-depth, etc.)

I agree with GP poster than compositors usually don't get very much attention, even though the compositors are often the ones who make a render 'work'. Granted, this is more true when it needs to integrate with physical effects or a live action plate, but it still applies to full-on CG shots as well.

Re:"Oops, no time for discussing compositing"

[jackbird]

All professional 3D is heavily composited. If anything, moreso than live-action greenscreen shots, as not only are foreground elements often rendered separately from backgrounds, but shots are rendered in a number of passes (diffuse color only, lighting only, reflections only, ambient occlusion only, global illumination only, etc.) in a way that would be impossible to photograph. This serves a variety of purposes:

-Much more flexibility to adjust the look of a shot in post without waiting for rerenders. Sometimes each individual light's contribution is rendered seperately so that the lighting can be completely altered in post by adjusting a couple of layer opacities.

-Decouple backgrounds/environments from characters. If separate artists are working on these, they don't have to wait for each other. Also, having less data in each artist's 3D scene makes their workstations more responsive, things less likely to break, and they get more done.

-Much less artist time spent tweaking parameters and waiting for test renders before rendering the shot. The compositor can accomplish fine-tuning of the color and lighting much more efficiently than a 3D artist can if the pipeline is set up properly, especially as rendering to unclamped floating-point color is becoming the norm.

-Render-intensive effects like global illumination and subsurface scattering can be accomplished just as well with 2D compositing tricks in a fraction of the render time for many shots.

-Effects like specular blooms and glows are so much faster to do in post, and without having to commit to a certain intensity or spread ahead of time.

-Ability to use photographic elements in conjunction with 3D. For one of the Warcraft 3 trailers they shot a matte using white paint dissolving in a black kiddie pool to do a blood-in-water effect that would have pushed or exceeded the limits of fluid rendering technology at the time.

-Ability to use different renderers for different aspects of the shot. Want the totally-smooth Sub-D's Renderman gives you, lit by mental ray's final gather? Or a background rendered in Maxwell or Vue behind rigged characters that render before you get old? No Problem.

-If passes like normal direction, pixel velocity, Z-depth, and UV coordinates are rendered, it is possible to make fairly radical changes in post, like replacing textures or adding depth of field/motion blur, without tying up the render farm. Clients being the indecisive creatures they are, this is a Good Thing.

On the Blizzard Starcraft DVDs they mention in the commentary that Starcraft was the last time they didn't use compositing and did everything in-camera, and it shows. On one of the shots they point out that the background (which they composited in-camera by using it as a background for another render) stops moving about 2/3 of the way through, but they didn't have time to fix it, as it would have meant re-rendering a bunch of layers.

There's a huge jump in the quality of the Blizzard cinematics between Starcraft and Diablo 2 - this is partially down to technology, as they moved off 3D Studio DOS to 3DS Max, but also largely due to the use of compositing as a fundamental step in the 3D production pipeline.

Re:"Oops, no time for discussing compositing"

[rcuhljr]

Could you have any less clue about Blizzards development track record?

Re:huh?

[Exawatt]

The "crappy prerendered cutscenes" are what we are discussing, and do "count," as those are considered the "cinematic" portion of the game.

Re:we want more

[fractoid]

Yeah, if only they were working on a WoW movie...

I also hear there's another movie called "Wh*res of Warcraft" if that's more what you're looking for. More like a naked blood elf chick dancing on a letterbox and less like an epic world in conflict though.

Re:Am I the only one?

[bigstrat2003]

Am I the only one who could care less about cinematics in video games?

No, but you are in the minority. Many people are very fond of video game cinematics.