Progress.. are we making any?

[JimDigritz]

After watching the Day Z behind the scenes I've just been thinking about motion capture which has become a proxy for a lot of other game development activities.

For the past 10 years we have been capturing live action actors motion and translating their actions into games. You know, a black one piece with ping pong balls attached to every joint.

So we should have a shared library of human motion.. after 1000's of hours of motion capture for 100's of games we should be able to represent the majority of human motion from crouching to jumping to punching.

My general feeling is that game developers are being asked to re-invent the wheel instead of creating unique experiences.

Why?

[Harlander]

Because the motion capture data is considered proprietary and thus not shared?

That said, Carnegie Mellon University has a mocap database: http://mocap.cs.cmu.edu/, as does Ohio State Uni http://accad.osu.edu/research/mocap/

and there's a couple of others.

[gundato]

Yeah. MoCap data is the same as any other asset. A lot of studios make their own, but there are databases (free or commercial) out there. That's the same reason why there was that big freakout that DOOM 3 "stole' resources from Half-Life 2. They just both bought the same library of textures.

[frosty2oo2]

so Google's singularity can hunt humans more effectively, or maybe its perceived market demand, where would you invest your money? Too put it another way, would you back the horse you've seen win several races, or that one you've never seen before?

[Bhazor]

I think the whole mo-cap thing will start to die out now automated animation is becoming more common. After GTA4 I can't help but look at canned animations the same way as FMV's. Capable of nothing but provoking nostalgia of bygone day.



In particular it allows you to animate non humans, which you just can't do with mocapping. Just add a few extra limbs, whack on a tail and twist the figure into whatever you want and let the physics do the rest. Bear in mind that example above is from 4 years ago.

Honestly though, the biggest improvement I've seen in the past few years is developers having the bravery to not try photorealism. Indies in particular are going into bizarre and glorious new art styles and I can't help but see that is infinitely preferable to

[Tei]

The source code of the 99.99% of the games you are played is lost.

[Sparkasaurusmex]

Motion captured stuff uses actors, so I would imagine reusing the capture would sort of be like re using recorded voice overs. I'm sure it's possible but you'll need to get a hold of the actor's agency.

Hollywood's calling it "performance capture" now, because I guess dialogue and movement are all captured at once. Also they just use cameras now, not little balls attached.

[gundato]

Motion captured stuff uses actors, so I would imagine reusing the capture would sort of be like re using recorded voice overs. I'm sure it's possible but you'll need to get a hold of the actor's agency.

Hollywood's calling it "performance capture" now, because I guess dialogue and movement are all captured at once. Also they just use cameras now, not little balls attached.

It depends. I know Avengers (the movie) mo-capped Mark Ruffalo's face with tiny balls.

[Velko]

Mark Ruffalo's face with tiny balls

There are at least a couple of ways to read that.

[MoLAoS]

There are at least a couple of ways to read that.

Based on my first thought and this post, I think there is only one.
>.>
<.<

[ambing1]

It would good to have new methods in creating games at this time. with the release of the next generations consoles, we can probably expect some unique or novel game-making techniques instead of the usual recycled stuff.

[khaz]

Portal 2 is proof that the most unique and rewarding method is still animation by hand. It's labour intensive and time consuming but the end result is always worth it.

[b0rsuk]

The source code of the 99.99% of the games you are played is lost.

The saddest sentence I've read in a while.

[SirKicksalot]

Hollywood's calling it "performance capture" now, because I guess dialogue and movement are all captured at once. Also they just use cameras now, not little balls attached.

Contrary to popular belief, Andy Serkis' facial expressions were not mo-capped in LOTR.
Performance capture is facial motion capture. Technically the term should be used only when dealing with the face of the actor, but since pretty much everything requires both body and face capture it replaced mo-cap.

In particular it allows you to animate non humans, which you just can't do with mocapping. Just add a few extra limbs, whack on a tail and twist the figure into whatever you want and let the physics do the rest. Bear in mind that example above is from 4 years ago.

Remember the old Spore creature creator demos? The final version was dumbed down.
Custom mo-cap will always be needed for things like The Witcher's medieval combat or Max Payne 3's weapon and cover handling. The true advantage of software like Euphoria is impeccable animation blending. The problem is in most games you don't want physically accurate animations and blending because they're cumbersome - not just for the main character.

[TechnicalBen]

Um, I think the two would be used in tandem. If mo-cap can be done quickly and cheaply, then it will be used.

[Zephro]

Remember the old Spore creature creator demos? The final version was dumbed down.
Custom mo-cap will always be needed for things like The Witcher's medieval combat or Max Payne 3's weapon and cover handling. The true advantage of software like Euphoria is impeccable animation blending. The problem is in most games you don't want physically accurate animations and blending because they're cumbersome - not just for the main character.

Why? Seriously why wouldn't dynamically generated animation reach the point of being indistinguishable? Or at least a fuck load cheaper?

[SirKicksalot]

The Witcher series mo-caps German medieval combat experts.
Until someone builds a licensing library of such esoteric animations, devs will still have to capture their own.
Even for mundane actions it's way cheaper and easier to hire a couple of guys, rent a sound stage and capture exactly what you want. I can't see that changing even as procedural animation software improves. You still have to tell the software what you want, and at some point it becomes easier to mo-cap it or make it from scratch.

For fighting games, arcade action or DmC-style titles every animation frame counts and the hitboxes must have pixel-perfect precision. You're always going to need people to clean up every animation. The more eccentric the action is, the more handmade animations you need. As smart as software can get someone still needs to supervise it all.

Recently there was a debate about Skullgirls. The devs launched an Indiegogo campaign to fund a new character for 150k dollars. A GiantBomb article explains why they need so much. Most relevant to this thread is that 2D animation is expensive as fuck. Here's an example of a Skullgirls character (and that's just the basic outline):

It's the simplest one with around 1500 handmade frames of animation. You can't build a software to do that. What you can do is you draw the key positions, hire a bunch of people and have each one work on a different stage and fill in the blanks between frame 5 and frame 10. A single frame can take two hours or more. You don't make 1500 drawings for 1500 frames, you make 4000 or 5000 because you need consistency and you keep tweaking. 2D animation seems simple but shit can get very real. Thankfully for 3D games iteration is much simpler than the hell Skullgirls goes through. It's still painful and animators are precious employees.

Until now memory was a huge problem for animations in console games - in ME3 you never holster your weapon because they ran out of RAM. All animations must be stored in the runtime memory, hopefully this won't be a problem in the future (I'm sure that's what they said back in 2005 lol).

Sure, you can get away with a basic mo-capped walk. You can build environment triggers so that characters look at things or reach out to touch them, like in The Last of Us. But when actual gameplay is involved there's a lot more going on behind the scenes than people realise. As you keep iterating on the speed and flow of the game, animations for basic actions will get altered on a frame-by-frame basis. I can't imagine a software that can automagically shave a couple of frames off and make the rest blend better so an attack feels more satisfying or falls in line with the rest of the combo. Someone has to do it manually.

[fiddlesticks]

I have nothing to add to this discussion, but I wanted to thank SirKicksalot for his last two posts. They have been very insightful.

[Sparkasaurusmex]

Contrary to popular belief, Andy Serkis' facial expressions were not mo-capped in LOTR.
Performance capture is facial motion capture. Technically the term should be used only when dealing with the face of the actor, but since pretty much everything requires both body and face capture it replaced mo-cap.

Yeah I don't follow Hollywood, but this movie Avatar is the one where the actors were going on about it being called "performance capture" and they basically just filmed the scene then later it was all graphicalized.

[kingmob]

That's a good point actually. Probably somewhere along the line either a company will start selling mocap data and/or some procedural motion stuff like speedtree, or it'll be open sourced. I'm assuming that companies already hire specialists to do the capturing for them now, so it is the next logical step. It almost has to happen with the ever expanding budgets, for several areas actually. That, or the AAA-game business implodes.