Natural Motion: The Next Generation of Animation

[bapenguin]

IGN (http://xbox360.ign.com/articles/651/651000p1.html) has an article up regarding an upcoming next generation animation technology called Natural Motion. One of the problems today is motion capturing all the animations for a game, as well as trying to have smooth transitions between those animation blocks. Natural Motion dynamically creates realistic human animation.

As Mr. Reil put it, "Once the technology is in the game for the run-time engine every tackle, for example, you would produce is unique, it's your tackle. Nobody else has done this before and nobody else will ever do that. The same is true for other gameplay elements. Games for the first time become unique. The way you play a game, what happens, really is something that you create since it has not been pre-produced. I think that is a huge shift in overall gameplay and the way games will look on the next generation."
Sounds great. We've said it before, procedural animation and graphics are really what will define next-generation games.

[Thenetcase]

Well, I don't know about you, but I think that this is a huge step for the next generation. This and the physics engines.
I'm so sick of crappy, unrealistic physics and animations. I don't want a football player that I stear down the field like a yacht. I want a football player that dodges and weaves like a real person.
I don't want a black spot on the ground that looks like someone dropped an oversized cigar on the battlefield, I want a crater that was rendered dynamically on the fly.

These things will go a long ways to help games look more realistic and feel more realistic. It's not the only thing to take into account, to be sure, but it's a major step in the right direction.

Open ended 'rules based' animation shouldn't be that hard to do. It might even take less time than recording 500 takes for every celebrity in the game and then incorporating those into the game. This way you can merely capture the celebrity once and apply the animation rules to their body, apply the textures that you captured and you're done. I think it'll make life easier, if anything. Surprised they haven't done it before, except for maybe the CPU intensive calculations (just another reason I look forward to the PhysiX cards).

Just my two cents. For whatever THAT'S worth.

-TnC-

[TRiLoGY]

I think this is good news..

I believe Spore will be using procedural animation.. :)

[SlamDunk]

Yeah. I've been following Natural Motion for quite some time now.

Here's a brand new Endorphin game video:

Possession by Blitz/Volatile
http://hardwired.hu/dl/3/10/Possession_X360_muvi_physicsbabe.zip

Press release: http://www.naturalmotion.com/files/possession_endorphin25.pdf

I wonder if AGEIA and Natural Motion will strike up a deal soon :)

[Xerxes]

Ok in that Demo video, why the ball carrier fumbled every freaking tackle. Now if he clutched tighter, then you raise a eye brow. But overall, this is good.

Didn't Tobal have this? ^_^

[PIPBoy3000]

This is admittedly a pretty big deal, though the problems aren't as simple as people might think at first.

It ties into AI pretty closely. If the game tells the person to "attack creature X with a sword", that can get into all sorts of complicated issues if the sword hits a wall, the person trips on a pebble, or any other sort of unexpected collision. Essentially you have to have pathfinding for each and every body part - human beings have a huge brain to handle all of that, while poor little computers have to make due with much simpler rules.

I think that's why they're focusing on football. It's a big empty field with only people and a ball.

[Citizen Philip]

This kind of thing isn't news, and is really an issure of who can afford Motion-capture, a large team of animators, or a small team of animators. Procedural animation working from a set list of physics and body deformations is about having immediate processing power.

If your average kinematic skeleton has a 30 point skeleton, that's 30 points with a potential XYZ movement or 90 transformations, rotations and pivots: if you need to apply rules to your 30 joints (ie. your 8 point spine doesn't go backwards) that's more processing power going towards maintance.

Now if you have a 8 man tackle, that's quite a few points of tracking information that has to be computed very fast.

Now add in AI, ranged weapons, melee weapons, potential kinetic collisions between static props and other jointed entities and you are getting into quite a few processes that need to be done.

I'm all for stream-lining this processing and trying to make it as efficent as possible, but its still about raw crunching ability than anything else.

I haven't directly animated in a bit, feel free to correct me.

[Heretic Machine]

This kind of thing isn't news

...How is this not news?

[mister_slim]

Well, props to the Natural Motion PR person. And thumbs down to IGN's whores.

[Citizen Philip]

...How is this not news?

Automated systems that do the 'in-between' animation have been around for awhile. They haven't been very good, and usually required someone to supervise them. I didn't read anything in the article that was revolutionary.

If they want to streamline and improve on transitional dynamic anmation, more power to them, but it all requires avaliable processing power, and processing power is a premium with console systems - and PCs for that matter.

It's possible right now to have dynamic collisions between jointed entities that react to mass, velocity, angle and what have you. The problem is: it requires a gross amount of processing to do it. Nothing in the article indicates they are even attempting to sidestep the processing problems, other than streamlining possible translations.

[bapenguin]

This kind of thing isn't news, and is really an issure of who can afford Motion-capture, a large team of animators, or a small team of animators. Procedural animation working from a set list of physics and body deformations is about having immediate processing power.

If your average kinematic skeleton has a 30 point skeleton, that's 30 points with a potential XYZ movement or 90 transformations, rotations and pivots: if you need to apply rules to your 30 joints (ie. your 8 point spine doesn't go backwards) that's more processing power going towards maintance.

Now if you have a 8 man tackle, that's quite a few points of tracking information that has to be computed very fast.

Now add in AI, ranged weapons, melee weapons, potential kinetic collisions between static props and other jointed entities and you are getting into quite a few processes that need to be done.

I'm all for stream-lining this processing and trying to make it as efficent as possible, but its still about raw crunching ability than anything else.

I haven't directly animated in a bit, feel free to correct me.

I think the point is, next generation systems CAN do this. The XBox 360 and it's graphics processor were designed for procedural content.

[F3nyx]

I think this could easily turn into the next ragdoll physics... you know, the cool new feature that every game's got to have. Imagine someone getting punched, doing a couple ragdoll-esque flips through the air, landing, and then getting back up again -- all simulated with physics rather than pre-animated. This will fix some of the big shortcomings of ragdoll physics.

[Roc Ingersol]

Automated systems that do the 'in-between' animation have been around for awhile. They haven't been very good, and usually required someone to supervise them. I didn't read anything in the article that was revolutionary.
It isn't simple animation blending or keyframe interpolation.
There is an AI component to their framework as well, that adds another level to the whole gig. The entities attempt to realistically maintain balance when acted upon, react to physical collisions, etc.

It's not 'new', except that this tech is now a feasible chunk of middleware for next-gen games -- helping bring down those next gen production costs we hear so much about.

[Citizen Philip]

I'll believe it when I see it. As previously said, they are talking abount a flat field with players that will behave almost identically: Perhaps good news for football and hockey sims, but for the time being, not much else.

I *do* agree that a set-kit of physics and animation algorithms will help to reduce costs in the future and is only good news for the gamer. This maybe the start of this trend, or a piece of marketing fluff. I believe it to be the latter.

[RMan]

I think the point is, next generation systems CAN do this. The XBox 360 and it's graphics processor were designed for procedural content.
Current generation can obviously do this as well, just not on as big a scale. I do think it's a little silly that people are attributing certain software advances that become more mainstream now as things that the next generation has bestowed upon us. This and normal mapping are good examples of unrelated software advances, and although more power is good, let’s not exalt the nextgen to magical status.

[Thenetcase]

I'll believe it when I see it. As previously said, they are talking abount a flat field with players that will behave almost identically: Perhaps good news for football and hockey sims, but for the time being, not much else.

I *do* agree that a set-kit of physics and animation algorithms will help to reduce costs in the future and is only good news for the gamer. This maybe the start of this trend, or a piece of marketing fluff. I believe it to be the latter.

Bill Gates also said that the maximum amount of memory that a PC would EVER need was 640k.

I think you're a fricking idiot if you really believe this isn't a big deal. Everygame imaginable could benefit from this. If you say that's not the case, why did Ensemble Studio add the Havoc Physics system to Age of Empires 3? Because it makes it a hell of a lot cooler. Adding something like this to a game like that would take environmental physics to a whole new level. You're thinking about humans, I'm thinking about trees and ships (sails, masts, etc.).
You need to stop thinking in your own tiny little box. Great things never happen to those who can't imagine them.

-TNC-

[Lint of Death]

Great things never happen to those who can't imagine them.

Will Wright himself has said that the biggest challenge with making Spore, a game that uses procedural animation, is convincing his team that they can make it. I believe Pixar has said that about their movies; almost every one has been an attempt at expanding the range of things that they can make. The Incredibles, for example, was their major attempt at rendering people; they did not think it possible at first, but look how the movie turned out!

Clearly, confidence is a major component of innovation and implementing new techniques or technologies.

EDIT: Added the quote to increase relevance to the thread.

[Thenetcase]

Thank you, Lint. I'm glad someone actually can see where I am coming from.

If no one strived to out-do themselves in this world, we'd be no better than the apes.

-TNC-

[Citizen Philip]

Thank you, Lint. I'm glad someone actually can see where I am coming from.

If no one strived to out-do themselves in this world, we'd be no better than the apes.

-TNC-

You started in the mis-informed and somehow travelled to self-righteous in 2 messages, congratulations! When you try re-reading my post, perhaps you can come to a more informed conclusion about my opinon on the matter.

I know its really hard to see from the top of whatever fanboy mountain you've climbed, but that's okay. It's even harder to accept your next console box isn't the second coming, and yet another iteration to be followed by further incarnations promising the same thing.

But feel free to vent your angst against any who would say anything but great things for... whatever is obviously extremely important enough to not bother reading what is posted.

[PIPBoy3000]

Like any new technology, you'll likely see procedural animations in very limited, high-value areas. Football is a great example, as the number of objects on the field at any time is fairly low and the AI is (relatively) straightforward. Later on you'll see procedural animation used in different areas - initially costs will be high as programmers take the burden of what was once an animation task. Eventually you'll see the cost of making games go down.

Personally I'm hoping for the nirvana of procedural everything - with procedural actors with emotions and voice acting, the ability to create models on the fly using fairly generic parameters, and so on. Early versions will be far less than perfect and likely require human intervention, but I'm hoping the incredibly steep art/content requirement for games starts going down again.

[TheKeck]

Yeah. I've been following Natural Motion for quite some time now.

Here's a brand new Endorphin game video:

Possession by Blitz/Volatile
http://hardwired.hu/dl/3/10/Possession_X360_muvi_physicsbabe.zip

Press release: http://www.naturalmotion.com/files/possession_endorphin25.pdf

I wonder if AGEIA and Natural Motion will strike up a deal soon :)

Those zombie takedowns were pretty fun.