Next generation animation

Q

Qroach

Veteran
I figured I'd start a thread about this since i'm seeing this on many forums lately. people are consistently complaining about motion capture and suggesting all sorts of things like Procedural animation, Ragdoll animation, Physical based animations, etc...

Why is it people think these choices are so much better than motion capture. how could any of these techniques (and i use the term lightly when it comes to rag doll) be better than capturing the direct motion of a human?

I'd like to see what peopel think teh advantages of these are so they can be properly noted or debunked.
 
That was motion capture... BTW, can one of the mods move this thread to the console forum? I put in in console games by mistake.
 
Well I responded in detail in the Madden thread about how unrealistic the motion captured animation is.

I just assume that another approach might actually have a chance to show more realistic action but maybe that's not necessarily the case. I remember an EA exec. a couple of years ago (forget his name but it was Chip or something like that) gave a speech at a developers event. The guy predicted that everything you saw on-screen would be modeled. This was long before the specs. for the next gen was revealed.

The other thing I noticed, as a long-time sports gamer, was that in general, the games of this generation gives you less control, as the controls aren't as responsive as the previous generation (or even PC games around 2000). One reason is that some of the motion-captured animations were too long and Madden in the last couple of years have tried to shorten them to return some of the responsiveness.

In a game like NBA Live, you could press a button to activate a spin dribble animation but sometimes, that animation would take you out of bounds so the player would furiously try to guide the ballhandler away or push other buttons but to no avail until that first animation was done.
 
Qroach said:
That was motion capture...
Click to expand...
Not all of it was mocapped, the little robot and the "metal gears" for instance ;)

I think motion capture is still a good choice for human animation, as long as thetransitions between movements are done seamlessly and there is enough variation.

But when the character moves in a very varied, uneven terrain, then I can see motion capture showing it's limits. For example if the character walks on a slanted surface, the varying degrees of slantness effecting the pose of the character would be time consuming to do with motion capture, maybe then some sort of ik combined with mocap would work, don't know really I'm no expert in this field either ;)
 
Qroach said:
I figured I'd start a thread about this since i'm seeing this on many forums lately. people are consistently complaining about motion capture and suggesting all sorts of things like Procedural animation, Ragdoll animation, Physical based animations, etc...

Why is it people think these choices are so much better than motion capture. how could any of these techniques (and i use the term lightly when it comes to rag doll) be better than capturing the direct motion of a human?

I'd like to see what peopel think teh advantages of these are so they can be properly noted or debunked.
Click to expand...

For one, motion capturing limits you animation to whatever you've captured. Anything else has to be hand-tweaked by the animators.
The problem is that we almost expect too much from videogames. Reality is a tough cookie to replicate, and it will take a LOOONG time till we can have the full range of human movements in a game. Even longer if you want a processor actually "invent" the animation, without any for of capturing or pre-recorded routines.
Even CGI is still kinda stiff, and that's all motion captured and/or hand-tweaked to make it look as perfect as possible.
 
I would not mind "motion capture" if it was designed better.

The problem with a "simple" MOCAP system is evident in Madden:

3 People hit a ball carrier. But the game only has 2 player tackles.

So what does the 3rd player do? He either 1. Stands there and the tackles hit him and fall at his feet or 2. he DIVES and does the magic "super leap in the same spot" lame dive.

The problem is

1. Repetitive
2. Uninteractive

Madden, for example, looks nothing like real life.

And then there is the common 3rd problem

3. Hitching. Transition stages between animation switches are often jerky/unnatural

And finally

4. The super sliding feet.

And I cannot say how annoying it is when I player dives, gets a hand on you and then "slides" in front of you for a full tackle. Real life? That arm tackle maygraze the jersey, maybe make the ball carrier twist some, but the sliding animation thing looks hokey and gake.

Overall, at least in sports games, I have found the animation systems to be very poor respresentation of the reality it wishes to recreate. I think more effort shuold be put into animation than visuals.

Graphics = Visuals + Animation

Right now the animation is a significant break in the "Believability" of the product. It also hinders gameplay.

Anyone not understand what I am saying? Watch a 1hr "Defensive Tackling Highlight" video and note how dynamic and fluid it is. Then watch 1hrs worth of Madden/NFL2K tackles.

The thread I started the other day with the "context" sensative animations was interesting. It used MOCAP, but it was designed with "Keyframe" begin and end points. So the "reaction" to a hit is dynamic, but the "End Pose" would be MOCAP.

That is an interesting avenue to explore.

Currently we are using the same techniques, for the most part, used on the N64 and PS1.
 
what was that game ,years ago ,with viking sword fights completly code drive animation (not a single keyframed animation *) ?.

*i considere mocap is keyframed animation ,since it requires keying.

edit : found it : Die by the sword
http://en.wikipedia.org/wiki/Die_By_the_Sword
The title's main selling point was the introduction of the VSIM control system, which allowed players to independently command the movement and swordfighting of their ingame avatars; running, jumping, and turning with one hand, while simultaneously slashing, stabbing, and parrying with the other. While many games of this type allow for a similar dualism, Die by the Sword remains to this day unique in its treatment of physically correct and force based melee combat, as described below.
Click to expand...
 
Last edited by a moderator:
ohh, I wish I coudl reply to everyone right now, but I'm very busy this afternoon. Basically there's a lot of misconceptions about the way mo cap works in this thread. I'll post some informaiton about this later on when I have a chance. Dang sorry guys...
 
Qroach said:
ohh, I wish I coudl reply to everyone right now, but I'm very busy this afternoon. Basically there's a lot of misconceptions about the way mo cap works in this thread. I'll post some informaiton about this later on when I have a chance. Dang sorry guys...
Click to expand...
NP :)

Btw, not all MOCAP is created equal. I have seen some done VERY well and others NOT done very well. I am sure you may highlight that MOCAP is not dead and can be VERY believable--and I would 100% agree with that. Even the method I had mentioned relied on MOCAP to a degree.

Just wanted to clear that up before everyone starts talking past eachother. Not all MOCAP systems and implimentations are created equal. Honestly, it is not even the technique that is the problem IMO, but the implimentation that is often a problem.

My comments were more directed toward Madden/Sports as I think that is where there is serious issues. Ditto fighting games. The precanned responses to actions is very limiting IMO.
 
I like ragdoll, when people hear ragdoll they think of a characters death scene(flailing arms 'limp' body)

Most 400lb linemen are not going to die or pass out when they are hit by a 200lb running back. :) their arms may recoil and they *might* be knocked to the turf, but it would be a 'big' hit.

bringing physics to a game..say football will be neat! causing a fumble will be the result of a well timed, well delivered blow to the ball carrier, not a random generated fumble number.
 
Nice Article on Endorphin.

Endorphin Zombies (use IE, FF is messed up on that page)

Endorphin (?) Football

I don't think this is necessarily the BEST or ONLY solution, but it does show how very believable results (more dynamic) are attainable. Note how every tackle would be different dependant on angle, velocity, and other attributes.

Btw, this DOES use some MOCAP (at least I think so). It has key frames for poses; the difference to mocap that transistions from one mocap (e.g. running) to another mocap (e.g. running to getting tackled) this system would have a mocap entry and exit poses, but inbetween it is more dynamic and dependant on outside forces for the animation.
 
Excellent find! Cool video!

It does look like a stop gap....but I could live with this! :)

Hopefully DOA5 will do something like this.....or some wrestling game.
 
The article is good too:

http://xbox360.ign.com/articles/651/651000p1.html

Too bad EA isn't considering it. Those tackle animations look good.

Now if the system was dynamic that one of the players being hit is heavier than the player diving into him, you would see a different result than if the players weighed the same, had same acceleration, etc.

Ideally, if a safety tried to hit a WR high with a kill shot, the WR would be laid out. If the safety tried to do that to a RB, the effect wouldn't be quite the same (assuming the safety didn't get a running start with no blockers in the way).
 
MMMMMMMMMMM barbeque man!

4036.jpg



damn i'm hungry
 
The problem with mocap is not the quality of the animation but rather the flexibility of some implementations of it.
A good flexible method would be capturing only the desired endpoints (or key poses/frames) of the moving limps, and then have reverse kinematics take care of the tweening.
That way, the motion is affected realistically by the mass of the body (which most likely does not correspond with the actors body mass), and is also affected by external physical resistance, like terrain and different objects.
One disadvantage of simple mocap is that animations sometimes "pop" into one another, but with smoothing between the animation cycles that isn't a problem in well made games today.
 
You guys do realise that these Endorphin animations are built from standard mocap running and jumping sequences, right? Endorphin takes care of the collision and what happens after it, until the next canned animation starts - but the rest of the animation still has to be created somehow.
 
You must log in or register to reply here.

Similar threads

I
The first Evil Within still has remarkably sophisticated lighting
Replies
2
Views
1K
Phantom88
P
F
3D Motion Capture Business
Replies
7
Views
4K
imtrobin
I
L
Next generation asset creation
Replies
19
Views
4K
Laa-Yosh
L
P
CoD3 interview + HD movies
Replies
9
Views
2K
NBK-Steve
N
R
Next Generation: CPU future...
Replies
12
Views
3K
DudeMiester
D
Back
Top