An Interesting Char Anim Locomotion System

[Synide 0]

For those interested in such things check out Rune Skovbo Johansen's character locomotion system.

[sparks50 0]

Yes, its definitely a big improvement from the standard "moonwalk and slide" technology presently being used in games.

[Dwarden 1124]

definitely amazing i heard about his work in past but this is definitely way better than i though :)

btw. on YT videos there is example of 9 and only 3 animations as source and the engine interpolating between to try keep fluid movement ...

[PhilippRauch 0]

if this principle works better in large multiplayer settings than ragdoll does right now, that would be quite awesome, i like the fluidity...

But couldnt this principle be already implemented in ArmA? Like checking the motion vectors and adding those animations as needed? For this there would be only:rolleyes: some more intermediate animations needed? Just my guess... not that it will be the same system, but somehow using its basic principles of 'animation estimation'??

[dissaifer 10]

Ran across this while browsing. I realize that it is for the Unity 3d engine, but I think it's pretty impressive.

Edited November 6, 2010 by Dissaifer

[Fox '09 14]

i would love something like this, it would make motion much more fluid in arma 2

[enad 11]

Very nice. Something like this would be amazing.

More fluid moment is one of the main things lacking in A2.

[Gunter Severloh 3949]

Fluid movement is pretty important, jerky, and stiff movements get you killed, or slow you down,

when you need to move. Having sticky and 2-3 step movement animations cost you time to do something

versus just freely moving into that position or place.

i think the movement anim are better in Arma2 compared to Arma, obviously, so you know BIS is working on these aspects.

In arma before going from crouch to stand it would be a 2 process movement like, and in Arma2 its more fluid, eliminating

the 2 seperate movments if thats what you call them into one motion.

Good suggestion, Im all for it!

[nkenny 1051]

if this principle works better in large multiplayer settings than ragdoll does right now, that would be quite awesome, i like the fluidity...

The principle is essentially the same AS ragdoll. Or rather ragdoll is a different way of utilizing inverse kinematics to dynamically affect the skeletons. In fact the (IK) capability is already in Arma2. Witness how weapons can have unique shouldering/holding animations.

Arma2 needs animation work badly. Much more so than petty engine upgrades adding questionable features*. Whilst getting experienced and skilled animators (motion capture is overrated) can be expensive-- keep in mind that this is an upgrade which DOES NOT REQUIRE ADDITIONAL CPU CYCLES -- and more to the point if the animations and the source is good. It can be retained and used for future projects.

-k

* I'm hard pressed to consider 'engineers' or 'shotguns' very significant engine upgrades.

[dmarkwick 261]

if this principle works better in large multiplayer settings than ragdoll does right now, that would be quite awesome, i like the fluidity...

But couldnt this principle be already implemented in ArmA? Like checking the motion vectors and adding those animations as needed? For this there would be only:rolleyes: some more intermediate animations needed? Just my guess... not that it will be the same system, but somehow using its basic principles of 'animation estimation'??

Well the attraction of this implementation is that it only needs to run client-side. The ragdoll stuff needs some amount of network synchronisation. At least the torso I would say, arm & leg positions can be client-side with no real impact.

[CarlGustaffa 4]

Which is why we get them as part of free patches related to DLC upgrades rather than new game (like OA and significant upgrades in sound and FLIR)? ;)

[Cookieeater 10]

ArmA II should stop with the step based animation system, it's too clunky and limits the player with clunky controls. Operation Flashpoint used a normal system like other games(BF2, SWAT 4) so I believe which was perfectly responsive and felt good and didn't sacrifice realism for fun.

[froggyluv 2130]

Alls I know is if BIS declared tommorow that this type of thing will be in their next project/release -well i'd start X'ing out the days on the calendar in profound anticipation.

[Cookieeater 10]

Also to think about it, this can already be implemented in ArmA II's animation. All the units have a walking animation, and a running animation for all directions. Imagine if we could harness this, we could basically have dynamic move speeds for all units that look incredibly believable and in any direction. And it would be more fluid as well. ALSO if any of you have installed Unity web player, you can see this implementation in the Bootcamp demo, which procedural animation I can say looks incredibly believable.

http://unity3d.com/gallery/live-demos/index.html#bootcamp

Edited November 7, 2010 by Cookieeater

[CarlGustaffa 4]

Fluid transitions is one thing. But getting variable speed requires a big change in the controls. With A2/OA already covering 6 speeds (I think), that ground is pretty much covered. I guess I miss finetuning my speed (to match a scipted AI), but I think the controls for it would get too complex.

[dmarkwick 261]

Fluid transitions is one thing. But getting variable speed requires a big change in the controls. With A2/OA already covering 6 speeds (I think), that ground is pretty much covered. I guess I miss finetuning my speed (to match a scipted AI), but I think the controls for it would get too complex.

The best solution to this that I've seen is in Hidden & Dangerous 2, where your avatar's speed can be controlled with the mouse wheel. Works particularly well when moving very slowly. I know that ArmA2 uses the mouse wheel for actions, but most are agreed that the actions interface needs a rethink in any case :)

[4 IN 1 0]

Totally agree that fluid movement is needed with newer gen game, ragdoll however is not what really needed, what is needed more is to find a better guy to do the motion capture and more reference material because BY GOD THE CURRENT ANIMATION SUCKS!:p

[CarlGustaffa 4]

The game already support fluid movement speeds I think, as I'm able to setup speeds for an AI leader that I can not match - I'll either be too slow or too fast constantly changing from walk to job to be able to keep up while not running past him. I agree on the current action menu, but I've seen some suggestions I believe could make it work better, like instead of instant removal of items you can't do, they become shaded instead where you can select them (and wait for them to become active) but not activate them. I've seen several mousewheel sugegstions, and I think fluid stance control and zoom would be better candidates than speed. But doing one thing might lead to the whole thing being redesigned, and that will be a problem for some.

[froggyluv 2130]

Would soldiers still die from 8 foot drops I wonder or merely be staggered/stunned like RL - oh the possibilities....

[Sith 0]

I've actually done a graduation project in Unity using this exact locomotion system. I'm no animator myself, but here's a few things I noticed:

• Yes, the system does greatly reduce the number of required animations for a full movement set. Especially the number of transition animations can be significantly reduced (or left out entirely for simple animation sets).
  
• Yes, the system does allow for relatively smooth interpolation, going from one animation state to the next.
  
• No, the system does not necessarily provide the instant movement feedback people here seem to expect. Some level of "clunkiness" often remains.
  

Give the implementation a try in this minigame.

As you'll see, similar to the locomotion system used in GTA IV and RDR, your character has some degree of inertia when e.g. shifting direction. In third person games like these, this isn't much of a problem due to the visual feedback you get on said inertia. But if you apply it to a first person perspective, the results could be in many ways similar to what people dislike in ArmA's anims right now: seemingly "sticky" animation switches.

It's not really my field of work, so I don't want to go too far into the technicalities. But the current discussion does seem to steer towards the "woah, awesome solution to all our problems in one package!" mindset, and I figured some personal experiences might make for a more balanced discussion ;)

[dmarkwick 261]

I've actually done a graduation project in Unity using this exact locomotion system. I'm no animator myself, but here's a few things I noticed:

• Yes, the system does greatly reduce the number of required animations for a full movement set. Especially the number of transition animations can be significantly reduced (or left out entirely for simple animation sets).
  
• Yes, the system does allow for relatively smooth interpolation, going from one animation state to the next.
  
• No, the system does not necessarily provide the instant movement feedback people here seem to expect. Some level of "clunkiness" often remains.
  

Give the implementation a try in this minigame.

As you'll see, similar to the locomotion system used in GTA IV and RDR, your character has some degree of inertia when e.g. shifting direction. In third person games like these, this isn't much of a problem due to the visual feedback you get on said inertia. But if you apply it to a first person perspective, the results could be in many ways similar to what people dislike in ArmA's anims right now: seemingly "sticky" animation switches.

It's not really my field of work, so I don't want to go too far into the technicalities. But the current discussion does seem to steer towards the "woah, awesome solution to all our problems in one package!" mindset, and I figured some personal experiences might make for a more balanced discussion ;)

That's really nice. I'm particularly impressed by the footwork when you make small movements on the stairs.

[sakura_chan 9]

This might be fine for a game with just the player and a few enemies on screen, but I bet it would fall on its ass as soon as you threw a hundred or so units into combat.

[Cookieeater 10]

Well the attraction of this implementation is that it only needs to run client-side. The ragdoll stuff needs some amount of network synchronisation. At least the torso I would say, arm & leg positions can be client-side with no real impact.

Didn't Battlefield 2 implement this?

I've actually done a graduation project in Unity using this exact locomotion system. I'm no animator myself, but here's a few things I noticed:

• Yes, the system does greatly reduce the number of required animations for a full movement set. Especially the number of transition animations can be significantly reduced (or left out entirely for simple animation sets).
  
• Yes, the system does allow for relatively smooth interpolation, going from one animation state to the next.
  
• No, the system does not necessarily provide the instant movement feedback people here seem to expect. Some level of "clunkiness" often remains.
  

Give the implementation a try in this minigame.

As you'll see, similar to the locomotion system used in GTA IV and RDR, your character has some degree of inertia when e.g. shifting direction. In third person games like these, this isn't much of a problem due to the visual feedback you get on said inertia. But if you apply it to a first person perspective, the results could be in many ways similar to what people dislike in ArmA's anims right now: seemingly "sticky" animation switches.

It's not really my field of work, so I don't want to go too far into the technicalities. But the current discussion does seem to steer towards the "woah, awesome solution to all our problems in one package!" mindset, and I figured some personal experiences might make for a more balanced discussion ;)

I think what people mean by sticky animation switches are that the transition from one animation to the next is way too fast. You have people sprinting at max speed and then once you stop moving forward, your character instantly stops, standing straight up, and all the momentum becomes null. It looks incredibly robotic, ridiculous, and unrealistic. In Half Life 2 and Counter strike Source, you can see the characters deaccelerating and making additional but small steps that really aid the transition between running and stopping and making it look much more realistic.

ArmA II

oYPH0OSkAt4

Notice how when he stops moving, he instantly stops, and loses his momentum. It looks pretty robotic.

Counter Strike Source

EZx2tssNEoI

When the characters stop moving, they deaccelerate for around 0.2 seconds and it looks much more realistic and "fluid".

Edited November 12, 2010 by Cookieeater

[froggyluv 2130]

This might be fine for a game with just the player and a few enemies on screen, but I bet it would fall on its ass as soon as you threw a hundred or so units into combat.

I was wondering about this as well. Would love to see a demo with an Arma sized force going at it :)

[Sith 0]

I think what people mean by sticky animation switches are that the transition from one animation to the next is way too fast. You have people sprinting at max speed and then once you stop moving forward, your character instantly stops, standing straight up, and all the momentum becomes null. It looks incredibly robotic, ridiculous, and unrealistic. In Half Life 2 and Counter strike Source, you can see the characters deaccelerating and making additional but small steps that really aid the transition between running and stopping and making it look much more realistic.

That's not the complaint I've commonly heard over the past years. The main difference between movement control in Half-Life and ArmA is the fact that the former lets you control a floating camera, which drags along an external player mesh performing walking animations roughly matching the camera's pace and direction (resulting in the infamous "ice skating" walk anims when done poorly). This allows you to make very small and rapid adjustments to your character's movement. The "few extra steps" you mentioned is simply a little momentum applied to this floating camera.

ArmA instead allows you to issue "movement orders" directly to the character skeleton, initiating a sidestep animation that moves the camera according to your character's head position, rather than nudging the camera and roughly animating an external mesh in its wake. As a results the character controls may seem "indirect" to people used to controlling a floating camera around the game world, refering to the animation-bound movement increments as "sticky" controls.

My point was that the use of a locomotion system also binds your camera to the dexterity with which your character can displace himself in real-time. The people in this thread hoping for it to allow them "floating camera"-style controls would most likely be disappointed by such consequences.