Page 2 of 2

Re: 4-legged animation testing / rag-dolling

Posted: Fri Dec 25, 2009 11:46 pm
by Dragonlord
Draconisaurus wrote:Pretty nice o-0 If you'd give him more blur in that image, he might be almost impossible to catch.
Yeah, it's not the most optimal photo-manip job. Back then I have not been that skilled yet with Blender so I had my troubles. And with the large reworking of the Armature of the dragon I can not render that scene anymore since with different Armature the pose it totally off. :(

Re: 4-legged animation testing / rag-dolling

Posted: Sat Dec 26, 2009 9:29 pm
by Dragonlord
So, here we are. Fixed some code to deal better with spring constraints. So here is the video to watch.
Dragon Rag-Doll Test 2 ( with spring-wings )

And for those not wanting to read there the description quoted in here:
Second test run of the dragon rag-doll this time fully rigged including the wings. For this I used additional spring constraints along the wing web to simulate the behavior of wing web without using cloth physics yet. Unfolded wing web is under tension. Hence if relaxing wings tend to retract to a non-tension state. That's what the springs are doing here. First part shows again the setup: bones=blue, collision-shapes=orange, constraints=purple. First test is using 0-gravity. The second test is using 4-gravity ( a bit less than half earth gravity ). This avoid tunneling issues with the Bullet Physics library ( until they bring back CCD... or until I figured out a way to cheat Bullet :P ) and helps with the slow-down induced by the capture software. In this test the spring-wing physics can be seen rather well. Last back of test the typical tumble-down-stairs scenario including first-person camera view.

Usually players are not supposed to see themselves ragdolled but in this game this is going to be a bit different. Players have a hidden stamina bar so it's possible they can get knocked unconscious for a short time as well as being on the receiving end of tranquilizer devices. Getting knocked out is not going to end the game and even triggers additional story parts but more I'll not say here :P . Now I just have to put up some transition animations and putting this all in-game for a test-run. Should be fun to play around with ^.=.^

Re: 4-legged animation testing / rag-dolling

Posted: Sun Dec 27, 2009 6:06 am
by Draconisaurus
:lol: Not perfect, but that's pretty sweet. 8) Betcha it'll be hell to make actual animations for, though.
Usually players are not supposed to see themselves ragdolled but in this game this is going to be a bit different.
Hmmmm... Halo 1/2/3 all do this I'm pretty sure :lol: One of the things that makes dying so fun. You get to see your body go through holy hell while you wait to respawn (when your body stops moving, the camera does switch to a third-person view of another random player).

Re: 4-legged animation testing / rag-dolling

Posted: Sun Dec 27, 2009 6:42 am
by RexHunter99
Uhh no, what Dragonlord has done is applied ragdoll to parts of the mesh, for instance, the feet, which then make sure they are orientated on the terrain/other bodies correctly.

Halo 3 does this, but you'll only notice it in the legs/feet of characters, when an actor dies, their biped becomes a ragdoll body entirely, where every portion of the biped becomes a tangible, moving entity that reacts with other parts of itself and with the environment.

Zelda Twilight Princess does this as well (very impressive) and a whole bunch of other games.

The term 'ragdoll' has become too widely abused and associated with a dead character that luls about on the floor.

Re: 4-legged animation testing / rag-dolling

Posted: Sun Dec 27, 2009 11:49 am
by Dragonlord
Not correct actually. I'm using procedural animation techniques ( inverse kinematics in general ) to align the actor with the ground as well as aligning feet with the ground as far as possible ( rules prevent over-bending if the ground is too steep ). This has nothing to do with rag-dolling at all but with procedural animation techniques. In the last video though what I did there is actual rag-dolling. This term is often misused but I use it here in the correct sense for what I'm doing there. The actor is approximated by a collision object linked by constraints. In my case I use analytical shapes as collision shapes as they are faster and less edgy ( half-life 2 for example uses triangle meshes as collision shapes and trespasser uses only boxes ). Rag-dolling is therefore the process of attaching an actor to such a collision object from a given point in time and optionally releasing it again. The term rag-doll comes from the fact that usually no motors are applied to such a collision object for speed reason which makes such a collision object behave like an actual rag-doll in the real world. With a real body muscles get cramped applying a retention force to the attached limbs. Using the spring constraints I sort of simulated this retention force on the wings. It would be possible to do the same on the other body constraints but such retention force setups are quite tricky to get working.

So in conclusion:
- The body/feet to ground alignment is done using procedural animation not physics
- The tumbling down is done using non-retention rag-doll physics
- The wings are done using retention rag-doll physics ( using spring constraints )

Now about the "for dead people only" comment. Using rag-doll only for dying actors has become the default in games today but is also boring. This is why I stated in the comment that I go beyond this. Players can get exhausted, knocked unconscious or tranquilized in which case the rag-doll is used. After some time though they come back to normal depending on the cause ( exhaustion is short to middle duration whereas the other two require a second player actor to take action ). Furthermore my system is generic. Hence you can create rigs with complex animation and physics properties and use them for anything not just models ( for example for AI, puzzles and more ). So I'm doing quite a few things different then what is going on in other games. I'm actually using the rag-doll physics for game mechanics not just dying-actor eye candy. I really have to stress this part as some seem to misunderstand what's going on here.

Re: 4-legged animation testing / rag-dolling

Posted: Wed Dec 30, 2009 1:44 pm
by Dragonlord
Put all together in-game ( same video, pick the one playing better for you ): moddb video, youtube video . Details about what you can see there can be found here.

Re: 4-legged animation testing / rag-dolling

Posted: Thu Dec 31, 2009 7:56 pm
by Draconisaurus
Whoa dude! True dragon's-eye-view! :| Nice. I've never seen that visually represented before. And yeah the ragdoll etc. is good :P

Re: 4-legged animation testing / rag-dolling

Posted: Thu Dec 31, 2009 8:26 pm
by Dragonlord
Well... most people are not really aware yet of what extraordinary and unconventional stuff is included in my game. "Dragon View" is just one of them with "Dynamic Investigation", "Realistic Interaction", "Cinematic Combat" and "Knowledge System" being some of the more interesting things :D

Re: 4-legged animation testing / rag-dolling

Posted: Thu Dec 31, 2009 9:34 pm
by Anna
Wooah, sounds nice! :D. Really nice, actually.