HOME | DD
axcho — Engine Prototype 05
Published: 2009-11-25 01:52:58 +0000 UTC; Views: 15623; Favourites: 48; Downloads: 2279
Redirect to original
Description
Engine Prototype 05 - by axcho on November 24, 2009This is my fifth prototype of the Dejeweled physics engine . The engine now supports collisions between circles and capsules. This version is simple, with an inefficient method of finding colliding pairs. And objects can pass through each other if they are traveling too fast. But that will all change soon.
This ragdoll is just like the one from Engine Prototype 01 , but with a ball to play with. Objects can actually collide now!
Nifty, eh?Click the window to start. Then use the arrow keys to move the stick figure around. You can also click and drag the mouse to move the ball, too.
left - rotate left
right - rotate right
up - raise arms
down - spread legs
For slow motion, hold down the space bar. Click the meters to adjust the physics settings.
Let me know if you find anything interesting.
Dejeweled Development Blog
Physics Prototypes by axcho
Related content
Comments: 36
I put on zero gravity then it only moves in place..?
👍: 0 ⏩: 0
press random buttons and hes breakdancing!^^ i love thus game!
👍: 0 ⏩: 0
rlly cool. give him a small ball and a sowrd. that'd be legend! baseball
👍: 0 ⏩: 0
Potentially fun thing.
Hmm it is kind of hard to make produce symmetrical results 
👍: 0 ⏩: 1
If you're talking about the ragdoll, this may be because of some inaccuracies in the spring lengths. If it is impossible to satisfy a set of constraints, there will be extra forces added inadvertently and the ragdoll will spin more to one side. You can see this more clearly if you turn off gravity and friction, and turn up rigidity all the way. Fortunately, this only happens with bad data - it's not inherent to the simulation.
As for tapping buttons randomly, well, that's what practice is for.
👍: 0 ⏩: 1
I did practice
👍: 0 ⏩: 1
Ah, good idea. I'll have to try that sometime.
👍: 0 ⏩: 1
Hehe
👍: 0 ⏩: 1
Hmm, yeah I can see that being useful. I'll try it.
👍: 0 ⏩: 1
Quite interesting when you have control all that way up and bounce turned down and press left to right quickly. Translates to: OMFG THIS IS AWESOME.
Lol this is really cool. Excellent job, but he stays on his side a lot :/
👍: 0 ⏩: 0
i had fun turning off the gravity
poor thing, just wiggling in mid air...
👍: 0 ⏩: 1
If you listen closely you can hear it crying for help...
👍: 0 ⏩: 1
set everything to max and drag the ball towards him
Aww , shaking pelvises
👍: 0 ⏩: 0
Been playing with it the last 5 minutes, mentally going "RAWR! GET BALL MUST ATTACK BALL RAWR! RARARARAWR ATTACK KILL BALL!!!11one!"
:3
👍: 0 ⏩: 1
...and it's great exercise if you do it in real life, too. Just don't hurt yourself.
👍: 0 ⏩: 1
Ah, nice (:
Pretty unstable I can see but you're getting some nice progress, good luck!
👍: 0 ⏩: 1
Thanks. Yeah, the instability is due to the added friction and restitution forces, as well as the discrete sampled collision detection. Once I get continuous raycasted collisions working, I imagine the stability problems should resolve themselves...
Physics engines don't always care too much about imagination, of course, but I'm hopeful.
👍: 0 ⏩: 1
When you implement some more softbody stuff it'll get real fun (:
👍: 0 ⏩: 1
Do you mean "fun" as in fun to play, or "fun" as in me smashing my head against a wall of physics formulas for several hours straight?
I just made a quick fix that should improve the stability of the collision response somewhat...
Let me know if you notice any difference.
👍: 0 ⏩: 1
Fun to play, hehe
Just noticed if you drop the ball on the little guy's head it sort of slides down like it's in butter : P
👍: 0 ⏩: 1
Oh good.
I just made another update, by the way - with new, more realistic friction! Now there's kinetic friction based on normal force, and static friction too, so if you balance the ball on the ragdoll it will stay there for a while before sliding down...
Try it and see.
👍: 0 ⏩: 1
Already looking way better, you're doing some fine work here axcho : P
What's your basic method for finding the collision? Is it just a plain line + the ball with an amplified invisible radius? Or am I just making no sense, hm..
👍: 0 ⏩: 2
By the way, I just made another update, fixing some issues with the static friction and getting rid of the two square roots with an improved kinetic friction algorithm.
I think that should be the last update on this prototype...
👍: 0 ⏩: 1
You're very skilled at applying fast updates, it takes me a long time to fix a lot of the problems I run into
Just noticed the ball seems very willing to balance on the edge of a limb, without sliding off. Not sure if that's intentional but it looks kinda weird : P
👍: 0 ⏩: 1
Ha, thanks.
The balancing is intentional. I set the static friction coefficient (stickiness) pretty high for this prototype, just so I could test that behavior.
👍: 0 ⏩: 1
Ah, I see! In that case, well done haha.
Do you by any chance have some sort of IM info - MSN, Skype etc. I could add you on? I feel like I've asked you this before, worth a shot though
👍: 0 ⏩: 1
Thanks. Yeah, that's the basic method pretty much, based on [link]
But the next phase is to implement continuous collision detection. I know how to do that with two circles [link] but I'm not sure how to do that when a line is involved...
I think I'll look up how to detect the closest point between two line segments.
👍: 0 ⏩: 0