Timeline for Resolving a collision with forces
Current License: CC BY-SA 3.0
21 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Sep 17, 2013 at 14:39 | vote | accept | Vittorio Romeo | ||
| S Feb 10, 2013 at 22:51 | history | bounty ended | Vittorio Romeo | ||
| S Feb 10, 2013 at 22:51 | history | notice removed | Vittorio Romeo | ||
| Feb 6, 2013 at 20:13 | answer | added | Leftium | timeline score: 2 | |
| Feb 6, 2013 at 9:48 | comment | added | teodron | Well, Disney/Pixar might be doing it like you want, but there's a catch - they have progressive penalty forces - just watch this youtube.com/watch?v=00DXNmjIbsM | |
| Feb 6, 2013 at 7:20 | answer | added | Leftium | timeline score: 15 | |
| Feb 4, 2013 at 21:14 | answer | added | Erin Catto | timeline score: 17 | |
| Feb 4, 2013 at 19:15 | comment | added | dennmat | I believe this is why a lot of physics engines run in iterations, correct me if I am wrong, does yours run in x amount of iterations recalculating positions as new forces/results of the first iteration change the expected result? | |
| S Feb 4, 2013 at 16:23 | history | bounty started | Vittorio Romeo | ||
| S Feb 4, 2013 at 16:23 | history | notice added | Vittorio Romeo | Draw attention | |
| Feb 3, 2013 at 15:38 | comment | added | Vittorio Romeo | @dreta: I tried many things, I tried setting the velocity to the penetration vector and stuff like that. But it never worked properly. Setting the velocity to the penetration vector fixes the "fall-through" bug, but bodies stop when they hit each other. I also tried adding or subtracting from velocity. How would I implement the "impulse based solution" you're talking about? | |
| Feb 3, 2013 at 15:24 | comment | added | dreta | @TravisG Oh, alright i see the issue. | |
| Feb 3, 2013 at 15:18 | comment | added | TravisG | Still, if you want to have simple separation of objects without having everything bounce around after a collision, somehow this must be dealt with (e.g. in the first frame after loading the scene, where things may overlap without having actually moved beforehand). Either way, I think the answer to this question doesn't have anything to do with physics, but instead with logics (code implementation). | |
| Feb 3, 2013 at 15:14 | comment | added | TravisG | @dreta his assumptions are fine. He pointed out that his mass for all objects is simply "1" for now, which makes his code sections valid. By the way, even though Box2D may deal with velocities directly, it somehow must deal with the same problem. If instead of applying a force, Box2D applies an impulse, it somehow must still deal with the fact that the impulse doesn't just go away once the objects are separated. Although, it's possible that it does in fact not deal with this at all and just lets the objects keep their energy (it would be like this in the real world after all) | |
| Feb 3, 2013 at 15:11 | comment | added | dreta | Personally, i'd suggest picking up a book on physics engines, at least read the first few chapters on Newtonian physics. Your assumptions are incorrect and trying to reply to this question would mean having to teach you basics of physics while trying to explain high level algorithms for solving collisions. | |
| Feb 3, 2013 at 15:06 | comment | added | dreta | Force does not equal acceleration, first of all. You need mass to calculate acceleration. If you're modifying positions to stop the two bodies from penetrating, you should use mass as well and move both bodies based on it. Applying a force equal to the penetration vector has no merit. Box2D is impulse based, it works on velocities directly, it may not be "correct", but it's good enough. Dealing with velocity changes in an impulse based engine is very simple, so could you specify if you definitively want a force based solution, or is the much simpler impulse based solution good enough. | |
| Feb 3, 2013 at 13:39 | history | edited | Vittorio Romeo | CC BY-SA 3.0 |
added 83 characters in body
|
| Feb 3, 2013 at 13:38 | comment | added | Vittorio Romeo | @TravisG: not yet, unfortunately. I'll add a bounty tomorrow if I don't get any reply. | |
| Feb 3, 2013 at 12:57 | comment | added | TravisG | I'm interested in this. Have you come up with a solution yet? | |
| Feb 3, 2013 at 4:07 | history | tweeted | twitter.com/#!/StackGameDev/status/297919237899550720 | ||
| Feb 2, 2013 at 16:13 | history | asked | Vittorio Romeo | CC BY-SA 3.0 |