Timeline for Ball not bouncing infinitely with perfect elastic collision settings
Current License: CC BY-SA 4.0
7 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 21 at 22:49 | answer | added | agone | timeline score: 0 | |
| Aug 21 at 22:29 | comment | added | agone | @DMGregory The Physics Simulation sub-step value as described by dev.epicgames.com/documentation/en-us/unreal-engine/… can exacerbate the precision errors exponentially from single step analysis. | |
| Aug 17 at 10:32 | comment | added | B.G. | @DMGregory Thanks, the ball is losing height much faster than what I'd expect from just floating-point precision errors - around 1% per bounce, which seems too high. I'm building a 2D simulation where users can place objects and constraints like springs, joints, etc., and then observe interactions. That's why I prefer to expose these features from engine to the user instead of implementing everything from scratch. I initially tried this in Godot 4, and the physics worked well, but I need to use Unreal Engine due to other features. | |
| Aug 17 at 9:23 | comment | added | DMGregory♦ | Generally game.physics engines prioritize speed of processing over perfect physical simulation, and they're limited by the precision of their number formats, which will lead to compounding errors from rounding. If you want perfect elastic collision like in Pong or Breakout, a physics engine might not be the right tool for the job, and you might want to calculate your own perfectly elastic bounce trajectories (to within numeric precision) using things like spherecasts to detect the next collision plane. We can help if you can show what kind of environment this ball is bouncing in. | |
| Aug 17 at 8:48 | history | edited | B.G. |
edited tags
|
|
| S Aug 17 at 8:28 | review | First questions | |||
| Aug 18 at 15:13 | |||||
| S Aug 17 at 8:28 | history | asked | B.G. | CC BY-SA 4.0 |