[project1dev] Re: octree optimzation for collision detection

  • From: Alan Wolfe <alan.wolfe@xxxxxxxxx>
  • To: project1dev@xxxxxxxxxxxxx
  • Date: Thu, 14 May 2009 23:27:11 -0700

oh nuts

hey kent remember that bug where objects moving were detecting themselves
for collisions and "stepping up" onto themselves so they flew off into the
sky?

same bug here the oct tree was forgetting which model it was supposed to
ignore (ie the one bein moved) hehe..

bugs are anoyin sometimes grrr....

On Thu, May 14, 2009 at 10:28 PM, Alan Wolfe <alan.wolfe@xxxxxxxxx> wrote:

> its not your fault.  If the things werent there, it would still be a bug,
> but we just wouldnt know about it.
>
> and later on when we found it, i wouldnt know whats causing it.
>
> no worries!
>
>
> On Thu, May 14, 2009 at 10:27 PM, Kent Petersen <kentkmp@xxxxxxxxx> wrote:
>
>> ah bummer. Sorry to make things complicated for ya
>>
>>
>> On Thu, May 14, 2009 at 9:41 PM, Alan Wolfe <alan.wolfe@xxxxxxxxx> wrote:
>>
>>> So i have it working on my comp
>>>
>>> On the cavemap, calling Model_Move (the function that moves models around
>>> with collision detection) used to take .24 milliseconds on average.
>>>
>>> Now, with the quad tree it takes .09 milliseconds on average so is about
>>> 1/3 as costly to move objects around in the world with collision detection
>>> (yay).
>>>
>>> Also with this optomization on, locally here on my machine i'm now
>>> getting 100-120 FPS where i used to get about a solid 65  (these numbers are
>>> with vsync off, which is the "real fps", vsync caps your FPS to your monitor
>>> refresh rate)
>>>
>>> So i will check it in real soon and hopefully it'll speed things up!
>>>
>>> The only problem is for some reason, the new peices kent put in aren't
>>> falling like they should anymore!
>>>
>>> i dont know what's up :/
>>>
>>> so i gotta fix that before i check in, hopefully it'll be an easy fix (:
>>>
>>
>>
>

Other related posts: