You can always EnableDrag(false) on your object to isolate the air
density/drag from the motion of the object. If doing this fails to fix
the problem, then it's not an air density issue.
Other factors that could be different:
damping, rotdamping: Make sure these are the same on your client and
server objects (hardcode them to something constant as a test).
The other factor that is almost certainly different unless you've
changed it in your mod:
The client and server apply user input at different points relative to
physics simulation. The server runs at a fixed framerate which is
always an integer multiple of the physics simulation rate. This can all
be configured by your mod, but the default behavior is that ticks are
15ms for both server logic and vphysics simulation. That means that all
user input is applied at the beginning of a tick.
The client runs at a variable framerate - as fast as it can render
frames. The vphysics simulation still runs at a fixed interval (15ms).
It intepolates between fixed frames to match the client clock. This
means that any impulses applied do not take effect until the next
simulation frame is reached. This will probably complicate your
client-side prediction or at least introduce additional error.
Jay
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of
Maurino Berry
Sent: Tuesday, September 12, 2006 9:29 AM
To: hlcoders@list.valvesoftware.com
Subject: Re: [hlcoders] Predicted VPhysics entity
It is a networked entity. They use a shared CreateVPhysics
method to initialize their physics objects to be identical.
If the clientside representation of an entity has its own
vphysics object it will not get forced changes from the
server (which is what we want) but I can use
GetAbsAngles/Origin to get the proper server position. The
idea was to eventually use sendproxies to write out
appropriate server physics data (inertia's etc) to correct
the client vphysics object.
Within the void CGameMovement::PlayerMove( void ) method I
call a simulate method on the players controlled entity, and
look at the most recent usercmd with
#ifdef GAME_DLL
cmd = pPlayer-GetLastUserCommand();
#else
cmd = pPlayer-GetCurrentUserCommand();
#endif
and then based on keys down (like IN_FORWARD) etc I apply
thrust, and use the mouse deltas to figure angular rotation,
but the point is even with a single keypress (I'm echoing the
amount of times this method is
called) the client drifts longer than the server, so repeated
keypresses of left and right end up throwing the client
object way out of sync with the servers.
On 9/12/06, Robbie Groenewoudt [EMAIL PROTECTED] wrote:
--
[ Picked text/plain from multipart/alternative ] The client
and server
have different times and ticks so it's quite possible the 2 are
different. Some more code would help.
Also, you should make it an networked entity so the client
predicts it
but receives the right position from the server.
On 9/12/06, Maurino Berry [EMAIL PROTECTED] wrote:
I'm attempting to simulate physics for an entity on the client as
well as the server (a type of vehicle) Everything works
fine except
what feels like the 'drag' is different on the client/server,
they're set to ignore gravity so they are hovering and a single
boost with ApplyForceCenter/Offset results in different positions
and I can't seem to figure out why, it's like the client has less
air density almost, but I've made sure it's the same. Is the
simulate rate of havok different between the client/server? or is
there anything I should know about regarding this? Thanks.
___
To unsubscribe, edit your list preferences, or view the list
archives, please visit:
http://list.valvesoftware.com/mailman/listinfo/hlcoders
--
___
To unsubscribe, edit your list preferences, or view the
list archives, please visit:
http://list.valvesoftware.com/mailman/listinfo/hlcoders
___
To unsubscribe, edit your list preferences, or view the list
archives, please visit:
http://list.valvesoftware.com/mailman/listinfo/hlcoders
___
To unsubscribe, edit your list preferences, or view the list archives, please
visit:
http://list.valvesoftware.com/mailman/listinfo/hlcoders
