On 6/4/05, Michael K�ser <[EMAIL PROTECTED]> wrote: > Please read the "Source Multiplayer Networking" Text to fully > understand the following explanation: > http://www.valve-erc.com/srcsdk/general/multiplayer_networking.html
Ditto. > As we all know, the models on the client are delayed by cl_interp > seconds to smooth gameplay. This is not entirely correct - the client models are delayed by a MAXIMUM of cl_interp seconds. > The lag compensation code uses the > following formula to move the hitbox back in time to the correct > position: > Command Execution Time = Current Server Time - Client Latency - Client > View Interpolation Re-read, this is 'Current Client View Interpolation' not Max Client View Interpolation > If a player sets cl_interpolate to 0, the models don't get delayed on > his system, but he can still set his cl_interp value to anything. But what is the running value of interp? (i.e. not the max, but the current) > The problem is: The server seems to ignore that the client has > disabled interpolation and still moves the hitbox back in time using > the formula above which makes the hitboxes move cl_interp seconds > _behind_ the model. Have you tried cl_lagcomp_errorcheck? > A lot of players abuse this bug (like cl_interpolate 0 / cl_interp > 0.05) to move the hitboxes behind the models, which makes hitting a > moving target ridiculously easy (just shoot at where the model was). I doubt this is easier than shooting at a target you can see, especially if the target has more intelligence than to run in straight lines. > This leads to unfair gameplay and a lot of complaining on public > servers from people who don't use this "cheat". 1. Turning off cl_interpolate on the client is a disadvantage at the client, as it requires mousing at ticrate speeds to deal with hitbox shifting. 2. The server is repsonsible for hitbox registration and this will always be flawed if you run a different configuration to the server (i.e. your above comment about the hitboxes not lieing on the model is correct, but this is not an advantage). 3. If someone has sufficient accuracy to hit and invisible hitbox behind a player model, then they should have sufficient accuracy to hit a proper target also. 4. Whilst there may be some 'advantage' to pre-visible models, the reaction time, sight time, and visible hitbox time are the same - the required reaction time has not changed, merely it starts and ends sooner. > A fix for the engine would be to set the "Client View Interpolation" > value to exactly 0 if the player users cl_interpolate 0, or, if this > isn't possible, to cheat protect cl_interpolate (like cl_extrapolate > already is). Alfred? I can see no justifiable need for this. > Michael > > > _______________________________________________ > To unsubscribe, edit your list preferences, or view the list archives, please > visit: > http://list.valvesoftware.com/mailman/listinfo/hlds_linux > _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
