Coding Multiplayer IO Pac-Man in One Week (Timelapse)

[Loonride]

 

[Krisztian Varga]

You've done a great job, especially in a week!

I'have a question regarding player movement prediction in multiplayer: when you use the velocity and direction and let's say a player or enemy is heading towards a wall. The distribution of network packages is uneven expectedly. Have you not run into an issue, where a package was in late, that describes a player/enemy turning before hitting the wall, therefore client predicting that it went through the wall, using existing velocity and direction for the prediction. ..I've personally run into this a lot and had to abandon the approach. Is there anything that I'm missing here?

[Loonride]

On 7/18/2018 at 9:27 PM, Krisztian Varga said:

You've done a great job, especially in a week!

I'have a question regarding player movement prediction in multiplayer: when you use the velocity and direction and let's say a player or enemy is heading towards a wall. The distribution of network packages is uneven expectedly. Have you not run into an issue, where a package was in late, that describes a player/enemy turning before hitting the wall, therefore client predicting that it went through the wall, using existing velocity and direction for the prediction. ..I've personally run into this a lot and had to abandon the approach. Is there anything that I'm missing here?

Here is how I solve this. I adjust the predicted position of an opponent when it runs into a wall in the same way that I adjust your own players position without going through walls. I am doing collision checking for the opponent predicted positions as well for your own player (but only if in visible area). I always try to keep the predicted position within a certain range of the actual position. If an opponent's predicted position starts getting out of that range, I move it the minimum distance possible to get it back into that range, account for any collision. So the most choppy parts are when an opponent changes directions, but choppy movement is minimized by having this acceptable range of inaccuracy.