If the server runs on an faster step rate (for example, in minetest.conf, setting dedicated_server_step = 0.05 instead of dedicated_server_step = 0.1, which doubles the server step rate from 10 Hz to 20 Hz), other player movement will look more responsive. This will increase the server's CPU and bandwidth usage, but it should be acceptable if the server isn't running on very old hardware or on a low-end VPS.ThatOneChris wrote:One thing I'm actually hoping for in 0.5 or later versions is smoother player movement in multiplayer. It looks kinda janky at times, and is quite annoying if you're in a PvP situation or are just trying to keep track of where a friend is. I've seen it in lots of servers, so I'm not sure if it's a server problem or just the game, but I hope it's implemented if it is a problem with the game.
We could perhaps default dedicated_server_step to 0.05, but it has to be tested on large servers first.
Actually, Minetest only interpolates other player positions (and doesn't extrapolate them). Extrapolating networked entity movement is not a very common practice, but it can be done to decrease the latency between the "smoothed" entity position and the "actual" entity position at a given time. When interpolating positions, there's always a tradeoff between low latency and smoothness – if you want both, you need to send entity positions more often, which is what lowering dedicated_server_step does.rubenwardy wrote:The issue is that Minetest uses extrapolation not interpolation IIRC.