It's interesting how it has unit possession. I'm basically doing that right now with the little drone demo I'm making. But the drone has both control and view latency. It's using server side physics, so I might actually put the physics loop in C. Then it stores game state in a ring buffer so I can access past states, and the server emits the correct state for each client in a loop. Then I have a babylon based demo that renders that state. It's not optimized at all at any layer. I just need to rewrite the physics and I should have the demo done.
Cool! That sounds dope - adding in multiplayer is always another layer of difficulty! I made a quick script to display my FPS and stuff, and with the lights it was around 15 (many light sources, using differed lighting.. so on) So i removed those and now am up to 30, which sounds bad, but its not, cuz the script that displays that information cuts it down by 20 as it is itself not optimized. So now im writing an approach to combine all walls strips into a single mesh, so that should help, and im debating on if i do the same with floor tiles. There is also an issue with colliders, and i think i may be able to shave off even more by editing those to be more performant. I also have to redo my walls/floor models due to height issues, which is fine, but since im there that means i can cut out some quads that the user will never see, or make a compound model to allow for voxelization of that object - like removing the sides of walls and only having a side "turn on" if its jutting into a room. So, simple optimizations like that can save as well. So i dunno, thats what im doing today. You should send some screen shots of use screen 2 gif to share what you got! I'm super curious and cant wait to lend some modeling time to it!
Log in
It's interesting how it has unit possession. I'm basically doing that right now with the little drone demo I'm making. But the drone has both control and view latency. It's using server side physics, so I might actually put the physics loop in C. Then it stores game state in a ring buffer so I can access past states, and the server emits the correct state for each client in a loop. Then I have a babylon based demo that renders that state. It's not optimized at all at any layer. I just need to rewrite the physics and I should have the demo done.
Cool! That sounds dope - adding in multiplayer is always another layer of difficulty! I made a quick script to display my FPS and stuff, and with the lights it was around 15 (many light sources, using differed lighting.. so on) So i removed those and now am up to 30, which sounds bad, but its not, cuz the script that displays that information cuts it down by 20 as it is itself not optimized. So now im writing an approach to combine all walls strips into a single mesh, so that should help, and im debating on if i do the same with floor tiles. There is also an issue with colliders, and i think i may be able to shave off even more by editing those to be more performant. I also have to redo my walls/floor models due to height issues, which is fine, but since im there that means i can cut out some quads that the user will never see, or make a compound model to allow for voxelization of that object - like removing the sides of walls and only having a side "turn on" if its jutting into a room. So, simple optimizations like that can save as well. So i dunno, thats what im doing today. You should send some screen shots of use screen 2 gif to share what you got! I'm super curious and cant wait to lend some modeling time to it!