Client performance rant

[Granger]

viewtopic.php?f=48&t=57609 - somewhere in that topic. Would need resources (and most likely a server that can be distributed), but doable.
And would be extremely cool.

The thing is - it does not.
Right now server has a LoS implemented. It is about 50-60 tiles radius.
So yes - server already has:
for(every object in 4 surrounding grids){
if(player.distance(object) < 60 tiles){
renderlist.add(object)
}}

If we pass any number instead of constant '60' we can achieve any pitch-darkness. With modifications I've listed before - seeing player behind a wall and having houses in same dimension.

The NOX game style is way more complicated because it is based on direction player looks. So we either have '360-degree los mod' possiblity or overload server CPU with player direction vs visibility checks.

Render range (as implemented) and Line Of Sight are vastly different mechanics, with the latter you can't look through walls.

[ekzarh]

Render range (as implemented) and Line Of Sight are vastly different mechanics, with the latter you can't look through walls.

True. But for fixed-shape structures (like inside/outside houses) you can get illusion of line of sight without actually doing what is needed for it.

Haven't seen 'real' line of sight in mmo games with server checks yet Does anyone have any examples?

[xdragonlord18]

i no that javascript has already been talked about and shot down but i figured it wouldnt hurt to mention that webassembly is coming along nicely and is something that loftar may want to watch as it could solve the cross platform c++ problem pretty nicely its currently on by default in current versions of chrome safari opera edge and firefox and has cross browser consensus on the w3c standard so it should behave in a standard way across all browsers and platforms unlike asm.js

[Gigglepuddy]

javascript cannot use hardware rendering. it does not happen. thats why you cant utilize the GPU of any users, because javas limitations require everything to be ran through software, i.e. the CPU. it claims to support "hardware acceleration" but that does not mean GPU acceleration, hence why no game or program ever using java has been able to actually use the GPU for more than the end render to monitor.

[Granger]

javascript cannot use hardware rendering. it does not happen. thats why you cant utilize the GPU of any users, because javas limitations require everything to be ran through software, i.e. the CPU. it claims to support "hardware acceleration" but that does not mean GPU acceleration, hence why no game or program ever using java has been able to actually use the GPU for more than the end render to monitor.

Additionally to confusing java with javascript you're also wrong as both can call into OpenGL that then makes the GPU render the graphics, including shaders.

Javascript in the browser runninc GPU accelerated graphics, one example: http://www.playkeepout.com/

[Granger]

On further thought:

[*]As a corollary, it would be nice if such a system could be formulated so that changes can be effected in parallel from the rendering loop itself, so that rendering wouldn't have to be blocked by objects popping in and out, leading to less stuttering.

You could 'double buffer' the dataset for the seldom-changing static objects (ground and non-moving stuff on it):
- one owned by the render thread and actively being fed into/used by the GPU
- one owned by the update thread that builds the static world with new data delivered by the server or, when complete, free to be taken over by the render thread
- possibly another one for the update thread to work on while the one before is still waiting to be claimed by the render thread

The render thread could happily render frames as fast as the GPU can (would only need to update the list of dynamic, moving objects - players, mobs, animation states like cycling through meshes, object rotation like the windmill, texture offset animation for fires and such) and would only have to check once per frame if a new static data set is ready for use (in case it isn't it would simply reuse the data from the last frame as the world hasn't changed). The update thread could also run freely and release the new state to the renderer when the network input queue is empty (for long enough) and/or a certain timespan has passed or a certain amount of new objects has been processed (to not stall the world update when the scenery changes massively, eg. passing through portals) and the prior completed dataset had been claimed by the renderer.

[linkfanpc]

I REALLY don't wanna sound antsy, but if i could ask, how is everything going along with the re-write? I'm super excited to be able to finally play Haven again.

[jordancoles]

How is the client work looking now Loftar?

[svino]

How is the client work looking now Loftar?

Leaked footage from loftars new test client:

[iamahh]

dis legit