The sound is crackling, maybe it is only for me. But maybe because you were recording it from the pc itself, so the heavy rendering interfered with the recording.
Yeah, I think it is just the heavy rendering (4K, 6-7M cubes). I might lower the amount of material simulated in future vids, use lower cost things like distance fields to fill out more space, and get back to 60-120 fps play.
@@GrantKot it should be possible to record from a separate machine by routing the output from the gpu via cable (display port/hdmi) to some recording device. I know it is the way professional content makers do it, but it shouldn't be too expensive. Main machine for computations, and a separate machine for recoding, to put it simply. And you will not have to limit the main machine's resources that way and can reach its maximum potential
Hey just wondering, how do you plan on using this system for flowing or large bodies of water? Was thinking that a some specific depth the simulation would edge more towards a LOD approach once full simulation isnt necessary?
There is the option of adding adaptive grid resolution as well. Or for an even larger world maybe I would just do something grid based like tracking volume and velocity of fluid per cell. But also, the simulation is pretty large at this point. This is 16x the size of the previous 2 demos, I think I just need to start adding things like terrain and laying out the world to give a better sense of scale.
Yeah it shows up in some form. Generally I am trying to stay on the accurate side of things and it should show when I add more obstacles, materials, objects.
Игры
![Real-time Trumpet Simulation [C++/Vulkan] [WARNING: Flashing Lights]](/img/1.gif)
Actual particle count is whatever shows up on the counter + 4 million for the structures system.
man that's so cool, the possibilities are insane !
GTA 10 early development footage
The sound is crackling, maybe it is only for me. But maybe because you were recording it from the pc itself, so the heavy rendering interfered with the recording.
Yeah, I think it is just the heavy rendering (4K, 6-7M cubes). I might lower the amount of material simulated in future vids, use lower cost things like distance fields to fill out more space, and get back to 60-120 fps play.
@@GrantKot it should be possible to record from a separate machine by routing the output from the gpu via cable (display port/hdmi) to some recording device. I know it is the way professional content makers do it, but it shouldn't be too expensive.
Main machine for computations, and a separate machine for recoding, to put it simply. And you will not have to limit the main machine's resources that way and can reach its maximum potential
Hey just wondering, how do you plan on using this system for flowing or large bodies of water?
Was thinking that a some specific depth the simulation would edge more towards a LOD approach once full simulation isnt necessary?
There is the option of adding adaptive grid resolution as well. Or for an even larger world maybe I would just do something grid based like tracking volume and velocity of fluid per cell. But also, the simulation is pretty large at this point. This is 16x the size of the previous 2 demos, I think I just need to start adding things like terrain and laying out the world to give a better sense of scale.
Are the Navier-stokes equations (or simplifying approximations) implemented somewhere in your code for simulation?
Yeah it shows up in some form. Generally I am trying to stay on the accurate side of things and it should show when I add more obstacles, materials, objects.