Could probably cut the time in half too with some small changes to the shaders. Some other issues need to be ironed out then I need to update mesh re/building and implement temperature followed by updating the overall sim logic and cell data model
Slowly moving the fluid sim to GPU and dayum it's so much faster! Full active sim area (3x3 world chunks) with 4 sim steps per run is only ~0.33 ms compared to hitting +10ms with a single full world chunk active and a single sim step #ue5#indiegamedev#indiegame
@Pirat_Nation Reminds me of Folding@Home. Biggest issue though is paying enough, for example the UK has ridiculous energy prices compared to many other nations
Very quick implementation of conversion of water to ice using a "spell". Water doesn't support flowing over it so I think the next step is to rebuild a few things so water can flow and adding temperature for dynamic phase changing #ue5#indiegame#indiegamedev
@WizardRobin8046 On normal actors yes though the freeze I'm talking about is more turning the water to ice which will be a simple case of getting the effect area sim data, seeing what's there and changing the simulation element
Narrowing down the feel for the gameplay, I quite like this so far. Also added basic UI, grenades and direct interaction with animated stamps, gonna add some basic magic casting next, a basic freeze I think #UE5#indiegame#indiedev
@solana_bandit Ahhh, many thanks! Too many variations of shallow water simulations😅 But this looks great and the method sounds like it fits my current setup. Would you spawn particles if flow rate gets high enough, to kind of emulate spray?
@solana_bandit Is there a any specific name for the method? I'm currently using a method that just stores height and uses vertical velocity for spreading waves but my current implementation is slow and doesn't work as well as I'd like
And this is what I mean by a slow fluid sim - 10ms - despite parallelisation on the CPU, as more chunks process water, the time it uses grows massively. I either need to find a much better algorithm or push fluid sim to GPU which I think I'll try first
Added the start of some actual gameplay, a bit of shooting, some particles and fluid interaction. Changed camera from ortho to persp so you can still see depth. Def going to need a different approach for fluid, it's too slow #UE5#IndieGameDev#gamedev
Updated the visuals, now uses a rough implementation of voxels (data is a 2D heightfield), a bit slow when the fluid starts to spread so I might need either a dedicated plugin or different solution #UE5#IndieGameDev#gamedev