Sound wave simulation

[@Alexander_Sannikov]

Please read the description if you're shouting "WTF IS THIS" at your screen after watching this thing. I always try to put some relevant information there, but it often ends up unnoticed.

[@mohammad8617]

That's cool!!! What is this software?

[@Alexander_Sannikov]

I wrote it

[@vitormoraes8320]

What solver are you using for the simulation?

[@Alexander_Sannikov]

it's a discontinuous Galerkin solver with first-order explicit integrator, equivalent to Godunov's finite volume scheme with HLLC Riemann's split flux solver

[@vitormoraes8320]

@@Alexander_Sannikov Nice. Ive been itching to write some since always but work and studies are in the way. Btw awesome talk at POE con, specially regarding shadows, Im still waiting for a more in depth writing of your ScreenSpace GI technique someday haha

[@kindlin]

@@Alexander_Sannikov I love this reply, but it meant next to nothing to me. Great video. Any further in depth discussions on the how's would be great. I did take a finite element class, or 2, in college and understand first-order explicit integrator and then I'm lost lol.

[@Alexander_Sannikov]

@@kindlin integrator in this example is first-order explicit too, because the system is not stiff and the simplest possible integrator here works just fine. the whole magic is in how you calculate time derivatives. explaining how godunov's finite volume scheme works in a youtube post is not trivial, but in short it's based on assumption that if you define solution within cells, you can construct a numerical scheme if you somehow reconstruct solution inbetween cells, it's called numerical flux. Riemann's problem is a very specific problem of finding solution on the boundary of two media with known properties and HLLC is one of the methods to approximate its solution. honestly this is not trivial to explain in two words, easiest way to learn more about it to just try solving different similar problems of varying complexity, read up how people do it and eventually you build up enough understanding of how it all works.

[@kindlin]

@@Alexander_Sannikov I knew a simple YT comment response would never be enough, but I'm hoping you may consider taking my interest as an impetus to add additional content to your channel related to the how's, and maybe showing some of those examples where as the complexity ramps up, such that the naive solver yields unreasonable results. Honestly, your further explanation added some insight to my understanding, but probably not as much as you would expect. Based on your posts in this thread I could probably research and fiddle with simulations for 10 hours and maybe _actually_ understand half of what you said. I do appreciate the effort tho. You know you truly understand a topic when you can break it down sufficiently to allow the lamen to grasp some of the important principles.

[@sythys_]

Should'nt it move a lot faster?

[@Alexander_Sannikov]

i set sound speed extremely slow for this simulation to see the waves in detail. consider it to be ultra-slow motion.

[@tommos1]

Any cool applications for this in games?

[@Alexander_Sannikov]

There might be. I'm working right now to make an interactive demo/game to play with this thing. It has nothing to do with PoE, however, so only in my spare time.

[@tommos1]

@@Alexander_Sannikov Hey, you never know. One day there might be a sonar league or something.

[@sythys_]

@@tommos1 Imagine being blind only abel to see the echos witch reach you.

[@AngelsNeverFade]

Could this be used to simulate sound reverb based on the environment in games?

[@Alexander_Sannikov]

This particular demo simulates way more than needed to achieve reverb simulation. For example, I simulate changes in temperature and pressure as gas contracts and expands. So if you need just reverb, there are more effective ways to simulate it.

[@AngelsNeverFade]

@@Alexander_Sannikov thats amazing! so cool

[@Florreking]

What's the scale of the box?

[@Alexander_Sannikov]

The size of the box is approximately equal to sound speed times 10 seconds. So if you imagine it's air in realtime then the size of the box is ~10 kilometers. Or you can imagine it's 1 meter but slowed down 10000x.

[@Florreking]

@@Alexander_Sannikov oh ok. I thought maybe it is a very zoomed in view of a pressure wave. The simulatuon confuses me because it looks like the medium moves with the wave but then isnt that bulk mass flow and not a soundwave? No?

[@Alexander_Sannikov]

@@Florreking Actually wavelength of a pressure wave can be arbitrarily long. For example a wind gust is a pressure wave that can be hundreds of meters long. So one way to imagine what's happening is for example think of it as we put a compressed square cloud in the middle of a 10km x 10km square box and look what happens from very far above.The cloud gets dispersed first, then it's reflected off walls of the chamber, etc. It's a very low frequency sound (infrasound) in this analogy, however. You are right that in general sound waves don't usually transfer much mass, but an explosion, for example does produce a wave that can have huge mass flow in it. Usually simulations only calculate 0-mass flow just because it's way easier (its equations are linear).

[@Florreking]

@@Alexander_Sannikov Alexander Sannikov I can flllow that logic, its similar to why i thought it was a "zoomed in view" because if you look at small enought scales youll see the atoms densify and rarefy, thus "mass flow". But its difficult to merge the two, it feels different somehow. Imagine a gas cloud or solid block of atoms getting pushed out in to space. How would one describe this "mass flow" as a wave disturbance? Is it an extreme case where the wavelength becomes infinite, and all particles have the same phase (they are at the same location of the wave otherwise the object would be ripped apart). Its difficult to explain 😅

[@Alexander_Sannikov]

@@Florreking so you brought up two classic cases: one is just mass flow (such is incompressible fluid) and the other one is wave dynamics (such as acoustics). What I wanted to do here is to simulate a process where both effects are significant: so mass is carried and elastically compressed at the same time. And yes, you pointed out correctly that in some sense when a gas expands into vacuum it creates an infinitely long wavelength disturbance.

[@obelisk9251]

cool

[@cirdec115]

#Delirium :P

[@vovanoknet]

Ещё бы технических деталей, а не только картинку

[@Alexander_Sannikov]

у меня есть планы, что можно сделать интересного с этим солвером, я постараюсь более подробно раскрыть в будущем видео, что он делает и зачем

[@angustin6590]

WTF IS THIS ...there is no sound

[@Alexander_Sannikov]

This is true, there is no audible sound. It is visible, however :) If you were to play it as audio, it would have to be sped up around 100 times and would sound like a sharp "clap" or "pop" in a tin can. I actually made a similar audible simulation for a guitar string earlier that can actually be listened to, but that one has no visualization, so there's no point publishing it as a video.

[@kindlin]

@@Alexander_Sannikov Not true! Provide the slowed down visual with a demonstration of the sound at normal speed. You could show multiple speeds and play multiple sounds, and show how it transitions from not actually noise (less than 20 hz) to audible noise (within typical human auditory range). That would be fascinating. It would remind me of the end of that one Daft Punk song where the the beat/buzz is brought all the way down to individual pulses. It's quite an auditory journey.

[@Alexander_Sannikov]

@@kindlin That'd be cool, but this whole sound wave simulation was just a side track from what I had actually planned with this, which is a fully-fledged thermodynamics simulation of heat machines. I have a video demo for it, but it's voiced in russian, so I'm keeping it as unlisted on the channel for now.