[HN Gopher] Real-time Trumpet Simulation C++/Vulkan [video]
___________________________________________________________________
Real-time Trumpet Simulation C++/Vulkan [video]
Author : tsujp
Score : 56 points
Date : 2023-09-07 13:00 UTC (2 days ago)
(HTM) web link (www.youtube.com)
(TXT) w3m dump (www.youtube.com)
| klysm wrote:
| This guy is going phenomenal work and publishing a good deal
| about it. Highly recommend checking out his engine videos
| tgv wrote:
| There are commercial physical models of instruments. They sound a
| bit better, especially when played using a wind controller. This
| one is for the brass family:
| https://www.youtube.com/watch?v=uRQEXAGGhgw
| dragontamer wrote:
| This isn't about having a good sound like a professional
| trumpet though.
|
| This is about building a 44100 frame per second fluid dynamics
| engine that just so happens to kinda sorta sound like a
| Trumpet.
|
| Then adding a particle system to the simulation and painting
| the particles in Blender to provide a nice visualization of the
| simulator.
|
| ---------
|
| A partial differential equation done cell-by-cell like this is
| called finite element analysis. This is basically the brute
| force methodology top scientists are doing at National Labs
| from my understanding.
|
| Just ya know, scaled down and hobby sized to this crude tube
| simulator (that arguably kinda sorta sounds like a Trumpet)
| thomastjeffery wrote:
| Very cool!
|
| I suspect the main problem here is that the input motion is too
| "perfect". There needs to be more randomness involved, otherwise
| you are just going to get really loud harmonics like you do when
| playing a perfect square wave.
|
| When I play the trumpet, there is a spectrum of pitch tuning I
| can do for each note by buzzing my lips differently; and the
| middle of that spectrum (the correct note) is the easiest to
| perform. Off-pitch notes are usually achieved by placing my lips
| off-center. It would definitely be worth exploring this behavior,
| since an infinitely perfect lip placement does not occur in
| nature.
|
| When I was taught to play the trumpet, one focus (for all wind
| instruments, really) was to always use _more_ airflow than
| necessary, especially when playing quietly. This results in a
| cleaner sound, probably because the lip movement changes less
| suddenly while still altering enough pressure to be loud. With
| that general sense in mind, I suspect the need for more airflow
| is to compensate for relatively square-wave-like pressure waves
| from the lips simulation.
|
| I wonder what sounds you could get if you recorded yourself
| buzzing your lips into your real mouthpiece, then using _that_ as
| an input. Of course, that would be tricky, because blowing into a
| mouthpiece feels very different than blowing into a trumpet.
| Could this model potentially be simulated in reverse to find a
| good input?
___________________________________________________________________
(page generated 2023-09-09 23:02 UTC)