4

Reverb volumes

Galen

October 04, 2017 13:17

None

Follow

I'd like to request the addition of a "reverb volume pseudo-object". Like trigger volumes, these are abstractions to represent where to compute and apply reverb in a scene. Everything outside these hand-placed volumes would not have computed reverb.

Naturally, if a scene contains no reverb volumes and the user chooses to enable reverb in the scene, the whole scene would be computed for reverb, like now.

I would also propose that the reverb for the whole scene and for each discrete reverb volume would have one or more sliders to adjust for the degree and other characteristics of the reverb effect, such as decay rate and delay.

2 comments

• 0
  
  Torley - Sansar October 04, 2017 13:33

  Thanks for requesting this, Galen!

  What would be some examples of where you'd use a feature like this?

  We've thought about situations where controllable "reverb zones" with different impulses would make a lot of sense, like... if you transitioned from being atop an island to underwater. Or say you went from an outside field into a really majestic cathedral — just like in the movies, you might want a far more exaggerated and long-tail reverb.

  In the meantime, for those reading this who don't know yet: the closest thing to "deaden" a fairly enclosed space in Sansar, acoustically, is to set the floors and walls to the Carpet audio material. It has the highest absorptive properties and such.

• 0
  
  Galen October 04, 2017 15:52

  Many kinds of natural environments call for different acoustic characteristics. I imagine you have the same thought and are not intending to only limit Sansar to computing only reverb forever.

  Also, I was assuming that the audio materials only affected the walking sounds. I had no idea they influenced reverb. That's awesome.

  I think I'm finally getting the gist of how you guys are computing reverb, too. I'm guessing you create a bounding box around the entire scene and then subdivide it into a 3D grid. At each grid point, you're computing a few reverb parameters based on the nearby surfaces and modulated by their audio materials. At runtime, you probably are doing linear interpolation for those parameters based on the position inside the cube formed by the nearest 8 grid points. And then plugging those parameters into a basic reverb engine for left and right ears.

  If I'm right about this method, then there's no limit to the other audio distortion parameters you could come up with to adapt to surroundings. Have you considered creating something akin to SL's particle systems? Let's call it an "audio distortion field" (ADF). That is, making it so a script can define a set of distortion parameters that adapts to some of the essential computations like distance to nearby walls and which has a defined region of influence? Then the script would not have to do any further updates; merely define the ADF's characteristics and letting the client compute what the user hears based on all the defined ADFs.

  Food for thought.

  Edited by Galen October 04, 2017 15:54