Hacker News Reader

Comment by rasz_pl | Hacker News Reader

It uses same algorightms 3d engines use to compute lightning and shadows

should be easy to implement in hardware in GPU.

BTW I seem to remember vaguely Carmack wanted to implement fully software 3D sound engine for Doom3, but some patent dispute forced ID to support Creative EAX.

View on HN

It's quite different from what we do. However we also do have working GPU implementations in CUDA and OpenCL :)

looks like you shoot rays around from the player

It's not that easy (unfortunately). Those are sound paths.

I have to admit, I thought it was just pretty standard ray-tracing as well. I'd be curious to find out how it's different. I followed the link through to the website, but didn't see much additional information on it.

It's different mostly due to diffraction which is neglible in case of light and very important when simulating sound.

Diffusion is important for light too; see https://en.wikipedia.org/wiki/Radiosity

Diffusion is not diffraction, they're different phenomena.

Radiosity simulates diffusion, though. Are you saying you also simulate diffraction? In visual rendering, that might fall under something called "caustics" or "photons" (I use quote marks because renderer photons are not quite the same as real photons).

Caustics are caused by refraction or reflection, not diffraction. There are many analogies between simulating light and sound, but you take different shortcuts to make those simulations fast. E.g. you don't simulate diffraction when simulating light (it's effect is neglible) and you discard small objects when simulating sound. Those differences exist mostly because visible light spectrum and hearable sound spectrum are so far apart.

Yeah. Carmack wanted to calculate the distortion of the ear-canal on positional audio and to calculate the effects of sound diffusion off of surfaces like light. Creative owns patents on both of those techniques. :(

Reply on news.ycombinator.com