What’s the correct way to use a reverb send on spatialized events? From what I understood (and tested), the standard FMOD reverb converts any source into mono and spreads the output into each channel. So a reverb send AFTER the spatializer is problematic, since it doesn’t respect the panning done by the spatializer, and therefore the more you send into the reverb, the more you reduce the spatialization depth. At the other hand, a reverb send BEFORE the spatializer doesn’t make any sense.
I finally found that the convolution reverb hadn’t this behavior, and I could use it as a send after the spatializer, it respects the panning. However, it’s ok for stereo output, but I haven’t tested if the convolution reverb accepts IR for 5.1 or more.
But the question remains: why does the standard reverb has such a poor behavior (which is not in line with what can be found on any stock reverb on any DAW) and cannot seem to handle a so commun case (ie, lots of spatialized SFX which, in one particular level, should be played within a dungeon ambiance)?
There are many ways to handle ambience and reverberation in games. From my experience, when I achieve a behavior I like, it usually restricts what I can do with my sound in the opposite direction.
Your question made me think of something. You say that when you send in-game real-time panning in the convolution reverb it respects the panning also in its output right?
My question is, if the player turns the head (camera/listener in a common FPS implementation) what does the reverb output do? Does it follows the player’s orientation or it counter-orients so that the reverb tail stays with the original sound source that fed the reverb?
I never tried to measure that in FMOD, but I have a hunch that the reverb output doesn’t change orientation no matter what. In that case, having the reverb respecting the panning would also smear the localization of sound sources around the player, especially in a dungeon setting that traditionally features RT60 tails over 3 seconds long. With discrete surround impulses it gets worse, as there is no standards for the production and usually 5.1 or 7.1 configurations suffer from holes between the front and back channels and even bigger holes between the left and right channels in the back, and turning them around produces questionable results.
Maybe I’m not helping enough, but from my experience, summing to mono and spreading around is a more elegant solution, even with convolution, not to mention the resources saving from using one channel calculations. I usually play with the initial reverb attack curve to give enough time (100-200ms) for my original sources to announce themselves and then introduce the reverb, so there is enough time for the human hearing sense to pick up the location. Psychoacoustics always do the trick. To boost that effect, you can try filtering out some of the middle frequencies form the reverb’s return, to help with masking, further boosting localization accuracy for the player.
Thanks for your in-depth thoughts.
Since our game is a 2D tactical in top-down perspective, we probably won’t benefit from surround sound and stick to stereo. Therefore, I’m perfectly fine with the stereo convolution reverb, which is even better for giving a sense of real space than the normal reverb. Instead of sending it from the events, I send it from the concerned SFX bus group, directly in the mixer, with a snapshot triggering the reverb return volume. It’s perfectly workflow efficient.
But I’ll try your trick of augmenting the reverb pre-delay to have a better sense of panning.
From what I get in the sandbox, it follows the player’s orientation, which is what’s expected: the sound fed into the reverb is already panned in the right direction (by the spatializer, set before the send in the effect chain).
The system you created sounds very nice and I agree that stereo is enough. I think stereo is more than enough in many cases if it’s done with enough detail and is enough dynamic to help the narrative and the information feedback to the player.
Best wishes for your project, 2D top-down tactical sounds great. I like that genre.