For nearly 2 weeks, I’ve been messing with the channel layout because everything worked great in Nuendo, Cubase Pro, Wavelab, Sequoia, Pyramix, and Resolve, but Reaper and ProTools were giving me issues. I hand to get creative to make Reaper’s channelwidth-only configurations work in my plugin that requires channels to be properly identified, but ProTools gave me bigger issues because, the bigger the channel-count, the more this digital noise creeps in. With ProTools sending audio to the wrong channels in the Dolby RMU, I thought it had to be in the core of the plugin or the output. Finally, to figure out if there’s an external reason ProTools was sending the 9.1.6 object bus LFE channel audio to the RRS channel in the 7.1.2 bed, I spent about 12 hours mapping every single channel order of every possible track/channel/bus in every section of ProTools and found my answer…
Its sending it to the RRS in the 7.1.2 bed for the same reason 9.1.6 objects in ProTools only map to 15 objects in the Dolby RMU-- IT LIES!
I’m somewhat joking but its true. When you create a 9.1.6 object in Nuendo, you get a 9.1.6 object flawlessly mapped to the renderer. When you create a 9.1.6 object in ProTools, it creates a 9.0.6 object and discards the LFE off into the 7.1.2 bus LFE. This was a stupid decision by Avid because it presumes to know what the engineer’s intentions are, but made worse by the fact that ProTools does not map to the renderer properly unless you manually change the channel order.
Anyway, once I figured out the the problem with the mapping is that ProTools is just a lot worse for Atmos work than I already knew, I realized that the issue might be located earlier in the path. So, I checked the parameter nobody that works in Atmos ever touches and finally found a big clue. Not only does the noise become greater as the number of channels in the bus increases, it also reduces as the crossfeed is reduced.
More importantly, turning it off does not affect wet signal until new dry signal enters and turning it back up does not affect the wet signal until new dry signal enters either.
So, the noise is injected at the crossfeed injection.
Any ideas what might cause that in AAX when VST3 and AU have no such issue? I’ve got an idea, but hearing someone else say it would be really encouraging while I work on this code and, in case I fail, it would be great to have a couple more ideas to look into next.