I'm curious about the way the bed channels are mapped onto speakers in a large theater. It's my assumption (feel free to correct me) that the 2 height channels map onto the entire overhead array and are time-aligned relative to the speaker position in the theater. But I wonder about those bed channels down on the main floor. For example, does the left front bed channel map only to the left front speaker in the room? Or is it distributed to left front, left-center, left-wide or something of that nature?
I'm in the UK so I'm a good few hours into the day already.
Glad to hear that my object based ramblings are of use. Always more than happy to talk about workflows and technologies.
I spend a lot of time going between big and small rooms with mixes. Starting Theatrical and moving down to Home is always much easier. Going small to big can be harder. The biggest things to look out for is the arrays/beds (point source in Home vs arrays Theatrically) and bed delays in a theatrical room. An object placed on the LSS speaker will sound the same as a bed routed to LSS in the home, but will sound very different theatrically. Unfortunately the old saying "if it sounds good, it is good" breaks down a little here when we're thinking about moving between rooms. Fortunately the metering in the Production Suite is very good and very useful. I always keep an eye on the meters and object view to make sure that I've got coverage around the room.
That said, with a little thought (and experience) it's relatively easy to make a mix that will work in both size rooms without much compromise.
The size control is very useful, not just for making an object bigger, but to blend objects together. Adding size doesn't just send the signal to more speakers, it has a decorrelation algorithm associated with it. The centre of the object remains un-affected, but the signal sent to more speakers is decorrelated. When you have 2 objects near each other with similar signals (i.e. reverb returns) it works very well to give them size to blend them together. When is comes to reverb's It makes it more realistic, especially in a theatrical room.
The great thing about Atmos is the rendering engine. This takes care of squeezing a big theatrical mix into a home environment. Once the studio is set up and the renderer know's what speakers are installed, it takes care of everything for you.
Life is definitely "different" at the moment. But on the plus side, it's giving everyone more time for learning and for discussions like this one.
Hope you're doing well and staying safe.
very similar. The LSS bed channel might cover from 3 to around 6 side speakers where the Lw might be the only speaker on the wall that is left out being only addressable by an object.
Bed's go to Arrays of speakers for the most part.
Objects can address individual speakers.
The surround beds are the same as the traditional surround arrays in 7.1 and 5.1. Switching from an object to a bed will not give more level, but will cover a greater listening area (which was the original idea behind arrayed surround speakers).
L, C and R beds will only go to the L, C and R loudspeakers.
I've been using your 3D reverb's (to great effect I might add) for many years now and have done a lot of experimentation with sending to only beds, only to objects and a mix of both.
All the bed channels are delayed so that they all arrive at the reference position at the same time, but the object outputs are not (otherwise when you pan an object around you'd get strange time changes). Mixing the same/similar signals between bed and objects can cause some strange delayed effects. In the mix position it might sound correct, but in the extremities of a large cinema it may not.
What I've found is that you use only the beds (with the overhead front/rear limitation) or all objects. Using all objects means that all the surround signals will arrive before the screen channels, so extra delay on the track is needed (I usually go for 1000 samples on the surround outputs to re-time them with the screen channels).
The great thing with using objects for reverb returns is the you can then give them size and move them around (within their allotted surround zone). This then gives an even greater sense of realism if done subtly, or can give some nice strange effects is the panning is not so subtle. You can exaggerate this even further with automating the send pans into the reverb track.
Hey Richard. I see you keep farmer's hours like I do! There's some solid experience in your posts and I genuinely appreciate it. My recent curiosity stems from mixing music in my room, and from the fact that I'm walking a group at iZotope through Atmos. In thinking about making a mix that works in a small room--a home theater--the idea of successfully translating to a large room brings a whole new set of considerations. That's the sort of thing I'm sure you've been considering for quite some time.
It may be that the best you can hope for is that a small room mix doesn't suck in a large room and vice-versa. I'm not even talking about loudness or dialog level--that's a can of worms I'm not remotely qualified to open. But taking into account things like the Haas effect or the downmix of objects into the smaller set of speakers in a home raises significant potential issues.
I had not thought about playing with the size of objects that might carry reverb. My background and recording experience are purely classical music and of course that's all about point-source and imaging. Marti and some of his cohorts have gently beat some sense into me about ways of mixing and this looks like another place where I'll need to think outside of my own particular box.
I can't say how much I appreciate this dialog. I do hope you and yours are managing to stay safe and sane through these peculiar times.
Thanks Ceri! You've been talking me away from the bed for a while and that's another good reason. I'd suppose that the distributed bed could give you more level if you needed it (and could save some objects). But you'd certainly lose some perceptual precision.
Thanks for the additional clarification, Richard. My interest is related to perception of reverb. The single overhead L/R actually diminishes any sense of space, since the front/back signal is correlated (except for the time alignment delay). In my verbs (Exponential) I provided a way to generate 4 additional fully decorrelated top channels. When those are assigned to objects pegged into the top corners, the perceived space is greatly enhanced. That's even improved when you avoid placing any reverb into the top bed.
So I found myself wondering about bed distribution down on the ground floor. It's probably not quite as bad, since each channel of side/rear reverb is roughly in a zone. I'd love to do some experiments with that, but my budget stopped at 7.1.4.
+1 to the remark about the rendering engine. Currently the group I'm training are all stuck at home with their laptops. They can only listen in stereo while they watch objects fly around the room. Amazing how well it works.
Wow, some amazing information there, thanks Richard!
That makes me wonder if we could open a subforum with just creative / mixing topics to exchange experiences?
I think that would make a lot of sense.