Dolby Atmos

[jeremiahmoore]

I was at the demo as well, and asked John Loose some tech questions that helped me get the lowdown. Thanks John and Kevin for the great demo! And someone please step in if I've got something wrong here.


Atmos is basically an abstraction layer for multichannel panning. Instead of panning to a speaker, you pan to a direction, and the speaker assignment is made on the fly during playout. The playback system knows where the speakers are in its particular theater, and sends the sound to the proper speaker(s). This way, theaters can be configured for different numbers of speaker channels and retain correct panning. (hence all the mention of "future formats like 9.1"). Baseline layout includes today's 7.1 (w/ rear surrounds flavor), overhead, rear, and full-frequency rendering all around the theater.

Here's the other thing: Dolby proposes Atmos as a way to automate the creation of multiple formats of deliverables. i.e. mix in 56.3 and render out in 7.1, 5.1, 2.0 LtRt, Dolby Headphone for the mobile crowd, whatever. (we will see how well their algorithms handle steering artifacts!)


In practice, Atmos printmaster deliverables will be something like:
- a 7.1 mix, entire film soundtrack MINUS what's assigned to Atmos.
- Atmos package, which consists of audio objects and panning metadata, which will be a computer file generated entirely within Dolby's Atmos RMU mastering box.

Atmos will be delivered to theaters embedded in the DCP. And played back with (you guessed it!) Dolby's cinema server. I gather there wil be a new Cinema Sound Processor for the audio playout.

During mix:
Three things stream into the Atmos mastering box.
- mono or stereo audio via MADI (e.g. from add'l pro tools outputs)
- pan metadata - e.g. from panner plugins in pro tools
- timecode (unsure what format this TC is)

Pro Tools panner plug-in runs on an insert and has typical PT automated parameters. You can control it from a panner joystick. It does NOT process audio; rather it outputs a metadata stream which is recorded in the RMU.


A workflow could look like: typical sound editorial, flagging sounds to be considered for Atmos panning. At mix, Atmos sounds will be pulled over to Atmos channels and removed from the main mix, a panning pass done,

Films could mix Atmos first, then (unless I'm mistake) render the Atmos output as 7.1 stem to be folded into the 7.1-only version. Or they could be mixed 7.1 first, with specials tracks later broken out and panned in Atmos. Probably the latter will happen because it's going to be expensive to outfit a theater; why not mix normal and hire another facility for an Atmos pass at the end of the mix.


Personally, I can imagine the format's challenge will be: directors will be able to mix EXTRAORDINARILY LOUD in it. And theater goers will complain, and Atmos will be known as the "loud" format. And/or Atmos theaters will be known as "quiet" because they all get turned down. Then again, the action film crowd might eat it up. Just remember to pack your earplugs


-jeremiah

[Felix Rost]

Originally Posted by jeremiahmoore

Pro Tools panner plug-in runs on an insert and has typical PT automated parameters. You can control it from a panner joystick. It does NOT process audio; rather it outputs a metadata stream which is recorded in the RMU.

I wonder how accurate this can/will be? We know that PT automation is not sample accurate (yet..).

[Frank Kruse]

Originally Posted by jeremiahmoore

Personally, I can imagine the format's challenge will be: directors will be able to mix EXTRAORDINARILY LOUD in it. And theater goers will complain, and Atmos will be known as the "loud" format. And/or Atmos theaters will be known as "quiet" because they all get turned down. Then again, the action film crowd might eat it up. Just remember to pack your earplugs


-jeremiah

I was also surprised to hear that Atmos has the same per channel SPL calibration as 5.1 so you can be confronted with 64x85dBSPL at -20dBFs.



Frank.

[thermisonic]

out of curiosity, did anyone at the demos talk about ambience channel counts? strikes me that if Dolby Atmos moves some productions to 9.1 beds that it may be time to work with more multichannel ambiences instead of frequently building up from stereo or occasionally 5.1. What do people think about multichannel ambiences in this context?

Rob Walker AMPS

[peter5992]

Originally Posted by thermisonic

out of curiosity, did anyone at the demos talk about ambience channel counts? strikes me that if Dolby Atmos moves some productions to 9.1 beds that it may be time to work with more multichannel ambiences instead of frequently building up from stereo or occasionally 5.1. What do people think about multichannel ambiences in this context?

Rob Walker AMPS

There are (currently) 128 object tracks which can be used for anything, whether it is specific hard effects, ambience, music or even dialogue (though that would probably make little sense in most situations).

One of the demos was a remixed scene from Mission Impossible IV, the scene where Tom Cruise runs out of a hotel in Abu Dhabi chasing the bad guy while a sand storm is imminent. They had the sand storm in Dolby Atmos, really made me feel like I was in the middle it (wanted to cover my face).

It is more spectacular with hard effects like the car crashing overhead and landing behind Tom Cruise but it certainly can be used for ambience effects as well. Somebody at Dolby correct me if I am wrong here.

[jeremiahmoore]

I see it as more creative options.

I'm sure there's a solid place for multichannel BGs if it works for the project. Of course with the understanding that it'll get collapsed to 7.1, 5.1, or LtRt (or mono distant TV speaker!) depending on exhibition situation.

If it sounds good, it is good

-jeremiah

[Brandonx1]

Does any oozy know if IRIS at the Dolby theater is in atmos? I saw it the other day and it sounded really cool but I wasn sure because it's a play type production

[Sreejesh]

There are 118 and not 128 Objects because the first 10 Inputs to the RMU are the Bed. I had a chance to Test Mix some of my material at Dolby's Facility in London. Mixing something extraordinarily loud, though possible is not that easy unless you really get so many objects playing simultaneously across so many speakers. And I think personally that the most amazing thing is the resolution of the pan and the real sonic atmosphere it can create. In fact, immersing the viewer in the film is very easy and effective now. Trying to start a mix in atmos will be a bit difficult because of the fact that you will still have to control objects individually and not with a master fader. (Unless you group the faders, in which case there isnt a way to have overall EQ or Compression etc like we use in Bus masters). Riding music against dialog or tweaking ambiences then will be a bit tedious. That being said, it isnt impossible and I am sure we will see movies done like that.
The automation is done via a plugin and has no relation to the Protools Panner. In fact, once you do a 7.1 or so automation, it is very well possible to paste the Front and F/R pan position to the X and Y parameters of the Protools Plugin and then decide to correct the height, saving a lot of time recreating the pans you had done.

[Sonopolis]

Originally Posted by Sreejesh

In fact, once you do a 7.1 or so automation, it is very well possible to paste the Front and F/R pan position to the X and Y parameters of the Protools Plugin and then decide to correct the height, saving a lot of time recreating the pans you had done.

This is good news! That's what I was hoping for.

[Sreejesh]

Sonopolis, I must tell you that it is not a direct method. Its the same way we can paste a fader level to Plugin using Cmd+C and then Cmd+Ctrl+V. :)