View Full Version : [Post] Mixing for the Silver Screen


Jon Fairhurst
February 3rd, 2009, 11:41 AM
INTRODUCTION

Yesterday we hand delivered a short film for the Seattle International Film Festival - minutes before the deadline, of course. The submission is on DVD, but if accepted, we want a great result in the theater. We can make adjustments between now and the May screening. Unfortunately, I don't own a theater to try it out in...

PREAMBLE

At least with video, we know exactly where black and white are. And, theoretically, we know where a neutral color balance should be, so we can tune those using scopes. Maybe the gamma and saturation in the theater will vary, but it should be in the ballpark.

It's not so simple in audio. How loud is loud? Where's the noise floor? Where should the dialog be? And then there are frequency response issues (is their system flat? are my monitors and room flat?), subsonics (I didn't notice that rumble on my NS-10s), not to mention surround sound...

QUESTIONS

* Where should the dialog be? I mixed it (the body, not peaks) to around 20 dB below full scale.

* How low should the noise floor be? Some of our shots were noisy. There was a combination of preamp noise (we used a MicroTrack II) and ambient noise, like distant traffic, the HVAC in a theater, etc. I tried to keep the 'hiss' at about 60 dB below full scale.

RESULTS, SO FAR

I did my final mix on large monitors, rather than nearfields. Jon's Speakers (http://www.fairhurst.com/jon/speakers/) I moved far away from the speakers and cranked the levels. I turned off my own HVAC system, so I could get a near theater experience. I don't have a surround system, so I stuck with stereo. It would be nice to move the dialog to the center channel though...

On my living room TV speakers, I normally set the volume to about '20' for primetime stuff. For good cinema DVDs, I boost the volume to '30'. For my mix, the volume needed to be closer to '40'. At that level, it sounded great. The dialog was clear. The noise was barely noticeable, the music levels were good. The dynamics were strong.

Will this sound good in the theater? Or will people be straining to hear it?

I've mixed for two 48-hour film festivals in the past. Their audio system was not great, and they didn't adjust the mix between films. The first year, the mix was too cold. The second year, the mix was too hot. Hopefully, this time, I'll get it 'just right'.

Steve House
February 3rd, 2009, 12:34 PM
The basic way of setting up to mix for theatrical is to get yourself a sound pressure meter - the iexpensive Radio Shack models are good enough for this - and a pair of mono test files, a -20dBFS sinewave tone and a -20dBFS RMS pink noise file. First load the sinewave file into your editor and set the software playback level for each channel so the output meters on the editor and any external meters read -20dBFS / 0VU. Without touching the levels in the editor, switch to the pink noise file and going one speaker at a time and muting the others, with the SPL meter at the listening position where you are working, adjust the monitor volume control for a meter reading of 85dB SPL. If you're mixing 5.1 the LS/RS should be about 3dBSPL lower. Note that when the sinewave playback meters at -20dBFS, pink noise of the same average level will read around -10 on typical digital peak-reading playback meters so don't turn down the levels in the editor to compensate thinking the meters are off.

How loud dialog, music, etc, should be and their levels relative to each other is an artistic choice at the director's discretion. Broadcast may have specific levels required in the deliverable but theatrical doesn't. Of course, you want to avoid clipping, etc, so you you should make sure levels don't get too close to 0 dBFS. And you have no idea what the playback levels are actually going to be in the theatre. But calibrating your monitor as above will let you hear in the mix at the same level a viewer should hear centre of the house about 30 feet back from the screen in a properly setup theatre. Remember to adjust your sound sync so it leads picture by one frame to compensate for the delay in reaching that viewer 30 feet away from the screen and the speakers behind it.

Jon Fairhurst
February 3rd, 2009, 01:01 PM
Great suggestions, Steve,

I had read about the 85 dB calibration before, but have never gotten around to it. I'll be getting a meter ASAP. Thanks for providing all the details, such as peak vs. RMS metering in the NLE.

I had never considered the sound sync issue before, but it makes sense. Sound travels at about 1ms per foot. As I understand it, the speed of light is a bit quicker. ;)

Getting the levels calibrated will allow me to optimize the noise reduction. It's necessary at times, and sometimes it works great. But overdo it, and things start sounding like they're underwater.

Steve House
February 3rd, 2009, 02:07 PM
By the way, here's a link to the calibration page on the Blue Sky monitor website where you'll find the required test files available for free download. Blue Sky (http://www.abluesky.com/asp/catalogue/catalogue.asp?linkid=172)

Steve

Jon Fairhurst
February 3rd, 2009, 05:21 PM
Thanks Steve. Those files will come in handy. I also have some on CD that I can rip for personal use. It will be interesting for me to compare them as a sanity check.

Andy Tejral
February 3rd, 2009, 06:03 PM
This sounds like a sticky, no?

John McClain
February 3rd, 2009, 10:49 PM
Jon, lots of excellent info in this post:

http://www.gearslutz.com/board/post-production-forum/326305-dolby-media-meter.html

Spend some time with it, quite helpful. And the info Steve gave you is great but keep in mind that you may want a lower level ref setting for your room if it is smaller, mine is set to 82 rather than 85. john.

Jon Fairhurst
February 4th, 2009, 02:17 AM
Thanks for the tip about lower levels in small rooms. That makes sense. Fortunately, my room (daylight basement and man cave) is good sized (26' x 16'), and reasonably damped (filled bookcases are great poor man's diffusers), so I can go for the full 85 dB.

BTW, I mastered our video this evening with slight compression and limiting, and the result on the TV speakers matches DVDs exactly - my TV volume is set at 30 and the levels are perfect.

I'm thinking that the right process is to use 24 or 32 bits, and to set the system at greater than 85dB when mixing. It gives you that much more headroom, so you don't need to even think about clipping. Just mix to taste, and let the peaks happen.

The time to set an accurate 85db is when mastering. By using a good multiband compressor and soft limiter (I'm using the Izotope tools with Sound Forge 9), you can maintain high dynamics while only the loudest peaks are sacrificed in the end. And with muti band compression, you only lose some of the frequencies on the peaks, not the whole signal. On the other hand, if you mix without enough headroom, you manage peaks by turning down the volume on some tracks, or drawing a series of strategic envelopes. It's more efficient to let the peaks fly and let the final processing handle it.

Anyway, my mixing on large monitors with the volume cranked up seems to have paid off. I'm happier with this mix than any I have done before.

John McClain
February 4th, 2009, 11:26 AM
I'm thinking that the right process is to use 24 or 32 bits, and to set the system at greater than 85dB when mixing. It gives you that much more headroom, so you don't need to even think about clipping. Just mix to taste, and let the peaks happen.

While this flies in the face of things I've read, rules are made to be broken! I generally mix at less than 82 (my room) and listen to final at reference level, I'm very curious to hear how it works out in the theater, I'll be waiting to hear your report. I might give this a try on a short I'm working on just to see how it flies. John.

Jon Fairhurst
February 4th, 2009, 11:58 AM
While this flies in the face of things I've read, rules are made to be broken!

As long as the mastering is done at proper levels, it doesn't really matter what the levels are when assembling things (as long as you don't clip or start hearing the noise floor). Personally, I like adjusting things based on how they sound, watching meters to make sure that I'm not drifting away from my target, and ignoring too many left-brain, technical issues.

What kills my productivity and creativity is when combined signal peaks hit digital full scale, and I start messing with things to stay within this artificial bound. (Being limited to 105 dB for single sample peaks seems overly constrained.) I'd rather ignore these technical details and get on with making the mix sound good.

Of course, the final product needs to be calibrated and within bounds (hence, this thread.) Leaving that for the mastering process makes a lot of sense to me - especially since the goals and limits are different for the theater, DVD/BD, television, CDs, radio and the web.

Frankly, it's the advent of 24-bit and 32-bit audio that allows this freedom. In the analog and 16-bit days, one had no choice but to mix within dynamic range and noise constraints; hence, the level guidelines for mixing as well as mastering.

Jon Fairhurst
February 4th, 2009, 12:08 PM
So, what about noise reduction? What are the limits of a reasonable noise floor?

During my mix, I pulled noise and his down to 60 db during my mix. (It's probably about -54 dB after mastering.) Is this about right?

For many of my scenes, getting down to -60 dB was a piece of cake, requiring little or no noise reduction. On some scenes, I had to be more aggressive, and the results still sounded great (using Sound Forge 9.) However, there were other scenes where the noise was a bit ugly (ironically, in a dance studio/theater), and the result after noise reduction has some underwater garbling.

If there is a guideline for the highest reasonable noise floor, that would be helpful. Reducing noise is a balancing act, and I like to hit the best compromise - and avoid ADR, if possible.

John McClain
February 4th, 2009, 03:39 PM
As long as the mastering is done at proper levels, it doesn't really matter what the levels are when assembling things (as long as you don't clip or start hearing the noise floor). Personally, I like adjusting things based on how they sound, watching meters to make sure that I'm not drifting away from my target, and ignoring too many left-brain, technical issues.

What kills my productivity and creativity is when combined signal peaks hit digital full scale, and I start messing with things to stay within this artificial bound. (Being limited to 105 dB for single sample peaks seems overly constrained.) I'd rather ignore these technical details and get on with making the mix sound good.

To the first part, I don't agree but there are many paths to the same destination. The beauty of setting your room up to reference level and then using that as your loudest playback level is you rarely need to look at your meters, you know by ear if you are mixing to proper levels or not. This doesn't take away from how things sound, it just assures your mix will translate.
As for the second part I've quoted, again, if your room is properly calibrated and you mix within that reference you will not run into overloads with the exception of perhaps explosions. But you'll know they are too loud and you can adjust accordingly. john.

Jon Fairhurst
February 4th, 2009, 04:08 PM
To the first part, I don't agree but there are many paths to the same destination. The beauty of setting your room up to reference level and then using that as your loudest playback level is you rarely need to look at your meters, you know by ear if you are mixing to proper levels or not. This doesn't take away from how things sound, it just assures your mix will translate.Still it seems limited.

As I recall, my speakers will do about 118 dB peak at higher frequencies. 105 db as a digital hard ceiling for single sample spikes seems much to low for mixing explosions with music and what not - unless I pre-compress. And, once again, I'd be compressing to hit a technical mark, rather than for the desired sound.

I think it's important to have a consistent setting though. Maybe 95 dB for the mixing phase. The only difference from an 85 dB reference is that I would have 10 more dB of headroom - and my entire mix would be 10 dB quieter than it would be otherwise. Then when it's time to master, I turn the system down by 10 dB, and use the mastering tools to bring everything back to the desired levels while paying special attention to the handling of the intermittent peaks and loud passages.

They say that the person who did the mix shouldn't master it (at least on the music side of the business.) I'm thinking that the 95 dB vs 85 dB thing would help to change my perspective between the two roles enough to be successful doing both tasks. With my budget, I'm the only guy I can afford. ;)

John McClain
February 4th, 2009, 04:23 PM
95dB is awfully loud. I didn't set my room ref at 85 because that's too loud to listen to in my space, my previous room was 85 but again, I only listened to it at 85 for final playback or intermittently throughout the mix. You are getting into levels much too loud to listen to for sustained playback when you are pushing 95dB as a reference level in the room, be careful. You should post your situation at the Gearslutz/post production section.

http://www.gearslutz.com/board/post-production-forum/

There are a lot of A list Hollywood and indie mixers who will give much better advice than I can and will happily work with you thru your questions, especially the idea of mixing at 95dB ref. I really recommend it. john.

Jon Fairhurst
February 4th, 2009, 05:25 PM
...You are getting into levels much too loud to listen to for sustained playback when you are pushing 95dB as a reference level in the room, be careful.Ironically, this doesn't mean that I will mix louder. I would actually mix much quieter. It's really a matter of giving myself headroom for inconsequential transient peaks. (It's similar to not mixing on speakers with too much bass. If one does so, the mix will ironically have too little bass.)

I certainly wouldn't be playing back CDs or DVDs at that level. I agree. It would be much too loud. It would only be for mixing original material while allowing for more dynamic range for the initial mix.

BTW, any comments on noise reduction? Anyone?

John McClain
February 4th, 2009, 06:11 PM
Understood, I misinterpreted you regarding levels. As for noise reduction, I'd say just use it judiciously trying your best to not hear the processing. I've never run into a figure regarding noise floor, most likely because it depends on the style of film. When I have to use noise reduction I go in small passes with a couple of different tools, I back off once I can hear the processing. I may have to mask with ambience but that too is dependent on the material. The feature I'm currently working on is mostly outside so I have little noise reduction I have to do as I can bring up backgrounds to compensate (that and it was recorded really well!). Hope that helps, this thread is fun as I've mixed features and shorts in stereo but I'm doing my first in surround, I'll be looking forward to hearing how yours turns out! john.

PS. Regarding noise reduction, if you don't have it and can afford another tool, Izotope RX is the way to go.

Jon Fairhurst
February 4th, 2009, 06:42 PM
Izotope RX could be the ticket. The Sound Forge NR is good on 80-90% of our material, but there are a couple of scenes where the noise is ugly, and the processing obvious.

I've got three situations: hiss, HVAC and traffic.

The hiss removal is really good, and I can cover things with music, background sweetener drones, and just the right amount of NR.

The HVAC is obnoxious. It sounds really ugly, and when removed, I can hear a muddy background, like people talking, a radio, or TV on some of our takes. It burbles in and out, and wasn't noticeable until the HVAC was removed. Unless I find a better solution, ADR here we come.

The traffic isn't too bad, but varies, due to our short takes and edits. Listen to the noise, and you can tell how we grouped our shots. While I could grab a single hiss sample, I had to process the traffic stuff line by line. It's usable though. The NR isn't too obvious. Maybe I'll mix in some crickets, more consistent traffic noises and dogs barking in the distance.

It's funny. The average person will claim not to know the difference between good audio and so-so audio, but play some bad audio in a movie theater, and they know it's bad straight away. Make it too quiet, and they'll yell, "we can't hear it." Make it too loud and they'll wince.

People are better judges of audio than they give themselves credit for.

Andy Tejral
November 25th, 2010, 12:52 PM
By the way, here's a link to the calibration page on the Blue Sky monitor website where you'll find the required test files available for free download. Blue Sky (http://www.abluesky.com/asp/catalogue/catalogue.asp?linkid=172)

Steve

Files were moved--they are now here: Blue Sky Blog Archive Blue Sky Calibration Test Files (http://abluesky.com/support/blue-sky-calibration-test-files/)