Noise Reduction in Production

So check it out, Cedar actually makes a portable and simple DNS. It's interesting because it means that a dialog mixer in the field might very well use it because Production is incapable of shutting the hell up for takes.

This is a quick and dirty solution for cleaner-sounding takes, but it ain't cheap -- it's $4000. And cleaning up with single-ended noise reduction while recording is not the "right" way to do it. But two things about that -- 1 it's likely that you'll have separated tracks of each microphone being recorded pre-noise reduction anyway and 2 so many productions go through post so quickly that the production sound person really is the last time a sound pro will touch the audio so you may as well make it sound like the finished product.

DAW, Transcriptions, Microphones, and Spaceship Panels

Here is my workstation where I am editing dialog and sound effects and music on Carbon Copy. I actually ended up with two workstations. The other one is half-set-up in the bedroom.

My little apartment seems weirdly spacious in this view.

Rev via Kangas. $1/minute transcription. This is relevant to my lifestyle choices.

§

Sanken COS-11 microphones. Without accessories. With accessories. Honestly, with the relatively inexpensive Sennheiser "Evolution" wireless systems I'm fairly happy with the quality of the wireless transmission and reception. I'm not as excited by the microphones that come with them (sort of a cheap-o version of the Sennheiser MK2). And in most scenes one shoots the difference between good mics like the Sanken and bad ones like the MK2 knockoffs really isn't that big a deal. But every once in a while you have a microphone in a bra or elsewhere that the location isn't quite perfect and bleh. The little bit extra by paying upwards of $400 for microphones makes a big difference.

§

2 Coast Customs makes props including spaceships panels. Sometimes a man needs spaceship panels.

Roboticide

I've been killing a lot of androids lately. I know that's going to come back to me.

Yes, Mother, you can export ProRes out of AfterEffects in Windows. It requires this free plugin from the company DuBon. And you can only export files, not compositions, so you have to pre-render first. But it can be done. It can. Be done. H/T Ian Hubert.

Pixel Plow

So I did an analysis of how much it costs me to render. I was surprised about how expensive it is. And now I am rethinking my experiment with Pixel Plow.
Pixel Plow is the cheapest rendering service for Blender 3D. Cheapest by far. And it is super-duper fast. Even at the slowest speed it is way faster than my fastest GPU accelerated machine.

Unless I put my i7 with Quadro4000 GPU in a place where electricity is "free"*, it's actually fairly inefficient for me.

Now in order to do that I'd certainly have to get Logmein Pro which is $250/year.

And all that brings me back to Pixel Plow.

They make it relatively easy to render with them. The frames come down to you automatically as they're created. And now that we're in 4K it makes a much bigger difference to get those frames done so that a fellow can get to actually finishing the dang movie (whichever movie that might be.)

So today I am pro Pixel Plow for big renders.

*The morality of such use is up for grabs but those places where one has/does pay a single rate for electric no matter what the use does/do exist.

Focusrite Scarlett 18i20

So, because every-single-thing has to be harder than previously imagined, we had to get a new audio interface for the new audio computer. We got this Focusrite unit. It is, I must say, really nice.

"Ice Nymph" is the name of the hard drive which sits atop. 

I don't know why so many people are into the RME units. They're quite pricey. May as well get Apogees*. But here's the thing: I've A/B'ed Focusrite vs. Apogee and they sound so close that when you invert the polarity the signal will oftentimes actually null-out.
Maybe I just like the red color that surrounds the unit (which you can see in the reflection of the wood support just above the converter in the picture above if you so care.)
But even the preamps are usable.

I haven't actually hooked up a 5.1 system to it. I'll tell you if there's any problems. But so far it's been very stable, which is critical (and maddening when things aren't stable).

*Traditionally considered the best converters for everyone but some classical music guys.

Distortion in ADR

So my new thing is putting some distortion on ADR tracks to make them fit better.
The amp simulator plugin in Samplitude works really well for this. But all by itself it's a bit too much. So I feed an aux channel from an ADR channel so that I can blend the inserted distortion all I like.

Like so.


Where did I get this idea from? I got it from this brilliant book on dialog editing. Oh man. This book is good. Read it. Live it. Ignore all the ProTools-centric stuff. ;-)

ADR

We had a couple noisy locations on this picture and there really wasn't any option other than some ADR to deal with them.

Sarah Schoofs doing ADR in the studio.

 I've been using an Oktava 012 microphone -- the same we'd be using on set (except that this movie was almost completely recorded with wireless lavalier microphones). And there's a bit of distance on the mic (you can't even see it in the picture above.)
It seems that for most people, seeing the picture while they record isn't terribly helpful. So we've abandoned having a picture monitor in the booth. I'll play the line three times and then go into record. No bloops or leader or anything.
I think this makes it easier. You just need to memorize the cadence of the original and you're not distracted by picture.
I do try to line up each line for sync immediately after they record the line. Just to, you know, make sure.
There's a standard impulse response in Samplitude -- a 1.1 second "garage" reverb. I've been using that to give a bit of controlled distance and room on the ADR. Honestly I've lost all perspective -- literally and figuratively -- but it seems like the right sort of sound for ADR. The tail of the reverb doesn't do weird things the way the tail of a reverb which is more appropriate for (say) guitars and pianos would be.
We have two more actors we need ADR from. And then (hopefully) we will deliver final picture.

Sound and the Furry

I'm gonna get this book. Dialogue Editing for Motion Pictures.
Also cool are the video tutorials available online associated with the book. Look, I straight-up do not agree on putting tones on any digital format. But I'm allowed to not agree with parts of stuff I otherwise like. It's in the Constitution.

§

Oh man. My nice Sennheiser 580 headphones are falling apart. A new pair of Sennheiser HD600's is more money than I want to think about right now.

What part of this design isn't awesome? No part, that's what.

Mixing too quiet means mixing too loud

If you're not trying to be really obnoxious to your office-mates you might think that it would be better to mix with your monitors turned down to a reasonable level. It's not. Indeed it means you have to mix over again.

In film there is a standard for how "loud" you should mix. Specifically, your monitors should be at 85dB SPL (measured in noise bands) from each speaker at the mix position with the meters at -20dB Fs. Does that make sense?
This is actually pretty loud. Louder than what you should be doing with music actually (my rule with mixing music is that you should be able to talk over it comfortably. That's how you know you're at about the right level.) Now the other thing with film mixing is that you typically have to hard-limit each channel at -12dB Fs. This may be more of a broadcast issue actually. But considering that -12dB Fs is 93dB SPL from each speaker it's a good idea to not go above that level anyway.
As an aside (isn't this entire post an aside?) I've often wondered why they fellows at Dolby or whomever decided to lose 12db of headroom from mixes. My only guess is that if all six channels of a 5.1 mix are limited to -12dB it becomes very hard to actually go above 0dB Fs when you mix all those channels together, even if you do it clumsily. That's my guess.
Anyway, remember about how 85dB SPL should equal -20dB Fs? Well even when I'm mixing "correctly" I don't even go that loud. I set my system to 72dB SPL at -20dB Fs. This makes mixes "louder" than if I'd done them at 85dB because I'm pushing the faders up more to make up for the fact that my monitors are turned down.
But this whole complaint is that I'd mixed the first pass of the Dead Residents movie even quieter than 72dB (somewhere around 60? Maybe?) and all of those mixes just slammed into those -12dB limiters way too often and too hard, losing any sort of dynamics in the mix and actually making it a bit distorted.
So, mixing quieter means mixing louder. Right?
Somebody re-write this so it makes sense.

Foley Kontakt

Alejandro Cabrera released a free Native Instruments Kontakt instrument of Foley footsteps. You launch the Kontakt player and then you select "indoors" or "outdoors" and then the type of shoes. Could you use a wider selection of each? Of course. But it's free. Is it worth paying for? Yes. Do you have to? No.

Signo makes a number of different Foley instruments for Kontakt too.
Note that you need the full version of Kontakt for these instruments to work for you (at least for more than about 15 minutes).
I haven't actually tried these in a post-production environment but it seems like you could really do a lot by "playing" your Foley on a keyboard. Frequently one is "playing" footsteps by "walking" with shoes in one's hands so it's not that far out a concept. It's certainly worth a try.
(Kontakt is a software "instrument" you play with a MIDI keyboard or similar.)

Production Based on Post-Production

I got the pleasure yesterday of hearing two actors talking about how great the script was. The script we're shooing, presently titled Dead Residents, was writ by the muse who directs the hand of Steven J Niles.
The script moves. And the characters are all specific and different. It's pretty awesome. And I'm digging how it looks.

+++++
I've also been really enjoying the 4-hour shoot days. I would like to keep us down to 4-hour shoot days. What does that mean? Well it means I'm lazy. But also:
We have to increase the number of shoot days.
+++++
The other thing is that we have some very hard deadlines for this picture.
But if we shoot the movie based on our post-production bottlenecks, we can do production and post at the same time. For practical purposes this means we shoot in such a way that entire acts in post can be edited. Which means we need to be sure to unload series of scenes which all go together.
+++++
I think this means we should immediately shoot:

• all the scenes which have composites
• all the scenes for the first act

So that's what I'm going to schedule.  A shooting schedule based on post-production needs, making sure post has the footage they need in order to work concurrently on the movie. This means we can actually be shooting picture right up until we need to make delivery. I mean, not right up until we need to make delivery. But, or our purposes, pretty darn close.

It'll eventually have to happen

Right now we edit on Final Cut Pro 7.
But...
But soon. We will have to hop off the FCP train.
Where does that leave us? I think not Avid. The only other option is Adobe Premiere.
One new advantage that Adobe has is this Adobe Anywhere thingy. My nominal complaint about these kinds of technologies is that they're tested under very weak conditions; commercials, music videos, and other short projects. A feature film is another ball of rabbits -- even at the low shooting ratios we tend to use. But being able to have multiple editors working on stuff simultaneously is rather intriguing. That is, if I could get multiple editors to work at the same time! ;-)

So. Yeah. I dunno. I suspect we're going to finish off this movie in Final Cut Pro.

CALM Down

You are utterly fascinated by relative broadcast levels of dialog, aren't you? Yes. Yes you are. 

The Angry Sound Professional explains the CALM act and what it really means. Here's part 1.

We used to have one of these jackets. It was stolen out of my car. 

In part 2 we go deeper into the meaning of dialnorm as Vince Tennant explains how the Act works. I suspect that part 3 will show how the act doesn't work but we're still waiting.

Fortunately/Unfortunately

Fortunately, Red Giant's BulletProof looks like a nice solution to our problems ingesting footage from the unhacked GH1 into Final Cut Pro.
Unfortunately, BulletProof is not actually compatible with Panasonic cameras.
Fortunately, Adobe Premiere will, as pointed out by Kangas in the comments below, work with Panasonic's .mts files.
Unfortunately, Premiere will not read and interpret those .mts files shot at 23.98 but in a 29.97 container properly.

Fortunately, the most recent version of JES Deinterlacer will read .mts files, reverse-telecine them, and create ProRes Quicktimes which can be read by Final Cut directly.
Unfortunately those same files don't seem to be readable by Premiere.
Fortunately there is a thing called Adobe Media Encoder.
Unfortunately it doesn't look like Adobe Media Encoder naturally does reverse telecine.

So... right now it looks like the JES Deinterlacer and Final Cut Pro. Unless it's the JES Deinterlacer and Premiere. I don't know. I'm sticking to FCP just because it's "the devil you know". But... I dunno.

New Workflow

Neoscene is dead. Also, it's just not really working for me. Following the post here I'm trying FCP and compressor.

It took two days to get this still of Julia Rae Maldonado without interlacing.

What's going on is that I'm having hellatious problems with interlacing. Which, seeing as how we're shooting in a progressive format, just shouldn't be happening.
I can't figure out why Neoscene is being such a problem with interlacing. For years Neoscene worked for us. Then suddenly it stopped working. I had to download a new version (and Cineform quickly got me a new serial number for which I am thankful) and I had trouble seemingly with the checkbox to filter 420 to 422.
But now even that doesn't work.
So the new way of bringing in footage from the GH1 is to

1. Log and Transfer into Final Cut Pro. (We use ProRes SQ). And then:
2. Set the footage to 23.976 using JES. Because, you know, why use one step when you can use two?

Final Cut Pro is terrible when it comes to data management. Why you would put final actual camera footage in a folder called "capture scratch" is completely beyond me. I mean seriously. Right?
Then with JES -- I can't actually find any instructions on the Internet which reflect the transcoding program with its most recent interface. So, uh, you kind of have to guess how to do a "reverse telecine" with it. If you click through all the menus and do what you think you should it seems to work out all right.

EDIT: it has not escaped my attention that perhaps this all means that we should be going to Red Giant's BulletProof.

Dialnorm

Dialnorm is like a briefcase full of bunnies. 

You are having trouble with Dialnorm, aren't you? Yes. You are. You're mixing for TV broadcast and you have questions about these new silly standards. You say "why can't we just make sure our peaks don't go above 0dBfs and be done with it?" Silly girl, you can't do that.

So... dialnorm. The idea is that we want to have all the dialog levels be the same across all kinds of television programs and stations.

Here's a thread about it.

Dialnorm is just the averaged perceived loudness over the course of the program. Dolby later refined their measurement method with the "Dialogue Intelligence" algorithm, but every program, with dialogue or not, has a perceived loudness.
You can use Audioleak to find the A-weighted Leq. This will be your dialnorm.

You want to know what LAeq is, don't you? Heck, Answers.com even has an answer:

LAeq is pressure level measurement parameter. Full form of LAeq is " Equivalent continuous A-weighted sound pressure level". It is widely used around the world as an index for noise.L_Aeq = 10*log[1/(t₂-t₁) * Integration of (P²_A/P²₀) between interval [t₁ t₂]]
where
L_Aeq = equivalent continuous A-weighted sound pressure level [dB]
p₀ = reference pressure level = 20µPa
p_A= A-weighted pressure [Pa]
t₁ = start time for measurement [s]
t₂ = end time for measurement [s]

Read more: http://wiki.answers.com/Q/What_is_a_LAeq#ixzz1HWm0D8BO

None of that is particularly helpful, is it now? No. It isn't. I tend to mix my dialog slammed into hard multi-band limiters peaking at about -12dB FS. The real question is where is the average level in the program? And that isn't something I bother to measure. I'm not saying it's good that I don't measure it. I'm just saying we don't do that here.