Wednesday, 6 April 2011

Audio and Filming with DSLR's

More and more I'm getting calls about filming with DSLR's. The problem is that the DSLR'S are really not built to record sound, because most have automatic gain control, and all have very poor mic pre-amps / A/D converters / limited dynamic range and are just generally quite noisy (the recording quality is, not the camera itself). It makes me inwardly weep when I see a Tweet saying that a D550 can sound "awesome" - no they can't; a guide and backup at best.

Sound devices did a great study of the audio capabilities of the 5D mk II, here's the link and it's worth reading.

So it's pretty clear that sound needs to be recorded on a separate system; and it can be thought of in much the same way as film. The audio and the pictures are sync'ed together before the edit;  via a clapper board - be it a digislate or conventional board.

However, there are other options. The audio on the 5D is perfectly acceptable for recording time code - so there is no reason that the audio timecode signal from a timecoded audio recorded can't be sent wirelessly to the camera (a sennheiser g2 / g3 sits very nicely on the top hotshoe mount) and then extracted in the edit for syncing with separate sound - although I still think that FCP has an issue with extracting T/C from an audio signal - more info here on that.

Pluraleyes

There are other options though. The software Pluraleyes has somewhat revolutionised the syncing of rushes, by matching the guide sound from the camera with the "proper" audio rushes from the audio recorder - and (as I understand it) it's all handled as a batch process, and it's irrelevant of whether the sound is in one continuous clip where the camera has cut at numerous intervals.

Problems with Pluraleyes generally come when the software can't match up the audio, from the on board microphone compared to the location audio recorder.The reason for this is so simple - the microphone on the camera is of very poor quality, and somewhat Omni directional in its response pattern, so it has a totally different sound perspective to a correctly positioned boom or lavaliere mic; never mind the time delay for the sound to reach the camera if working on a long lens. In a nutshell: the sound will be completely different.

A neat way to get round this is to send a wireless mono mix to the camera (again sennheisers g2 / g3 mounted on the hot shoe): This sends a copy of the audio from the sound recordist to the camera, and even though the quality of the onboard sound maybe for guide purposes only, it is absolutely ideal for pluralise to match up the sound files because both sets of audio files now have the same perspective.

Many DSLR productions I've worked on have now adopted this production technique, even to the extent of not using a clapper board (which I don't necessarily agree with). With this in mind, there is also another advantage in working with the wireless audio link to the camera: the sound recordist can send a blip of tone at the start of the take which is then recorded by both the camera and the audio recorder, and can help pluralise that little bit more.

Additional edit - 27 April 2011

One big problem with Pluraleyes comes when trying to encorporate it into a conventional workflow. It works fine with stereo audio files; but problems creep in when using multitrack audio recordings, especially ones that use iso tracks - from multiple radio mics, spot mics etc. When this is the case, you can't beat time code and a board for allowing both the editor to sync the mix files, and then the audio post facility to link the sound rushes via meta data - either through an OMF or conforming software.

-end of edit


Input Levels and Connections

The only other gem of wisdom I've learnt with working with the DLSR's, is one of RTFM. The audio input is an unbalanced 3.5mm mini jack at MIC level - so I always make sure that the signal from my G2 receiver is at mic level; either by attenuation within the receiver, or by using an attenuated cable to step down the headphone output level from (nominal) line to mic.

There are some very swish boxes by Beachtek for DLSR's, but I have to admit part of me sees these as being slightly pointless: Is the audio quality on the DSLR itself good enough to warrant them, when you'll inevitably recording the sound separately? That said though, I can see certain advantage for specific tasks. Another thought is that many of the DSLR rigs I now work with have either the Red Rock / Movietube mounting frames on them, meaning that it would be difficult / impossible to connect the Beachtek box anyway.


So why go to the trouble?

The whole aim of developing this work flow, was two fold:
  • Firstly to get the best possible sound when working with a DSLR 
  • Secondly to save production time and money when syncing the rushes during editing. 

As a particular supermarket, who I have no affiliation with, says "Every little helps"

G - 6 April 2011

2 comments:

  1. Good musings, a great start into the blog world. I'm enjoying the DSLR movement at the moment, I feel that on the most part it's brought respect back to our discipline.

    ReplyDelete