Canon DSLR interview workflow
Before describing my DSLR workflow for shooting interviews let me first explain why it’s worth doing all of this rather than sticking to my EX1 and making life easier. When shooting an interview with something like the EX1 it’s best to put a lot of distance between the camera and the subject and have the subject a good distance from the background in order to achieve a shallow depth of field. Having a shallow depth of field is advantageous in an interview because it draws the eye to the subject and makes the background less distracting whilst still being visible enough to set the scene.
Canon DSLR’s allow you to achieve this look without so much room so they are a great tool for shooting interviews. They do however have need some special thoughts when it comes to workflow.
For the purposes of this description I’ll go into detail about the workflow steps but not necessarily the detail about each step, certain parts I’ll cover in more detail later if required. Where possible I’ll include links to other resources where the individual steps can be learned in more detail.
My usual setup for shooting an interview using DSLR’s is to have my 7D as the main camera with a follow focus, matte box, Rode video mic and an external Marshall monitor attached. I use the 7D mainly due to it working better with the Marshall monitor than the 5D2. It’s such a shame that even after adding the frame rates of the 7D and a few additional features Canon didn’t fix the 5D’s inability to keep running an HDMI monitor at the same resolution during recording.
I then use the 5D2 as a second camera on a glidetrack grabbing subtle dolly shots and close-ups during the interview.
Of course both cameras are set to the same recording format, now that the 5D2 can shoot 25p this is generally 1920×1080/25p in my case. I’ll normally set a shutter speed of 50 and then adjust aperture and apply ND using filters as necessary.
Getting correct exposure can be a bit tricky with DSLR’s so I utilise two main methods. Firstly I have a light meter which is usually a reliable way of getting settings for correctly exposed skin tones. The other method that I use a lot is to use the false colour function on my Marshall monitor. After using this for a while you soon get used to exposing correctly.
One of the biggest changes in workflow that results from shooting video with DSLR’s like the Canon 5D mark II & 7D is handling the audio. Even with the 2.04 firmware update for the 5D2 that allows manual gain control I still don’t record my main sound in camera and prefer using the Zoom h4n.
The main reasons for this are mainly due to control and monitoring rather than the actual recording quality, although I’m sure the sound quality of the Zoom is probably better than that of the DSLR’s as well. Even with the new manual gain controls on the 5D2 you can’t adjust levels during a take, and without the use of an external pre-amp device there’s no way of monitoring the actual sound being recorded.
I record sound on both DSLR’s using Rode video mics and also on the Zoom h4n using my main mics which are either a Senheisser 416 shotgun or Sony ECM-77 Lav mics connect to Sony UWP-V1 Wireless systems. This allows me to monitor and adjust recording levels on the fly. I do a levels check and start the Zoom recording before rolling the cameras. The zoom then continues to record audio regardless of how many times the cameras are stopped and started. Quality wise I record sound as 48Khz 24bit wav files.
If using wireless lavs I’ll often carry the zoom and wireless receiver with me allowing me to monitor the audio through headphones at the same time as being able to move around working with both cameras. Of course you need to be careful not to pull any cables etc when doing this and I always use the hold feature on the Zoom to make all of the controls inactive.
I’m always paranoid about making sure my media is protected so as soon as possible so when card is full or I reach the end of a shoot I copy my cards to my Nexto DI NVS2500. It’s always very reassuring to know that my footage and audio is then on the original cards and on the NVS2500′s hard drive.
Another nice thing about working with the NVS2500 is that when it comes to setting up a project and ingesting footage I can get all of my media from this one device. The first thing I do is copy all of the media from the shoot onto my macs media storage drives. I like to copy the entire directory structure from the CF and SD cards over to these drives, this is especially important if using Canon’s plugin-E1 for Final cut as the plugin expect to find original directory structures.
Ingest and transcode
When it comes to ingesting the media for use in final cut there are a few things to consider. Final Cut does not like working with the native H.264 files that the Canon DSLR’s produce, it works but it makes editing a painful experience. So the first thing I do is transcode the footage into a more usable format. I like to convert the footage to Apple’s ProRes 422 (LT) format. This can be done in a number of ways.
The first option is to use a free application called MPEG Streamclip. Using this software is a simple process that involves manually dragging all of the .MOV files generated by the DSLR’s into the app and having it convert them all to ProRes using it’s batch list feature. This process is by far the fastest way of transcoding footage I’ve found so it’s the one I use if pushed for time and I don’t need metadata or timecode in my clips.
If using MPEG Streamclip I create the ProRes files in a ‘Prores’ directory on my scratch disk (The disk defined in Final Cut as my ‘Video Capture’ location). As with the EX1 I always keep the file structure from my original cards intact on my ‘Video Store’ drives. I look at the ProRes files as replaceable as they can always be recreated at a later date as long as you keep the original files and directory structure from the CF cards.
It’s important to me to keep my workflow with DSLR as similar as possible to my EX1 workflow as I’m then working in a way that I’m familiar with and I work more efficiently.
The second option, and one that’s only recently become available is to use Canon’s E1 plugin, this utilises the ‘Log & Transfer’ feature in Final Cut and gives you a lot more logging options than the previous method. The E1 plugin allows you to preview clips, mark in and out points, make clip notes and much more. Any imported footage using this method also contains timecode based on the time the footage was shot so it makes working with DSLR footage a lot more like using footage from a dedicated video camera like the EX1.
The downside to using the E1 software though is speed, in my experience so far it seems slower at transcoding to ProRes than MPEG Streamclip so you need to weigh up the benefits of both options for yourself. The E1 plugin has only been out for a few days at the time of writing this so time will tell if I start preferring this method. I do like the idea of keeping my solid state workflow as similar as possible between cameras and this certainly makes using DSLR’s very much like the XDCAM-EX EX1.
To give you an idea of the speed difference I ran a quick test using a 5 minute 16 second 5D2 file on my Mac Pro. To be fair this mac is getting towards the end of it’s lifespan but it’s a quad core 2.66Ghz machine with 8GB RAM so still fairly beefy.
You can see that that MPEG Streamclip transcoded this clip in 7:38 whereas the E1 plugin took 10:27.
When using the E1 plugin the ProRes files will generally be created on whatever scratch disk you define in Final Cut as your ‘Video Capture’ location. If you save a project before importing the footage the files are stored within a subdirectory matching the name of your project, if not they will end up in an ‘untitled project’ folder. I don’t tend to use in and out points when importing footage using the log & transfer plugin, I import everything as complete clips and worry about choosing what I’m going to use in Final Cut. I find that to be more efficient than deciding which part of my clips I’m going to use in the log & transfer window. It also makes it easier if you have to recreate the ProRes files at a later date.
One thing to note is that by default the E1 plugin transcodes to ProRes 444, which makes huge files that are really not necessary for the DSLR footage. To change this click on the gear icon (middle top of the window) and change the EOS Movie option to ProRes 422 (LT) or whatever you prefer.
You can also transcode the footage using apples compressor software but I’ve found it ti be quite unreliable when running large batches. Even using droplets can be problematic so I recommend both of the options above over compressor.
Once I’ve transcoded and imported the footage from both cameras into Final Cut, the next thing to do is import the audio from the Zoom and sync everything up. For this job there’s a fantastic plugin for Final Cut called ‘PluralEyes‘ made by Singular Software. The plugin costs around £100, or $150 respectively. There’s a 30 day trial of the software available so if you haven’t tried it out yet then I highly recommend giving it a go.
Have a look at Singular Software’s tutorial videos, they do a great job of explaining how simple pluraleyes is to use.
This is where recording sound on the DSLR’s comes in because what pluraleyes does is look at the waveforms from each track and line them up with the wav file from the zoom accordingly. I’ve posted a full review of pluraleyes if you’re interested in finding out more about it.
The alternative to using pluraleyes is to sync everything up manually, this is fine if you’re prepared to mark each shot in with a clapper but I find it much easier not having to worry about doing that.
Storage & Archive
I leave the transcoded files on the scratch disk while working on the project then at the end of the project I use the Media Manager to copy all of my sequences and any clips used over to an archive drive. Media Manager allows you to only archive the parts of each master clip you use plus a defined amount at the start and end of each clip, known as ‘handles’. This Means that my archive of the project contains all the source files it needs should I need to re-export it in the future, but it doesn’t contain all of the unused clips. The files on the scratch disk are then deleted because I know I can recreate the transcoded ProRes files from the originals at any time in the future should I need to edit the project at a later date.
I use an 8TB Drobo to backup both my video store drives and my scratch disk throughout this process, my files are never in one location alone.
Hopefully that explains everything, if there’s something you’d like more detail about or have any questions please leave a comment.