In 2014 I was asked to shoot a documentary for Purplecoat Productions. They were taking on their biggest project to date, performing Shakespeare's 'Twelfth Night' five times, in five cities, in six days; and it would be up to me to capture that week as best I can and put together whatever I shot into a film. It was a huge feat and neither myself or Karl quite knew what we were getting ourselves into when we went ahead with it; it was big.
It's possibly the biggest project I've been part of in a variety of ways, mostly because I was totally on my own to do everything involved whilst on the road, I had to get creative with not only the content but how it would be captured as I can only be in one place at one time. These challenges included:
- Kit limitations due to travel, and flying to Ireland - I could only bring the essentials
- Having only on-board audio - if my subject was at a distance I couldn't always move closer for the mic
- The audio was also recorded separately - every recording required two button presses
- Having enough cutaways for edit, knowing I would need to splice up pieces to camera (PTCs)
- The physical demand of being on your feet all day, and holding kilograms of weight, for a whole week
- Using a GoPro for timelapses, and remembering when to stop/start/move and charge that too
- Keeping all kit in working order - charged, cleaned, not rattling
- Data wrangling on the move, at the end of every day - to two drives, never just one
- Working with a cast you've never met, who don't understand what you're doing; but you are required to interrupt daily
- Having the right bag to carry this all in - my bag contents can be seen in this blog post
To shoot the film I had to go with whatever kit I owned, could borrow, or could buy for cheap. Each person was covering most of their own costs, I had no budget to buy or rent gear; whatever I paid for was with my own money. This added limitations to start with, all I owned to shoot at the time was a Nikon D3100 and a Canon 600D, the Nikon doesn't do manual control whilst filming so that was out of the question; 600D it was then! I also mostly owned Nikon lenses back then, and they were all catered for photos - I didn't shoot video on my own gear enough to buy two sets of lenses - and you can't adapt Nikon autofocus lenses onto Canon cameras, the flange distance is all wrong.
I ended up using my Mum's old Olympus film lenses with an adaptor. These lenses have higher f stops (28mm 2.8, 50mm 1.4 and 135mm 2.8) so definitely better than what I had, but also primes (all zooming by feet then) and attached by a £10 lens adaptor - we were getting somewhere. On top of this, to get the best out of what I had, I installed Magic Lantern to give me 1/3 stop ISO increments, and Technicolour's cinestyle image profile - in hindsight 'Vision Colours' image profiles would have likely got a better image than Cinestyle, I find it noisy and often so de-saturated its hard to come back from.
One issue I came across with this setup was the general sharpness and highlights being clipped at times. In general the lenses were sharp enough, but definitely not L-glass sharp, this combined with the lens adaptor and a low budget consumer DSLR, lead to the image being fine, but not as good as I'd have liked. If I was at f1.4 on the 50mm the image wasn't sharp and the highlights would often look clipped in camera, so I tried to stick to f2.8 and raise the ISO instead when necessary - this is very visible in the footage shot prior to the tour starting at the pub, look at the windows and harsh edges of highlights.
In terms of sound I had no recording gear, no voice recorders, no microphones etc, I didn't shoot audio enough for it to be worth the investment. Also with the 600D recording audio internally was a nightmare and always had a high gain hiss, regardless of input, so I knew I'd be recording externally regardless and syncing later; wonderful. Ideally I'd use a shotgun mic of sorts to be sure I could isolate subjects where needed. I asked anyone I knew, if I could borrow their audio recorders/mics for a whole week of travelling and insist I won't break it. In the end my friend Andrew Larner lent me his Zoom H1, it's small, light, runs off a single AA battery and records to a Micro-SD card. It does record in a cardioid polar pattern (so no subject isolation) but it was better than nothing, and performed well; even when it was dropped once, sorry Andrew.
So far I had my camera, microphone/recorder and lenses; now for support. I had owned a cheap-as-chips chest rig for a while and albeit not great for support and could give you a sore back over time; it was small, light, folded down and most importantly I already owned it - no extra purchase necessary. The only things I added was a sponge to the chest mount to give it more comfort, and a '5mm rod to tripod mount adaptor' to attach the Zoom H1 to the rig. The rig is a little like the Zacuto Run-and-Gun bundle but for a tiny fraction of the price.
This rig was so bare bones, and just pieces smushed together, I was one step off tape and cable ties holding it together, there was no frills: no secondary monitor, no follow focus, no counter-weight, no lens hood or filters, I did have a clip-on 'Loupe' to see the screen more clearly on sunny days but this was rarely used as it often just got in the way. But overall this setup worked and in its hours of use I got to learn it's strengths and it's foibles; I think handing it to anyone who didn't know it, would have been bemused at this Frankenstein contraption i'd put together.
Beyond this for kit I used a GoPro Hero 2 (already owned) with some mounts, that were often taped into place. The GoPro was used entirely for timelapses (such as the one out the aeroplane window), as leaving the Canon clicking away for an hour or more just wasn't feasible: battery usage, noisy shutter, I may actually need to shoot etc. Although the GoPro 2 gives you minimal control and doesn't have the best image, it worked, and in a way the limitations gave me freedom to just tape it into place, press go and leave it alone without wondering if i'd chosen the right settings or the right angle or the right lens; it has on/off, auto everything, one lens, and the JPGs give me room to digitally zoom in post - easy peasy.
Lastly I used an iPhone 4S for shooting on, using the 'Filmic Pro' app to give me the best image out of the phone. The phone had only 16GB of storage so I could only film so much, then had to offload to a laptop quickly. But filming on this was necessary at times of travel, being crammed in a minibus or a London tube carriage with luggage up to your ears doesn't allow for a lot of room for you to get a rig out. The footage did sometimes screw-up and I had to cut around this. This was probably because the phone was a bit old and always nearly full and Filmic Pro cranks out a higher data rate, and the phones storage just struggled, but nevermind, it managed to some degree when required.
The phone was in a drop proof case all week (never had to find out how drop proof fortunately), this did make it bulky but I was less worried of it being broken - priceless.
Overall then, the core kit went as follows: Canon 600D + Olympus lenses, Zoom H1, chest rig, iPhone 4S and a GoPro. It wasn't exactly a no frills bag of kit, but through various tests to see what was absolutely necessary, I knew what to do to get the best results from this, and the lack of options when on the road meant I didn't have to think of anything other than if I was exposed properly and about the story I was trying to find. But if I missed something off the list before leaving for Day one, I had to make do. Always, always plan, you'll inevitably forget or lose one item on a big shoot like this (a battery, a lens cloth, a USB cable) but with lists and tests you should at least get there with what you need.
DIRECTING, AND GETTING TO KNOW THE CAST
What I found harder during this project isn't the camera and tech side of things, but in fact the people aspect. When first meeting everyone, it became very real I knew none of the cast, they didn't know me, and no one quite understood what I was there to do. The first set of footage I shot was of a rehearsal above a pub, this involved interviewing everyone briefly to get on tape who they were, what part they were playing and how they felt about the production. This was useful for me to meet each member of the cast and crew, and to gauge how they were towards me, as some of them may be totally up for being on camera and others may not be interested in the slightest; and it's good to find that out now and not in a week or so when they shove the camera out my hands - that didn't happen by the way. It could be difficult to make yourself part of this tight-knit group who have already been working together for weeks and have their own dynamic, to then move into that could be very tricky, and if it turned out no one was interested in the doc, then it could have been a week of hell for us all.
Luckily this wasn't the case, it took a couple of days for people to understand what I was doing (using Louis Theroux as a reference point) but once that settled in and people warmed to it, and to me, it became much easier. I learned when was appropriate to film them, and when wasn't, but always with the understanding that if they were uncomfortable or wanted to be left alone they can tell me to leave, and I would. I was an accessory to the production, to observe and to film what happens and see what we have at the end and if it could be a good doc; I wasn't the primary focus, the play was, I was secondary - unlike Louis Theroux.
So the edit definitely took it's time, two years of time in fact. This was down to various reasons, mostly that this was an unpaid side project for me, and there was no deadline; if it was my full time job I could have done it much much faster. I was doing the edit in spare time, when other projects weren't in the way. Working in your spare time is fine with various projects (I find photos the easiest for this) but this project required concentration, time and consistency, and the 5 hours after work each day, plus weekends, don't always give you that. I started the main chunk of the edit by taking a week's holiday from my day job and each day making sure I had a rough edit of each city's story, which sometimes was hard to find, and other times was easy. For example Stratford had the flyering to use as the story, with all the cast and crew trying to get the general public to be interested in the play and to come see it; whereas Birmingham was a day of rest as it was the day after Dublin. We had between 1 and 4 hours sleep each and had to fly from Ireland to then get a Minibus and setup the stage from scratch, not a lot of interesting things happened that day.
Other obstacles with the edit was the amount of footage, and working on an ageing machine. I had over 12 hours of footage collectively once you put the 600D's footage alongside any iPhone footage, timelapses and the interviews - where do you start, I shot it all and had an idea what happened, even then it was a challenge. This was why I worked through the cities individually first, giving each one their own story thread. Also whilst trying to find the gold that was buried beneath hours of irrelevant content, there were several parts that were hilarious or interesting, independently, but didn't work in the grand scheme of things, and so were scrapped. There was partial ideas we had along the way that weren't fully there, so fell flat when put into the story, whatever went into this had to make 100% sense to the audience. They had to understand what we were doing, what the journey was, and the work that went into the tour, there was room for exploring funny moments, but we couldn't be too involved with inside jokes and anything too personal to us; it wouldn't make sense to anyone outside the group.
The edit started with a long process of transcoding all footage to ProRes 422 for a consistent edit codec throughout, minimising stress on the CPU swapping between codecs for playback. This was done with Apple's Compressor 3 as it can use attributes 'from source' such as frame size. Whereas Media Encoder CS6 is faster in some regards (64 bit, multi core processing), but needs to be told every, single, attribute, which can lead to mistakes very easily. When it was all transcoded I synced all the footage using Plural Eyes, I had to do each section separately: interviews, Dublin, Stratford etc, or else Plural Eyes would just slow down to a crawling pace. I also edited with all 4 audio tracks, both the 'in camera' and both the 'external Zoom tracks' incase one was better than the other (and that was useful at times in the sound mix). Although this whole process lead to over 500GB of footage (over 1TB including a backup), it was easily editable and didn't tax Premiere too much - try cutting 100GB of H264 footage on a 2012 iMac and see how far you get - if you want to know about codecs, I wrote a guide on video compression in 2014, which can be found here.
In terms of post hardware, I was working on my 2012 iMac (basic specs: i3, spinning boot drive, 256MB VRAM, nothing fancy), initially I was running off firewire 800 drives, and later an internal SATA drive (down to some iMac surgery) but it struggled at anything other than playback so I had to be smart and pragmatic as to how I worked. I was cutting with Premiere CS6 throughout and very quickly the project took about one minute to save, and under fear of a crash (which were not uncommon) my autosave was every 15 minutes; essentially every 15 minutes I lost a minute of time I could cut. Now, I like a render, I genuinely do, I used to enjoy the 'under a minute' FCP7 used to take to do pretty much anything. FCP7 rendering was a choice though, it gave you just under a minute to think about the decisions you've just made, and i'd often cancel the render and change something, then render again. But a Premiere autosave is always as the worst time: right when you are about to press a button, or have just stopped playback, or worst of all, halfway through naming a folder - can it not be in the background, maybe it is now.
Once I had a rough edit of the cities and refined it down to the core story, I re-interviewed the cast and crew. I got together who was available for one day in Liverpool (I live in Leeds so wanted it done in one day, I couldn't be making several trips back and forth, it wasn't practical or cost effective). After knowing who was available I asked specific questions about what had been included, to cue up the next part of the film. I often asked two or more people for each segue so I could cut back and forth between them, not only to give me edit room, but to give more context and to give a clearer view (or even contradictory view) of what was about to happen.
These interviews weren't shot with the same gear as the doc, hopefully this is obvious. They were shot using a Nikon D610, a Nikkor 50mm 1.4, and sound was recorded on a Zoom H4n and a shotgun mic held onto furniture with a gorillapod. I also borrowed LED panels to light them, one of them died part way through Rhea's interview so her face is a bit dark at points; we didn't have time to reset.
Locking the 'final edit' was sort of a big decision, as once I progressed beyond this step I would be in the colouring and sound mixing phase, which would be much harder to unravel and re-edit, than If I made any changes before then. So I watched through the edit several times and made the odd tweak, left it for a week or so, then watched it again with fresh eyes, made more changes and adjustments - then it locked.
Once the edit was locked I had a big process ahead of me: colouring the pictures, mixing the sound, adding any subtitles, making graphics for name straps, and various other little decisions that can make a big impact (font choice being one). To start this process I wanted to be in a new project, the one used so far had become so bloated and big that it was taking several minutes just to open and I knew I could trim that down in one way or another. My first try at this was to use the consolidate option in Premiere, which would essentially copy only the media used in my last edit and then I could delete anything I hadn't used, which could free up several hundred GBs of storage. I'd seen this work before when colouring projects for Pat and Aki but no matter what options I tried or what tutorials I followed it would not work for this project. My other option was to manually weed out anything I didn't need and remove from the project, and then delete from the drive. I started by duplicating the project, labelling it with the suffix 'Online Edit' and then deleting any sequence that I didn't need in the colour, mix and online process. Once these were all gone (and I keep a rigorously organised project so it wasn't too hard) I then used the columns 'Video Usage' and 'Audio Usage' to locate any footage I hadn't used in my final cut; horribly only a handful of clips hadn't been used at least once in the edit, meaning deleting even the few that could be was pointless. The project was made smaller though by removing anything that I didn't need now, such as sequences, the project was now taking seconds to save - yay.
Now with the project slimmed down I reorganised my final sequence for the purposes of colouring and sound mixing, essentially layering the video into types: 600D, iPhone, timelapses, shorts (the two Bond scenes we made on the road) and interviews, these types extended to other layers over the course of the online edit. The audio was also split up into: dialogue, other and music. This process made it simpler to understand the sequence's many layers and kept it easier to manage.
Normally for colouring I use Red Giant's 'Looks' but for the 600D footage this was going to be overkill and waste hours in render time. So I opted instead to use adjustment layers with a general look for each city, using Premiere's own tools, and then any fine tweaks (once the first pass was done) were made by dividing up the adjustment layer and changing just that bit; made the process much faster than selecting each clip individually. The small amount of iPhone footage was done on it's own layer above the 600Ds footage as it needed a different look adding to it - a flat 600D file looks very different to a clip from an iPhone. Above both the 600D and iPhone I added a layer of unsharpen mask (another adjustment layer) as it all required it so why not make it easy in one big go, also means I could turn it off in one go rather than drilling down each clips effects.
Anything above this was done independently as again, there wasn't much of it. There was only a few timelapses and they were all in different places so needed a unique grade. Next was the post-tour interview footage which was all shot under the same conditions so were slightly different depending on where they were sat or what lighting I'd done but essentially very similar; these were then done using Magic Bullet Looks as these would benefit from it. I made the one look initially and copied it to one clip of each interview, tweaked where needed for that person then copied it to each of their clips. Any layers above this were all graphics or subtitles so didn't need colouring.
During the colouring process I made a few tweaks to the edit which were done very carefully as it was laid out in a particular order by then and needed to be done with accuracy or other bits could drop out of sync or be moved under the wrong adjustment layer, or even overwrite other footage. Also I decided to render the sequence as I had finished for the day so I had preview files I could playback the next day; the footage un-rendered with sharpening and colour wasn't working.
The audio mix for this was in a way, was more important than the visuals, if you could see but not hear (or read) what was being said then it would make no sense and there were many times when background noise got in the way of the audio that should be heard; partially down to the cardioid pickup pattern of the Zoom H1 rather than a shotgun mic which would be more direct. Mostly I just used the audio from the Zoom and raised or lowered it's volume, and used crossfades and further editing (extending or shortening cuts etc) to make it sound seamless; this was all done on eight tracks of audio. I had previously left in all four tracks of audio from each clip and in Stratford it came in very useful as there was a wind outside, the Zoom caught every bit of it, but the Canon didn't catch as much so I used them both, the Zoom for the low bassy tones and the Canon's audio for the higher tones, it worked well but I still used subtitles across most of it to be sure.
The music used was chosen as it was used in the play itself, so this gave me a selection of music to pick from, it wasn't copyright free however so that has it's own issues, but this project was mostly going to be enjoyed by people involved and friends/family beyond that so being fully copyright wasn't too important in a way - it wasn't going to festivals. Also I mixed wearing a set of Turtlebeach gaming headphones, not the best audio monitors admittedly as they aren't 'flat' but were all I had and was better for hearing small noises than internal iMac speakers.
On top of the colouring and audio I had many, many subtitles to make as there were various parts that were difficult to hear or understand, and would benefit from a subtitle. Each one of these was a new title card in Premiere (which took it's time), I implemented a similar process to past processes, by making the first one, deciding on it's style and then copying it, knowing that if I decided to change one I had to change them all, manually. As a note for these, each title card needs to be duplicated in the project browser NOT ON THE TIMELINE. My workflow was to make a folder with the timecode of each section that needed subtitles and then number the cards, followed by the text in that card, e.g. 01 "Twelfth Night on tour is the biggest...". I'd then duplicate card 01 and label that 02 etc. If you duplicate the cards on the timeline, the original's content will change when you change the duplicate; you only want to make that mistake once. Duplicating the cards meant the formatting stayed the same and didn't require manual rebuilding each time; also the cards with one line of text would be moved down a few points (using Motion under effects, not in the text editor, you can then copy and paste the effect) to sit in the middle of the other cards; the text was too high and looked weird if it wasn't moved down. Lastly, the name straps were just made in Premiere and animated on the timeline, one layer for the purple background and another layer for the text.
When I had finished a pass of all processes and was 'happy' with it, i'd often export it out and watch it, wth a notepad and pen, in the living room on a bigger TV to hear and see it in a different light (not everyone is going to watch it on my iMac, in my bedroom) and then go back to the edit to adjust accordingly. There were many 'final' files, I didn't quite get to 'final final final USE THIS' but I got close; I think there were four FINAL MASTERS in total, each only slightly different to it's predecessor, but different enough.
For the final export I did a few versions for future proofing the film, and this itself is a process that needs to be done with expert precision. Initially Premiere refused to export and I worked it out to be the audio for some reason (who knows) so I exported the audio as a stereo WAV file and re-imported it (taking it from 10 layers to one) and then I could export. Instead of exporting the sequence and waiting for Premiere to render it out, I used the tool 'Use Previews' in the export menu; to do this I rendered my FINAL MASTER in it's native codec of ProRes 422 (took over 11 hours) and when exporting the finished file, I made all my settings match the sequence manually (not using the match sequence settings ticky box, it doesn't work, everything seems to come out in MPEG) and then ticking the 'Use Previews' option which means if your preview/render files match your export Premiere will just copy those instead of re-rendering everything. This then took a further hour (quicker than 11 though).
The different versions were made for different purposes and some of these were made by removing elements that could cause a problem or could need altering in the future, these version were as follows:
FINAL MASTER - ProRes 422 and uncompressed audio
FINAL MASTER - H264 1080p and AAC audio
FINAL MASTER - H264 720p and AAC audio
MASTER - video only, without subtitles and graphics
MASTER - audio only, without music
The reason for these versions is for different viewing purposes, and or if I need to change a subtitle or name card in the future, or remove a song, I can't keep all 500GB+ of footage for that, this way I can splice in the version without the offending part and change it - the total comes in at just over 100GB, so it's still big but it's under a fifth of the original size.
To sum it all up, this project was massive, it cost me several hundred pounds, two years of time and will largely go unnoticed by the wider world, but, I learned so much, not only in shooting and being totally self dependant when in the field, but in edit also. I'd never cut a documentary like this, I cut one in college about immigration, whilst learning the basics of filmmaking, but this was a different beast. Currently I mostly cut short films and more often than anything, promo videos for ITV's Emmerdale. This was a very different pace, I had to learn to let the footage breath, don't force the pace, it will find itself; and it did. I enjoyed the edit, once I found my feet with the reams of footage, and knew what I should do with it, it was an enjoyable experience. Most of all it was fun as a whole, because of the people, we were like a family at the end of it. We were all going through this hard week together and I think at times having a camera there lightened the mood and gave people something else to focus on at times; almost like a mental rest.
After this tour, Purplecoat went on to do another run of plays but only in Dublin, and we shot that also. Without the travel it was much simpler, and looks a lot better on a technical level; I took better, albeit heavier, kit with me - 15kgs in total, We also shot a short film, a 30 minute zombie-comedy which was filmed over several weekends in November and December in 2014; again another challenging but rewarding project.
And I would do each one of these projects all over again, you'll never learn if you don't push yourself.
You can watch both the Documentary's trailer and full film in the links below, if you have any questions then please do get in touch, either via comments on this post or on Twitter.