Of B-RAW Tests, Some Green Screen in Outer Space, and Week 1 of Principal Photography
This week saw the beginning of Principal Photography for both FMP projects “The Tragic Case of Gilly O’Connor” and “The Pillars of Creation”. Already the previous weeks have been extremely busy with organisation, kit prep, an entire project rejig for Gilly’s as well as camera, material, and lighting tests for both FMPs.
Since a lot has been happening behind the scenes, allow me to expand on the progress of both projects. This summary might become a bit incoherent compared to my usual setup of my blog entries, but this is due to the fact that a lot has been happening simultaneously, which informed and re-informed a lot of my creative and organisational decisions along the line.
So let’s dive straight in and please bear with me until the end.
Monday, 17 February 2020 - Camera Tests in ProRes HD and 1080p
The beginning of this week saw the long-overdue latitude and dynamic range tests for the BMPCC4k. As I already blogged about in my blog entries “Camera Tests, Pt.1: The Theory of Camera Testing” and “Camera Tests, Pt. 2: The Practice of Camera Testing” for my Minor Project, I conducted camera tests for my FMP in order to know how to push these cameras and, quite frankly myself, in obtaining better imagery.
Now, ever since then it has been revealed to us that we won’t be using the cameras that I conducted these tests on, but that we would instead go on a BMPCC 4k, which was ordered last November together with a Ronin-M gimbal and a Metabones speed booster to make up for the smaller sensor of that camera.
Now, sadly, this particular camera has not arrived yet as the manufacturer Blackmagic Design is currently having issues in supplying enough of them to please the demand, so we have instead been working on Has’ personal BMPCC4k.
However, since it does not seem likely anymore that we will get that particular camera in time for our FMPs (the current delivery time for the BMPCC4k lies at approx. 4-6 months), I decided to re-conduct the camera tests of my Minor to test Has’ camera and this particular sensor.
As such, Hannah, Has, and I dedicated Monday evening to conduct these camera tests following the procedure that I applied within my Minor project. Like back then, I wanted to test how the BMPCC4k deals with colours, skin tones, blacks, whites, and green screen in under- and overexposures.
Camera and Material Tests (correct exposure at f.5.6, ungraded)
However, since we also had a chat with our costume designer for Pillars and needed to see how some fabrics play out on camera, we also combined these camera tests with material or fabric tests.
Camera and Material Tests (underexposed by -1, ungraded)
Although the footage still needs to be processed in post in terms of latitude, simply judging from the in-camera results, it was already obvious that this camera fared much better in underexposure than in overexposure and was thus more closely to the results of the BM URSA 4k camera than the older model of the BMCC 2.5k. As such, and just judging from the in-camera results, the BMPCC4k seemed to be safe by overexposing for 1 stop and underexposing for 2 stops.
Camera and Material Tests (overexposed by +1, ungraded)
However, there’s a caveat to this, which needs explaining.
Whilst these tests were replicated the same way like my previous one, there was one major difference to them – namely the application of a speed booster. I will not venture into too much detail about it as I already blogged about this in my blog entry “Of Camera Sensors and Speedboosters”, but since the speed booster is essentially adding an additional stop of light to the dynamic range of a camera, it is safer to deliberately underexpose the image by that stop.
However, this also has a disadvantage to it, as this additional stop of light is often considered to be of inferior quality, making it easy to irreparably overexpose footage and make the image blurry with light (which, in turn, makes it even harder to focus). Which is why I would base my observations above solely on the fact that we used a speed booster and not carve them in stone.
An Unexpected Surprise
Ever since the beginning of this project, I was repeatedly being asked whether I wanted to shoot in a higher resolution than HD, which I repeatedly denied due to the fact that we hitherto had a limit to digital space on our edit suite that prevented us to do more 4k footage, especially for our longer drama pieces.
Anyways, having thus conducted the abovementioned tests in HD and in ProRes HD (the Apple codec with the smallest compression rate and thus slightly more colour bit depth), I realised that we quickly ran out of SD-card space. Being unhappy with the amount of card space that we would burn on only a slightly better bit depth than with ProRes 422, I decided to have a look at how much time and space would be taken by the proprietary Blackmagic RAW (B-RAW) codec and with its individual compression rates of 12:1, 8:1, 5:1, and 3:1.
And this was the first surprise: While the quality of the image in-camera didn’t look too different in HD between B-RAW and ProRes, the field of view was naturally scaled down for HD (known to be the windowed effect). As such, I then decided to just pick 4k for fun to see what the image would look like.
And lo! and behold!
Not only did the colours and image in general look better, we also had more estimated running time on the card itself shooting in B-RAW than shooting in ProRes HD, at least for the 12:1, 8:1, and 5:1 compression rates. Even with the lowest compression rate of 3:1, we we’re looking at just a few minutes less on card, but the image looked so much better, nuanced and full of colour.
Sadly, however, we did not have enough time left to conduct these tests again in 4k B-RAW 3:1 before closing, so we left it for another time.
Tuesday, 18 February 2020 – Considerations
Having considered that outcome for the entirety of Tuesday, I finally decided that we could indeed, possibly, move on to filming in B-RAW with a 3:1 compression rate. Since I have already worked on RAW footage previously with Digitalfire and was amazed by the amount of flexibility it allows in post (and by the nicer colour space within the image itself as well as a nicer fall off of light in general), it was a tempting suggestion.
This would not only make “The Pillars of Creation” look more like a Sci-Fi piece, it would also help alleviate some of the colour correction and colour grading issues for the void scenes for “Gilly O’Connor” and give a bit more latitude on both my end as DoP as well as on Has’ end as the colourist of this project.
As such, I decided that we would redo the tests and at the same time conduct a stress test on the edit suite to ensure that we are not only able of capturing the footage, but also editing it. Thus, I sat down with our technician demonstrator and discussed this tests to which he agreed providing support to our editors as editing in B-RAW will come with a different procedure.
Wednesday, 19 February 2020 – Camera Tests in 4k, B-RAW 3:1
Thus, we redid the tests of Monday on Wednesday morning, capturing a total of 40 clips to be ingested into the editing process and see how our Edit suite would handle the pressure.
At the same time, we also added some additional LED lighting tests into the mix. As you might remember from my blog entry “Creating an Artificial Sunrise: Of Lighting Workshops and Research on LED Strip Lights” last month, we ran into some issues with the refresh rate of the LED strips that I ordered, causing the image to flicker at certain stops and colours.
Whilst I thought that this issue could not be solved without us going massively over budget in obtaining the professional RGB-LED strip light setups for filmmaking and thus thinking about a different solution, conducting the camera tests revealed another surprise.
Another Unexpected Surprise
Whilst going through the settings of the BMPCC4k, by accident I detected that the camera had an inbuilt option of swapping between a refresh rate of 50Hz and 60Hz – and the camera was set to 60Hz itself!
So the reason why most of the lights – apart from the primaries – flickered in our previous test was due to that setting! Immediately swapping around and playing a bit more with the LED strips, we realised that we had a lot more to work with.
Whilst certain colours and especially the lower intensities still did flicker to some degree, we had a lot more room to work with as we moved from a maximum of 2 stops of brightness to choose from, to up to 4 stops depending on the colour. We were even able to get more amber into the mix as well, whilst purple – however – was a lost cause entirely. Good thing though we did not plan on using purple for our lighting design!
Later that Wednesday – Green Screen for “Pillars”
Now, whilst “The Pillars of Creation” will not really go into production until mid-March, we already started filming some of the exterior scenes that take place in outer space in order to give our editor some time to work on the footage before the big workload is approaching.
As such, we planned of filming in the studio using the green screen, a model of the space ship and a model of the asteroid. However, this plan was cut short mildly by the fact that our model maker did not have the models entirely finished yet. Whilst we were able to get the model of the space ship done by 12.30, we had to wait until the evening (almost closing) to get the model of the asteroid. Which meant that we had to reschedule that part of the shoot to another day.
Looking at the amount of lighting tests I have conducted so far, it might seem odd that there were almost no lighting tests done for that particular setup yet. However, this was due to the fact that it was practically impossible for me to test and practice the lighting or rehearsing the shots until the model was fully finished as we could not borrow it and could not test the lighting properly without seeing the final outcome of the paint job done on the model.
As such, we started by decking out the studio in green fabric, using the curtain and floor as a backdrop. For reference images, we used the storyboard and reference images from NASA and carefully started crafting the lighting scenario by adding lights piece by piece, starting with the key, then fill, and lastly kicker before gelling them with their respective colours.
In order to be able to pull this off, I looked into reference images from Sci-Fi films to see how they emulated highlights in outer space and made them my basis for my study.
And whilst a seemingly single-lit solution would indeed be truthful in our galactic neighbourhood (where the Sun is the only close object of light), this would not hold true for the area around the Pillars of Creation. Here it is important to point out that – according to my research – nebulae in outer space actually do emit visible light, and that this visible light needs to be replicated on the object as well.
Since we wanted to make sure that all green screen footage that we capture is actually usable, we thus lit for the longest time using the false colour function of the BMPCC 4k to ensure that what we pick up as a unified colour in the green screen is also picked up as such by the camera. This, while taking a bit of time in the setup, would save us much more time in post – time which will be sorely needed anyway. This decision was furthermore enhanced by the fact that we now decided to film in B-RAW 3:1 and thus would have a lot more colours contained within the image (thus also a bigger variety of green pigment and noise when Chroma-keying).
After more than 2,5 hours of lighting, we finally started liking the result and started filming. As such, we initially wanted to use a combination of sticks and gimbals to recreate different styles of shots that are both static, slightly moving, and moving. However, these plans again had to be post-poned as the model did not fulfil the requirements we had.
The first issue that we encountered was the fact that the model hadn’t been suspended from strings, which we initially asked for so that we could have it hang freely and thus help us with the free-floating look that we aimed to have. Instead, the model was put on a stick that requires to be keyed out in post, which limited us in terms of movement.
However, after another couple of hours of filming – and the static shots of the capsule covered – we wanted to move on to using the Ronin-M gimbal.
Here, we ran into more issues.
As mentioned in my blog entry “In Equilibrium: Balancing the Ronin-M and Ronin-S Gimbals”, and even though Has and I pre-balanced the Ronin-M and worked hard trying to suss out the measurements for each and every lens setup, it quickly turned out that these were not applicable for this particular setup with the batteries from the stores: Turns out that the identically constructed batteries for the Pocket coming from the stores do have a different weight and thus our pre-balances did not work out as planned…
Looking at the possibility of losing a lot of time (we only had 45min left until we needed to be done in order to set up the studio for the Gilly shoot), we instead tried to cover some more movement shots on sticks by alternatively moving the model instead of the camera and see whether we would land anywhere suitable.
To cut a long story short and give it an anticlimactic end, nope we did not.
Since time was now almost done and since we were not able to either film with the gimbal or the model of the asteroid, we decided that we would reschedule these shots for another day and return to them with a new master plan. As such, we started decking out the studio using bin bags and carpet protection to create the floor that we would need for filming “The Tragic Case of Gilly O’Connor” on Thursday and Friday.
Thursday 20 February and Friday 21 February 2020 – Gilly Goes Into Production
Considering the Principal Photography for “The Tragic Case of Gilly O’Connor” there was nothing remarkable to state for the first day. Since we had previously decided that Hannah would take care of assembling the shot list for Gilly in her function as a camera op, thus freeing time up for me to get the equipment and studio bookings, as well as the final lighting plans and all the other prep of my departments in order, we breezed through the first day like a charm and even managed to grab some more shots than initially planned.
However, some lighting circumstances could sadly not be tested by me as there were a lot of very short term changes to the script and production design for which I could not accommodate anymore tests within the time and restrictions on equipment and studio space given and without compromising camera tests and principal photography for my second FMP project.
Looking at the lighting, this still took a bit of tweaking for the initial setup but worked like a charm once done and only needed minor tweaking for some particular shots. In order to make the process smoother for everyone involved and keep as much continuity as possible regarding the lighting and placement of characters and objects within the Void, I set down markers and shot everything against the same wall to avoid shadows in the background.
In order to make the shots workable in the edit, I furthermore ensured that especially the character of Lucifer had a visible backlight and thus outline and that he wasn’t melting too much into the background for the shots where we wanted him to stand out.
For this, and in my function as Head of Department for Camera and Lighting, I decided to ask Has to help out on the shoot in the function of stills photographer, grip, spark, and most importantly as colourist. As such, we made sure to have the correct exposure at all times and tweak lighting where necessary.
Sounds like a lot of smooth sailing, eh? Not so much for Friday though.
Friday turned out to be more of a challenge as we were not only a couple of hours shorter in time due to the university closing at 4.30pm, we also had a photo shoot and a voice over session scheduled that needed to happen for both FMP projects and that could not be rescheduled to another day but had to be done then.
Furthermore, we were derailed by bad atmos sound due to a raging storm that paused most of our takes. Lastly, this day had a lot of complicated setups that required time to get the symmetry right in order to make the shots work seamlessly in the edit without causing any jump cuts.
Today, Sunday, 23 February 2020 – Of Rain, Coldness, and a Changed Hordon Beach
Lastly, this week’s round of principal photography was rounded off by a session of early morning filming at Hordon Beach, which we recced a couple of weeks prior. This session however was met with cold weather and rain. Since we were not being able to reschedule without running into serious scheduling issues across all projects, we decided to keep filming despite the weather.
Since the temperatures and especially the number of people required had to be kept to a minimum whilst still being feasible to work with, I decided to prep for a minimum of kit to be used only by a skeleton crew that consisted of Chloe, Hannah, Katie, Lynley, Has and me, plus our actress.
Apart from the weather issues however, and despite our light kit today, we ran into issues with our location, for the tide and weather had worn away a lot of the really nice embankment lips that we initially wanted to use as a cliff when we recced the place the first time.
Furthermore, due to the bad weather lately, a lot of earth has loosened around the embankment and broken off, which made it not only difficult to obtain the lovely imagery that we initially obtained during our location recce, but was furthermore and more importantly dangerous to film: Upon quickly re-scouting the area on my own, I could hear and feel some parts of the embankment crumbling underneath my feet alone, making it unsafe for an entire crew with kit to be perched atop.
Upon looking at some other alternatives, I found some areas that were safe to stand on due to the amount of rocks and wall rocks underneath it. Sadly, these were the areas that did not have as lovely an imagery as the previous ones and I quickly learned that even a very simple looking beach in the UK can look completely different within only a few weeks.
Anyway, I thus made the executive decision to prioritise safety over the lovely location that we wanted previously. This was also enforced by the circumstance that the tide had also washed away a lot of the bigger rocks and stones that we wanted to use, which forced us to relocate to a different part of the beach anyway and make do with what we received.
As such, the first half of the beach was filmed with the camera department (Hannah, Has, and myself) on top of the embankment, whilst the rest was underneath it capturing the sound and preparing our actress for her role.
In order to ensure that all of us were safe standing atop the rest of the crew, we decided to place our actress a bit further away from camera and instead use a longer lens so that she would not lie underneath in case the camera should drop down on her or the embankment start to slide down. At the same time, Has and I ensured that we spotted Hannah as our camera operator by me holding onto her backpack and Has holding onto my belt.
Lastly, since it was raining down and since the BMPCC’s vents are atop the camera (making the camera very sensitive to rain and sand), we protected the camera and any other equipment using see-through bin bags. Since the rain became stronger, slating and logging the footage on paper was futile as the ink would just run off either the slate, so we decided to log digitally instead using the proprietary function of the BMPCC 4k.
Anyways, despite the abovementioned issues, and despite the fact that our actress was freezing cold, we tried to capture as much footage as possible without endangering the equipment or ourselves. However, one tremendous thing that I learned on that day was to bring some more additional jackets and towels on set to, firstly, keep our actors warm and, secondly, to dry off our actors as well as the equipment. Whilst we were bringing stuff on set, we underestimated that particular aspect in the attempt of keeping the kit and crew light and easily manoeuvrable.
Anyways, we will see what the following week will bring.
References:
Svea Hartle. (2020) Camera and Material Tests for the BMD Pocket Cinema Camera 4k (HD, Correct Exposure). [online] Available at: https://youtu.be/hHH6V4SXNiw [Accessed on 24 May 2020].
Svea Hartle. (2020) Camera and Material Tests for the BMD Pocket Cinema Camera 4k (HD, -1 Underexposed). [online] Available at: https://youtu.be/ylk6e2Q9DhU [Accessed on 24 May 2020].
Svea Hartle. (2020) Camera and Material Tests for the BMD Pocket Cinema Camera 4k (HD, +1 Overexposed). [online] Available at: https://youtu.be/ir76AozgLgk [Accessed on 24 May 2020].