Night Vision

Corporate Revolutions by Isaac Marchionna

Following up on the earlier discussion of the BE Meyers 40th Anniversary video, this entry will be more of a technical breakdown of a key effects sequence involving a rather large machine gun, dolly, multiple passes, and a little bit of hair pulling.

7 Second Storytelling

BE Meyers not only builds their own products, but helps supply warfighters with products that enhance existing weapon systems, which then help integrate their own laser products, increasing lethality and survivability. Early on in the planning stages we paid special attention to one key product, which was a chassis system for the venerable M2 .50cal machine gun, affectionately known as ‘Ma Deuce.’ Generally in the planning phases we scale our visual storytelling to the requirements of the client, this sometimes tends to keep things conservative, especially when on a narrower budget. However when we sat down to scope what was possible for the anniversary video, and we laid all the possible product stories out, we were struck by the key visual of the M2 in all its built up glory.

Typically as someone involved in planning, production, and post-production I tend to err on the side of restraint when proposing complicated visual sequences. However putting myself in the shoes of the audience I knew that seeing something as modular as the M2, built up, in motion rather than a static build (or a CAD/CAM build), would instantly become memorable, and a wonderfully challenging visual high watermark. So rather than this being a client request we proposed a sequence starting from the pintle mount, up through the chassis, gun itself, accessories, and optics, all while the camera was rotating around it.

However the challenges were these:

  1. Budget clearly did not allow for a MoCo (Motion Control) rig.
  2. The M2 is exceedingly long. Almost 6’ in length.

The first challenge was the main issue. A motion control rig would have made this effect a snap. However given that budget was limited it meant improvising and creating a human powered motion control rig as best as possible. The other was length of the object being filmed. The M2 is almost 6’ long, which meant that lighting the machine gun would be a challenge as well as controlling focus. 

Tylor Jones prepares the camera to rotate around the M2.

Tylor Jones prepares the camera to rotate around the M2.

The solution we settled on was to utilize our Fisher 10 dolly, which would be used for the majority of the corporate location filming, on a 360 length of track. We would then push the camera around each stage of the gun, with 2-3 passes per stage to allow for coverage in case of a camera issue. We would start with the machine gun fully built up, as it would reduce the chance of the gun being bumped or changed in position as little as possible. After each pass a stage would be removed, another 2-3 revolutions filmed, remove another stage, rinse and repeat. Each pass could not be any shorter in length than the pass before it, so using a simple timer app on a phone we were cautious to count out each step, and focus on smoothness of motion, erring on the side of being too slow than too fast.

Because this would be a complicated effect for so little on screen time, it meant that we had to slot in all of our lighting and camera prep while still trying to juggle A-roll interviews. This meant that this B-roll effect was severely under the gun (pardon the pun) to get right, and get done, in a time faster than normally allowed. In addition we were also limited on track diameter, as to complete a 360-degree revolution meant using a smaller track, which limited us on lens choices. This goes back to problem number 2, which is that the object in camera was incredibly long, but would drift out of focus given available light, which limited our depth of field. Ultimately given the restriction on time, and how many lights we could use, it meant being more than a few T-Stops less than where you'd want to be given how ginormously long the M2 is in camera.

The one key thing that a lack of time didn’t allow us to accomplish was to bust out the rulers, lasers, and bubble levels, to optically true up the camera, dolly/track, and machine gun. Another deficiency was in properly leveling up the track to reduce or remove any low or high spots during the rotation. The end result is lovely looking footage, but one where the camera and gun were moving eccentrically to each other. Clearly this wouldn’t work for the final effect. See example below, particularly take note of the shake, and rises/dips in frame:

Original source footage. No stabilization.

How to solve this? Given that each pass would be give or take 55 seconds in length, with 6 individual passes, there was only one solution…basically a lot of key frames. No stabilization tool would be able to make sense out of a long object during a 360-degree rotation moving eccentrically. It essentially would have warp-stabilized the M2 into looking like a bow tie. The most direct solution would be the most time consuming, which would be to take the one constant, in the form of the pintle mount, and using that as our optically true center point. By essentially throwing a pair of vertical guides, and going through roughly 1350 key frames (per pass) the M2 would rotate smoothly during its entire turn. The same had to be done horizontally, due to the aforementioned dips in the dolly track. All in all this amounted to just a little over 8100 key frames. Due to each pass and the resulting key frames you could essentially see that the camera was making a figure 8 as it moved down and around, whereas it should have been a single point.

Each pass had to be frame stabilized to avoid the M2 dipping or shifting.

Each pass had to be frame stabilized to avoid the M2 dipping or shifting.

Once this was accomplished for each pass all 6 passes were overlaid and transition points determined. I knew I didn’t want to have jump cut transitions, but rather I wanted seamless crossfades. This meant avoiding any visual ‘tells’ such as a tension crank that may have moved during installation/removal, ammo belt shifting, etc. After these 5 transition points were determined, the entire sequence was then sped up to a crisp 7 seconds. We didn’t want to make too big a deal out of this effects sequence, enough so as to let the viewer appreciate the effect, but not do an effect for an effects sake that overstayed its welcome.

Rigging lights and gun.

Rigging lights and gun.

Because we wanted to use this sequence to illustrate all the products that BE Meyers either supplies, or creates, we then overlaid product name call outs and product numbers, all tracked to the camera, so that 3D type would follow each component.

3D Product Names and Part Numbers overlaid.

3D Product Names and Part Numbers overlaid.

Overall I’m incredibly happy with the end result, especially given the time and technical limitations. It was also a firm reminder that even 15-30 minutes during production to measure twice, and cut once can alleviate postproduction time by a factor of 10-20x. Ultimately what we set out to do was to create an effect that could have been accomplished rather uninterestingly in a normally throw away line of A-roll, and some B-roll montage, but in a far more compelling visual manner. The end result was technically challenging, but something that was instantly recognizable for it’s addition to the overall piece’s production values, as well as the entire production team striving to raise the bar on what a corporate piece could be.

Zero Dark Nerdy by Isaac Marchionna

IMG_3228.jpg

Fundamentally there's a few ways to film into the darkness of an evening. Option 1, bring a lot of lights. Option 2, bypass the whole "visible spectrum" and go right into Infrared. That isn't to say Option 2 isn't without it's quirks. All of which presented an interesting learning curve when utilized on a project for a company that specializes in illuminating the darkness. 

IMG_3226.jpg
IMG_3230.jpg

Our client asked myself and my creative partner to film a night 'live fire' rifle shoot using night vision, utilizing their own in-house night vision system. This client specifically creates IR aiming and illumination systems from individual soldier rifles, all the way up to crew-served machine guns. The advantage is that this allows us to give 1-to-1 feedback on how we think their night vision system can be improved from the standpoint of filmmakers.

The BE Meyers OWL night scope is fairly unique in a few regards:
- It utilizes C-Mount lenses
- It creates a non-vignetted image on full-frame sensors

Why the two aspects listed are important is that normal systems, for example the AN/PVS-14, is limited to 40 degrees of field of vision. Suffice it to say this isn't very helpful. The PVS-14 also has two modes of adjustment, either a front objective lens for focusing, or a diopter end for adjustment to the human eye. Things get a bit hairy when you have to adapt a system developed for fighting, to one for fun. This involves step up rings, adapted to a donor lens, so now you have 3 systems of focus to nail, an objective end, a back-focus end from the night observation device (referred to hereafter as a NOD), and finally the focus of the donor optic connected to the NOD.

Simply put...this is a mess. And is an absolute soup-sandwich for filmmaking.

So when presented with the BE Meyers OWL we overcame a few problems right off the bat. First, we get the ability to interchange relatively low-cost lenses, with adjustable IRIS', repeatable and fixed focus system, and one step mounting. In the case of the OWL it was as simple as specifying that we wanted an Canon EF mount (swapping from PL and EF on the RED's side was as simple as removing 4 screws, swapping, and reinstalling).

Overall this means one system to achieve a desired result. As long as the back-focus of the OWL is set correctly this means that you can remove and reinstall reliably without any issues on the EPIC or Optic.

 

Example: 'Zero Dark Thirty' (ARRI Alexa + NOD)

Example: 'Zero Dark Thirty' (ARRI Alexa + NOD)

Unfortunately we also learned a few downsides. The biggest being is that in this project's case we didn't learn till the night before that it used C-mounts, which while a pleasant surprise, wasn't a type of lens we had kicking around. So we were locked into the supplied 50mm. On the night of we simply rolled with the focal length we were dealt. But it does allow for intriguing future opportunities to source more C-mounts in shorter and longer focal lengths. C-mounts also tend to be fairly affordable and fast (this lens was an f/stop of 0.95).

The other technical issue is that it's all focus by hand. This means pulling focus or iris pulls becomes a fun game of guess work. A good quality NOD tube, the photo-electric plate that actually pulls light in and amplifies it a million times to a visible level, is running at about 600-800 lines per inch. This means you're essentially filming a barely 720p image through a 5K sensor, then outputed to a field monitor. As a result it's pretty tricky, but not impossible to get solid focus.

The C-mount lens is also un-geared, which prevented us from utilizing our ARRI Follow Focus, which in combination with a larger diameter focus wheel, would have made focus pulls less of a head scratcher. And because we only received our optic the night before, we weren't able to source any zip gears. After shooting that evening our hope is to see about either sourcing small enough zip gears, or having a set of delrin gears machined, and press-fit, on to the C-mount. This combination would essentially allow us to drive the lenses in the same manner as a normal DSLR or Cinema lens.

Another interesting facet of night production is that you're constrained to using either the natural moonlight (albeit amplified by a factor of a million) or additional infrared light. The OWL has it's own IR illuminator, but this amounts of an IR LED. As a result it tends to overexpose anything within a short throw of the OWL, without really getting the light where's it needed in the case of outdoor environments. Thankfully the products being filmed were incredibly powerful IR illuminators/designators, which meant that by essentially bouncing the light using one of these weapon-intended lights sources, we could kick additional IR light where we wanted.

Our schedule was fairly tight, and future testing would be nice to see how these IR illuminators would behave if used in conjunction with flags or reflectors. There's a whole world of possibilities we just didn't have time to test against. Next time for sure.

Example: PVS-14 + Canon 5DmkIII Source: Roy Lin / Weapon Outfitters

Example: PVS-14 + Canon 5DmkIII
Source: Roy Lin / Weapon Outfitters

None of these are really complaints, but rather areas I see as improvements to what is normally a very finicky proposition when using other systems (PVS-14). But the results are rather spectacular. The OWL doesn't create the typical 'image in a donut' effect that we normally associate with NODs hooked up to cameras (example: Patriot Games). Instead the OWL fills the entire frame with a glorious green image. As a result this gives us the option to later on add back vignetting if we're trying to simulate the point of view of a soldier seeing through a NOD. But for the purposes of a product video it's certainly a requirement not to waste half your image with just straight blackness.

At the end of the day, or rather night as this case may be, it's an interesting experience to eschew conventional optics and visible light, in favor of systems designed for the military. The results were exceptional, and combined with the RED Epic allowed us to get some shots typically not seen on video. We feel that our 36 hours with the OWL were short, but provided valuable take-aways that we can come back to, and improve on.