Metering With Heat: Shooting My Love with a Thermal Camera
Over time we've been fortunate enough to build up the reputation as something along the lines of "the guys you call when you have a technical challenge and you don't expect to find anyone with expertise in the matter". As a bunch of self-described huge nerds, this is a (admittedly rough) title that we very much appreciate. One such occasion was when our friend and director of photography Adam Donald gave us a call for this unique project - the concept of the video conceived by director Ryan Staake centered around photography captured through thermal imaging and he was looking for technical support in making the production happen with the use of such an unfamiliar camera system. Since we worked on this video there have been several high-profile videos made that used thermal imaging (Coldplay's Midnight, Earl Sweatshirt's Grief) but at the time of the production such a concept was truly unheard of and we were incredibly excited about the prospect of entering such uncharted territory.
For the project the production team had sourced a Flir SC8200, which is a top-end science-grade thermal imaging camera. The camera was brand new at the time of the production and was important for its resolution capabilities. Thermal cameras had previously been limited to resolutions in the middle hundreds but with the SC8000 series Flir introduced cameras that were able to record at resolutions in the high-definition range by video standards, allowing us to use it in the context of modern video production without distracting pixilation and low levels of detail.
As we researched and tested the camera it became immediately apparent that the design of this imaging system is not suited to video production. The camera itself is a large metal box, with I/O ports on one side and the comparably small lens port on the other. There are not mounts beyond a simple tripod mount on the bottom. There is no battery system of any sort, the camera instead must be used while powered by the A/C adapter. The camera itself has no buttons or user interface - instead the camera is designed to be used with the bundled software package. The software allows the user immense control over the camera, including a huge amount of high-speed flexibility which is typically what scientific thermal imaging is needed for. However, the software itself does the recording so the camera as such needs to remain tethered to the computer while in operation. All of these design elements make perfect sense for scientific observation in controlled circumstances but presents obvious issue for the much more fast-paced and improvisation-heavy world of video production, and especially in the case of this video which was planned to be shot on steadicam.
After testing we were able to find a way to use the camera that suited our use-case. The camera has an SDI-out port and we found that after setting it up while connected to the computer we could unplug the ethernet tether and the camera would remain outputting signal at our desired settings. We used a Sound Devices PIX 240 ProRes recorder to capture this signal during the production. This allowed us to operate the camera much more like a conventional camera, leaving only the power tether to deal with which was a much more tolerable problem to deal with and one that the camera assistants had dealt with often in the past.
Beyond these technical matters, we had to learn how to work with the camera in a more theoretical sense. Because the camera recorded temperature and not light, we needed to completely rethink the way we understood imaging. Directing photography was no longer about thinking of light in terms of footcandles or how light travels around the set and then through the lens and into the sensor. Probably the strangest idea was that heat doesn’t actually look like anything to the human eye, and the way the camera depicts it is just a representation of the information it records, and as such we could choose how the information was represented. We picked a color scheme for the information (we chose the “Iron” color setting which is reminiscent of how thermal cameras are typically shown in cinema). The information was mapped linearly by the camera’s software, using different colors in the scheme to represent different temperatures. The “warmer” colors represented higher temperatures. “Exposing” the camera then became a matter of setting the range of temperature values to be mapped. Setting the low bound essentially set the black point for the video, and setting the high bound set the highlight limits. In order to get the image we were looking for, we manipulated the high and low temperature bounds in order to get the temperature range in the frame where we wanted it.
This had interesting implications on set. The set itself suddenly became less important, as it was impossible to effectively illuminate it. Because of this, we shot in an empty white cyclorama studio. The video is primarily set in a club, so we utilized the empty space and colder set to simulate the dark room. Instead of using light meters, we found ourselves “metering” with heat; we were thinking about how the various temperatures of things would appear on the camera, and how to add heat in places to craft the image similarly to how we would work with light. We had a handheld Flir temperature gauge, the kind typically rented from Home Depot to find areas of low insulation in homes, to act as a meter, but we didn’t end up using it much - we found we were able to fairly intuitively think about heat without it.
Because the camera was on steadicam and we didn't have to wait on lighting, and production design was minimal, we were able to move very quickly. We got the cast into position, set up the frame, plugged in the laptop briefly to set up the temperature range to eye, and shot. As the day progressed we found that setting the temperature range was very intuitive and we developed a rough language to talk about temperature in reference to the image.
One interesting point in the shoot was when we shot a wide shot of the club from a balcony overlooking the dance floor. We didn’t have enough extras to completely fill in the crowd, so we instead planned to shoot the crowd in sections and composite them together, similar to how the DC reflecting pool crowd shot in Forrest Gump was created. Because we were working in terms of heat, we were able to use the cold floor as a luma drop and key out the black (as the subjects were much warmer and thus brighter).
The shoot went very smoothly and proved that the medium can be adapted well to video production. We hope to have the opportunity to work on a project like this in the future and explore further the possibilities of working with technology and physics this way!