I just watched Justin Timberlake's "Say Something" video, and then, I watched it again. No doubt that's professional production work. Knowing the technical challenges of such a video, for me it felt like I was watching a reality show. The almost "unplugged" vibe of the song was so right for that video that the rest of my senses could be focused on how those guys pull that work off (pun intended). This article will be a humble attempt to reveal how they shot the video.
When I watch such work, I'm always trying to figure out how they made it: was there any trickery, how was the audio synchronized, what cameras were used, what lenses, etc. In this analysis, I will talk about what I saw from a technical standpoint, but feel free to leave your comments to correct me if I'm wrong.
The most obvious aspect of the video is that it was shot as a single take on some sort of a stabilization system. The motion was quite stable. I have two options: a classic Steadicam or a motorized gimbal. Let's look at the shot of the elevator:
This is a pretty tight shot. There are other tight shots too. If a motorized gimbal was used, similar to the Movi, it would require significant space for the operator, because both hands are usually stretched forward. Take a look at this demonstration:
Even if they can fit a motorized gimbal in the elevator, it will require a very wide angle lens. You won't notice a cinema wide angle lens by a barrel distortion, because those lenses cost a fortune to not have such. When you try to guess a type of cinema lens, you have to look at the perspective it gives. Obviously the elevator shot does not resemble an ultra-wide angle lens, but probably something around 24mm.
My guess is they used a classic Steadicam, because of this shot here (go to 2:50 in the video):
If you have used a stabilization system similar to the Steadicam, you know that even for an experienced operator, it's difficult to make a shot that looks like the camera was on a tripod. There will be always a slight floating motion. Another difficulty is using a long focal distance or a long lens. Every imperfection in the way the stabilization system is balanced and in the way it is operated is greatly multiplied and visible in the video. That was the only place where I saw a slight, really very slight floating motion of the camera like a reed shaken by a light breeze. You don't usually see that on a motorized gimbal. Also, the classic Steadicam requires a much smaller space to be operated. I must admit that the stabilization system was balanced almost to perfection. They may have used something like the ARRI Trinity too, which is a combination of a classic Steadicam and a motorized gimbal:
Update: By the time I wrote this, I recalled there were credits at the end of the video. There you can see the name Ari Robbins, who was also the "A" camera operator and a Steadicam operator in "La La Land." Obviously, not an everyday guy.
Let the speculations begin! Can this be shot on a mirrorless camera? Of course, but from the magnitude of the production, I doubt they've used stills cameras that are not designed for video work. My guess is either a RED or an ARRI. Most US-based video productions use RED, but many of the Hollywood movies are shot on an ARRI. The image is not that sharp to be a Sony, but I would be surprised if it is a Canon cinema camera, which Shane Hurlbut used to shoot "Need for Speed." I vote for a RED, not because I'm sure, but because I need to "Say Something."
Most of you have guessed. They've used a zoom lens which was operated by a very skilled first AC (Assistant Camera operator). In my article about the differences between cinema and stills lenses, I mentioned the smooth zoom operation of cinema lenses. Obviously, the zooming motion was incredibly smooth and precise, so much so that they've managed to make a push/pull shot synchronized with a Steadicam operator:
Is it an anamorphic lens? I don't think so. We can have an anamorphic lens without the signature horizontal light streaks. Usually the anamorphic footage is stretched horizontally in post and the circular highlights that are not in focus look elongated. I looked at the blurred lights in the background of close-up shots and they were perfect circles (see the first screenshot too). My guess is they used a non-anamorphic fast zoom lens in the range 18-80mm, at probably T/2.6 or 2.8 at the whole time.
Resolution and White Balance
Knowing it's very risky to shoot with a zoom lens on a Steadicam, I think they used a camera with a sensor allowing 6K or 8K, because they may have needed more room to stabilize some of the zoom shots in post. That's why I guessed they've probably used a RED camera.
The video seems like it was shot at a color temperature greater than the usual tungsten light color (3,200 K) and with less than 5,600 K, which is the typical daylight-balanced light. In the beginning of the video, the light looks very blue, while the tungsten ambient lights in the interior look warmer. So, my guess would be a white balance at around 4,000 K.
Working with a Steadicam or a gimbal usually requires someone else who focuses the lens, and in this case, operates the zoom. It is obvious that this person is not following the Steadicam operator, because there's not enough space in the elevator. In this case, the signal from the camera is wirelessly sent to a monitor outside the set where the 1st AC is sweating and turning the knobs, making sure the image is in focus. The choreography of the dance between the camera operator and the first AC is brilliant. I like it especially on the part where they zoomed into Chris Stapleton on the other side of that floor and then they did the push/pull when he was coming towards the camera.
When you light a big set, you need a ton of light. In this video, they used very dim lighting. These seem to be the practical lights of the interior. They may have changed the light bulbs to match the camera sensitivity without bumping up the ISO too much. Having many spotlights does the job of lighting a huge set, but there will be also lots of pure dark corners. In order to work around this, they used haze to dissipate the light and make it softer.
In the places of the interior where there were no suitable practicals, they've installed lights on light stands and proudly showed them in the video.
If you don't know what ADR is, that's the process of replacing audio by re-recording it in post, trying to lip-sync the speaking or singing in the video. That's probably useless information for this article, because I don't think they've used ADR.
Obviously, they've used lots of microphones to capture the ambient sound, as well as the two moments where Justin Timberlake said something (pun intended). You can hear the singers and the musicians cheering at the end too. Here is probably the wireless transmitter and receiver attached to the guitar strap:
In their monitors, they've listened to the audio from the other singers and instruments. The question is: have they used a studio recording playing in their ear monitors and added some of ambient sounds for the video? I don't know. The audio work is very convincing, and it looks like they've recorded it live. It is not impossible, but I think they've used a studio recording both in their ear monitors and as a base for the audio in the final video. I'd be pleasantly surprised if they did a live recording.
Well, I think they can save some money on the editing by hiring a very good colorist and an underpaid film school graduate student.
I've probably missed lots of technical details, but I still think that even these are overwhelming enough. Say something in the comments. I'm curious to read your thoughts on the analysis.