How Did ‘Mr. Robot’ Pull Off That Audacious Single-Take Episode?

It took me several minutes into tonight’s Mr. Robot before it occurred to me that there hadn’t been a single visible edit.

This was a very promising sign at the start of one of the most purely entertaining Mr. Robot installments ever.

Scenes presented as a long, continuous take — a “oner,” in industry parlance — were once so difficult to accomplish that they were only done for very special occasions, usually by master directors: the famous opening sequence of Orson Welles’ Touch of Evil, or Henry taking Karen into the back of the Copacabana in Goodfellas, or Alfred Hitchcock’s Rope, which was an entire movie presented as a oner, even though each take was only as long as the amount of film the camera could hold.

Advances in both camera and visual effects technology have been so huge, and so fast, that oners have begun to feel both common — on TV alone, True Detective, The Knick, Quarry, Better Call Saul, and even It’s Always Sunny in Philadelphia have all done memorable oners over the last few years, plus we’ve had plenty of all-oner films like Birdman — and at risk of seeming, as Mr. Robot creator and chief director Sam Esmail puts it, “gimmicky,” where the showmanship takes precedence over the story.

With “Runtime Error” — which joined The X-Files‘ “Triangle” as the rare TV episode to be presented entirely as a oner — Esmail found a way for the gimmick to enhance the story, rather than overwhelm it. It’s logistically audacious, following Elliot and then Angela up and down multiple levels of Evil Corp’s high-rise headquarters, out into the street and then back in during a chaotic riot, yet at the same time feels incredibly compact because of the caper movie-style premise of it: Elliot wakes up from a long stint as Mr. Robot to discover that 1) Mr. Robot and Tyrell are about to blow up the building he’s been trying to protect all season, which he can only stop if he has access to Evil Corp’s internal computers, and 2) He is about to be fired and kicked out of the building. We are with Elliot for every second of the story until things pass off to Angela, who’s on Mr. Robot’s team and terrified of the violence happening all around her, and while there are a few visual flourishes that call attention to themselves, for the most part the lack of visible edits only makes the story more exciting and intense. (It also makes it — in a rarity for a show this dour by design — more fun.)

Earlier this week, I got on the phone with Esmail to discuss how an episode conceived only as taking place in real time instead became a oner, the many production challenges involved — including a disastrous first day of shooting that nearly forced him to abandon the conceit — where you can look for some of the edits (particularly since the interior and exterior of the building are in different places), how he convinced USA to air the episode without commercials, and how he and his team pulled off some of the more challenging shots:

How did you decide you wanted to present the episode this way?

The conversation initially started with telling a story in real time. From a story perspective, I wanted to experience Elliot the minutes after a switching [from Mr. Robot] occurred. Does he remember everything right away? Is it foggy? Initially, it was just really wanting to be in the minute to minute with Elliot as he’s disoriented and trying to find his footing. And then adding the pressure of this attack on his shoulders while he’s in the midst of this fogginess. The real time doesn’t have to dictate a long take, because one of the movies I referenced early on was Nick of Time with Johnny Depp and Christopher Walken. It’s actually one of my favorite movies — very underrated. That was obviously edited, not a long take. And then in talking to my cinematographer, we looked at all the scenes and we decided that there was something more visceral about staying in this long take and being with Elliot, walking with Elliot. Because we have the stream of consciousness through his voiceover, it felt to match that best to film him as if we’re just following him around in one take.

Then once the episode shifts POV to Angela, we were going to do our first edit there. And then we decided to opt to keep the take going. We discovered that with Angela, if we just shifted the camerawork from Steadicam that would reflect the floating feel that reflects Elliot’s fogginess, once we got with Angela, we’d go to handheld, and it would reflect Angela’s more kinetic storyline of trying to break into the room and get into this air-gapped computer. When we figured out that, we realized we didn’t really need an edit point to shift POV, that it would, in fact, be more interesting to keep the long take going and shift the camera style and camera movement. That’s how it evolved from this conversation about real-time storytelling to how we filmed it.

Have you generally been a fan in the past of movies or TV shows that were presented as one long take?

I remembered loving Rope, but loving it for the camerawork. And I do love a long take because of the visceral experience. The Copacabana shot from Goodfellas being an example of one, or Boogie Nights — a lot of P.T. Anderson’s shots have this sweep-you-off-your-feet energy when he engages you in this long take. But it really is dictated by the story. Sometimes, it can come off as very gimmicky. With the advent of digital technology, where you don’t have to do breaks for film reels, a lot of people want to do a whole film as a oner. It started to be more about the gimmick than about how it viscerally gets you into the story. I was obviously in awe of when the masters did it right, and a little bit more turned off by the people who turned it into a gimmick.

I will say that I did not know that this had happened, so I had to go back and watch it: The X-Files did one. I only went back and watched it after I filmed this episode. I thought it worked great. I loved it. I thought it was great, especially when they’re in two different timelines, and two Gillian Andersons are walking down the same hallway. In an example like that, that to me was riveting and amazing. But I felt like it was right for the story, but I wasn’t all about the gimmick of pulling it off for a TV show, or anything like that.

In both pre-production and actual production, was this a lengthier process than for an average episode of the show?

It was a crapshoot. We were blockshooting [filming multiple episodes at once], so we didn’t shoot 305 all sequentially. It was sporadic around the first block. But the days we were shooting a scene from 305, we knew we were all going to go home early and be with our loved ones for dinner, or be there till six in the morning. It all depended on, “Did the take come off well?” Sometimes it took four takes, and sometimes it took 27 takes. That was the max that we hit one day. But even with those shitty odds, where you have no idea what to expect that day, we were all super excited, from the cast to the crew. We were always up for the challenge. There was just something electric about doing it. It united everybody. My production designer had to be on set, my costume designer. The camera operator and our cast had to really get into our choreography. Generally, in filmmaking, everybody’s off doing their own things, and then you shoot a scene three, ten seconds at a time. Here, it was like putting on a play every day. It was great.

You can hide edits more easily these days. What was the longest continuous physical take?

The longest take was the second scene: Elliot walking from the elevator bank on his floor, sitting down at his desk and then having that conversation with Samar. And that lasted all the way until he first logs in with his computer. And that was the first scene that we shot. I believe our camera operator collapsed at one point, and we had to break for lunch. That got up to 24 or 25 takes. We had not done this before, no one really knew what to expect. We were supposed to shoot that in three hours, and it ended up taking most of the whole day. It threw our whole schedule out of whack.

When that happens on the very first shot of the episode, did you think, “Maybe this wasn’t such a great idea, guys. Can we shoot this more traditionally?”

[Laughs.] Absolutely! Everybody was scared, people were yelling at each other, we were all just panicking. By the way, this was the second day of shooting of season three. I think it was a very long lunch break that day, because we also didn’t know if our camera operator was going to be okay. I’m not joking when I say he literally collapsed and we had to stop shooting. After lunch, I kept saying, “Let’s just see how it goes after lunch.” And then, I think it took our camera operator two takes, and he got it down. And then I think we did another scene from 305 later that day, and it went much faster, and all those fears went away. We were like, “We knew this was going to be challenging, and we’ll ride it out.”

In Rope, the camera just sits on somebody’s back or on a piece of furniture while they changed the film and started the next take. Where did you figure out where you could put the seams for this?

The easy ones are people wiping the lens. It’s always about finding a way to get in front of having something wipe between the camera lens and Elliot or Angela. That generally seemed to be the rule of thumb. We had our VFX supervisor on set to figure that out. There were moments where I said, “I don’t want to do that.” When we would shift from handheld to crane or Steadicam to handheld, those were the tricky moments, and you just had to be very meticulous. There was a shot in the hallway that was empty. There was no one in the shot, no one to cross the frame, no door beam to pass by. To create a stitch like that, you would literally have to match the frame from the A side to the B side, which means every bit of debris on the floor, every crumpled piece of paper, every bit of graffiti, had to be exactly in the same spot. It would take literally 20 minutes for the camera operator to sit there and match exactly the frame prior. But generally speaking, it’s about finding that thing between the camera and the main character.

I want to talk about a couple of shots in particular. The first is Elliot and Darlene in the street, and then you circle all the way around to the TV news crews getting ready to film live reports, and then push in on the riot as it starts and on some of the rioters as they break into the office building and start attacking people and property. How did you pull that off?

The Elliot/Darlene scene was obviously shot all in one take. That actually started from Elliot leaving the building. The exterior of the building is not the same as the interior of the building. So we had to deal with the fact that we’re on sets in three or four different locations and we had to find ways to make it look like one location. Once we got the Elliot and Darlene scene done, the camera would blow past them, follow somebody walking away, and then that would create a stitch as we wipe off of him and we go to the news crew. All of that, we’re on Steadicam still, because we’re on the Elliot chapter of this episode. Once we finish with the newscasters, we then follow what will become one of the Dark Army rioters into the crowd, and that’s when the camera shifts into handheld, and that’s when we created a traditional Hitchcock or Chris Carter-style going behind someone’s back and creating the stitch there. Now, when the riot starts going apeshit, and they’re raiding the building, remember that the exterior of the building is not the interior that we shot, so that stitch was created with the smoke that comes up when the tear gas goes off accidentally.

The second shot is the one where we’re following Angela, and the camera tilts 90 degrees to shoot her as if from above the ceiling as we follow her from room to room, eventually winding up with her in the computer room where we are seeing her in that room but also the view from just outside the exterior wall of the building. How did you pull that off?

That was the shot I was referring to when we’re on an empty hallway. We’re handheld following Angela on the floor. They’re beating up the security guard, and we had to get on the crane to get above Angela and follow her through the ceiling cam mode. That was the stitch where we had to match all the props and elements exactly right to make that work. Then Angela comes in, and we’ve moved from a location to a set, because it’s very difficult to shoot through ceilings on an actual location, and we’re above Angela and following her through this. I remember pitching the idea of showing the interior and exterior at the same time in the pre-production meeting, because I thought it would be interesting to show Angela, this small figure in this larger scheme, and the highrises of this evil corporation we’ve been touting since the first season, to show how small of a cog she is in this larger wheel. Everyone was so confused every time I pitched this idea, that the office would be on the edge of the building, half the frame would be the exterior. Everyone looked at me like I had three heads and they weren’t quite sure how it was going to work out, including the VFX people. But we did it, and that was our biggest high-take count, because that was a very complicated shot that relied on a lot of choreography, from the crane, from the guy working the crane, from the dolly grip, and from the camera operator, not to mention Portia hitting her marks perfectly. It took about 27 takes, but we got that right. And then the VFX guys really came through with the exterior.

My initial assumption about the episode’s split POV was that it was done in part because it would be too much of a burden for either Rami or Portia to have to be in a whole episode being shot this way. But since you had that idea even before it was going to be a oner, why was it important to devote half the episode to his POV and half to hers?

Initially, when we were thinking about it as real time, it was an intercut between the two. One of the things about this episode is that a lot of things come to a head: Darlene fesses up to how she’s betraying Elliot, and Elliot realizes that Angela has been betraying him. We wanted things to come to a head between Elliot and Angela, this is the faceoff that’s been brewing since the beginning of the season. When we were discussing the real-time version of this, we were going to cut back and forth between the two, where Elliot is working in his void trying to figure out who’s doing this to him while stopping Stage Two from executing, and then showing the other side of that, and Angela is coming back at him and undermining everything he’s doing. Interestingly, that’s a parallel of what Elliot and Robot are going through this season. When we decided to do the long take version, we realized it was better that the episode gets divided into two, and one leads to the other. Elliot’s attempt to stop Stage Two is what triggers Angela to retaliate. Structurally, it worked more organically than trying to figure out how to intercut the two. But to speak to your point, I think it would have broken one of them if they had to do that for an entire episode. So luckily, it made sense for the story to split it into the two halves.

Which was the simpler half to shoot: Elliot or Angela?

It’s a weird one to answer. With Elliot, it required a lot more choreography with the cast. There’s a lot more interaction with cast members, a lot of them new ones working just that day, and it required the camera operator to operate a Steadicam and to be still a lot of times, because there’s a lot more dialogue in Elliot’s half. In Angela’s half, it’s mostly handheld, which doesn’t require as composed shots, but there was a lot more running, a lot more chaos. They both had their challenges. I will say that the two most difficult shots is that one overhead, and that first scene with Elliot, where the camera operator collapsed, and was the one that was so hard to choreograph, and it ended up being just two characters talking. Ironically, that would be the most difficult thing we shot.

At what point in the process did you and USA realize that if you were doing it this way, there couldn’t be commercial breaks for the episode?

That was a very interesting conversation. I love and appreciate the support that USA has given me, and I’ve asked them to do a lot of difficult things. But to say “no commercial breaks,” that’s really hurting them where it counts. I was scared to bring it up initially and let the episode speak for itself. But I’ve gotta be honest with you, it was barely a conversation. They knew when they saw the episode that it was special, and that breaking it with commercial breaks would ruin the experience. They agreed almost immediately, and wholeheartedly that it should be done without breaks.

And going back to what you said earlier about how you often find oners to be gimmicky, how did you assure yourself that doing it this way was going to enhance and not detract from the story?

I wasn’t opposed to ever cutting, and I was never going to lie to people and say it was a oner, not that I think it would go over these days. I didn’t actually know when we were shooting, and left it open to the idea that, “Look, if we have to cut, we’ll cut. We’re not at the service of us as technicians; we’re at the service of the story.” The one thing where I realized it was working was that I screened the first cut for a few people, and a lot of them didn’t realize it was a oner for the longest time, and that was the biggest compliment. That meant that part of it wasn’t important. They clearly sensed the real time experience, which is what we were aiming for, but they didn’t realize we had not cut the camera or made an edit point. That’s when I realized, “Okay, this is working,” because the gimmick of it being a long take has just disappeared and you’re just embroiled into the story of Elliot and Angela.

Alan Sepinwall may be reached at He discusses television weekly on the TV Avalanche podcast. His next book, Breaking Bad 101, is on sale now.