John Refoua and Stephen Rivkin discuss editing Avatar The Way of Water

Today on Art of the Cut, we speak with two of the four editors of Avatar: The Way of Water. Joining us on the call are Stephen Rivkin, ACE and John Refoua, ACE.

Rivkin won an ACE Eddie for Pirates of the Caribbean: The Curse of the Black Pearl. He was also nominated for ACE Eddies for Pirates of the Caribbean: Dead Man's Chest and Pirates of the Caribbean: At World's End. He was nominated for an Oscar, a BAFTA, and an ACE Eddie for editing the first Avatar. Rivkin’s other credits include Fantastic Four, Ali, and My Cousin Vinny.

Refoua was also nominated for an Oscar, a BAFTA and an ACE Eddie for the first Avatar. His other credits include Transformers: The Last Knight, The Magnificent Seven, and Balls of Fury, plus TV including Law and Order, Ally McBeal, Dark Angel, Reno 911, and CSI: Miami.

One of those not on the call is David Brenner, ACE. David has been a previous guest on Art of the Cut for Zack Snyder’s Justice League and Batman versus Superman. He won an Oscar for Born on the Fourth of July. His other credits include Transformers: Age of Extinction, Man of Steel, Pirates of the Caribbean: On Stranger Tides, Independence Day, and The Patriot. Sadly for the entire editing community, David died earlier this year, during production of Avatar: The Way of Water.

The other editor not on the call is director James Cameron, ACE. James won an Oscar and an ACE Eddie and was nominated for a BAFTA for editing Titanic. He was also given the ACE Golden Eddie for Filmmaker of the Year in 2000. And he was nominated for an Oscar, a BAFTA, and an ACE Eddie for the first Avatar.

John Refoua and Stephen Rivkin discuss editing Avatar The Way of Water

I wanted to also point out there are two other editors on Avatar Way of Water: One is James Cameron, who I did not realize was an ACE editor.

RIVKIN: He's been an ACE editor for many years and was very much a part of our team on Avatar 1. John and I worked very closely with him as an editor on that film as well— besides being a director — and all the other hats that he wears — writer, producer, you name it.

REFOUA: Makeup artist.

He’s got that big tank of blue paint… I hate to bring the interview down with this, but during the making of the movie, we lost David Brenner, ACE, who was the other editor on the film. I don't want to go any further in the interview without having you guys have a chance to tell everybody what a great friend and what a great editor he was.

RIVKIN: Well, if you didn't mention it, we certainly would have, because David was an integral part of our team and his contribution was significant. We tragically lost him last February. We're all devastated. The entire Avatar production was devastated by his loss. I'm disappointed that he wasn't able to see the final film and all of his great work — along with everybody else’s — come to fruition. I truly believe that he's got the best seat in the house somewhere and he's seeing it now.

REFOUA: David was just such a great guy. He was really easy to be with and was always thinking about other people. He was very empathetic and just a joy to be with and a great editor at the same time. So he's definitely missed.

RIVKIN: A tremendous, tremendous talent and so dedicated to his work. He loved his work and he was so good at it and we were so lucky to have him. We're just incredibly saddened by his loss and his work will live on in this film.

Let's start off with a little bit of craft. Talk about the macro pacing of the story elements. There're a couple of big turns, a couple of big sequences of things that happen — big chunks of the story. How much were you cognizant about how long to be in each section of the story?

RIVKIN: All films start with a script and the script was very elaborate and very detailed. As most of Cameron’s scripts are. Of course, scenes get pared down as we go and how long to be in each is ultimately a Jim decision.

REFOUA: The thing about this movie that is kind of different, because you're there in a 3-D experience. So a lot of things where we would normally say, “Hey, let's get on to the next thing” we would stay in it longer and Jim would insist that we have to stay in this longer because people want to see this. People want to experience being underwater and seeing all the new creatures and all that. So that's what's different about this movie. You can't judge it by other movies because people just want to sit and watch something beautiful sometimes, you know?

RIVKIN: That's a really good point. Jim often said, “I don't just want to go story point — story point — story point. The Avatar experience is something different and people want to just immerse themselves in this.” One of his arguments for maintaining length in sections was that on Avatar 1, the studio fought him and fought him to cut the flying down, and the flying sequence in Avatar 1 is one of the highest-rated, most-mentioned sequences in the film.

So, with a film that is this visually and aurally rich, how do you think about staying in those scenes long enough when at the beginning you're probably not seeing something that's that beautiful, you're not seeing the finished gorgeous 3-D visual imagery?

RIVKIN: We should talk about the process just for a moment because one of the unique things about Avatar and the Avatar saga is that we cut the film multiple times and as a basis, we cut it twice.

First of all, Jim works with the actors and we had like 16 reference cameras around them. Each reference camera operator is assigned to a character and they shoot a big close-up of their face. And then some of the cameras will shoot a wide shot. After careful review with Jim — we cut a “performance edit” and that's based on all the best takes of all the actors. We put together a mosaic of all of their best takes. Sometimes they're not from the same take, but we have to build this performance edit. Then we go through the process of preparing it to turn over to our internal lab where they process all of these performances with the virtual characters and bring in the environment, the wardrobe, the props, the machinery, the creatures, you name it.

So then at that second phase, Jim creates these virtual cameras — what we call virtual cameras — that actually make up the shots that you would now cut like a movie. So that's the first time we actually have dailies.

In the performance capture stage when we're cutting reference cameras, these are not the shots that are in the movie. Any performance could become a closeup, a medium shot, a wide shot, a crane shot, a dolly shot when virtual cameras are created. So in a way, it's liberating because during capture Jim is just focusing on the actor — working with the actor and the actors with him. He's not worried about focus, about lighting, about laying dolly track, about the shot. It is just the performance. So in the latter stage of creating these shots for the film, the actors aren't even there anymore and he can shoot to his heart's content to get what he wants. So that's the basic process.

But at that phase — what we call the template phase — the template for the shots are created. We do see the landscape and the environment and the wardrobe and the props and everything, so you really get a sense of what those shots will be when they're ultimately rendered at high resolution. But you don't really see the faces. You get an indication of the eyes and the mouth. You know it's in sync, but you don't see the detail of the expression until way later when these final renders come back.

REFOUA: It took me a few months to figure that out. It's all catered to Jim's desire to control and his ability to control every aspect of it. So the first aspect, only has to do with performance and nothing else. Then the second aspect has to do with lighting and shot creation, and he doesn't have to worry about the performances because he got the best performance that he wanted.

That's what we do. We put it together and make a scene out of it with sound effects and music and all that. And he says, “Okay, that's the scene. So now let's shoot it.” It's a complicated thing and every editor, it takes them a couple of months at least to understand because it’s so different. It's different from everything else.

It's a lot of work for an editor because you’ve got to imagine what the scene is going to be without having shots, which is the opposite of what we do. We always have a shot. We react to the shot. We choose, “This is a great shot for this moment.” But we don't have that. We do have a performance that we think would work best. I don't know if he's going to put it in a close-up or whatever he's going to do at that time. That's up to him.

That makes it interesting and also makes it more complicated for an editor. In the performance edit, you can't do the typical cheats that you do editorially where you want to speed things up because Jim may not want to be in those angles when he gets to his virtual camera. So at that point, you have to make sure that you allow him to shoot whatever angle he wants.

RIVKIN: One of the interesting things you bring up: when you're talking about cheats, you're talking about cut points. In any given scene, there may be a take that goes so far and then there may be another take that picks up from there. So Jim may not want to cut between those two and we may have to find ways to stitch those performances together. Inside our internal lab they can smooth those out and we can blend takes together in case he doesn't want to cut at that point.

Unless Jim designates, “I can accept a cut here,” what we call a “load” where we prepare files for each scene to shoot. If he says, “It's okay to force a cut here. I know I'm going to cut to this close up here” that actually gives us the freedom to say, “This is where we'll have a load break. This is where we'll go to the next part of the scene,” because a lot of times we're limited. We can't always do the whole scene in one build. Although there are scenes that are like that.

Many times we have different performances stitched together — combo’d one take from one actor, one take from another. It gets very complicated.

Getting back to what you were saying, John, about the editors on Avatar having a learning curve. It's funny because when David came on — John and I went through this process already on Avatar 1 — and when David came on, Jim said to him, “You're going to have a steep learning curve. I don't know how much you'll be able to contribute for six months or something because you'll be learning all this.”

David really picked it up quickly, which is a testament to his ability, how smart he was, and how much he could adapt. Within a couple of months, he was up to speed. Jim was impressed. It is a very complicated process. I can't imagine there is a more complicated way to make a movie than what we do on these films.

When you described a “load” is that load all of the different motion capture videos that you're editing with to choose performance?

RIVKIN: Yes. The capture itself is recorded by these cameras. I hesitate to call them cameras because they're not doing shots, but they're all over the ceilings in what we call a Volume, and they track every movement of the suits that the actors wear. Let's call them recorders, because I think that's less confusing. So there are recorders all over the ceiling and they track the actors’ movements and there's a face-cam in front of each actor that is recording in high definition every facial movement.

Then we have reference cameras to cut with. We do have a wide-angle face cam, but that's not something you can judge performance by. That's really for WETA FX as data. The reference cameras give us a documented video record of the performance, something that we can look at as editors and put together and we can see the best moments, but that is our only reference to what is recorded by the capture cameras. So that's what we use to put together that performance edit.

The loads that you're asking about — load 1 may be a combination of one actor from take one and one from take five and load 2 to maybe what we call an “as shot,” which is basically — if I'm using a two-actor scene as an example — the two actors in their native capture as they relate to each other. Now it gets into all kinds of complications where we can slip the timing so that one line can follow quicker to another. If he wants to Jim can manipulate the capture spatially and temporally, and he can literally move mountains behind them to compose the shots that he creates.

It's an amazing process with an enormous amount of flexibility. For a director that is intimidated by the infinite possibilities, it could drive one mad. But for a director who knows what he wants and is able to create shots and move things around, it's incredibly liberating.

So you've got this “load” — you've got all of these performance captures — you're seeing this emotional performance from the actor. You've got multiple actors and you've got maybe a wide shot. How are you constructing that in the timeline so that you know what the performances are and Jim can judge those in a scene.

REFOUA: While we present him with what we call a pre-cut, which is of the reference cameras. We actually put that in the reel. So for a while, in the reel you're looking at people with dots on their faces and a black capture suit on a gray Volume. That's the first layer. So that stays in the reel. The loads don't go in the reel. The loads are basically an order to the lab that says, pull this take from here to here, pull that take from here to here and combine them and give it this name and have it ready for Jim to shoot it.

I was imagining maybe a two-shot with a little picture in picture of people's faces. But that's not the way that you would proceed.

REFOUA: Sometimes we do that. As he’s doing the capture, he's also shooting one camera just to see what the place looks like. So you can show the actors: you're here, you're doing this, and this is what's going to be happening.

We can cut to that and then we put faces in picture-in-pictures so when these two are talking, this is what's going to be happening. But we try not to do that too much because it's disconcerting to change from one way of looking at the movie to another way of looking at the movie.

RIVKIN: John brings up a good point that during every captured session, Jim picks up the virtual camera and shoots a reference camera of the virtual scene, so we integrate that into our performance. If there's an establishing shot that shows the environment or a tree or they're in this location or that location, we can actually cut that into the performance edit and either put picture-in-pictures of the performances or just pepper those shots in here and there. So we know what it is.

Also, Jim’s doing his initial scout during the capture to determine which side he's going to shoot the scene from. So it helps inform us for later. There's an interim stage where we do a rough camera pass where we're testing all the loads. It's something that comes in between the performance edit and the virtual camera edit. It's a test edit that goes in between. We're checking the contacts while Jim reviews the lighting and gives notes on that, then he's looking at the props and environments and approving that.

To answer your question about the size of the shot or what type of shot it is: any reference shot from the capture can become a close-up. Any shot can be a wide shot. And we don't necessarily know what that's going to be until Jim shoots the virtual cameras.

I love the way Jim describes it. A lot of times in traditional filmmaking, you may have an actor that gives their best performance in a wide shot. And by the time they got to the close-up, it just wasn't as good. How many editors have been frustrated with something like that? “If only I had that in the close-up.” Well, now any shot could be that. You have the idealized performance that you've selected carefully and built to be shot. Jim can decide at any time during the virtual camera creation process what he wants in close-up and what he wants in wide. And it's always the same performance.

You don't have to worry about an actor blowing their lines or hitting their marks. You don't have to worry about focus. “This was a great take, but the guy bumped the dolly when they were moving.” None of that matters anymore. It's all up for grabs with the best performances available.

REFOUA: Yeah! See, it's simple! :-)

RIVKIN: Yeah. Simple. Right? Right. (laughter)

I don't know if there's a more complex way to make a movie. That's why it takes four editors. And we actually had two additional editors that I should give a shout-out to: Jason Gaudio and Ian Silverstein. They were working along with us. And we had an army of assistants — a literal army of assistants — VFX editors and VFX assistants as well. And this crew was so dedicated.

The ones that came on this show that had some experience in the live-action world had never experienced anything like this. Everything from the way we build our dailies of performance capture; the way we have to stack the reference for every character in the virtual shots. All of these things had to be accounted for when eventually the virtual edits of the scenes had to go to WETA. Everything had to be vetted and we had to have proper sync checks for every character and make sure that the reference was turned over properly. It is endlessly detailed and our crew did a wonderful job.

REFOUA: It's like making a sand painting.

RIVKIN: “Every grain” is that you're meaning?

REFOUA: Yeah. Every grain had to be in the right place.

In the Avid timeline were you carrying all those mocap references on different tracks?

REFOUA: For a while we did that. At a certain point you jettison it because the virtual camera is there and it has references that go back by itself. When you're cutting the virtual camera, it doesn't really matter so much.

As editors, we learn the performances pretty well. So we knew, “Okay, this is where he blinks.” “This is where he has a little tiny smile.” Even though you can't really see it yet. So if we wanted to emphasize that, we would put it in a picture-in-picture to remind Jim, “This is why we're cutting; because of this expression.” But, there's too much material to carry all the time.

RIVKIN: Ultimately, though, when these scenes get turned over to WETA, we rebuild all the reference performances that are involved in every single shot in the upper part of the timeline. And all of those get turned over — each track meticulously labeled: “character face.” And sometimes the face would be offset. Sometimes we would do our equivalent of ADR which is called FPR which is Face Performance Replacement.

For example, there were times when Jim wanted to rewrite a line and have an actor say something else so we could use the original captured body, and the actor would come in for an FPR session and say the new line. Then we’d actually put their face — complete with the expression — on their own character body again. So, a lot of times the face tracks would be separate from the body tracks.

Everything had to be accounted for. And each shot in a scene — when it's turned over to WETA — had to have every character and face accounted for in the timeline that went.

That’s amazing. There are a couple of places where you've got storylines that are intercutting: either between the bad guys and the good guys or you're cutting between one family member and another family member as they're separated. Can you talk about any changes between the intercutting in the script and the way that intercutting ended up in post, and why that happened? What was driving the choice of when to transition to a different storyline?

REFOUA: Many times we changed things.

RIVKIN: When something's on the written page — as we know from just general editing processes — it always ends up quite a bit different than how it reads, and by necessity, you have to make changes.

There were a number of scenes that were intercut differently than how they were scripted. And part of it was just balancing all the cross-cutting stories. And how much screen time you could spend on each and how long it took to tell each little story.

The underwater capture was a whole new deal. I don't think it had ever been done before. So, they built this huge tank and all the actors were captured underwater. That presented a whole other layer of editorial challenges as well.

John worked very heavily on the end battle section.

REFOUA: Like any movie, at a certain point, you’ve got to set the script aside and say, “What works? What doesn't work? What can I do to make it better?” At that stage, the inter-cuts that were in the script… that’s a script and this is a movie. It's a guide, but it's not the Constitution. It's just an idea of how it should go together. So we follow that to a certain extent but at a certain point, we decide, “It'll be better if he learns how to jump on this creature four scenes earlier so that we show his experience longer so he's better the next time we see him.” Every time you cut away from a scene, the audience thinks time is passing by for that character.

The battle scene was a lot of fun. There was stuff in there that I have never seen before. I was just having a great time cutting that stuff. There’re underwater chases that no one's ever done before. You just don't do that because you can't be underwater that long with actors.

It's that question of how long do we stay in there? Do we cut to the bad guy saying “There he is,” or do we just ignore the bad guys and then they shoot anyways? We had it one way for a while and then we took it out and just made it from the perspective of the people being chased, which was our heroes.

You pointed out that at some point you don’t know whether something is playing in a wide or a close shot, so how did the dailies process work for you?

RIVKIN: When Jim's creating virtual shots for the scene — now, this is months, sometimes even years after the capture — no actors are there. Jim is creating the shots as if he's on a set, making shots to put a sequence together. Sometimes he would cover things with an idea of how he thought the scene should go together, and other times he would explore. That's where he does his most creative work.

What's liberating is not worrying about the actors anymore, being able to just explore the capture that exists and find the shots.

Because it's 3D Jim likes to keep the camera moving. He likes to experiment with drifting camera shots or moving shots that boom down and reveal things. All of these things are possible from this static capture.

So we end up with dailies and the editors will make a basic cut that we think plays and flows. And a lot of times there's dialogue involved because we have an interaction with Jim while he's shooting. He's creating shots and we're there because we can help navigate the “loads.” We're there assembling as he's creating these shots.

We'll show him a work in progress. He'll say, “I like this. Maybe I don't need this shot,” or “Maybe I'll get a shot that does this.” So looking at it as it evolves as an edit will inform the way he shoots the rest of the scene, which is an interaction that seldom occurs in filmmaking where the editorial process and the editor’s cuts — as they evolve — are being viewed on the stage and are influencing what’s being shot.

And then we create alternate sections because he'll go down multiple paths to create shots. He'll say, “Well, this version could go here, here and here, and another version could do this and another version could do that.” So it's up to us to present alternate sections, and then he'll go through all of these possibilities that we present and create a hybrid of them, as well as his own exploration in editing.

He's very much hands-on in terms of the editing process. So now he puts the camera down and puts on his editing hat and he takes all of these suggestions that we provide for him and makes a cut that he'll show us and he'll say, “What do you think?” Then that becomes the locked scene that gets delivered to WETA to take it to the next and final stage.

How much did any of those “locked” edits change once you saw the look of the final footage.

REFOUA: I think it changed only because of editorial concerns about length and, emphasizes in storytelling.

RIVKIN: There were a few times where — looking at the shot — he said, “I'd really like ten more frames of this or another second.” But the cost is astronomical when you talk about per frame, so adding frames is a big deal, but he's the guy that can decide that, right? If he says, “I want another 2 seconds,” we're going to get another 2 seconds.

REFOUA: Or he’ll say, "Let’s throw out this shot because I don't like it anymore and it slows down the scene". Even though it’s been turned over, and they're already working on it — out it goes.

RIVKIN: We put shots on hold as they drop out and eventually they are official “deletes.” Obviously having to turn shots over as we go, there were going to be shots that got cut out.

Here's the other thing about this film: we were in a constant state of deadline. Normally, you'd work on a film, you'd finish then cut, then you'd turn all the visual effects shots over and there'd be a crunch time at the end.

Because of the necessity of nearly 4000 visual effects shots, we had to be turning shots over on a monthly basis, even sometimes bi-weekly basis, sometimes weekly. There was constant pressure to feed the visual effects house these shots so that they had a possibility of completing the film. There was no way to delay. We were turning shots over for literally years. That's how it had to be.

REFOUA: WETA would sometimes take a year to show us a shot. They would work on it for a year before we would see it back and say, “Oh yeah, it looks great” or whatever it is. So the visual effects process itself is very complicated.

How long were both of you on this film?

REFOUA: Five and a half years.

RIVKIN: I was on over seven years because I was on early to do story reels and art reels. The art department was generating literally thousands of paintings and drawings and sketches; design for both characters, creatures, machinery. We put together presentation reels for the studio to get everybody fired up.

I worked on a story reel using a cast reading of the script that we would start to piece together a framework for the film that would ultimately evolve into replacing everything with capture. But it gave us the first sketch of the film.

After you have worked in this very specialized way, THIS LONG, how do you go back to a normal edit?

REFOUA: It’ll be simple! It’ll be easy! Everything is easier after this.

RIVKIN: That's right. But the normal edit is when the virtual shots are there. That's a normal edit. So you just skip all the front-end stuff and you're cutting a movie. There just happens to be an enormous amount of front-end work that has to happen.

You did this in the middle of COVID, or at least overlapping with COVID. What happened with that and how did you deal with remote work?

REFOUA: We all went into remote work. Everybody's remote now, and it's great. So we work from home. For us, the reason as editors that we go into work is to work with our director. But our director happens to be in New Zealand and he connects with us remotely, so there’s no reason to go into work for us.

When COVID hit in March of 2020 and they said, “We're going to shut down and set you guys all up remote,” it's hard to describe those feelings. I'm sure we all went through that. But it was very strange. So, I'm sitting in my garage and they deliver these laptops and monitors and they show me how to set it up, and then off they go.

RIVKIN: Zoom. Slack. BlueJeans. Evercast. All of these became our daily routines. Oddly enough, we had a greater ability to communicate with each other than if we were all in the same building in different rooms. We found ways to collaborate, like jumping on each other's Avids and reviewing things together and working together.

The pandemic kind of accelerated the technology that was already in place but not embraced. Thank goodness it happened in 2020 because ten years before everybody would have been just dead in the water.

We had the technology to move forward and to collaborate and to keep going. One of the most amazing things was for us to be able to log in to our Avids at Manhattan Beach Studios in Los Angeles or log into the Avids in Wellington, New Zealand, and we could work in either place. When Jim was doing virtual cameras, we would be logging in to Avids in New Zealand and working there. The only downside was that we had to keep New Zealand hours, which meant that we'd start in the afternoon and work until the wee hours of the morning. And they were on the next day, by the way. So our Monday was their Tuesday. It was just amazing the way we were able to adapt to this situation.

I'm really interested in the creative part of that collaboration between the four editors. Can you talk about how you were speaking into each other's sequences or as the film evolved, what the discussions were between the four of you or the three of you?

RIVKIN: We would have review sessions where we would be talking about the shots preparing for a virtual camera. Everybody kind of had their scenes they were working on. There were a number of times when the three of us would get together to say, “Let's have a unified front on what we want to present to Jim as some of these alternatives for our possible cuts.”

There were sessions where all the editors were on with Jim, and we'd review sequences, review performance edits for approval, review various parts of the movie that were in virtual form. There was just no one way that this happened.

We did more communicating amongst ourselves than with Jim because he was very preoccupied with his own deadlines. We work very well together on Zoom and Slack. And Evercast stepped up and played a huge part in being able to review sequences.

After the virtual edit was approved by Jim, Jim would turn things over to us to go to WETA, but there were a lot of things that had to be fixed, and a lot of it was the typical kind of continuity issues that you might have in a normal film. Many of them could be corrected before they went to WETA. If a chunk was cut out of motion and we needed to match something across a cut, we could actually correct the motion to do that, which is kind of unique in this form of filmmaking.

REFOUA: This whole remote thing didn't affect the most important part of what we did on this movie, which is being on the stage and watching Jim work and watching him talk about the point of the shot and why he likes this shot in particular. Because we were logged in to the same Avid that was on the stage, we could cut it into that section and show it to him. “Is this what you were thinking about? Do you like this?”

He would sometimes look at it and say, “Oh, I’ve got to change what I'm doing,” or sometimes say, “It's great,” or sometimes “It sucks.” But being able to be there — which I thought was the most important part of what we did — is that interaction with Jim when he's shooting virtual camera on a continual basis.

RIVKIN: There was a period of time, through the latter part of 2020 and a good part of 2021, where we were rotating going to New Zealand to be present for that process.

REFOUA: You had to sit through the New Zealand quarantine.

RIVKIN: That's right. In 2021, I took two trips over there and each time the government had a mandatory quarantine of two weeks. So I had to sit in a hotel room for two weeks and I brought my little portable laptop. I was able to continue working — the first time better than the second because the Internet service was not great the second time. So I spent a month in quarantine in 2021, but all of us had a turn going to New Zealand and working with Jim. Then in 2022, we were doing it all remotely and talking to Jim on Zoom during camera sessions.

How were you monitoring both visuals and audio? Were you in 5.1? Were you in LCR? Were you monitoring stereoscopically for the 3D?

RIVKIN: No, even when we were in our cutting rooms all together in L.A., we were not watching the film in 3D. We were cutting into 2D. We had a pretty good idea of what the 3D looks like. Even on Avatar 1, it was a multi-stage process where we would look at it in 2D and before it got turned over to WETA, we would check it in 3D and make sure that no adjustments had to be made. That was similar to this. We had a department that was dedicated to seeing things in 3D as they were turned over.

If something was an issue, it was flagged and Jim and the crew would make adjustments. As far as monitoring audio, on our home systems we were just using stereo.

When did you start seeing high frame rate stuff and did that affect editorial?

RIVKIN: Again, like the 3D, it was a post process. A lot of the film was shot at 48fps so anything could be a high frame rate ultimately. But that was a finishing process and it had very little impact on the editorial process.

REFOUA: Yeah, I mean, this movie is so big that they had departments set up for things that normally an editor oversees. They had a whole department of maybe 20 or 30 people set up to take care of that. So it was just this giant, giant production.

RIVKIN: I think upwards of 3800 people worked on this film. A lot of those are on the visual effects side.

REFOUA: Our assistant crew were all great. Justin Yates was our first for most of this, and we had over 20 assistants, and they were all crazy busy.

With a team this large, were you more “being managed” than “managing?”

RIVKIN: We had our First managing the assistant crew. Our head visual effects editor, Justin Shaw — who was on the first film — had the visual effects editorial crew working under him. Some assistants were assigned to editors specifically, and then there were other assistants that would work with all of us. There was never a shortage of things to do, I'll say that.

REFOUA: And we were all busy as editors ourselves because, we're going through, trying to figure out these scenes, these camera loads, all that stuff. So at a certain point you're happy that you know how to take care of it.

Screening something of this length to make sure that you know the overall flow of the film when it's that long how difficult is that or does it limit you in the number of times you can watch the whole film?

REFOUA: We couldn’t watch it as often.

RIVKIN: We would reserve the full screenings so that we weren't watching it all the time. Obviously, we're deeply involved in sections of the film on a daily basis, but Jim was also big on saving the screening for special occasions where we'd look at it and evaluate.

He also had a little trick that he did to kind of renew the experience where he'd flop the film and look at it with fresh eyes. So you expect someone to come in on the left side of frame and they come in the right and the first time it's very disorienting. But it does force you to look at the film differently. He did that on Avatar 1 and he did it again here. It's just a way to keep a fresh perspective. And you can only do it a couple of times and then it becomes the norm again.

We did a friends-and-family screening, and then we had a couple of audience screenings at Disney where there were things that Jim wanted to address based on some of the feedback. He's a big believer in if something is a consensus that it's worth dealing with, but you can't overreact to a couple of comments here and there. But it's the normal, you know, preview process, except that Jim decides what he wants to deal with and what he doesn't. He's not going to be told that he has to take something out. If he doesn't want to take it out, it doesn't come out.

I love that idea of keeping objective about it through flipping the film. I've heard that one from Joe Walker on Dune. He also said he does black and white sometimes.

RIVKIN: Oh, yeah, that could do it too. More than anything, the flopping of the film does create an objectivity that doesn't exist just seeing it in black and white.

REFOUA: Yeah, because your eye is used to “this guy enters from here, and this is where the action is.” As an editor, you know it so well that your eye goes there automatically. You have to be careful to recognize that the audience doesn't know what's going to be happening, so it's going to take them a while to shift their eye to where you want it to.

I used that flopping trick on a few movies that I did between Avatar 1 and 2, so it has its advantage and use.

We couldn't really screen the movie till much later. It wasn't until July and August of 2022 that we had enough movie done in a way that you could screen and have it be meaningful. You can’t show scenes of actors with dots on their faces on a gray background. Then when you see it all finished and they're swimming through this amazing landscape with bioluminescence coming off of them, you go, “Wow, that's amazing!” Even I hadn’t anticipated how amazing it would be.

It’s pretty incredible that you were able to imagine the quality of shots enough that you stayed with them long enough when most of the time you’re working with renders that aren’t nearly as pretty as what the finals looked like.

RIVKIN: That’s where I think Jim's expertise for what it's going to look like in all its splendor and glory comes into play. We learned from Avatar 1 that there are things that you just want to experience. But as far as the viewing process, John raises a good point that because the film was in various stages of completion for so long, it was a mosaic of performance capture, a template of final render and it wasn't like the beginning is one way and the middle is a different way. It was peppered throughout. So it wasn't really screenable until very late in the game.

One of the joys of seeing the final renders was in witnessing those original performances in all their detail. Looking into the actors eyes and seeing every nuance of their facial movement and expression come back and you think back to when you originally picked those performances, and here it is years later and you're seeing the incredible performances that were delivered by these actors in kind of a vacuum — like a black box — where they had to use their imagination and envision themselves in this world.

The actors did an amazing job. I can't say enough about the performances, the performance captures and their individual contributions.

Quite often people confuse this with animation, and it's not. It is purely performance-driven and every performance is acted by an actor. Not voiced later — with very few exceptions — but an organic performance from that capture session.

REFOUA: Kudos to WETA. They were able to do this better than Avatar 1 because the technology is better; they know more and they're able to capture that performance and put it on the screen. So really, kudos to Weta and the team and great job.

RIVKIN: Yeah, they did an amazing job. The technology evolved over the years since Avatar 1, so they developed a whole new way of doing things. There is not a shot of real water in this film. It is all CG: on the surface and below. They cracked the code on water.

We haven't really talked too much about audio. When you're trying to build a world like this, especially in the early days when you're looking at mocap or something. How much sound work are you doing, or are you leaving that to assistants or the other crew?

REFOUA: On this movie we ended up doing a lot of that work ourselves because the assistants were too busy and they had things that they needed to do.

RIVKIN: We had the benefit of all of the stems from Avatar 1. So you need a Pandoran jungle? We got that. We would tap effects, creatures, backgrounds. And we used them extensively.

REFOUA: And music too. We had to track the film ourselves. I like to do that anyways and I know Steve likes to do that, too.

RIVKIN: We had a sound effects editor on for literally years during the editorial process, and a sound designer. Our sound designer, Chris Boyes, who was feeding our library so that we could have creature effects — both for the tulkun (whale-like creatures) and the various other creatures that could live in the film. You get these effects cut in and it's important to live with them and have them evolve over time so that by the time you get to the final mix, you already have a pretty good idea of what Jim’s interested in.

We also brought a music editor on as well to experiment with different temp tracks. We had many months of music editors as well. It's just that when you're talking about years of work, it's very difficult to carry someone that would normally be brought in for the last several months to carry them for five and a half years. So, we had to — within our crew — come up with a temp, and Jim's very involved with that as well. It was a team effort. I really wish that David could have seen the film finished, but his work will live on in this film.

Do you have a philosophy about temp music? Did you stick with the same composer, for example? Or how much of Avatar 1 could you pull forward?

REFOUA: We temped with everything we could find. We had a bin of James Horner's music. Fortunately, he's done a lot of movies. We had a bin of other composers. We temped with whatever we could. There's no limit. There's no restrictions.

RIVKIN: We try to keep it in a similar wheelhouse, so that it felt like Avatar. Jim would pull cues that were different for shape, but not necessarily orchestration. There's a whole art to building temps. It evolved over time.

Gentlemen, thank you so much for joining me.

RIVKIN: Thanks, Steve. It's a pleasure to be here. Thanks for asking us to do this.

REFOUA: Thank you, Steve. Take care.