Search form

'Flags of Our Fathers': Making War With VFX

Ellen Wolff reports on Digital Domain’s extensive photoreal CG efforts in recreating the horrific Battle of Iwo Jima in Clint Eastwood’s Flags of Our Fathers.

Digital Domain was entrusted with the creation of photoreal explosions, crowds, surf and terrain. Above, the vfx house tackles the Times Square celebration. All images © DreamWorks Llc and Warner Bros. Ent. Inc. 

Digital Domain was entrusted with the creation of photoreal explosions, crowds, surf and terrain. Above, the vfx house tackles the Times Square celebration. All images © DreamWorks Llc and Warner Bros. Ent. Inc. 

Oscar-winning director Clint Eastwood took on an unprecedented challenge for his Paramount/DreamWorks/Warner Bros. film, Flags of Our Fathers. He ventured into the realm of extensive vfx in order to bring this World War II story to the screen. Collaborating with visual effects supervisor and second unit director Michael Owens, Eastwood was able to pursue his typically swift, low-key style of moviemaking entrusting the creation of photoreal explosions, crowds, surf and terrain to the CG artists at Digital Domain.

Based on the memoir by James Bradley and Ron Powers, Flags of Our Fathers required Eastwood to recreate the horrific battle on the Pacific island of Iwo Jima, which killed thousands of U.S. and Japanese soldiers. Among the key visual effects scenes was the raising of the U.S. flag on Mount Suribachi, which became an iconic image that mobilized Americans support for the war. The story also follows the heroes welcome that the surviving flag-raisers faced upon returning to America. That required huge (and virtual) crowds in places such as New Yorks Times Square and Chicagos Wrigley Field.

To tackle this ambitious project, Eastwood surrounded himself with familiar collaborators: cinematographer Tom Stern, editor Joel Cox, producer Robert Lorenz and special effects supervisor Steve Riley. Because of DreamWorks involvement, Saving Private Ryan director Steven Spielberg also co-produced with Eastwood and Lorenz.

But visual effects supervisor Owens (whod previously worked with Eastwood on Space Cowboys) asserts that Clint followed his own instincts when it came to in planning and executing the effects in Flags of Our Fathers. Once a plan was in place, he gave Owens and DDs team the latitude to do what they do best. His driving intent was to be respectful of the true story behind the film.

Eastwood wanted a photoreal look to the film. He and vfx supervisor Michael Owens agreed early on that the characters would be the foreground and that hardware would be around them.

Eastwood wanted a photoreal look to the film. He and vfx supervisor Michael Owens agreed early on that the characters would be the foreground and that hardware would be around them.

The Planning

Owens recalls, I did storyboarding and conceptual artwork for key sequences to get a feeling for what Clint was thinking. I needed to understand what he was after, and he wanted to understand what it could look like so hed know what we could do with it. I hired two storyboard artists and they flushed out the thumbnails that I had drawn. Animatics werent necessary. We also used stock stills that showed us the look of the beach at Iwo Jima and the size of the army. Then I devised a plan for the best way to go through this, considering Clints style and the subject at hand.

As a filmmaker, Clint studies a great deal, but then on set he wants to be very spontaneous. Hes very appreciative of the fact that I needed something to figure out where we were headed.

The visual effects storyboards in no way tied Eastwood to anything as specific as camera angles, notes Owens. I didnt want to limit him at all. After wed looked at stock footage and photos and the storyboards, I felt safe giving him complete freedom. If I nailed it too tight, it wouldnt be Clints movie it would be about me making a visual effects movie.

The look that Eastwood wanted was one that was as photoreal as possible. As Owens explains, Its as if you were a combat cameraman. We tried to limit very objective shots, especially platformed objective shots. Platformed shots make it twice as hard for the visual effects to not look in your face, and trying to say something. In this movie, they are environmental, not in the forefront.

Once we knew where we were going visually and emotionally, it was much easier to determine our strategies and techniques. We agreed early on that Clint would provide as much foreground as possible with his characters. Then wed put as much hardware as we could around them, and use as much pyro and physical effects as possible but for just one or two takes. Because Clint doesnt want to slow down for that kind of stuff.

Digital Domain was challenged to create the virtual landscape of Iwo Jima. During the location scout, it became clear that DD would have to build the island in CG. 

Digital Domain was challenged to create the virtual landscape of Iwo Jima. During the location scout, it became clear that DD would have to build the island in CG. 

The Shoot

And thats the way it worked out while shooting on location on the black sand beaches of Iceland. The physical effects guys did a great job, says Owens. We were all in cahoots. So take 1 would be with full effects, and take 2 would have partial effects and the others had no effects at all.

Because physical effects add so much, they set the pace. We thought: Whats the least we can do to get us down the road, but be sure were not screwed later? In watching it happen a few times early in shooting, we saw that we could be bolder. We could live with the first take of physical effects and be able to add to it, and put ships behind it all.

That provided for a fairly streamlined shooting schedule, observes Owens. I directed second unit war footage and I was able to go back in and pick up more, using the same sort of hardware, personnel and extras to create more mid-ground elements.

While Owens was on set with Eastwood during filming, he says, I tried to be almost invisible on set. I didnt want to interrupt Clints method of filming. I would just watch over his shoulder, and at times I might step in and say: Let me remind you of the effects that will be going on in the background. I would interject myself whenever I thought it would enhance the crews knowledge about what would be placed in the shot later on. Even though there was stuff that Clint couldnt physically shoot, it had to remain his movie.

Owens says Eastwood became totally un-intimidated by how we go about creating effects. I give him so much room that if there was something I was really concerned about, hed say, OK, Ill tell you if I can live without it. Or Ill give you more time to do whatever you need to do. Hes very reasonable in that sense.

Nonetheless, Eastwood wanted to shoot as efficiently as possible and then move on. With Clint, youre lucky to get a rehearsal take. So the actors knew that they had to do what they do naturally. All of the performances in this movie are extraordinary in their subtleties. Clint is famous for saying: Do it, and something will come of it. If a soldier tripped, Clint would see that as an asset. In fact it was, and we could capitalize on it during post. We played off of anything that happened.

The Virtual Worlds

One of the boldest decisions Owens made was to not use bluescreen to stage the iconic raising of the flag atop Mt. Suribachi. I knew the best thing for Clint and Tom Stern and the actors would be if they were physically on a mountain somewhere. But I wasnt going to put a 400-foot bluescreen on top of a windy mountain. I knew that the actors had relatively smooth-surfaced costumes and helmets, and didnt have hair blowing in the wind. So I knew that we could get away with rotoscoping the whole thing, and that was what we did. It was a huge undertaking the people at Digital Domain were thinking I was almost crazy to do this. But I knew that was the best scenario in which the actors would feel natural. Of course, the surrounding terrain in the valley below was completely wrong. But it made for a much better scene. Im sure there were in excess of 300 shots that were rotoscoped, where traditionally they would have been bluescreen.

Along with the immense rotoscoping assignment, Digital Domain was challenged to create the virtual landscape of Iwo Jima. Owens knew early on that DD would have to build the island in CG. As we were scouting locations, we realized that shooting on Iwo Jima wouldnt have worked. Theres still live ordinance there. Its a memorial to the 20,000 Japanese soldiers who died there, thousands of whom havent been found. So its sacred ground. Fortunately, Iceland couldnt have been a more perfect place, and a complete survey was done of our locations there.

One of the boldest decisions made was to not use bluescreen to stage the iconic raising of the flag atop Mt. Suribachi. Putting a 400-foot bluescreen on a windy mountain wasnt feasible, so more than 300 shots were rotoscoped instead.

One of the boldest decisions made was to not use bluescreen to stage the iconic raising of the flag atop Mt. Suribachi. Putting a 400-foot bluescreen on a windy mountain wasnt feasible, so more than 300 shots were rotoscoped instead.

Working with that survey data and with historical footage, DD was able to create a landscape that closely resembled Iwo Jima. Supervisor Matthew Butler explains, Our version of Mt. Suribachi is a mathematical rendition. Its a combination of high elevation data, hand modeling and a lot of procedural modeling. A key software tool used to do this is called Engine (originally developed at DD under the name Terragen.) It was used previously to create the terrain beneath the high-flying jets of Stealth, but Eastwoods movie presented a different challenge. We werent moving over this landscape quickly enough to get away with murder, says Butler. We used it for shots where we were locked off and staring at it.

Since Iwo Jima is an island, and significant action happened on its beaches, DD also had to generate believable ocean water. Butler, whose credits include Titanic, remarks, You try to utilize your skill set from past movies, but for this show we did come up with some new tools and techniques. Weve done tons of movies with generic ocean surface water or fluid simulation and white water volumetrics. In this movie, weve done all of that. But Flags of Our Fathers also required us to emulate the behavior of surf crashing on the beach. The way that water curls has a very characteristic behavior that were all used to seeing. So we put effort into emulating that very specific natural behavior. In many shots in Flags, you can literally see the water curling, breaking, becoming white water, ebbing and flowing. That was a big achievement for us.

DDs technique employed what Butler calls, A hybrid that involved internal mathematics. The driving person behind this was Greg Duda (X-Men, A Beautiful Mind). He used Houdini [from Side Effects] as well as our water renderers and our volumetric renderer called Storm. It used a multitude of effects.

Butler notes that throughout the past decade since Titanic, The audience has gotten more advanced in what theyre willing to accept and their expectations are raised. We cant get away with things that we got away with in the past. One of the hardest parts of doing this is that we are all programmed with the knowledge of what looks right or wrong. As humans, we notice motion more clearly than value or hue. So we had to get past a very high standard right off the bat.

Emulating the behavior of surf crashing on the beach with the water curling, breaking, becoming white water, ebbing and flowing was a big achievement for DD. 

Emulating the behavior of surf crashing on the beach with the water curling, breaking, becoming white water, ebbing and flowing was a big achievement for DD. 

Populating Those Worlds

Owens remarks that when the live-action shots were handed to DD, The plates were practically empty. Few actual tanks still existed and no operable battleships or planes could be found. In addition, notes Owens, The historical photographs were terrible. We took measurements and did scanning of some objects to establish a starting place for CG textures and forms.

Then it was up to DDs modelers to create believable CG versions using Autodesks Maya. Butler recalls, We only had six landing crafts and six LVTs, so we had build hundreds and hundreds of them.

But the biggest challenge revolved around populating the battle scenes, and the crowds, which filled U.S. stadiums to welcome the Iwo Jima returnees. Owens and DD made a determination that for this type of animation they would use Massive, Stephen Regelous A.I.-driven software that animated the armies in Lord of the Rings and The Chronicles of Narnia. You give the crowd members their timings and then let them go, says Owens. We used it for crowd scenes in Times Square, Wrigley Field, Soldier Field and Central Station in Chicago. You need enough volume of work to amortize the cost of doing that, but we knew wed be crazy to do that any other way.

Butler notes, We used Massive to populate the stadiums with crowds. But we also used it for thousands of Marines getting off of landing crafts onto the beach, and for putting them onto troop carriers in the ocean, and having them blow apart.

Massives usefulness extended beyond groups of humans. DD used it to animate the boats and it allowed landing vehicles to become sentient. 

Massives usefulness extended beyond groups of humans. DD used it to animate the boats and it allowed landing vehicles to become sentient. 

DD had used Massive previously for commercials but not for a large feature, and Butler observes, A lot of upfront effort goes into using it. You need to train people before it becomes a plug-and-play tool. So youre looking for a skill set from artists that is not only a hand-animating skill set. It takes a technical and procedural planning skill to create the mechanisms that allow you to do these shots with thousands of sentient beings.

But Butler stresses that Massives usefulness extends beyond groups of humans. Armies or crowds of people represent a small subset of what Massive can do. We used it to animate the boats, too. It allows objects like landing vehicles to become sentient. We used that methodology where appropriate, and also used hand animation where appropriate.

Remarkably, Eastwood never looked at any sims, says Owens. He looks at the edits and the previs mockups. We worked through the edits, and then he looked at the finals. Butler adds, somewhat in amazement, Clint pretty much accepted everything that we showed him. Wed have to twist his arm to pull something back. He is the coolest guy.

The Rendering Challenges

Butler explains that the CG in Flags of Our Fathers was pretty much all rendered with RenderMan. We also used mantra and mental ray [from mental images], but because of the size of this show, we built a pipeline where we had to make compromises. DD was able to take advantage of the latest version of RenderMan, which supports radiosity and global illumination, and Butler concludes that, fundamentally, RenderMan was the right choice for this show.

But DD also used its proprietary software NUKE to handle some of the rendering. While NUKE is primarily known for its compositing capabilities (and it was used for the extensive comp work in Flags), Butler maintains, NUKE is now a fully 3D and 2 1/2D rendering package. So it was used along with RenderMan.

But the biggest breakthroughs on this project came in the area of volumetric rendering. Because the number of on-set explosions were limited, DD had to create a significant amount of digital pyrotechnics, which they animated using Houdini software. Butler notes, Pretty much all of our effects pipeline goes through Houdini, including the sand bomb explosions and gun hits.

The biggest breakthrough for DD on this project came in the area of volumetric rendering. It had to create a significant amount of digital pyrotechnics, which they animated using Houdini software. 

The biggest breakthrough for DD on this project came in the area of volumetric rendering. It had to create a significant amount of digital pyrotechnics, which they animated using Houdini software. 

Houdini has an open enough architecture for us to get in and tinker with it. We always invent tools for a show, and theres always a new need to manipulate images above and beyond what you can get from off the shelf software. Houdini allows us to pull the hood up and get in there and manipulate it to do specific tasks. The most important thing for us is the ability to augment shots.

Rendering those explosions was a complex challenge, he continues. We had many explosions going on in the battle scenes, which we rendered with our volumetric renderer Storm. Its a tool that hooks into our Houdini pipeline and allows us to animate dynamic behaviors and then bring them in as full volumetrics. The lighting interactions are complex. One of the advances of volumetric rendering is that if youve got a CG tank thats exploding you can light the tank with that volumetric lighting. It adds a level of reality.

Butler credits the creation of Storm to Alan Kapler (X-Men, Stealth). He started writing it on a whim, and back then we called it Voxel Bitch. Once it became publicized we thought: We cant call it Voxel Bitch, so it became Storm. Its very useful in rendering all types of realistic phenomena, observes Butler. Its something that pops its head up many times. The crude rendering of particle systems doesnt cut it now. That doesnt mean this work doesnt involve particles or other simulations fluid simulations or gas simulations are dynamic forces behind it. But the ultimate rendering winds up being Storm because of its volumetric capabilities.

The Mirror Movie

At the same time Eastwood was making Flags of Our Fathers, he was also envisioning a companion film called Letters From Iwo Jima, which tells the same story from the Japanese point of view. As Owens recalls, It was originally supposed to be a complete Japanese production, with a Japanese director, shot entirely in Japan. But Clint was loving everything on Flags and when the script came in for Letters he loved it. His experience on Flags led him to think We need to complete this.

When he began the project, Owens couldnt predict the number of effects on a non-effects film. In the end, the vfx team did 450 shots for this project, though most will think there were a only a handful.

When he began the project, Owens couldnt predict the number of effects on a non-effects film. In the end, the vfx team did 450 shots for this project, though most will think there were a only a handful.

From a visual effects perspective, notes Owens, Letters was bound to be a simpler project. It was more contained, and a lot takes place underground or at night. It needed some bridgework to get it done, and we kept playing around with it. What is interesting is that there are shots in both movies that are exactly the same because they cover the same moment.

But the challenge was still significant. Owens admits, For a while we were all getting a little slap-happy, thinking Which movie is this? Its hard to keep all the shots in your head for one movie. And Letters is completely in Japanese. There were two translators and script supervisors on the set, as well as a cultural A.D. set person. Joel Cox edited Letters with some translations and timings but he worked off what felt natural, knowing what the dialogue was supposed to be. I never would have predicted it would have gone that smoothly.

The Final Analysis

In the end, notes Butler, We did 450 shots for this project, but most people will only think there were a handful of visual effects shots, which is a good thing. Butler sees major parallels between Apollo 13 and Flags. Apollo 13 was a movie that people didnt associate with visual effects, and that was a tribute to the storytelling. When the effects work isnt showcased thats a double-edged sword on one side youre happy that you fooled the audience, but on the other side, nobody knows to credit you!

For his part, Owens admits that on a non-effects effects film you cant predict how much work will eventually be required. We didnt know the volume of shots in the beginning, but we were saying, Lets call it 250. Thats how it always goes. But we werent fooling ourselves. We knew Clint was going to make up his mind as we went through this, but we said, This was the minimum that it could be. I guessed it would be as much as 500 shots, and I came pretty close.

Seeing how well we were able to integrate the effects, Clint told me that it surpassed his expectations of what he thought we could accomplish. Clint would probably never choose to direct a total fantasy film, but I think because of this experience he has a new confidence in visual effects. It wouldnt phase him for a second if he had to do a movie with a great deal of visual effects.

Ellen Wolff is a southern California-based writer whose articles have appeared in publications such as Daily Variety, Millimeter, Animation Magazine, Video Systems and the website CreativePlanet.com. Her areas of special interest are computer animation and digital visual effects.

Tags