BATTLESHIP: Chris Harvey – VFX Supervisor – Image Engine

After he talked about his work on TRON LEGACY, Chris Harvey then joined Image Engine and explains to us his work on BATTLESHIP.

How did Image Engine get involved on this show?
The groundwork to getting involved in this show really started way back with DISTRICT 9. After seeing the film, ILM became very interested in our creature work and came up to Vancouver to discuss the potential of working together in the future. When the opportunity came up to collaborate on BATTLESHIP, and the need for a specialized crew to handle the look and performance of the Thug it seemed like a perfect fit and we all jumped at it.

How was the collaboration with Director Peter Berg?
That’s an interesting question really. The way the show worked, as an outsource vendor from ILM, we did not deal directly with Pete. Instead we worked with the ILM crew very closely and our work was simply presented alongside the rest of the work on the film… creating a single source of contact for Pete. Now that being said we were certainly “working” with him to create the sequence albeit not directly, and for us it went very well. Peter really liked our sequence and some of the things we were doing with the Thug, to the point that we often got notes like, “wow, that’s great… lets add it to all the other shots!” or “I like it, can we double or triple the length of the shot so I can see more?”. Its a lot of fun and very rewarding when the director is happy with what he is seeing and asking for more.

How did you collaborate with VFX Production Supervisors Pablo Helman and Grady Cofer?
It was great… I often talk to people about how collaborative and honest working with the ILM crew was. At Image Engine we dealt primarily with Pablo. He and I got along very well and enjoyed a similar sense of humor and vision for where we wanted to take the Thug. Both ILM and Image Engine has Thug shots to complete on this film so there was a lot of share in both directions. It was very refreshing to feel like there were no egos involved in the artistic collaborative process. If they came up with something cool they would share it with us and when we came up with something new we would share it back to them. And having that trust and creative feeling to be able to run with something and know that if it worked it would get rolled back into not only our work but the film in general was really a joy.

What have you done on this show?
Image Engine handled the Engine Room and Ship Deck sequence. After a small group of Thugs rescue a Regent, one stays behind to gather information about their human enemies… it is here that our sequence picks up. Initially the Thug is completely unconcerned by the humans on board, as he doesn’t see them as any kind of threat, instead he is simply more interested in gathering information. It might be interesting to know that this was something that changed during production – when we first started on the sequence we had a number of shots that showed the Thug smashing and cutting the engine room apart. Peter Berg decided he wanted to shift the focus, so our shots changed into repairing the practically damaged engine and coming up with a scanning tool instead. However, once the Thug is attacked… it flips back to all hell breaking loose and you really get to see the damage one of these guys can dish out! Throughout the rest of the sequence we see the Thug reign down havoc on the crew…that is, until Rhianna says “Mahalo” and blows him to pieces with a 5″ gun.

Can you explain to us the asset sharing with ILM teams?
As I mentioned earlier the development process and sharing with ILM was very collaborative, with assets, design and ideas flowing in both directions. Initially we received a Thug asset with some look-dev which was at about 75% completion. The material included a model, textures (with a texture breakdown document since we both have different pipelines), a few early motion studies and some turntables. The model we were able to pretty much use as was initially, however the design changed in a number of ways that I will get into in a moment. In terms of look-dev and model/texture ingestion that was spearheaded by Nigel Denton-Howes (CG sup on the show) and Rhys Claringbull (then Asset Supervisor), we took the textures and the turntables that ILM provided and translated everything into our 3Delight lighting and rendering pipeline. The rig was used only as a way to translate on-set imocap data as IE has a robust animation and rigging pipeline with a lot of proprietary tools… so it was better for us to build our rig from scratch. Jae Cheol Hong was our Lead Rigger and together with Jeremy Stewart (Animation Lead) he did a great job building a complex yet very animator friendly system for dealing with the Thug and all its Armour. Once we had these things “in the system” the actual work to bring him to the final level really began.

How did you take the Thug look development to its final result?
Once we had the basic asset ingested we started the long and highly iterative process of “finding the Thug”. Sometimes look development is thought of as shaders and textures. But we treated it far more holistically, it involved everyone, animation, rigging, shaders, textures, lighting and compositing. You needed to understand how he moved, his presence … how he would be lit and composited into the shots, the mood you wanted to convey. So as you can imagine the process took some time and was not completed before shot production began. Obviously the usual lookdev fundamentals apply, checking him in actual and multiple shot lighting conditions and compositions. But in this case the arching holistic approach worked well because I think it was how Pete looked at the development cycle as well…he never arrived at a, “ok the Thug is signed off on” prior to deep into shot development. And as you can imagine that meant design changes we would have to roll with throughout this process and into shot work. He got “thicker” through the torso and legs…giving him more bulk and mass, we added armoured boots to previously unclad feet, his tool hand went through a number of design changes/modifications, his visor changed, etc.. However because we planned for this style of development we have built a very flexible rigging and animation system and likewise lighting and compositing partnerships, so even though all of these changes came throughout actual shot production we were able to update and have most things just ripple through the shots without an immense amount of pain. In the end, getting to put so much creative design work back into the Thug was very rewarding. ILM gave us lots of space to come up with things we thought would work and in the end we would simply package it up and send it to them to integrate back into their Thug shots as well.

How did the DISTRICT 9 experience helps you with this show?
Having not been here during DISTRICT 9 this is a tough one to answer except to say that the crew that came off of that and subsequently THE THING had the right mix of experience to go into this show. In addition to that Image Engine’s pipeline is very well set up for this kind of work, and we continue to improve upon and push (and I should say push with an explanation point) everyday! The thing about the experience of DISTRICT 9 is that people knew what they had to achieve, they knew how to do it… all that was left to do was to raise the bar.

Can you tell us more about the impact of the changes for the Thug helmet?
The impact here had less to do with the overall helmet and really more to do with the visor itself. In the beginning the Thug was never intended to have a lot of facial animation due to the fact that initially the visor was supposed to be almost fully opaque. That changed one day, when we took one of the close-ups and decided to give more of a hint at the emotion the Thug was going through… revealing more of his eye and face. And much to the ‘chagrin’ of my producer (not really) Pete liked it so much that he wanted more of it, not only in that first experimental shot but in every shot…it opened up a lot of previously completed shots. But we all knew it was the right choice and that in the end it would only strengthen the character and strength of the Thugs performance.

Can you tell us more about the facial animations?
After the significant change to the visor it became very apparent that we would need a lot more performance out of the Thug’s face. It was up to our animation team to go in and add full facial animation (with some last minute adjustments to the facial rig) to a slew of shots. Jeremy and the team would carefully look at the actors facial performance, taking cues from both him and from the body animation they created for the Thug…and with small eye darts, blinks, nostril flares, scrunches, twitches, massage that character into the Thug. It was very important for Pete that the Thug could not only smash things but also emote. After we had the new animation, we needed to create some new lighting adjustments to add internal helmet lights. It became a balancing act of art direction between animation, lighting and comp to find the right mix of performance, light and visibility for each shot. Lighting provided separate key, fill and ambient lights as well as interior helmet lights which could be dialed in on a per-shot basis to get the best mix of reading the facial feature while keeping the Thug looking mysterious behind the visor.

How did you create the breath and condensation effects?
This was one of those things that we started experimenting with early on and Pete ended up liking it so much it ended up in pretty near every shot. Sometimes together with the body animation we could use this to help show emotion and character, and at other times it simply became a tool for adding visual interest to a shot. It was handled almost entirely in compositing with a couple of utility passes from CG. It became an interesting balancing act with the newly revealed facial animation I mentioned above.

Here is what our Lead Compositor Bernie Kimbacher had to say about it:
“On one hand we had to enhance the look of the visor and add some life to it, which we handled with cold breath and condensation elements and on the other hand we still had to get some of the facial expression across while not revealing too much of the Thug. In order to stay on top of our tight schedule we had to make most of these comp tricks as automated as possible, as setting it up for each shot would have taken us significantly longer. For the visor we mainly used a pass we get from lighting called ‘positional reference’. This allowed us to map different textures on the visor, which we could then animate to replicate the Thug’s breathing cycle.”

Can you tell us more about the transforming weapon hand?
The tool was a lot of fun, and its an area we got to play a lot with. It was pretty much an open book in terms of what it could do or become…so that’s how we handled it. But instead of handling it from a modelling or design perspective, we approached it from what initially might be considered a backward approach…we looked at its motion first and then backtracked from there. We pulled the animators into a meeting and basically gave them free reign to see what they could come up with… telling them to take the existing model and just start animating it into different configurations with only one rule, it shouldn’t feel like its something coming from nothing. And that’s just what they did… we produced lots of different variations and configurations of potential weapons and how they might translate into them. This allowed us to get a lot of flavours very quickly without long or tedious modelling… or considerations on the ‘how’ of things, and instead just focused on the coolness of the motion and silhouettes. We took the best and showed these to ILM and Peter. Once we had a design settled on, it went back to modelling to actually create the components that fit into what the animators had done, and the rigging team to create a rig that could actually support it. From there it just grew… Peter loved the tool so much that he kept looking for new opportunities to show it off, resulting in a number of different configurations and movements.

How did you achieve the lighting challenges?
HA…actually for me this was a lot of fun because in an early meeting with Pablo he made a comment about pushing a plate around a lot! This is not something you hear everyday in visual effects, though I think it speaks to Pablo’s experience … I think its something you should hear more. But more often than not the plate is king… MATCH the plate. Well after the call I had a room of people looking at me like they couldn’t believe what they had just heard… CHANGE the plate? I remember laughing and saying something like “This is great…we are free to be more artistic!” Essentially it boils to this: If, on the day, an actual Thug was on set… you know what, it might have been lit differently. And yes all the old rules still apply, you still have to sit it into the plate, but it meant we could alter the lighting in the plate to not only force the Thug into the plate but we could also move the plate to fit with the Thug.

Rob Bourgeault, our Lighting Lead put it like this:
“The resulting output from lighting was visually incorrect as compared to the plate with cheated intensities and positions; however, it gave compositing the necessary range to apply a post processing effect to the renders and thereby achieve the desired stylized look. Most of the artistic license existed in painting and shaping the character with very intense reflection cards to achieve the result that Peter Berg was looking for. We were often requested to add additional cards to accentuate the key and rim reflections of the character. This was done with special attention to the plate, looking for any opportunity to enhance the interactive nature of the character within its surrounding environment.”

Can you explain to us step by step the Thug destruction?
Blood – Sweat – Tears. We actually completed this to final twice. Originally the Thug was more or less engulfed in flame, utilizing a lot of practical effects… however, as our sequence progressed and with Peter Berg’s enthusiasm for how it was playing out overall, he requested a change. He wanted to make more of an event of the final destruction of the Thug, he wanted to see a visceral moment where the audience gets to see it torn to pieces and can cheer for. And in order to do that the effect became almost entirely synthetic, and to make sure we really nailed what Pete was looking for the second time out, we spent some time essentially postvizing out how the death would play out and the visual beats. This in itself was a bit of a process involving primarily Animation and Compositing with a bit of FX. However once everyone knew where we were going then it really was turned over to the FX team to start figuring out how the hell they would achieve this look. And to speak to that it’s better I let Jacob Clark, our Lead FX TD and Daniel Elophe, Senior Compositor, describe the process:

Jacob Clark: “With regard to the Thug destruction shots, nothing quite like this had been undertaken at Image Engine before. This was a very ambitious shot from day one, involving a fully computer generated character in broad daylight being torn apart in slow motion by a fire spewing cannon. The fire and smoke not only occupied a large section of the screen, but the fire had to interact with the intricately detailed armor of the Thug, which had almost 800 different interlocking metal shapes. The project required some heavy hitting fx power, and Houdini was chosen to handle the job. I have had a long relationship with SideFX software and have been very happy with their product Houdini in the past. With this in mind, I approached SideFX to use their new Houdini 12 beta version, which has great tools for addressing our volumetric needs in development. Our IT team was able to quickly integrate into the software into the Image Engine film pipeline. In the end, we used Houdini’s new volumetric technology quite extensively in the production of the Battleship Thug destruction.”

Daniel Elophe: “The most challenging aspect was that the fire needed to interact with the breaking thug, and move around/through pieces. There was a good solid month of back and forth, brainstorming and trying lots of variations of fire within the context of the explosion with Jacob running fire simulations from various body parts of the Thug. For me as an artist, this was the most rewarding part of the process, working extensively with Jacob and being given a lot of creativity and flexibility from Chris and our producer, Vera Zivny to explore what would work well.”

How did you create the various simulations for it?
Well, I am sure I looked at hundreds in dailies. But when I asked Jacob what the final tally was in the shot, here was his response:
“By the time the shot had finaled we had run over 80 different simulations of Fire and Smoke in the scene. With all of these simulation tests coming through, I have to mention Senior Compositor, Daniel Elophe’s fantastic ability to gather all of this together for the final completion. Each fx render had multiple passes, and if you were to add all these passes up with the renders of the Thug, Daniel was able to corral well over 100 different image renders to complete the shot!”

What was the biggest challenge on this project and how did you achieve it?
That’s tough to say on this one… aside from the final explosive death, the work across the board was fairly consistent… so one single item or effect is hard to pick, rather it was more the show as a whole single entity. I guess for me it might simply have been navigating a new facility. Being my first project here at Image Engine, a lot of the crew was new, it was a new pipeline, new tools and methodologies to how certain things are approached here. And so while I brought with me some of my own ideas and approaches I needed to grow with the people here and learn to trust them, and them me. We were lucky enough to be able to assemble the ideal crew for the show. All of the lead artists and production team had come from shows like DISTRICT 9 and THE THING, and along with their respective crews each played an instrumental role as part of a great team in making the sequence what it is.

Was there a shot or a sequence that prevented you from sleep?
No to be honest the show ran pretty smoothly and it was a lot of fun so I didn’t lose a lot of sleep over this one. That being said we did have a few shots that were more, shall we say “challenging” than others. In particular the Thug explosion/death shots. But in the end you have to trust your team to do a kick ass job and they did exactly that!

What do you keep from this experience?
Well, it was a great introduction to Image Engine for me, I got to get more familiar with their pipeline and the crew here. I also had a great experience working with Glen and Pablo from ILM, and I certainly hope to keep those relationships alive and work with them again in the future. Beyond that the thing you should always take away from a production: new ideas, new ways of looking at things and new friends… all of which live on to improve future work and your life in general.

How long have you worked on this film?
Image Engine was involved in this for about 10 months and it broke down roughly into 3 months of assets and look development, 4.5 months of shot production, with an extra 2.5 months of extra shot production using a reduced crew at the end.

How many shots have you done?
In total we have 77 shots in the film and worked on an additional 10 or so that didn’t make the final edit.

What was the size of your team?
Approximately 40 at its peak.

What is your next project?
There are a number of new projects going on here at Image Engine that the crew has moved onto. Neill Blomkamp’s next feature ELYSIUM (Sony Pictures International and Media Rights Capital), “R.I.P.D.” (Universal Pictures), and the show I am Visual Effects Supervising, Kathryn Bigelow’s yet untitled film.

A big thanks for your time.

// WANT TO KNOW MORE?

– Image Engine: Dedicated page about BATTLESHIP on Image Engine website.

© Vincent Frei – The Art of VFX – 2012