How did Rodeo FX got involved on this show?
We had worked for the studio in the past on other productions. We also were in contact with the studio and the VFX producers on the film and we got awarded two scenes. One where Jack walks in a room filled with treasures, and the last shot of the movie where we pull out of the Tower of London and go through the clouds.
How was the collaboration with director Bryan Singer?
Depending on the size of the show we deal with the director & VFX supervisor or sometimes only the VFX Supervisor. In this particular case, our collaboration with Bryan Singer was very limited. We had a very close creative collaboration with Hoyt Yeatman, the VFX supervisor of the movie. He’s has a great eye for detail and always had some great stories to tell.
What was his approach with the visual effects?
The approach with the VFX was all about photorealism. The environment we created for London had to look like a filmed environment. The brief was simple, it had to look like it was shot like any other shot of the movie. The treasure filled cave had to look massive, old and abandoned. We had to create the illusion that the gold was sitting there for the longest time in this huge cave that the giants built and carved out the rocks.
Can you tell us more about your collaboration with Production VFX Supervisor Hoyt Yeatman?
Working with Hoyt Yeatman was a real honor and pleasure! He has an extraordinary eye. He understands VFX extremely well and guides you very well technically and artistically while making sure the storytelling aspect of the VFX is always present and serves the movie properly.
What have you done on this show?
On this show, we worked on two sequences. One sequence was when Jack walks in a cave filled with treasures and the other sequence was the last shot of the movie, the big pull back from the Tower of London. These were internally supervised by Sebastien Moreau, Rodeo FX’s founder and VFX supervisor.
Jack arrives in a huge cavern full of gold. What references and indications did you received from production for this treasure room?
The idea with this scene is that Jack inadvertently finds himself in a cave filled with gold and treasures. It’s an old forgotten room where the giants stored all the treasures they took from the humans over the years. Our references were classic illustrations of treasure filled caves. We also looked a lot at natural caves for lighting and topology reference.
Can you tell us more about its creation?
This environment was a big challenge. The fact that the movie was shot native Stereo with the RED Epic meant that we had to have a very solid 3D geometry to create the set. We started with a lot of concept work to get the look and scale down. Once we had a good concept to use as reference, we started with building the geometry of the cave. Scale was very important at this step, because the rocks and built nature of certain parts of the cave needed to look like it was built by giants. All the construction needed to have a giant scale while the treasures had a human scale. We then proceeded by filling that room with a massive amount of treasures and carefully light the scene to give it a very dark and moody look.
One of the main challenges in this shot was to properly and accurately showcase the massive scale of that cave in a stereo world. Things can look very quickly out of scale in stereo.
How did you create so many gold elements?
Our approach with the creation of the gold piles was procedural. We knew that doing it “by hand” would work but it would take an eternity to meticulously place them in a natural way. So instead, CG supervisor Mikael Damant-Sirois decided to model a huge database of golden treasure props, instance them and use a particle system within the ICE system of Softimage XSI to “drop” these props and create piles and mountains. It made the scenes very light and simple and they looked very natural.
We shaded, textured, and lit all these elements and rendered them out using Arnold as a renderer.
One of the challenges was that we needed to send these elements to our matte-painting department to finalize their look and to integrate them into the cave environment that was being painted. Since they were created using an instanced particle system, they did not have a physical geometry on which it could get re-projected. We decided to use World-position passes to create the geometry “after-the-fact”, and reproject the matte-painted treasures back on themselves. it was a very ingenious way of doing it and the result was seamless.
What was the real size of this set?
The size of the live plate was pretty small, it was only the foreground part of the shot, where we see Jack walking. it had the FG harp, and the FG pile of treasure. The environment we created was the equivalent of two football fields in length!
What was your approach for the impressive pull-out shot?
We knew this shot had to look as photoreal as possible so our approach was a very photographical one. We took a ton of photos and recreated digitally London.
How did you get the reference pictures and footages for London?
We sent Robert Bock, our VFX director of photography, to London where he shot a TON of references that were used by our environment department to faithfully recreate the Tower of London, the London Bridge, and London.
Can you tell us in details about the creation of London environment?
Our crew shot a ton of reference photos to be used as reference and as textures. I believe it was around 3000 photos. Production provided us with a Previz to use as a guide for the shot they wanted us to create. We recreated that previz and carefully decided what angles we would need to cover to digitally recreate the environment. We rented a helicopter and shot the angles we needed. We actually blocked the traffic on the London Bridge for 20 minutes to do this! Sorry London (laughs).
We selected 8 “master” photographs that would allow us to reproject on geo. One was a close-up of the outside window of the Tower of London, the next one being one of its courtyard, the next one being close to the bridge, and so on until we were high up in the sky looking down on London. We basically tried to recreate the camera move with still photographs from a helicopter that was doing an approximation of the path that the CG camera would end up doing. We couldn’t do the shot with live photography from the helicopter since we needed to start very close on the Tower of London and fly underneath the Bridge of London and go very high in altitude rather quickly. It would have been impossible to do with live action photography from a helicopter.
Once we selected our master photos for matte painting purposes, we used photogrammetry techniques to very accurately recreate the geometry of the environment. We used 3D Equalizer to generate the point cloud that we exported to XSI where the much more accurate geometry was built. Meanwhile every single building was extracted out of the photographs and missing parts were faithfully recreated.
These were all projected on the geo and we carefully transitioned from different photos throughout the shot to accurately have the proper perspectives. This part was a challenging one since we were starting from a frontal view of the Tower of London and ending with a top view so we had to gradually mix from one projection to another one throughout the shot.
We created the trees using SpeedTrees and rendered them out of Softimage, using Arnold. The crowd in London was all CG, as were the cars, the boats, and the Thame river. We also created volumetric clouds in the background. Basically the whole shot was a digital recreation, the only live action part of this shot is the first part, where we see a kid looking like a young Stanley Tucci looking at a crown in a showcase.
How did you manage the stereo aspect of the show?
This movie was shot native stereo using the RED Epic. Managing it was a fairly easy job. We had already done a good number of stereo projects and our pipeline and tools are very stereo friendly. Production sent us the scans in SXR format with all the necessary info in the header of the files, it helped us a lot.
How did you split the work on Nuke and Flame?
For this show, all the shots were composited on our Flare systems.
What was the biggest challenge on this project and how did you achieve it?
One of the big challenges of this movie was creating fully digital environments in stereo and giving them the proper scale in stereo. In a stereo world, there is very fine line between having a good and logic stereo depth and making something look miniature and out of scale. Especially when building a huge cave in a Giant world scale or a massive pull-out of London.
Was there a shot or a sequence that prevented you from sleep?
No not really, but the London pull-out shot was a good challenge because of the amount of layers, its length and the fact that everything had to be rendered twice because of the stereo. Also, since we were seeing objects in very close-up and seeing those same objects from high above in one continuous move, anti-aliasing issues kept creeping up and had to be dealt with.
What do you keep from this experience?
This was a great experience! Working with Hoyt Yeatman and Bryan Singer was a very rewarding adventure and would love to repeat it. The movie was a great ride filled with beautiful shots, I’m very proud of the work we did and to have had the opportunity in creating these great looking shots.
How long have you worked on this film?
We worked on this film for almost 6 months.
How many shots have you done?
We delivered 7 complex shots.
What was the size of your team?
We had a team of around 40 on this project.
What is your next project?
We just finished working on NOW YOU SEE ME and PACIFIC RIM. We’re currently in production on HUNGER’S GAME: CATCHING FIRE and on an IMAX film entitled JERUSALEM.