Rhythm & Hues visual effects supervisor Raymond Chen discusses his studio’s vast visual effects work for Night at the Museum: Battle of the Smithsonian (outside the USA: Night at the Museum 2), with fxguide’s Ian Failes.
fxg: What did Rhythm & Hues work involve for the film?
Raymond Chen: We worked on a number of CG characters – Lincoln, Octopus, Squirrel, Einstein Bobbleheads, Rexy, Thinker, Venus, Jonas Cherubs, and a number of background sculptures and animals. We did Diorama Jedidiah and Octavius, the Times Square sequence with CG crowds and CG set extensions. Also the Teddy bust sequence with a partial CG bust. We also worked on the Air and Space sequence which had a CG Wright Flyer and green screen/miniature comps, a Lincoln Memorial set extension, Natural History Museum set extension, a number of CG planes – Amelia’s Vega, Wright Flyer, model planes in Air and Space, Horus heads, the Netherworld and the transformation of Horus.
fxg: The effects in Battle of the Smithsonian are so much fun. How did that impact your approach to the vfx work?
Raymond Chen: The biggest impact was that, for a good number of the shots, humor or comedy was the most important thing, so timing and performance became more critical.
fxg: What kind of reference or previz did you look to for your shots?
Raymond Chen: The production had previz done for a majority of the sequences for the film. For some, like the Air and Space Escape sequence, we stuck extremely close to the previz, since that was also what determined the miniature BG plates, and as it was done relatively later in the process, it was close to what the final cut was. For other sequences, we used the previz as a starting point, and deviated from it as the edit evolved and as the direction from Shawn Levy changed.
fxg: Can you talk about your on-set involvement?
Raymond Chen: I was on set for most of principal photography in Vancouver with overall visual effects supervisor Dan Deleuuw. In addition to the usual set-data reference and HDRI gathering, I was involved in some of the motion base and motion control camera work where what was being done on-set was being driven by data provided by Rhythm & Hues (either from animation or tracked plates).
Raymond Chen: There is a fair amount of interaction between the octopus and Larry and Amelia in the Turner Gallery sequence, so we had to do quite a bit of work to have something in each shot for the actors to work against. For the shots of the octopus picking up and dragging Larry in for a hug, for example, we had several gags that needed to be set up to put that interaction of CG creature with live actor into the plate.
For the wrapping around shot, we had a quick-ratchet strap that went around Ben Stiller’s chest area to get the compression and squeezing that would come from the tentacle. The shot where Ben is dragged closer also used a pick wire and harness to suspend him slightly off the ground, and another guide wire to drag him across the floor. And for the close up shot of Ben’s face colliding with the octopus, we used a green pad covered with methyl cellulose for slime, which we enhanced and partially replaced with CG slime.
The scene you are asking about, the water splashing shot, was one of the more complex shots in the sequence. We had a large styrofoam buck created, based on the design and model that we had for the octopus. This was painted black, so that we would be able to capture the highlights of the water as it hit and rolled down the surface. This was placed into the gallery, along with a water cannon to shoot water at the forehead of the buck.
Ben Stiller used an empty frame to perform the throwing motion, and the water was shot through the empty frame and onto the buck. Since the final CG octopus was moving around a bit, we couldn’t use all the practical footage, of course, but it was also great reference for the many layers of CG water we ended up creating for the shot. These shots were also shared shots, with CafeFX doing the work for the animated paintings.
Raymond Chen: The squirrel shots where there was little interaction were pretty straightforward. We had two methods for representing the squirrel for Steve Coogan to play against – a foam core cutout head for the more static shots, and a ball on the end of a long painter’s pole for shots with movement, such as when the squirrel rears up to full height. For the riding shots, we created a styrofoam buck based on our CG model of the squirrel, using the section from the back of the skull to the middle of the back. This was mounted onto a motion base, and rigged with an additional piston to get beyond the range of motion constraints of the base.
We used both motion curves derived from animation of our CG squirrel, as well as live control of the base using a waldo (some of the animation, such as our run cycles, ended up being too fast or violent for playback). For the shots of Octavius on the squirrel, we also ended up using partial CG replacement for his legs and cape.
fxg: The lighting on the statues has a real quality about it. What techniques did you use to light or render those shots?
Raymond Chen: For all shots with CG elements, we used our cube HDRI camera to capture the lighting setup for the shot. This is a camera that Rhythm & Hues developed that uses six lens and CMOS sensors to stitch together a 360 degree high-dynamic range image of the environment. This image is used as a basis for our CG lighting and also is used for reflection maps. For the statues, we also did a lot of work during our look development phase to create specific looks for each statue. For example, The Thinker had multiple, overlapping specular highlights with slightly different coloration to simulate the patina of aged bronze. The Venus statue incorporated a significant amount of subsurface diffusion of light to simulate the translucent quality of white marble.
Raymond Chen: The shots within the Lincoln Memorial used a lot of CG; the main thing we used from the live action plate were the actors. There were also a number of fully CG shots. The green screen stage had a floor and several columns which were only around 15-20 ft. high, so most of the Memorial was done as a CG extension. We created the entire Memorial, so in some shots we even replaced the practical columns and floor that were on set. The exterior of the Reflecting Pool and the Washington Monument were done as matte paintings.
And Lincoln himself, of course, was CG. The only shots with a CG Lincoln that used actual footage from Washington, DC were a couple of wide shots showing the exterior of the Monument. For the shots with Larry being held up by Lincoln, we also used a CG jacket that transitioned from Ben’s jacket to allow for interaction with Lincoln’s fingers and for controlling the stretching and folds on the cloth that would result from being suspended.
fxg: Finally, for the Rexy shots, were you able to utilise the previous model and animation setup from the first film? How did you go about adding such personality to Rexy?
Raymond Chen: We did use the previous model and animation setup from the first film, but needed to rework some of the textures and lookdev to add more detail, because there were a number of shots in this sequel that were more close-up than in the original. Using the same setup, however, meant that we ran into some of the same problems they encountered working on the first film.
Rexy is a character that was distinctly defined in the first film and Shawn Levy had very specific ideas about what Rexy was supposed to be like, so understanding the character was very easy. Getting that personality to read clearly in the shots was the work of Craig Talmy, our animation director, and a team of very talented animators.