<I>The Last of Us</I>: Inside Weta's post apocalyptic VFX
Kendra Ruczak
August 14, 2023

The Last of Us: Inside Weta's post apocalyptic VFX

Set two decades into a devastating global pandemic, HBO’s The Last of Us is a post-apocalyptic story of survival, perseverance, and hope. Inspired by developer Naughty Dog’s acclaimed video game franchise of the same name, the series explores a world that has collapsed after the outbreak of a massive fungal infection that transforms humans into violent, zombie-like creatures.

The series, which received 24 Emmy nominations, follows hardened smuggler Joel (Pedro Pascal) as he takes on the daunting task of protecting Ellie (Bella Ramsey), a tough 14-year-old girl whose immunity may be the key to saving humanity. The unlikely pair must brave the unknown and traverse the uncharted ruins of the United States, encountering terrifying hordes of infected fungal creatures and viciously dangerous human survivors along the way.

Visual effects played an essential role in bringing the post-apocalyptic world of The Last of Us to life on screen. Wētā FX delivered close to 500 VFX shots for the project, focusing on the creation of infected Clicker and Bloater creatures, CG animals, environmental transformations and extensions, and pivotal battle sequences. Two key members of the Wētā team, VFX supervisor Simon Jung and animation supervisor Dennis Yoo, shared insights into their Emmy-nominated work on the series.


Can you provide an overview of Wētā’s work on The Last of Us?

Simon Jung: “Wētā FX delivered 482 VFX shots in total, ranging from simple set extensions and the weathering of buildings to full CG environments. The creature work consisted of hordes of Infected, which we see in the battle scene in Episode 5 and in the Faneuil Hall environment, the Clickers inside the Bostonian Museum, the Bloater, the Child Clicker and a digital double of Perry [Jeffrey Pierce] in the battle scene outside Kansas. We also created animals, namely the monkeys at the university campus and the lab, the deer that Ellie shoots, and the giraffes in the final episode.”

Dennis Yoo: “On the motion side of the show, we had to find and create the performances for all of our CGI creatures. As Simon mentioned, this included finding performances for crowds of Infected, animals, and monsters like the Clickers, Child Clicker, and the Bloater. Most of the motion work included staging crowds and finding / creating performances with realistic animation using various techniques in motion capture and key frame animation.”

Can you tell us about creating VFX for the show’s infected creatures?

Simon Jung: “We utilized scans delivered to us of the live action Clickers we see in The Bostonian Museum and built those to a highly detailed level. That allowed us to use them very close to camera in the shots where either augmentation or replacement was required. For the VFX shots in that environment, the Clickers were either all digital or partial replacements. In some cases we did full head replacements, and in others we replaced only the cordyceps growth while holding onto the rest as it was captured in camera. 

“The prosthetics of the Clickers already looked amazing, so we mainly focussed on material changes like subsurface scattering and adding little details, like pockets of fungal fuzz in the cracks of the fungal shapes and the overall level of sliminess. All other Clickers that we see in the battle were variants of those two original ones. Using them as a base, we varied their gender, ethnicity, wardrobe, scale, and grooms. We also ran variants of the signature cordyceps growth on the Clickers’ heads.

“Additional scans of multiple Infected were built to a hero level, which allowed us to simulate their tissue, muscles, clothing, and hair. All of those also had sub-variants to avoid obvious repetition in shots where we see a lot of them in camera, during the battle, for example. The first wave of Infected that burst out of the sinkhole are fully CG creatures. We also made crowd elements that were used as Massive agents in the big wide shots during the battle, as well as for the shots in the Faneuil Hall environment. These were single mesh assets for our Massive simulations and were used further away from camera, in addition to the motion captured, higher resolution ones closer to us."

Dennis Yoo: "For animation, we relied heavily on using The Last of Us game footage as reference and inspiration for the performances of the Clicker and Bloater. 

“We used motion capture as our base of motion for our Infected. Ike Hamon was the mocap performer for our Bloater. We used his performances as our starting point of the Bloater’s movements. This motion was altered with keyframe animation to sell the disjointed performance of our Clickers and to create the sense of strength and weight of our Bloater.

"Our senior animator Jung-hoon Park stood out with these keyframing tasks as he single-handedly keyframed our extensive Clicker shot in the Bostonian Museum where Joel hides behind a glass display.

“The Child Clicker’s body performance was “rotoscoped” in a way, where we copied frame by frame the footage of the real performer with our animation puppet. The facial performance was a mixture of blend shape animation and “shot sculpt” modeling from our lead modeler Pascal Raimbault.

“The giraffe facial performance was solely keyframed using blend shapes, which our lead facial artist Jacob Luamanuvae-Su’a rigorously tested against the live action giraffe to make sure our CGI giraffe could perform one-to-one against the real footage.”

What types of environment extension and transformation work did you complete for the show?

Simon Jung: “We did a fair amount of augmentation of buildings that were captured in camera across multiple episodes. That included adding ivy, breaking windows, adding damage and dirt to facades and roofs, and adding debris to roads. 
The Bostonian Museum is a good example of augmentation. The main building was partially live action and we surrounded it with CG rubble piles and destroyed high-rises. The actual location where the scene was filmed was a menswear store in Calgary, Canada. We used Google Earth images to reconcile it with the real-world location in Boston and matched the layout of the surrounding buildings there. 

“Using LIDAR scans as a base, we then built a CG asset of the building and added a collapsed extension to it. Having a detailed model allowed us to grime up the facade, add the dried out cordyceps growth to the walls, and add ivy that covered parts of the building, adjacent rubble piles, and the roof. We also changed the lighting of the plate building in post to get more drama with our CG asset.

“It was important to the story that the only way forward for our protagonists was through the museum, which is why we had to add giant piles of debris and collapsed buildings around the entrance area. Those were achieved by a mixture of simulating instanced geometry and hand placing individual pieces of concrete, metal girders, rubble, and smaller debris. On top of that, we added overgrowth and added ambient motion to leaves and smaller vines. All the surrounding buildings were CG assets that we augmented using traditional matte painting and compositing techniques.

“Full CG environments included Faneuil Hall, also in Boston, where Ellie looks down at the market buildings and learns about how the Infected are connected through a mycelium layer underground. In this scenario, drone footage of the area was used to extract photogrammetry data that assisted in the modeling process. 

“Having a renderable environment allowed us to find the best camera angle and distance so the audience would understand that we are looking at a group of Infected that are connected through cordyceps and move in unison, like a hive, as the sun travels over them. One of the biggest challenges for that sequence was the amount of detail and scale. If there was too much detail in terms of dressing and overgrowth, for example, the scene became cluttered, and it was hard to make out the characters. 

“We used photographic references to get close to the look of the market buildings in their current state, then demolished the area in a way that is coherent with a bombing as referenced earlier in the series. Adding a wet look to our digital set gave our shots a nice contrast and helped direct the viewers to the infected on the ground. 

“To give the environment a sense of having been inhabited by humans, we added dressing like shop awnings, banners, outdoor dining furniture, lamp posts, water pipes, broken windows, scattered glass and debris, shrubs and grasses, trees, roof tiles, aircon units, and organic overgrowth like ivy. The cordyceps breaking out from windows and bomb craters were used to illustrate the hostility of the environment and the material that connected the Infected, but they also helped as a compositional element due to its distinct orange color.

“The park in which Joel and Ellie come upon the giraffes was also a fully CG asset, apart from the more distant buildings, which were matte painting projections that were stitched together from multiple panoramic images taken by drone in Salt Lake City. We researched the most common trees in Utah and built CG versions of those based on reference. These were planted around the perimeter of the park to establish a boundary and to blend into the existing trees in the matte painting. 

“We gave the inner park a little league baseball field, with the diamond barely visible. An overgrown batting cage as well as weathered bleachers helped to establish that look. In homage to the game, we built the scoreboard to match, and used the same birch trees and placed water puddles of varying scales throughout the park. Further dressing included many different types of grass and flowers, and all kinds of shrubs at different scales. Most of the foliage, like trees, bushes, and grasses had ambient motion on them to add life to the scene. Surrounding the park were roads and walkways, with abandoned vehicles and urban debris serving as dressing.” 

Can you tell us about Wētā’s VFX work for the major battle scene in Episode 5?

Simon Jung: “A substantial amount of VFX work across all disciplines went into the battle scene. In terms of creature work we generated the Bloater, the Child Clicker, hordes of Infected and Clickers, digital doubles of rebels, and a hero level Perry asset for when he gets his head ripped off by the Bloater. Other assets included multiple vehicles, including a highly detailed version of the truck that crashed into the house, as well as a full extension of the cul-de-sac environment including a road and additional suburban homes, trees, fences, and streetlights. 

“We also built a CG version of the house from below which all the Infected emerge and simulated fire on it, so we were able to add more fire and destruction to it as the sequence progressed. The forming of the sinkhole was achieved through simulation also.

“The aim of the battle scene was to convey complete chaos and mayhem, and for us, as the viewer, to witness it through the lens of an almost documentary-style camera. That relentless onslaught of more and more Infected flooding out was used as a backdrop for distinct beats within the battle, like the Bloater emerging and going after Perry, or the Child Clicker climbing into the car that Ellie is hiding in. 

“Maintaining that energy meant that everywhere we look something needed to be happening, people fighting for their lives and being overrun. A lot of the performance was captured in camera but even the 50 plus stuntees in makeup and prosthetic suits couldn’t convey the required scope. We had built a substantial library of assets of Infected, Clickers, and rebels, which we layered in areas of the frame in shots that didn’t have enough action. That could be filling in background action or also really close to camera, like the Infected that get hit by a car, for example. 

“Creating an iconic character like the Bloater was a really exciting task for us. Once it became clear that we had to change the performance from what was initially captured in camera, that also freed up his design and materials. That allowed us to make him look more menacing by adjusting the proportions and giving him a more defined musculature in areas like the legs, arms, and chest. We were also able to give him a more caved-in face and a much sturdier silhouette, which really paid off in the backlighting we ended up seeing him in. 

“Another advantage of having him as a CG character was that we could define different materials more clearly. Like his skin compared to the thick, plate-like cordyceps that cover large areas of his body, serving as armor that is impenetrable to even high-powered rifle bullets. He also has spore pods on him that look like barnacles and have their own material properties again. All of these materials were simulated differently as he moved. The cordyceps plates needed to stay a lot more rigid and slide over each other, while the skin needed to stretch. The fat layer needed to have a believable amount of jiggle to it, while the spore pods were either pinned down in place or slid against underlying tissue, depending on where they were placed on the body.”

Dennis Yoo: “The battle scene had most of our crowd animation and all of our Infected creatures and monsters. We postvis’ed several of the shots in order to quickly find the numbers of crowd we needed with basic staging. This was helpful to do in order to find the animation that needed to be pushed into our final work. 

“The last shot of our work in the sequence is my favorite shot, and is fully CGI, except for the foreground plate the camera pushes past.  It’s the wide shot of the destruction that has unfolded throughout the sequence. We also had a hand in the camera, where we crane up over the scene where the Bloater is center frame, with Infected streaming past him racing toward the city.”

What are some of the tools your team used to create the show’s VFX?

Dennis Yoo: “On the motion side of things, the software we used for the capturing of motion is primarily MotionBuilder, where we stream the motion data live on set to review our takes and performances. The takes get prepped and mapped for our animators by the motion editors using Nuance. This is a process to ensure the motion properly sits onto our characters.

“We also use Nuance to edit our motion, using the software’s unique tools / ability to manipulate a rigid FK [forward kinematics] puppet (solely using joint rotations). The keyframe animators use Maya, where they can also edit the motion using a puppet with animatable controls with a mixture of FK rotations, IK [inverse kinematics] translation controls, and blend shapes, usually for facial motion. The animators also use Maya to keyframe animation (pose-to-pose animation) where we need to create realistic motion without motion capture, usually because the motion is difficult or impossible to capture. “We also use a proprietary real-time renderer that we call Gazebo. We render all our animation presentations with Gazebo, which includes image plates of the shot footage, our CGI characters, base shading, lighting/shadows, bloom, depth of field, screen space ambient occlusion, and FX cards.”