Back in 2024, Andreas Giesen stepped out of the vault to walk us through RISE’s work on the first season of Fallout. Now he’s returning from the wasteland to talk about the new season—and the fresh challenges lurking beyond the horizon.
What was your feeling to be back in the Wastelands?
It was an incredible feeling and a huge honor. Season 1 was already a career favorite, especially since I’m a massive Fallout fan myself. Coming back meant we could build on the great trust we had already established with Jay Worth and Andrea Knoll. Everyone at RISE was super excited to return to the Wasteland, and we were lucky to have nearly the exact same team back—including my partner in crime, our VFX Producer Annelie Dangel.
Season 2 significantly expands the scale of the series, at what stage did you realize that the aerial and destruction sequences would surpass what you achieved in Season 1, and how did that impact your initial planning?
We realized the massive jump in scale quite early on. Because we were working with a similar team size to Season 1, we knew right away that we had to divide and conquer. We set up specialized teams under dedicated leads to tackle these very different and challenging sequences independently. To pull off that expanded scope, we had to lean heavily into procedural approaches and really maximize everything we learned from the first season. Knowing exactly the grounded, retro-futuristic visual style that Jay Worth, Jonathan Nolan and the other producers were looking for was a huge advantage—it allowed us to hit the ground running and focus entirely on execution rather than finding the look.
The Caswennan dirigible feels more detailed and imposing this season, how did you redesign and upgrade the asset, and what were the key technical or artistic improvements compared to Season 1?
We completely reworked the Caswennan this season so it would hold up in extreme close-ups, especially for the complex docking sequence. We also had scenes on the flight deck that involved virtual production. It was actually a funny full-circle moment—we received an Unreal asset to combine with the LIDAR scan of the partial practical set, and that Unreal asset was originally based on our dirigible from Season 1!
Ultimately, we had to bring all those pieces together and inject a massive amount of detail to reach cinematic quality from every angle, including a huge establishing shot from above. Our asset team did a brilliant job under the supervision of CG Supervisor Jenny Leupold and Asset Lead Joshua Koilpillai.
From a pipeline perspective, USD was an absolute lifesaver here. Because the Caswennan is basically a massive, high-res flying set, USD enabled multiple artists to work on individual parts in parallel, with everything updating flawlessly in the background.
You also introduced faction variations of the Caswennan, how did you differentiate them visually while staying consistent with the established retro-futuristic aesthetic of the show?
We received fantastic concepts from production that outlined the specific logos and color palettes for the different groups. The core engineering and that classic retro-futuristic look of the dirigible remained the same across the board, so the real differentiation came down to the granular details. We focused on swapping out the specific insignias, adjusting the wear and tear on the paint jobs, and adding dynamic elements like custom faction flags. It is a subtle but highly effective way to expand the world, and die-hard fans of the games will immediately recognize those emblems and exactly what they represent.
The vertibird sequences are particularly dynamic, especially Xander Harkness’ arrival at Area 51, how did you approach the choreography, previs, and final animation to achieve both clarity and scale?
We always prefer to ground our animation in reality, so we started with a precise LIDAR scan of the practical hangar set. Having that accurate spatial data was absolutely crucial to make sure the scale, distance, and flight speed of the Vertibird felt completely believable. Our animation team started with a rough blocking pass just to dial in the timing and spatial relationship. Once that was approved, we pushed into final animation, adding all the mechanical secondaries like the intricate rotor rotations and weight shifts.
What made this sequence really fun, though, was that the flight itself had to be a character performance. The animation needed to reflect Xander’s personality—he is a genuinely fantastic pilot, and he likes to show off. We made sure his arrival, the flight path, and the way he handles the Vertibird had that distinct swagger and precision to tell you exactly who is flying it. Nailing that specific attitude took quite a few iterations until everyone was completely happy with the motion, but seeing the final result, it was totally worth the effort.
The vertibird/Caswennan docking sequence is quite complex, can you break down the technical challenges of integrating large moving CG assets within a believable aerial environment?
For this sequence, our first step was going straight back to the source material. We gathered a lot of reference from the game for both the Vertibird and the Caswennan’s lifting arm because we wanted the mechanics to be as authentic to the lore as possible. Of course, the in-game models are built for gameplay and can be a bit simplified, so we had to go in and engineer a lot of practical cinematic details—like heavy hydraulics and complex mechanical joints—so the arm would hold up on screen.
The biggest technical challenge, however, was conveying the sheer mass of these assets in the animation. The Vertibird is a massive, heavy vehicle, so it absolutely cannot move abruptly. We had to find a really fine balance: giving it that heavy, lumbering momentum while also adding procedural micro-vibrations to make it feel believably buffeted by the air currents. The trickiest part was nailing that exact moment of transition—the specific second the mechanical arm connects, takes over the full weight of the ship, and the rotors finally power down. Getting that physical transfer of weight to look completely natural was crucial to selling the whole sequence. As a final touch, we simulated some cables dangling from the arm. It’s a subtle thing, but having that dynamic secondary motion really brought the mechanism to life and made the whole shot visually much more interesting.
The pre-bombing establishing shot of Shady Sands required a pristine version of the city, what references and world-building considerations guided the creation of this environment before its destruction?
Our main starting point was actually the fantastic practical sets built for the scenes taking place inside the city. Production provided us with live-action plates and extensive LIDAR scans, which were incredibly helpful for locking in the exact scale. We also had a bit of a head start, having built a portion of Shady Sands for a dream sequence back in Season 1, which we upgraded to perfectly match the new plates.
From there, we had to expand the world significantly to match the specific mid-century, retro-futuristic architectural style. Our Layout Lead, Karin Wolf, and her team did a phenomenal job here. They handled the layout for the non-destroyed city parts in Maya, and then we brought everything into Houdini to scatter all the granular details and surrounding vegetation—especially those endless, lush cornfields in front of the city. Our USD pipeline was a huge advantage here. It allowed us to flawlessly mix and match the layouts between Maya and Houdini without any friction, meaning we could rely on the best tools for each specific task.
In contrast, the procedurally destroyed Los Angeles surrounding the city conveys massive devastation, what pipeline or simulation tools did you rely on to generate destruction at that scale while keeping art direction under control?
To ground the devastation in reality, we based the entire surrounding environment on actual map data to stick as closely as possible to the real topography of Los Angeles. We started with extremely simple geometry—just basic cubes representing the buildings and flat planes for the roads.
From there, we brought that data into Houdini. Our incredibly talented TD, Kilian Baur, built a highly complex procedural tool specifically for this challenge. It took that super simple block-in geometry and automatically converted it into highly detailed, destroyed buildings. The tool procedurally generated collapsed roofs, damaged walls, ripped-off fire escapes, and broken-down AC units. The brilliant part was that it remained completely art-directable; we could essentially ‘paint’ the specific degree of destruction across different zones of the city to control the composition.
Our USD pipeline helped immensely here. It allowed us to start blocking the shots and even dialing in the lighting using just that initial simple geometry. Then, district by district, the high-res destroyed assets would seamlessly update in the background until the whole city came together.
Finally, to really sell the messiness and chaotic busyness of a post-apocalyptic city, we scattered a massive amount of secondary detail on top of that generated geometry. We layered in endless piles of rubble, broken streetlamps, road blockers, rusted-out cars, and debris to give the environment that tactile, grounded Fallout patina. While the procedurally generated destruction geometry was quite heavy on its own, all of these scattered elements were instanced objects, which was absolutely crucial to keep a scene of this massive scale renderable.”
The Shady Sands atomic explosion is a pivotal moment — how did you balance physical realism with the stylized tone of Fallout when designing the blast, shockwave, and aftermath?
For the Shady Sands explosion, we stuck closely to the visual style of the atomic blast we created for Season 1, but we enhanced it with significantly more detail. To upscale the actual volumetric simulation, we utilized some machine learning tools within Houdini.
The timing of the shockwave and all the secondary effects—like the dust kickup, the destruction, and the bending vegetation—was crucial. If the shockwave traveled too fast, we would lose the feeling of massive scale in the scene. However, we also had to work within the strict constraints of the shot length and hit specific story beats. To find that fine balance, we relied on simple representations of the effects to lock in the timing before committing to the heavy, complex simulation work.
The vegetation closer to the camera was a particular challenge. Our FX Supervisor Jens Martensson and FX TD Luca Schippmann did a great job here. They built a KineFX-based setup for the more complex vegetation, which allowed for very fast iterations. For the simpler elements, like the grass, we used a more straightforward approach of just rotating the instances. As a final layer, we added destruction to the barn and replaced the practical cars with CG versions to get proper interaction with the blast. One small but highly important detail we included was the back-and-forth movement created by the shockwave—a specific physical reaction we referenced heavily from archival atomic test videos.
The reveal of Area 51 combines vast exterior builds, dune environments, and extensive set extensions, how did you manage scale continuity between practical sets, digital environments, and large-scale sand simulations?
The sequence was firmly grounded in the really cool plates production shot out in the actual desert. From there, we built out additional dunes and the entire Area 51 complex, specifically focusing on the two main hangars. Because we were working directly from LIDAR scans of the location, we could guarantee a completely seamless transition between the practical set pieces and our digital extensions.
The challenge—and also a very interesting part of the work—was managing the different reveal states. We had to track and art-direct exactly how much of Area 51 was still covered in sand leading up to the actual reveal. This required highly detailed sand simulations. To achieve the necessary realism and scale, we blended different simulation passes together, combining more billowy, volumetric dust with actual sand particles in the air, and then the contact sand swirling on the ground and physically blowing off the geometry.
To bring it all together, our Compositing Supervisor, Michael Lankes, and his team did a great job taking those roughly 12 different FX layers and performing the final balancing. By utilizing deep compositing workflows in Nuke, they were able to seamlessly integrate all that complex simulation data into the live-action plates.
The dramatic dirigible crash and explosion in the desert is one of the season’s visual highlights — how did you stage the sequence in terms of asset destruction, FX simulations, and editorial rhythm?
This is actually one of my favorite shots of the season. It is just so epic seeing the dirigible come down. For reference, we looked heavily at historical footage of the Hindenburg crashing and burning. To pull this off, we had to make sure the dirigible asset was built destruction-ready right from the start. Before initiating the actual simulations, we blocked the sequence using simple placeholders to define the timing and ensure we maintained the massive scale of the airship.
The simulation process itself was heavily layered. First was the initial crash into the dune, which involved computing plastic deformation for the metal bending, alongside the different sand layer impacts. Then, the pyro simulations for the explosions triggered a secondary RBD simulation so more parts of the hull could be physically ripped off. Finally, those debris pieces were fed into yet another pyro simulation to attach trailing flames and smoke, as well as secondary sand kicks when they hit the ground.
To keep everything manageable, all of these sims were clustered to ensure fast iterations and turnaround times. The sheer amount of FX layers for this shot was even higher than the Area 51 sand reveal, and our Compositing Lead, Kader Bagli, did an incredible job balancing all of them perfectly.
The cold fusion and diode glow effects add a distinctive sci-fi signature, can you elaborate on the look development process behind these energy-based effects and how they were integrated into live-action plates?
For the cold fusion and diode glow, we were actually able to reuse the core effect we established and built for Season 1. However, because we had to add it to so many more shots this season, we needed to create a highly efficient setup across Houdini and Nuke to keep the workload manageable.
Our Comp Lead, Jannik Walzer, really pushed Nuke to the absolute limit for these effects. Integrating these energy signatures seamlessly into the live-action plates required a heavy amount of rotomation and rotoscoping work. The biggest factor in making it feel grounded was the interactive light. Because that distinctive blue glow naturally affects everything around it, we had to do a lot of precise relighting in compositing, paying special attention to adding accurate reflections across any shiny metal surfaces in the practical sets.
RISE Berlin and Stuttgart both delivered work under your supervision — how did you structure collaboration between the two teams, and did your experience on Season 1 allow you to refine your workflow or creative approach this time around?
We had very strong teams in both Berlin and Stuttgart, each with their own dedicated leads. To really push the team feeling and ensure everyone was on the same page, we held one joint VFX dailies session for both offices. Because of that, it never really felt like we were working across two separate locations—we operated completely as one unified team.
From a technical standpoint, the collaboration is incredibly seamless because both sites work with the exact same tools and pipeline. We rely heavily on our in-house production management tool, risebase 3, which helps immensely when managing a project efficiently across different facilities. But last and certainly not least, making that dual-location structure work comes down to our great production group. Having Annelie Dangel as our VFX Producer, Scott Kunysz as Production Manager, and the whole production and coordination team keeping everything on track was absolutely vital to our success.
How long have you worked on this show?
We had our core, larger team on it for about six months, starting in May 2025 and running right through to the end of October. But the actual timeline was a bit longer than that. We started earlier with just a handful of artists tackling some previs and lookdev to lock in the workflows and the visual style. After our main delivery in October, we kept a small crew on to handle a few final straggler shots and late turnovers, wrapping everything up by mid-November.
What’s the VFX shots count?
In total, we delivered 244 shots. Our work was spread across seven different episodes, with the vast majority of our heavy lifting concentrated in episodes 2 and 3.
What is your next project?
I am currently working on something very exciting, but unfortunately, it is strictly under wraps right now, so I can’t speak about it just yet. But I am definitely looking forward to sharing it when the time comes!
A big thanks for your time.
// VFX BREAKDOWN
// TRAILERS
WANT TO KNOW MORE?
RISE: Dedicated page about Fallout Season 2 on RISE website.
© Vincent Frei – The Art of VFX – 2026
