Death Stranding 2: On The Beach is a fancy videogame. You can tell it is fancy because it is 70 dollars. Though technically a third-party title, it is a timed PS5 console exclusive developed on a first-party engine, one of the two AAA releases for the platform this year. It is a single-player game that can take anywhere from 25 to 90 hours, with lots of high-production-value filler. It has an overwhelming number of difficulty modifiers and quality-of-life features. It has meticulous Dualsense haptics and skill trees and base ranks and a satisfying combat model in addition to the traversal, and the stealth, and the crafting. It has the bandwidth for playing bumper cars with demonic motorcycles and a fully built-out in-universe MP3 player. All of these things signal to the player that this is a luxury product—the kind of game with the resources and expertise to be everything to any player.
But you don’t need to know the name of its developer, or even pick up a controller, to know that Death Stranding 2 is a fancy videogame. You can just look at it.
As advancements in tech become more incremental, leaps in resolution comparable to previous generations demand exponentially more labor, time, and raw processing power. Between the goldfish memory of gamers and the dwindling number of blockbuster games released in a given year, it’s easy to lose a sense for the current upper limit of visual exactitude unless you’re playing Call of Duty campaigns. Even then, those are such artistically neutered endeavors that the real possibility space for the medium’s current-gen machines is obscured. This is all well-trodden rhetorical territory, prone to subjectivity and hyperbole. But I speak with total sincerity when I say that Death Stranding 2 is the best-looking videogame I have ever seen.
The game opens with real footage of a desert. The landscape grows increasingly populated by CGI elements, eventually transforming into a full in-engine render. I could not tell you where this switch occurs. When the scene cuts to gameplay, framing Norman Reedus’ Sam Porter Bridges against a white rock mountain range, it’s nigh-impossible to register any image degradation. DS2’s nature maintains photorealistic fidelity but hyperrealistic atmosphere. Plumes of dirt kicked up by Sam’s stumbles rise and dissipate against the clouds. The mountaintops diverge like tesselating palm lines. It is difficult to tell whether a particular route is at an incline or decline until the player is right up next to it. The developers at Kojima Productions leverage their decades of experience in the field to produce spectacle impossible at any other time in history, in any other medium.
This level of awe is maintained throughout the rest of the game. From the objects scattered around Sam’s cozy bunker apartment to the sparking pinwheels of the first encounter with Neil Vana, DS2 approaches its art direction with maximalist intensity. Often, the camera pulls back and a song begins playing, flexing the ludicrous draw distance on the foliage, how every minor hill has been placed with consideration for its mark on the skyline. Buildings and landscapes are evocative of the “real world” the way it looks in one’s memory, where a space’s most distinct attributes consume any mundanity.
This is all, admittedly, garish, but at some point during the intervening six years since the first Death Stranding, its aesthetic (which I have affectionately termed “Chainsmokers kitsch”) has turned from grating to charming. The first game was set in the United States and looked like EDM Iceland. This game is set in Mexico and Australia, but basically looks like Denver. The underlying technical wizardry makes the aesthetic of DS2 a suitable container for its narrative, one that can withstand any goofy or extreme setpieces. Kojima Productions commits hard to its tar-and-chrome post-apocalypse. Like '80s greasers or Hot Topic goths, it feels like the apotheosis of a vestigial cultural moment. There are games with better art direction and similar ingenuity. But like the Gamecube remake of Resident Evil or the original Shadow of the Colossus, I would bet DS2 still makes my jaw drop twenty years down the line. I did not know you could make games at this scale, at the vanguard of modern graphics, where the innovations are integral to the core artistic vision.
This aforementioned digital infrastructure is perhaps so pronounced because of how often I could do nothing but look at the game. I beelined the game in 25 hours. This is fast, especially because I did not skip a single unique cutscene. According to a compilation by Youtuber Shikkaro, nearly 40% of my playthrough was spent with the gamepad limp in my hand. More than any other single component of the game, my main interaction with Death Stranding 2 was merely watching the screen.
This is not unusual for Hideo Kojima, whose games have always featured extended and frequent cutscenes. The meme is that he secretly wants to be a filmmaker. He uses bodyscans of his favorite directors for various characters in his games, directors whose influence on the work in question is often obvious (One character is modeled on George Miller and careening through the desert on a tricruiser is certainly Mad Max-esque). Even so, I think the assertion that “he makes games so he can put movies in them” is inaccurate. Everything shot is framed like it’s coverage (i.e. the secondary footage used by editors to stitch between shots if need be, in a way that maximalizes information in a frame over form). This signals that the exact composition of the shots isn’t what matters. I think Kojima just likes watching things happen in the game without the player’s input. He likes to force a conversation between passive observation and the interactive portions of his work.
The modern AAA cutscene, with its simulated camera and pretensions of realism, is a weird and uncanny creation. Something that often goes unmentioned, even in discussions of “cinematic” videogames, is that cutscenes do not work the same way as movies. They do not draw the viewer’s eyes to the same elements, they do not follow the same pace, they do not use the same principles of editing. An entire performance cannot simply be recreated with total precision in-game; developers must choose what is most important from the reference material. Thus, cutscenes are compelling not just for their story and dialogue or wholly invented assets, but also for the density of information communicated by the motion-capture.
This has loads of cascading effects on the form. For example, games typically avoid having a character who is physically present speak off-screen. One of the pleasures in watching cutscenes is to see how the models move their mouths, or how the animators translated data from optical motion sensors. The assets and rigging take precedence over shot composition. This means cutscenes move slowly. Characters don’t talk over each other, and there is often time in between sentences for the audience to ogle bespoke gestures or facial expressions. That pace puts a greater focus on the performances as aesthetic considerations.
Performance capture also has far more variables than other genres of acting. A character’s body, gestures, and speech can hypothetically come from three different people. Emotions that read well on an actor’s face don’t fit quite right when mapped onto someone else’s skin. A solid line reading might make the mo-cap look awkward. A developer must make sure everything registers within the confines of the game’s fidelity. The inclusion of microexpressions that can’t be efficiently or proportionally depicted in-engine, or a comparative absence of authored animation, can throw an entire exchange off-kilter.
Good motion capture will often look good long after being outpaced by later innovation. Take Naughty Dog’s The Last of Us Part I, both the 2013 and the 2023 incarnations. For the remake, the cutscenes were re-animated to be more accurate to the original mo-cap. Side-by-side comparisons show obvious discrepancies in the amount of detail, though neither is “worse” than the other. The original is still dramatically legible; any signs of age feel irrelevant. Both versions of the scene may incorporate different individuals' movement, each to the same goal; displaying the core of these performances without including anything that might lie beyond the limits of the hardware at the time. The original Last of Us doesn’t look bad because it used less of the mo-cap material; it looks good because it didn’t. It operates with confident clarity of purpose.
The first Death Stranding, on the other hand, has bad cutscenes. They are long and full of dead air. The writing is anemic and the performers seem totally directionless. This is less of a problem for the industry veterans in the cast, who occasionally make deft work of a horrible script. Movie stars like Léa Seydoux, on the other hand, are particularly shafted by Kojima’s approach. Perhaps the greatest and most famous actress on Earth, her approach to acting seemed completely irreconcilable with videogames. So much of her performance as Fragile relies on imperceptible facial twitches bubbling over into iconic catharsis. The PS4 could not technically realize that range; attempts to paper over this only result in a recurring problem where Fragile would bare her gums or raise her eyebrows seemingly at random. Her line readings weren’t bad, but coupled with such a lurching performance, it was downright confusing.
The cutscenes in Death Stranding 2 are weird. Kojima has written a slavish mirror to the plot of the first game, and any novelty in setting has worn off. Every plot beat is foreshadowed plainly, and up until the last few hours, there is just not much going on. This leaves a lot of time for the player to spend idly studying character models.
It is jarring how much more evocative the new animations look, how much more leeway there is to capture the emotions of its actors. Neither Death Stranding game feels like it was written or directed with its limitations in mind, but the sequel takes this miscalculation as an opportunity for experimentation. The game’s exteriors are perfect and timeless, a synthesis of artistic cohesion and technological innovation. Yet the real spectacle of DS2 lies in the ersatz spasms of its characters’ faces.
Norman Reedus benefits immensely from the generational leap. Sam’s head once moved like it was a Memoji of Reedus’s face. Now, Sam’s mournful stoicism is omnipresent; you can watch every grunt reverberate across his face. Reedus has repeatedly failed to demonstrate any dramatic chops in his filmed work, and I would wager there is little more to his performance capture than what can be found in-game. But that digital layer allows for a degree of suggestion that makes his performance more striking than what plenty of more skilled actors could whip up.
In the first game, the cast's standouts were characters who borrowed their likeness from others. Characters like Guillermo del Toro’s Deadman (voiced by Jessi Corti, performed by Justin Leeper) were buoyed by veteran game actor prowess, livening up otherwise dull exchanges. Here, characters like Tarman and Heartman (Miller and Nicholas Winding Refn, respectively) are played floatier than feels correct. They don’t walk like they’ve carried the same face their whole life. Six years ago, you couldn’t see that. You can now.
On the opposite end of the spectrum is Seydoux. Her performance in Death Stranding 2 is remarkable. Lines that would have previously left her model a blank slate are now accompanied by an avalanche of familiar patterns. It’s the best performance capture I’ve ever seen, mainly because she seems completely unmoved by the notion that she should be emoting at a certain intensity. By opting to remain in her more reserved, watchful demeanor, she delivers the first mo-cap performance I have ever seen that does not feel purpose-built to be scanned into a computer. After a quarter-century of playing games which feel like they are trying to obfuscate their own lapses in rendering power, watching Léa Seydoux completely disregard an entire philosophy of digital performance, and seeing that choice supported by the game itself, is shocking.
In one image, I registered her mouth open in furious desperation, her jaw twitching, a tear streaking down her face, and her suit’s magic gloves each gripped in fists, all at once. I don’t know if she cried for the capture session. I don’t know whose hands were the reference for her gloves. But I know that only foolish ambition would lead a developer to include something as minor as the twitch in her jaw when the scene would have been just as clear without it. The story does not demand that level of nuance, and yet the game provides it anyway. It is here that I believed, for the first time, that some feature of a performance had spontaneously occurred within a videogame. For a split second, I was convinced they had not meant to render that, that this had not been chosen consciously but had arisen from somewhere within the reference files like divine intervention.
I am sure Grand Theft Auto VI will have more natural and complicated facial animations, though I doubt there will ever be a passing moment where it feels like anything but the product of a thousand workers over thirteen years. No moment in that game can afford to feel like a miracle. Death Stranding 2, though, has a dozen such moments.
Marionettes appear often in Death Stranding 2: Sam’s new sidekick is a miniature wooden puppet named Dollman; Die-Hardman speaks to the Magellan crew through a talking mannequin called Charlie; Deadman even commandeers Heartman’s body in one scene. Dancing, too, appears repeatedly. Rainy is first seen galavanting to a certain B.J. Thomas song. Die-Hardman performs a joyful act of defiance while dancing to his own version of "B.B.’s Theme," his body glitching between plastic skin and nylon tights. The illusion of a body in motion, and the joy of that illusion being given the capacity for expression, is the metaphorical force of hope in the text of the game.
Troy Baker is the most famous videogame actor in the world. He plays Higgs, the antagonist of both Death Stranding games. He deliberately mixes exaggerated, fluid movements and subtler but isolated flourishes. He seems to have a knack for sensing what will translate to the final product. His performances seem useful.
In the second game, Baker plays Higgs like he’s at a convention cosplaying as himself. It is notable that in the game’s final mission, amidst a shower of chiral kaiju and tar mechs, Kojima chooses to spend several minutes holding on a close-up of Baker’s face. In a way, he is up to the task; Higgs’ final monologue is gripping without even listening to the words. But Higgs is nothing more than a sadistic glam-rock samurai, and Baker, consummate professional, rests on safe tics and predictable choices. There is a sense that the game has trained its eye on both a performance and character that cannot withstand its gaze.
And yet, after a minute or two, the player begins to notice something more tangible about the scene. Higg’s body looks real and permeable. You can see his tendons. Light flashes through his skin, like when you hold a flashlight to your cheek and see all the veins. Unlike the vistas of the intro cinematic, nobody could mistake this part of the game for “real life.” Yet it is this moment that offers me the same buzz in my heart as looking at a real person.
This is why the Polaroid minigame is such a highlight. Unlike the game’s proper, expansive photo mode, the player is limited to a few slight zooms and a few looping scenes. Yet the way Sam’s shipmates pose and shift their weight, the way their expressions falter and re-enliven every few seconds, is a level of intricacy in interactive animation that I would have previously deemed impossible. I didn’t take my first photo for ten minutes. I just wanted to watch them all. They weren’t characters I cared about; they are, still, basically strangers. But they felt like people. When Elle Fanning appears at the very end, holding the photo I took, it bowled me over. I remembered every inch of it.
This thrill is inconsistent. Fanning’s performance as Tomorrow is baffling. Her face is constantly shifting, yet every time it feels like a mistranslation of whatever the sensors were registering. Her body only carries weight when it’s clear a stunt double is doing the mo-cap. Dollman is the most annoying guy I’ve ever encountered in a video game. I wish there was an option to mute only him. What’s worse, the choice to have him move at a lower FPS to mimic stop-motion-animation is a useless stylistic flourish.
Worst of all are the times where the aforementioned pleasures feel futile. Luca Marinelli plays Neil Vana, a somber redux of Mads Mikkelsen's time-hopping vet in the first game. Marinelli’s performance as the husk of a man is devastating, an illustrative counterpoint to Mads Mikkelsen’s dissociated swagger; Death Stranding 2 is capable of quieter moments than the first game, is invested in what flesh pulses behind its mythos. But this subplot is – barring a few asinine implications – irrelevant, a pure indulgence as Kojima reconfigures his latest game into a self-reflexive vortex.
I’ve been holding off on saying whether I like the game, because I don’t. Gone are the days when Kojima could tell a politically trenchant story with experimental mechanics. DS2 is a frictionless toybox of total player empowerment. It improves on the first game in basically every way, and yet it somehow makes them both feel redundant. The game’s politics only appear insofar as they can be an allegory for his own preoccupations with legacy, the future, children, and grief. So Higgs makes Sam do the worst QTE from Metal Gear Solid 4, the game declares AI an anti-labor Trojan horse of imperialism before promptly stomping this subject out with the power of community, and Tarman says the phrase “phantom pain.” Kojima’s signature cutscenes could once be sustained on a kind of drone-y abstraction. But now, with such vivid performances, his severe limitations as a writer cannot be ignored. All that’s left is his latent misogyny and a few muddled rehashes of his earlier work’s political assertions. Face it, “unseen plot-critical brain-dead pregnant Mexican women” is exactly the kind of thing the dude behind Metal Gear Solid 2 would make up if he were washed.
So, what’s the point to any of this? The game looks amazing and plays effortlessly, which is an accomplishment worth celebrating but not worth writing about. Kojima has never been more out-to-sea as an artist. I am left picking through scattered fragments, hoping for another glimpse of an impossible game.
My favorite moment in Death Stranding 2 involves both of these things. It comes when Daichi Miura plays in Sam’s room aboard the Magellan. Dollman, who begins bouncing to the beat before the player even selects the prompt to talk to him, will whisk Sam away to watch a dance performance set to the entire song. Daichi Miura himself shows up to perform and lipsync, along with five extra Dollmans, each of whom have unique choreography. The player can toggle different lighting setups, and spin the camera around to spot an out-of-focus Sam bopping along in front of the miniature stage. This is entirely missable, with only a single line of ambient dialogue hinting at its existence. It has no narrative or mechanical purpose. It is one of the few moments where Death Stranding 2, as I will remember it, actually exists—as a game whose characters begin to move on their own volition under the guidance of their creators, like a baby learning how to walk.
***
Sam Bodrojan is a freelance writer based out of Chicago. She has written for MUBI Notebook, The Kenyon Review, LARB, Reverse Shot, Hyperallergic, Filmmaker Magazine, and elsewhere. She runs a weekly newsletter, cc: helmet girl. She can be found on X/Twitter @sbodrojan, and at helmetgirl.bsky.social