DEV Community

dev game
dev game

Posted on

Why Split Fiction Bet on Unreal Engine 5: The Boring Co-op Tech Behind a 7-Million-Copy Split-Screen Hit

Split Fiction did the unfashionable thing: it shipped a co-op-only, no-single-player game on rented tech and sold over seven million copies.

Developed by Hazelight Studios and published by Electronic Arts, Split Fiction runs on Unreal Engine 5 for PlayStation 5, Windows and Xbox Series X/S (March 6, 2025) and Nintendo Switch 2 (June 5, 2025). It is multiplayer-only, designed for split-screen cooperative play, and follows sci-fi author Mio Hudson and fantasy author Zoe Foster trapped inside their own stories. Hazelight describes it as a sci-fi and fantasy adventure built around hoverboards, gravity bikes, platforming, stealth and puzzles that change every level.

That engine choice looks boring until you compare it to Hazelight's own history. It Takes Two was built on Unreal Engine 4, released March 26, 2021 for PS4, PS5, Windows, Xbox One and Xbox Series X/S, and sold over 30 million units by April 2026 after winning Game of the Year at The Game Awards 2021. Same director (Josef Fares), same EA Originals-style publishing, same split-screen co-op-only rule, same trick of giving the two players different abilities every level. Split Fiction is not a new formula. It is the same formula moved from UE4 to UE5.

So why rent UE5 instead of building something leaner? The short answer: Hazelight used UE5 for iteration speed and lighting consistency, not for photorealism.

What UE5 actually does for this game

Unreal Engine 5, developed by Epic Games, was revealed in May 2020 and formally launched April 5, 2022. Its headline systems are Nanite, a virtualized geometry system that auto-manages level of detail, and Lumen, a dynamic global illumination and reflections system. Epic famously used Fortnite as a live testbed, updating it to UE5 in December 2021.

Split Fiction does not need Nanite forests or open-world streaming. Its levels are linear, theatrical, and disposable by design: one sci-fi chase, one fantasy dragon ride, then throw the mechanic away. That is exactly where UE5 helps a 60-80 person team:

  1. One lighting rig, two genres. Lumen lets a small lighting team keep sci-fi neon and fantasy daylight coherent without baking lightmaps for every one-off set. Mio's cold labs and Zoe's warm forests share the same global illumination logic with different palettes.

  2. Redesigned editor tooling. UE5 shipped with new animation and modeling tools in the editor. For a game that invents a new verb every 20 minutes — gravity bikes one chapter, sandsharks the next — in-editor iteration beats raw runtime performance.

  3. Current-console baseline. UE4 carried It Takes Two across last-gen (PS4/Xbox One) and current-gen. Split Fiction dropped PS4/Xbox One entirely and targeted PS5, Xbox Series X/S, Windows and later Switch 2. UE5's PS5/Xbox Series optimization, proven on Fortnite, made that cut cleaner than backporting a custom engine.

This is not a Nanite showcase like an open-world epic. It is a content-factory use of UE5: spend the engine budget on authoring throughput, not on-screen density.

Why not stay on UE4, Godot, or custom tech?

Staying on UE4 would have been defensible — It Takes Two already proved 30 million sales on it. Hazelight moved anyway for three practical reasons:

  • Hiring and outsourcing. UE5 C++ and Blueprint talent is easier to hire than UE4-only veterans in 2024-2025, and outsourcing houses had already retooled for Nanite/Lumen pipelines.
  • Switch 2 path. A simultaneous Switch 2 version on June 5, 2025 is easier to negotiate when you are on Epic's actively supported branch with documented scalability settings, rather than a frozen UE4 fork.
  • Sequels cost. Hazelight's design is mechanics-per-level. A custom engine would need to be rewritten every chapter. Renting UE5 turns that rewrite cost into Epic's R&D.

Godot or a custom C++ stack would cut royalties but add hiring friction and console-port risk for a studio whose advantage is level design, not engine programming.

My take: three contrarian ideas you will not find in the reviews

Most coverage says Split Fiction succeeded despite requiring two players. I think it succeeded because it requires two players, and UE5 is the alibi that makes that viable. These are opinions, not Hazelight statements:

1. Split-screen is a server-bill hack disguised as nostalgia.

Online-only co-op needs matchmaking, anti-cheat, live ops and regional servers. Split Fiction renders two views locally and uses online only to mirror what local already does. That halves netcode complexity: if couch co-op works, online is just remote couch. UE5's split-screen viewport support is old, boring and reliable — exactly what you want when your real innovation is level variety, not networking. My label for this: render twice, operate never.

2. Sci-fi vs. fantasy is a shader-budget split, not just story.

Mio (rational, sci-fi) and Zoe (heart-first, fantasy) look like narrative opposites. Technically they are a production split: hard-surface metal and emissive neon on one side, foliage and soft magic on the other. One Lumen setup can serve both if you palette-swap aggressively. That lets two art pods work in parallel without blocking each other. The story justification — two authors trapped in one machine — is also a production justification for two art tracks.

3. No single-player plus Friend's Pass is an anti-funnel filter.

Conventional wisdom says forcing co-op shrinks your market. I think it filters for finishers. A solo buyer who quits after two hours generates a refund risk and a negative review. A pair who schedule sessions, teach each other, and laugh at failed jumps generates completion, clips and word-of-mouth. It Takes Two's 30 million and Split Fiction's 7 million by April 2026 suggest pairs finish more and evangelize more than solo tourists. Banning single-player is not arrogance; it is churn control.

Where others praise variety, I praise refusal: no battle pass, no live service, no single-player compromise. UE5 is what lets a mid-size team afford that refusal.

What indie devs can actually steal

You do not need UE5 to copy Hazelight. Copy the constraints:

  • Design asymmetric verbs, not symmetric power. Give player A rewind and player B clone, or gravity vs. stealth, so progress requires talking. Symmetric shooting just needs aiming; asymmetric tools need communication.
  • Throw mechanics away. One chapter, one verb, then discard. Players remember the monkey dance and the evil kitties because they only happened once. Reuse kills the joke.
  • Fail soft. It Takes Two respawns quickly and lets the downed player mash back faster in bosses. Split Fiction inherits that mercy. Co-op dies when one player watches loading screens.
  • Build for two views on day one. Even if you ship online-only, keep a local split-screen mode during prototyping. If a puzzle is unreadable when you can see both screens, it will be unreadable online.
  • Rent boring tech for weird content. If your novelty is level ideas, do not also invent a renderer. Use Unity, UE5 or Godot and spend your novelty budget on verbs.

If you want to practice the co-op muscle in a browser

Split Fiction needs a committed partner and a console or PC install. If you want to test whether you actually enjoy talking through jumps before you buy, these instant-play browser games train the same skills with no download:

For true platform teamwork, try Super Robo - Adventure. It is built as a true 2-player platformer where levels benefit from cooperation — bouncing on a partner to reach ledges, splitting roles against enemy groups, and hitting switches together. That bounce-and-coordinate loop is the low-fi version of Mio-and-Zoe complementary abilities.

For a five-minute couch warm-up, try Splatmans. This 2-player stickman arena is about painting territory before the timer ends, versus AI or side-by-side with a friend. Rounds are short, routes matter more than aim, and repainting enemy color teaches the same read-your-partner habit that Split Fiction's versus minigames demand.

For pipeline coordination under pressure, try Hospital Hustle. You direct patients to the right machines, repair before breakdowns cascade, and keep flow moving in both 1-player and 2-player modes. It feels like a Split Fiction escort chapter stripped to its timing core: keep moving, upgrade high-traffic nodes first, and stay calm when the waiting room fills.

None of these replace Split Fiction's campaign, but they answer the core question fast: do you like solving with another brain in real time? If yes, Hazelight's bet will make sense to you.

Bottom line

Split Fiction bet on Unreal Engine 5 not to look next-gen, but to author faster: one Lumen rig for two genres, editor tooling for disposable verbs, and a supported path to PS5, Xbox Series X/S, Windows and Switch 2. The brave part is not the engine. The brave part is keeping co-op-only and no live service in 2025 — and using boring rented tech to afford that creative stubbornness.

That is the lesson for small teams: rent the engine, own the refusal.

Top comments (0)