Magrunner: Dark Pulse

A 3D action-puzzle on Unreal at Frogwares — my first Unreal project and my first time owning a GDD end-to-end.

After the hidden-object work this was a different direction for both me and the studio: a single-player puzzle game built on UDK. I came onto preproduction and stayed through to ship.

What I worked on

• Full game design — pre-production concept, systems, mechanics, every level’s plan, final GDD.
• Forty levels — prototyped each one in Kismet, then led production through art and scripting.
• Multiplayer prototype — covered below.

First time on Unreal

Magrunner was my first Unreal Engine project. We worked on UDK and I learned Kismet — Unreal’s visual scripting language at the time. I loved it almost immediately: it let me build logic on a level by myself, without queuing a programmer for every prototype. A lot of the game’s character came from being able to try things cheaply.

How the mechanics emerged

In preproduction I was paired with a programmer. He built the engine-level magnetic-force mechanic — two magnetised objects pull or push each other. I built on top: spending time with that one piece, trying combinations, looking for gameplay hiding inside it.

Most of the game’s main features came out of that period — including Newton, the robotic dog that lets you stick a magnet to any wall. Newton wasn’t in the concept document; he emerged from a “what if the magnet wasn’t tied to a fixed surface?” experiment.

The polarity dilemma

The first really difficult professional dilemma I hit was on this game.

Real physics: same poles repel, opposites attract. We started there. But during prototyping I noticed playtesting was much more interesting — and dramatically easier — when we flipped it: same colours attract, opposites repel.

Why does the inverted rule feel right? Because “similarity attracts” is the broader human intuition. We group ourselves by similarity in everyday life; we describe friendships as “we’re alike, we clicked”. Players reach for that model without thinking, and the gameplay clicks with it.

The cost is real, though: every viewer of the trailer sees the magnets and asks “hey, your physics are wrong — didn’t you take school?” It’s the kind of feedback you can’t fully argue with in a thirty-second clip.

But the design space is also bigger. With our (inverted) rules, you can stack many same-colour objects into one magnetic cluster. With strict real physics, three same-colour objects start fighting each other — you can pair, but you can’t build. That ceiling on puzzle complexity would have changed the whole game.

We shipped with the inverted physics. The “wrong physics” complaint stayed in reviews; the puzzle range came from the choice.

The multiplayer prototype

On top of the single-player game we built a multiplayer prototype. Quake-style movement — fast, vertical, void-and-gravity arenas — but no weapons. Each player had a magnetic field they could shape and direct. Movement used magnetic jumps in place of rocket jumps. PvP damage was indirect: pull explosions toward your opponent, drag them off platforms into the void.

It was fun. It would probably have shipped as DLC if Magrunner had broken out commercially. It didn’t, so the prototype stayed a prototype.

What I took from it

First Unreal project, first end-to-end GDD ownership, first design call where the cleaner rule (real physics) lost to the better-feeling one (similarity attracts). I still bring up the polarity choice when teaching — it’s the simplest example I have of correctness on paper losing to fit-for-purpose, and the importance of being able to defend that choice when it gets pushback.