Legend of Grimrock 2

The first Legend of Grimrock was a stunt that worked. In 2012, the Finnish studio Almost Human resurrected a design language that had been embalmed since the early nineties — the tile-by-tile, party-of-four, real-time dungeon crawl of Dungeon Master and Eye of the Beholder — and sold it to a market that supposedly wanted open worlds and cinematic set pieces. The stunt became a business.

Legend of Grimrock 2, released on October 15, 2014 and still the studio’s calling card, is what happens when that business gets ambitious. The original was a single tower, descending, linear, claustrophobic by design. The sequel throws you onto the Isle of Nex — a shipwrecked party of four convicts, no exposition dump, no tutorial hand-holding — and then does something the genre almost never does: it opens up. Beaches feed into forests, forests into swamps, swamps into a castle, a pyramid, a mine, a sunken temple. The island is interconnected and largely non-linear. You can wander into a zone that will atomize a low-level party, retreat, and come back later with better boots and a grudge.

This is not a nostalgia product wearing modern shaders, though the shaders are there. It is a genuine expansion of a formula that had been sitting untouched for two decades, made by people who clearly played the originals obsessively enough to know exactly which parts deserved preservation and which were just 1987 hardware limitations calcified into “tradition.”

The game found its audience and kept it — helped considerably by a shipped Dungeon Editor and Workshop integration that turned a finite campaign into an open-ended platform. Positioned against the RPG mainstream, it’s a deliberate counter-argument: no quest log, no waypoint, no map marker telling you where the fun is.

• A non-linear island, not a tower: Forests, beaches, swamps, castles, pyramids and underground mines, most of them explorable in whatever order you can survive. Progression is gated by your knowledge and your gear, not by a plot flag.
• Grid-based real-time combat: Movement and attacks resolve square by square. Positioning, formation and attack-cooldown timing matter more than raw stats. A large bestiary, each entry with its own attack cadence you’re expected to learn by dying to it.
• Deep party construction: Four characters drawn from multiple races and classes, with skills, traits and equipment loadouts that meaningfully change how a fight plays. A Minotaur Barbarian and an Insectoid Alchemist are not the same game.
• Rune-drawing magic: Spells are cast by tracing rune combinations on a grid in real time. It’s a manual-dexterity tax on every fireball, and it is one of the best mechanical ideas in the game.
• Alchemy and crafting: Herbs gathered in the world become potions and bombs. The Alchemist class turns foraging into a genuine resource economy rather than inventory clutter.
• Secrets everywhere: Pressure plates, hidden switches, illusory walls, trapped vaults. Secret-hunting isn’t a side activity here; it’s roughly half the design.
• Full Dungeon Editor + Steam Workshop: The tools the developers used, shipped to you, with a distribution pipe attached. Plus achievements and Steam Cloud saves.

The Gameplay Loop

The loop is elegantly stupid in the best way: walk into a room, get hurt by it, understand why, come back. Almost Human refuses to explain anything. There’s no journal, no objective tracker, no NPC standing in a doorway describing the mechanic you’re about to encounter. What the game gives you instead is an automap you can annotate yourself, and the strong implication that you should.

That annotation feature is the design thesis in miniature. The game assumes you are taking notes — about the pressure plate you can’t reach, the four-symbol sequence carved into a wall three zones back, the locked gate that needs a key you haven’t found. Most modern RPGs outsource memory to the UI. This one hands it back to you and treats that as the reward. When a puzzle finally clicks long after you first saw it, the satisfaction is real, and it’s real precisely because nothing in the software helped you get there.

The cost is equally real. Some of these puzzles cross the line from obscure to obnoxious. There’s a difference between a puzzle that withholds its solution and one that withholds the fact that it is a puzzle, and Grimrock 2 lands on the wrong side of that line more than once. The community wiki exists for a reason, and plenty of players will end up consulting it — which is a design failure, however much the hardcore crowd frames it as a badge of honor.

The Combat Waltz

Grid-based real-time combat produces a very specific movement pattern that players have nicknamed the Grimrock waltz: attack, sidestep, let the enemy’s swing whiff into empty air, sidestep back, attack again. Learn it and you can beat things far above your weight class. It’s positional chess played at the speed of a rhythm game.

It is also, by the mid-game, the answer to a great many problems. This is the sequel’s most legitimate criticism: the tactical vocabulary of combat is narrower than the puzzle design’s. Almost Human clearly saw this coming and countered with monster variety — enemies that close distance faster than you can circle, ranged attackers that punish sidestepping, and packs that pincer you in open outdoor spaces where the tile grid gives you nowhere to dance. Those encounters are the best fights in the game. The rest are choreography you’ve already memorized.

Difficulty spikes compound it. The open structure means you can wander into a region calibrated for a party twice your level with no warning sign, and the game’s response to that is a lethal one. Veterans read this as honest. Newcomers read it as arbitrary. Both readings are defensible.

Magic, Alchemy and the Build Problem

The rune system is the standout. Casting a spell means drawing its glyph pattern in real time while something with claws closes on your front rank. It converts spellcasting from a hotbar click into a physical act with failure states, and it makes a Wizard feel fundamentally different to pilot than a Fighter — not just numerically different. A decade on, almost nobody has copied it, which is a shame.

Alchemy works nearly as well. Herbs you pick up in the swamp become healing potions and thrown bombs, and because the world’s resources are finite, every brew is a small strategic decision. It’s the rare crafting system that doesn’t feel like busywork bolted on to justify a loot table.

Character progression is where the design gets muddy. Skill descriptions are vague about actual numerical impact, some class-skill combinations are quietly far stronger than others, and a bad build won’t reveal itself until deep into the campaign — with no respec to fix it. In a game this punishing, opaque progression isn’t difficulty — it’s an information problem, and it’s the one area where the retro fidelity should have been sacrificed for clarity.

Interface

The UI is spare and it works. Drag-and-drop inventory with real spatial constraints, a right-hand weapon panel per character, the annotatable automap. Everything is one click away. There’s no menu diving, no submenu nesting, minimal onboarding friction on the mechanical side — all the friction is deliberate and lives in the world design instead. That’s the correct place to put it.

The custom engine is the unsung hero. Almost Human built their own tech, and it produces an island that looks considerably better than a small studio’s output has any right to. A 3D game of this scope and scale is light on hardware by design, and the engine’s lean footprint means no bloated install and quick transitions between areas.

Aesthetically it’s a study in restrained mood. The dungeon interiors use hard directional light and deep shadow to make torch radius genuinely tactical — you frequently cannot see what’s hitting you. The outdoor zones, new to the sequel, are the bigger risk and mostly pay off: fog-heavy swamps, rain-slicked beaches, a pyramid interior that shifts the palette hard enough to feel like a different game. Physics on thrown objects and rolling boulders is simple but tactile, and it’s wired into puzzle solutions rather than being decoration.

Sound design deserves specific credit. Grimrock 2 is quiet — long stretches of ambient drip and wind — which makes the audio cue of something skittering in an adjacent corridor land hard. The audio is doing horror-game work in an RPG’s body, and it’s the single biggest contributor to the atmosphere everyone praises.

Steam is effectively the whole story, and the platform features here aren’t checkbox filler.

The Dungeon Editor with Workshop integration is the reason this game still has a player base. Almost Human shipped their actual authoring tools, and the community built full-length campaigns with them — some of which are, by consensus, harder and more inventive than the base game. If you finish the island and want more, there’s a decade of user content waiting, one click from the store page. That converts a finite purchase into something closer to open-ended.

Steam Cloud matters more than usual for a game where a single save file represents hours of accumulated map annotations. The achievement list maps onto the secret-hunting loop reasonably well, though a few are the tedious “find every one of X” variety.

The control story is mouse-and-keyboard, full stop. Movement on WASD with Q/E strafing, everything else driven by the cursor — inventory drag, weapon panels, and critically, the rune-drawing for spells. There is no meaningful gamepad experience here; the rune grid alone would be miserable on a stick. This is a desk game, played sitting up, probably with a browser tab or a paper notebook open beside it. Treat that as a system requirement.

Grimrock 2 is the rare sequel that correctly identified what its predecessor was hiding. The first game’s linearity read as focus; here, with the tower replaced by an island you can attack from multiple angles, you see how much the format had been holding back. The exploration is better, the monsters are smarter, the secrets are more devious, and the rune system is a mechanic that deserved to be stolen by an entire genre and somehow wasn’t.

Its flaws are structural and honest. Combat runs out of new ideas before the island does. A handful of puzzles are gated behind leaps of logic that no reasonable player will make unassisted. The build system asks for commitment while hiding the information required to commit intelligently. These are not small complaints, and anyone telling you they’re actually features is selling you a difficulty ideology rather than reviewing a game.

But the thing holds up. A decade on, it remains a high-water mark for grid-based dungeon crawling and a steady source of community campaigns. A small team made this, and it’s better than most of what much larger studios ship.

We recommend it — with a notebook.