Hazelight

Hazelight prototyped mechanics and settings separately, then paired them into levels. Art explored places a shrunken couple could visit while design prototyped many mechanics at once, including the sap and match guns. Each mechanic was matched to a setting, and a designer or pair got a level to fill.

Oliver Granlund and Robert Johansson built the Tree and the snow globe. A calm section Henrik Sandin designed late now looks like the start of the level; art added ants by accident, players who followed them off a hill died, and Henrik used that on purpose.

A prototype takes a week to play, but far longer to make fun. Hazelight built playable pieces in about a week, like Respawn's action blocks, but polishing them took far longer than expected, a lesson the team learned painfully.

Half the work was making mechanics clear and the other half making puzzles smooth. Ideas kept similar concepts but ended with wildly different execution: the vault door first exploded when sprayed, then became a turning lid to keep the camera tight. "it might take like a week to prototype and get playable, it might take way longer to polish."

The boat reached its final controls only through repeated rework. It went from recoil propulsion, where you couldn't see where you were going, to a rudder, which was fun but broke the no-UI rule, to simply pressing fire to go.

Oliver Granlund credits the team's cycle of making, playtesting, talking and changing for getting the boat to its final form.

The guns had to do more than explode before they could carry puzzles. The first sap gun only sapped and exploded, great for combat but useless for puzzles, so weight became a property of sap, letting it pull things down.

The team let sap land anywhere rather than on marked spots, which created places that felt bad: in the first wasp nest cage Cody could sap the outside where May couldn't reach, so version two had Cody sap the roof. Much of the work was removing spots where you could do things that don't feel nice.

Iteration turned a mini enemy into a boss and cut its skill demands. The beetle was meant as a mini enemy, but later passes made it one of the most intense fights. Adding grates reduced aiming, replacing a version where players shot its butt, which needed positioning, aim and communication.

Players mourned the beetle's death though it began as a one week prototype.

Stop a mechanic once players have used it long enough. The boat could have been a 30 minute segment or a whole game, but it lasts five minutes; Oliver Granlund says a longer boat would have made a worse game.

Variety sits beside 'finding good enough' as a pillar, so the team put its best finds into the level instead of dragging them out. "depth doesn't mean it's good if you don't have time to use it."

Variety replaces repetition, but intensity still needs a curve. Storyboarded as one hour, the Tree grew to nearly two, diluting the story, so cutscenes were added to carry it. The team admits it chased cool things and variety and should have watched the intensity curve.

Hazelight used Nintendo's four step level design rule only in places and crammed in as much variety as it could. Nothing limited the strangeness; meeting feedback was to make it crazier, "like Avatar on LSD", and players ask whether they are still in the tree.

Mechanics carry the story through theme. Magnets are about attraction and clockwork about time; the team calls these emotional levels. A Way Out was story focused, while in It Takes Two gameplay won but the aim was to keep both together.

A segment's story weight only shows once the game is nearly whole: too long without story bores players, too long as a bare cutscene loses their interest.

Mix three kinds of co-op to keep a pair busy. Hazelight mixes parallel co-op (two solo games side by side), step-by-step turn-taking, and simultaneous co-op where both act at once.

Pacing these moments is the big challenge. A side-scroller camera, added fairly early, fixed camera problems in the spinning lab section.

Each player must understand their own role at a glance. Without a separate tutorial, both players shot every shooting sign and were confused. In combat Cody is the instigator and May the detonator, and wasps are tuned to attack May while she waits.

Each level has its own grounded rules, so a weight going down must visibly connect to something going up, with no Mario style floating platforms. The Tree is the only aimed-combat level and the hardest for new players, especially those who can't use both sticks.

A respawn timer kept unequal players both in the game. A shared health bar failed because the weaker player took damage until both died, and a Gears of War style revive failed because players are often apart or platforming. The final timer runs while both are alive, giving each player effectively two lives.

Aim assist is lenient so dodging matters more than skill shots. Dying is fun but replaying is not, and the timer turns into cheering: "don't die, I'm almost back."

Side interactions and rivalry filled the gaps playtests exposed. After each milestone, playtests showed too little to do, so the team went on a spree of minigames and side interactions; hidden minigames became the collectible, since collectibles had no good purpose.

Competitive minigames break up cooperation, and when the characters bicker the players do too. Minigames suit different players, including a chessboard, so the same person doesn't always win.

Give artists context before asking them to build. When the lab's big spinning wheels had no context, artists turned them into an elaborate lab that shaped parts of the story.

Handing art an abstract piece of gameplay and asking "what could this be?" is not fun for anyone. "it's very true that game development is, you know, it's a team sport."

Expensive systems must earn their cost. The swarm technology, the hammer and the seamless fight-into-slide took a lot of work, so costly features need a clear final look first.

The team's strength is reacting fast: make, playtest, talk and change, which Oliver says helped make the game great.

Cluttered puzzle patches often signal a missing mechanic. Reviewing Mark Brown's Mind Over Magnet, whose rooms piled up doors and lasers to block shortcuts, Oliver Granlund said the real problem is often a missing mechanic that would solve it simply.

He also suggested camera framing to show only small parts of a puzzle at once. "often the problem is a gap in your toolbox as a designer"