Readable by Design

The governing rule behind the game's look is easy to state and hard to hold to: the board has to stay readable, and it has to do so without becoming ugly. Those two demands pull against each other. A grid that is emphasised enough to be instantly legible can turn clinical; a scene that is atmospheric enough to be beautiful can obscure the information a puzzle depends on. Most of the art direction in Order of the Sinking Star exists to satisfy both at once.

The game's levels are rendered from a fixed top-down camera, and the reasons for that are partly practical. A movable camera would demand that every asset look convincing from every angle, which means every pipe and wall and prop would have to survive close inspection. Fixing the view means the art team only has to make things look right from where the player actually is. The director is candid that this is something of a trade: he would sometimes like to fly the camera down and look at a piece of level geometry from another angle, and some puzzle games map the second stick to exactly that. He resists it anyway, because a freely rotating camera does not serve the puzzle, and because the team is relieved to have fewer viewpoints to satisfy.

Readability is not only a question of clarity but of restraint. The design keeps the view where it needs to be rather than where it would be fun to look.

The game's structure began as a thought experiment during the director's previous project: what would happen if, instead of combining abstract puzzle symbols, you combined things that are themselves already complete — entire games, in effect. That idea became the overworld. Four territories sit on the map, each with its own title screen and its own completely different mechanics, and at first everything you uncover belongs to one of them.

Each territory is based on a small, free puzzle game that existed on the web years before development began, and the designers of those originals worked on the regions they inspired. One world turns on teleporting mirrors and the puzzles of position and momentum that follow from moving objects at right angles. Another is built from the rules of a browser game about characters with role-playing abilities recast as puzzle abilities — a warrior who can push an indefinite number of blocks, floor buttons that can be weighed down, and so on. The regions are recognisably expansions of their sources, and the point of building them was always the collision that comes afterwards.

What the four-territory structure buys is a way to teach a great deal without overwhelming anyone at the start. Each region is self-contained and internally consistent, and only once the player has internalised a region's rules do the worlds begin to meet.

The grid is the game's most important visual element, and it is rendered in a way that is more sophisticated than it first appears. Faint grid lines can be painted in and cleared out, and they are built from an animated flip-book texture, an intensity overlay, and a multiplying noise layer — the kind of channel-packing that 3D rendering has used for years, repurposed here to keep the grid's presence adjustable rather than fixed.

There is also a subtle three-dimensional quality to the lines. Close up, the grid waves carry a normal-mapping effect, so they read as slightly raised rather than painted flat onto the floor. It is the sort of detail that most players will never consciously notice and would immediately miss if it were gone.

The atmosphere around the board is doing similar work. The scenes lean on what look like volumetric effects drifting through the background, and the team spent a number of iterations and guesses arriving at them. The effect itself was not especially hard to build; getting it to read as depth rather than as haze took the time. Layered over a fixed camera and a legible grid, it keeps the board readable while giving each region a mood of its own.

The game runs on an engine the studio wrote itself, in a programming language the studio also wrote. The walkthrough makes the practical consequences concrete: levels are edited inside the running game, and a single keystroke plays exactly what is being edited, with no separate build step and no wait for the project to start up. Notes can be left inside levels for the art team — reminders that a particular hole needs to stay visible, for instance — so that design intent travels with the geometry instead of living in a separate document.

The engine is not general-purpose in the way a commercial one is. You cannot make a game in it without programming, which is a real limitation and one the director states plainly. What it buys is control: because the team writes every line that makes the game run, a system that is not performing well can be drilled into and refined rather than accepted as it is.

Asked whether the engine will be sold, the answer is that it will be given away for free. It has no name yet, and one will have to be chosen before it is released. The plan is consistent with the reasoning behind the whole technology stack: build the best thing rather than the thing everyone will use.

Taken together, the art direction reads as a set of decisions made in service of one goal. The camera is fixed so the assets only have to be convincing from one place. The grid is adjustable and quietly three-dimensional so it can be emphasised where clarity demands it and softened where it does not. The atmosphere is layered behind the board rather than over it. Nothing in the look competes with the information the puzzle needs the player to see.

That is a harder target than it sounds, and it explains why the game can be described as both beautiful and plain. The beauty is real, but it is never allowed to win an argument with legibility. In a game whose entire verb is pushing blocks around a grid, that discipline is not decoration. It is the design.