Designing Spatial Interfaces

Start with the most concrete change: on a screen, every control shares one flat plane at one distance. In a spatial interface, each element has a position, a distance and a size in three dimensions, and you choose all three. A menu can hover at arm's length in front of you; a reference drawing can sit on a virtual wall two metres away; the building model can stand at full scale filling the room. Nothing is automatically visible - to see something you turn toward it, so the designer is really deciding *what belongs in front, what belongs to the sides, and what belongs behind*, knowing that anything behind you is effectively out of sight and out of mind.

This spatial freedom is genuinely powerful and genuinely dangerous. Powerful, because you can put a tool exactly where the hand expects it and let the design occupy real space at real scale, which is the whole point of the medium. Dangerous, because a flat page politely contains everything inside its border, while a room has no border - it is easy to scatter panels all around a person until they feel surrounded, lost, and unable to find anything. The discipline is to treat placement as a deliberate act with a budget, not a free canvas. A small number of well-placed things beats a cloud of floating widgets every time.

It helps to think in terms of what should be head-locked, body-locked or world-locked. Head-locked content follows your gaze and is almost always a mistake for anything you want to read or study - it feels stuck to your face and induces fatigue and unease. Body-locked content stays around your torso as you turn, useful for a small palette of tools that should travel with you. World-locked content stays fixed in the space like a real object - the right choice for the model, for reference drawings, for anything you want to walk around and treat as part of the scene. Choosing the right anchor for each element is one of the first real decisions in designing a spatial interface, and getting it wrong is one of the first ways to make people uncomfortable. Any binding dimension shown in that interface still comes from the verified model and drawings, not from the placement itself.

A spatial interface has to respect the plain physiology of the eyes, neck and arms, because unlike a screen it can actively hurt. The first constraint is distance. Content placed too close forces the eyes to converge hard and quickly becomes uncomfortable - a problem made worse by the vergence-accommodation conflict, where your eyes converge on a near object but stay focused at the display's fixed focal depth, a known cause of eye strain in headsets. So there is a comfortable near limit; text and controls you will study for any length of time belong out around half a metre to a couple of metres, not right in front of your nose. Far away, content stays comfortable to look at but becomes hard to read and hard to touch.

The second constraint is the field of view and the neck. People have a comfortable cone in front of them - roughly the range they can see and lightly scan without turning - and a much smaller range they can reach without moving. Anything important should live inside that comfortable cone at a natural viewing height, slightly below the horizon rather than up high, because looking up for more than a moment tires the neck fast. Putting a critical control overhead, or far to one side so the user must crane around to use it, is a classic beginner mistake that turns a short session into a sore one. Frequent actions belong dead ahead and within easy reach; rare things can sit further out where fetching them costs a deliberate turn.

The third constraint is the body over time. Holding arms up to manipulate floating panels causes what people call gorilla-arm fatigue within minutes; a spatial interface that assumes constant mid-air reaching will exhaust its user. Favour resting positions - hands low, short interactions, the option to sit. Design for a person who may wear the headset for ten or fifteen minutes, not an idealised user with an infinite neck and tireless arms. None of this replaces the manufacturer's own comfort and safety guidance for a given device, which is where the real health limits and wear-time advice belong; the designer's job is simply to place content so that the ergonomics work with the body rather than against it, and to treat comfort as a first-class constraint rather than a detail to fix later.

Because a spatial scene has no edge to hold things in, the default failure mode is clutter, and clutter in three dimensions is far worse than on a screen. A busy web page is at least all in front of you; a busy spatial interface surrounds you with floating panels, labels, buttons and gizmos until the person cannot tell what matters, cannot find the one control they need, and feels a low-grade stress at being crowded. The single most useful principle is restraint: show the few things needed right now, and let everything else be summoned when wanted. An interface that reveals tools on demand and hides them again keeps the space calm and keeps the design - the thing you actually came to see - unobstructed.

Fatigue compounds clutter. Every element the eye must track, every label that follows the head, every animation that moves in the periphery, adds a small tax on attention and comfort, and in a headset those taxes accumulate into tiredness and sometimes queasiness surprisingly quickly. Motion in the corner of the eye is especially costly, because peripheral vision is highly sensitive to movement and reads it as something to worry about; drifting, pulsing or constantly reorienting UI is a reliable way to tire and unsettle people. Prefer stillness. Let the interface sit quietly in the world and move only in direct, brief response to what the person does.

There is also a legibility tax that clutter hides. Text in space must be large, high-contrast and held at a steady, comfortable depth, and there is far less room for it than a screen affords - so a spatial interface cannot carry the dense panels and long menus that a desktop tool leans on. This is a genuine limitation, not a stylistic preference: the honest response is to move most of that density back onto a companion screen or tablet and reserve the spatial interface for the few controls and the spatial content that truly benefit from being in 3D. A calm, sparse, legible spatial interface around an unobstructed model is not a compromise; it is the correct design, and it is also the one that people can bear to use for more than a few minutes at a stretch.

The deepest shift is one of mindset. It is tempting to think the job is done once the model is loaded and floating in front of the client - but a model dumped into space with no thought is often confusing, uncomfortable or underwhelming, and it wastes the medium's real power. What you are actually designing is an experience: where the person starts, what scale they first meet the design at, how they move through it, what they can reach and change, what is explained and what is left to be discovered, and how comfortable the whole thing feels from the first second to the last. That is closer to designing a short guided visit than to arranging a screen.

Concretely, this means asking experiential questions the screen never forced you to ask. Where does the person stand when the scene opens - outside the building looking at a model on a table, or already standing in the entrance hall at full height? Should they begin at doll's-house scale to grasp the whole plan, then step to 1:1 to feel a room? What is the one thing you want them to understand first, and how does the placement of content lead their attention there? A well-designed spatial experience answers these on the person's behalf, so they feel oriented and in control rather than dropped into a floating void with no idea where to look.

This is exactly where spatial computing earns its place for designers, and exactly where a thoughtless one squanders it. For a client who cannot read plans, a well-designed experience - start here, look this way, now you are standing in your kitchen at true size, reach out and change the finish - can be genuinely transformative, the moment they finally understand their own home. A raw model with no such design is just a slower, dizzier version of a screen render. So treat the spatial interface as the staging around a valuable experience: place the content to serve the design, keep the interface calm and comfortable, lead the person gently, and remember that the goal is understanding and communication, never a technical readout - the binding dimensions and decisions stay with the verified drawings and the specialists. Design the visit, not just the object in it.