Nova Studio

Spatial media

Spatial media

Images and video are flat records of a space. Spatial media keeps the space itself — which is what lets a viewer look somewhere the director did not point, and what makes authoring it a different craft rather than the same craft in a wider frame.

What the term means here

A frame is a decision about where to look, made once, by somebody else. Every convention in film exists because that decision is the author's: the cut, the reveal, the rack focus, the reverse. They are all ways of moving attention inside a rectangle you control completely.

Spatial media moves part of that decision to the viewer. The scene has extent beyond the frame, and looking around is something they do rather than something they watch somebody else do.

That changes authoring far more than it changes viewing. Composition stops being a rectangle and becomes a question of where attention should land when the audience can point their head anywhere — which means the tools that used to do that work are gone and something has to replace them.

What the product does today

Stated precisely, because this is the column where vagueness would be most convenient.

CapabilityStatusNotes
Scene-consistent generationWorksReferences that hold a place, a light and a subject across many shots — the prerequisite for a coherent space rather than a set of unrelated views.
Multi-view framingWorksFramings covering more of a scene than one shot needs, so the surrounding space exists rather than being implied.
Director stage sequencingWorksA sequence read from the same graph, with order, duration and notes on each beat.
Consistent depth between viewsPartialViews agree on light, material and layout. They do not agree to the millimetre, and nothing here computes geometry.
Headset-native exportNot in this buildNo USDZ, no Reality Composer project, no stereo pair output.
Volumetric capture ingestNot in this buildNo point cloud or gaussian input or output.

Where it goes

The direction is a scene that survives being looked at from more than one place — depth consistent between views, and a delivery format a headset can read.

The order matters and it is not the obvious one. Export is the easy half. The hard half is consistency: a scene that reorganises itself when you look away is worse than a flat image, not better, and shipping an export format before that is solved would produce a very convincing demonstration of a broken idea.

So the work happening now is the unglamorous part, and it is the part that makes the rest possible.

Questions

Can I make something for Vision Pro or Quest with this today?

You can design and previsualise it here. You cannot export it to run on one — that gets built in the platform's own tools. The visionOS page walks through where the handoff sits.

Is this 3D?

No, and the distinction is worth keeping. This is generated imagery that is consistent between views. A 3D package gives you measurable geometry. They are complementary and neither replaces the other.

When will spatial export ship?

No date. It appears on the status page the day it works. A date on a page like this would be a promise made by somebody who cannot keep it.

So why publish this page at all?

Because the consistency work that spatial media requires is what the canvas already does, and it is useful now for flat work. This page explains where that leads without pretending it has arrived.

Related