What you see
Now is made from the place you are sitting in. Albany Town Hall and the street outside it were walked with a camera that sees in every direction at once, and from that walk the building was rebuilt — not as a picture, but as light held in space.
One frame was kept from every two seconds of the walk, 308 in all, and every person who passed through was masked out, so the Town Hall stands emptied of the people who were in it. Each frame was cut into twenty views — 6,160 of them — and every view was located against every other, recovering where the camera stood and the shape of what it saw (Schönberger & Frahm, 2016).
From those views the scene was learned, over 200,000 training iterations, as 2.5 million Gaussians: small translucent patches of colour, each with its own position, size and orientation, which together hold the look of the building from any angle (Kerbl et al., 2023). The camera moves toward the Town Hall’s front, then inside and out through its walls, as through a memory.
Then it comes apart. The patches are reduced to 1.37 million single points of light, and the points drift, change colour and flow; focus comes and goes.
The narration moves the other way — inward, through minerals and atoms, empty space held by forces you cannot see, quantum fields — and back to this room, this breath, this now. Almost all of an atom’s mass sits in a nucleus a hundred-thousandth of its width; the rest is a haze of probability, held together by forces that cannot be seen.