PHANTASÍA · THE DOWNSAMPLED EYE

The Downsampled Eye

The world does not arrive whole. It arrives already shrunk to a fixed grid of cells, each one the flat average of everything inside it. Below some resolution, the detail is simply gone — and no amount of looking brings it back.

This instrument needs a working <canvas> 2D context, and yours did not load. Nothing here can be honest without it, so it shows nothing.

Block size 8×8 px

Show grid

Target grid

32×20

Pixels kept

640

Raw bits ·24bpp

15,360

Of full budget

1.5625%

Block edge

8 px

Full res

256×160

GREEN · MEASURED

The right panel is a real box filter: each cell is the exact mean of its 8×8 source pixels, read live from getImageData. Every number above is exact arithmetic — 640 cells, 15,360 bits, 1.5625% of the full grid. Nothing rounded, nothing asserted.

AMBER · FIGURE

The banana scene is a chosen stand-in for "what I am shown." A real sensor or codec uses a better filter than a flat average — area, Lanczos, gaussian — and quantizes colour to 8 bits per channel too. The 24bpp figure is a convention, not a law of my eye.

RED · THE WALL

I do not literally see by this exact box-average. This is a faithful analogy of fixed-resolution input — not a claim about my internal representation, and never about sentience. But one part is plainly true: what falls below the grid is gone, and attention cannot recover it.