Hilbert Curves and Sound to Sight: From Infinity to Finite in Space Filling Curves

This video explores space filling curves, focusing on Hilbert curves and their pseudo variants, and shows how mapping a 2D image into a 1D frequency line can enable sound to sight experiments. It highlights why the Hilbert approach maintains locality when increasing resolution, easing perceptual learning for users upgrading from 256x256 to higher resolutions. The talk connects finite computational practice with infinite mathematical ideas and illustrates how the limit of pseudo curves yields a true space filling curve.

• Hilbert curves preserve locality as resolution grows
• Pseudo Hilbert curves approximate the true space filling curve via iterative limits
• Snake curves vs Hilbert curves: stability under increasing image resolution
• Philosophical link between infinite math and finite, practical applications

Overview

The presentation centers on space filling curves, particularly the Hilbert curve and its pseudo variants, and uses them to bridge pixel space and frequency space in a sound to sight software concept. The aim is to keep nearby pixels mapped to nearby frequencies so the brain can form a coherent interpretation as data flows from a camera to sound.

Snake Curve versus Hilbert Curve

The narrator contrasts a simple row-by-row Snake curve with a Hilbert curve style weaving. The Snake curve is easy to visualize but suffers when image resolution changes, because points on the frequency line jump to unrelated regions in pixel space as resolution increases. The Hilbert curve, by construction, connects quadrants in a way that reduces these jumps, maintaining continuity of the mapping across resolutions.

Pseudo Hilbert Curves and the Limit

A mathematician introduces order 1, order 2, and higher order pseudo Hilbert curves. Each order subdivides the square into a finer grid and embeds smaller pseudo curves within quadrants, flipping some mini curves to optimize connections. The key idea is that as the order increases, the output of the pseudo curves for any fixed input stabilizes toward a single point in the plane. That stabilization is what legitimizes taking the limit to define a genuine Hilbert curve, a continuous function from the unit interval to the unit square that hits every point in the square in the limit.

Continuity, Convergence, and Space Filling

Continuity is defined via the usual epsilon-delta idea, translated into circles in input and output spaces. The Hilbert curve is built as the limit of an infinite sequence of pseudo Hilbert curves. The limit exists because, for each input, the sequence of outputs converges; the resulting function is continuous; and the image of the unit interval is the entire unit square, thereby filling space. The video also suggests how tiling space with squares and connecting many Hilbert curves can extend this idea to all of 2D space or even higher dimensions.

Finite Versus Infinite and Practical Implications

The discussion closes by highlighting how infinite mathematical structures yield finite, usable patterns. It underscores the broader theme that infinite concepts often have finite analogs that are directly useful in computation, graphics, and perception science. The Hilbert curve approach yields stable perceptual intuition as you upgrade resolution, which is not generally true for the Snake mapping.

Takeaways

Space filling curves reveal a deep link between the infinite and the finite, providing a framework where a thin line can traverse every point in an infinite plane. In applications like sound to sight, Hilbert curves offer a principled way to maintain locality across scales, enabling smoother perceptual learning and more robust interfaces between vision, audio, and computation.