Researchers at ETH Zurich have built a pixel that both emits and reads light. The advance, detailed in Nature, rests on sculpted surfaces precise to a few nanometers. These structures harness interference and Fourier analysis to manage amplitude, phase and polarization in one compact element.
Call it a Fourier pixel. Incoming light turns into a surface plasmon polariton that travels along the chip. At another spot the wave scatters back into free space. The resulting interference patterns form images or reveal hidden properties of the incoming field. Reverse the math and the same structure analyzes what it sees.
From Display or Sensor to Both
Conventional pixels pick one job. Screens emit. Sensors capture. The new devices refuse that choice. “In addition to light intensity, meaning the bright and dark areas from which images are created, our Fourier pixels can also control other properties of the light waves, for example their polarisation,” says doctoral student Yannik Glauser. Polarization marks the oscillation direction of the electric field. The pixel generates any desired state by overlapping surface waves of different polarizations whose overlap depends on surface shape.
Phase control follows the same logic. Researchers create doughnut-shaped beams with a dark center simply by dialing the right phase profile. The approach works across wavelengths, so full-color operation arrives without extra hardware. And the math stays simple. Fourier transforms replace complex models. That simplicity matters when thousands or millions of pixels must share one chip.
Postdoctoral researcher Sander Vonk explains the payoff. “We can also, however, apply the principle of interference and Fourier analysis in the opposite direction to analyse light using the Fourier pixel.” Incoming light mixes with a reference wave on the pixel. A camera records the interference pattern. Phase and polarization drop out of the calculation. All three properties—amplitude, phase, polarization—sit inside a single element.
Professor David Norris leads the Optical Materials Engineering Laboratory behind the work. “Our new pixels for control and analysis could, therefore, become a useful tool in many areas,” he says. Norris sees near-term progress in large matrices of these pixels that function as both camera and display. Longer range, he imagines pixels that react to what they capture without routing data through a separate processor. Surface waves perform the math directly on the material.
The ETH Zurich team published its findings on June 24, 2026. The paper, “Fourier pixels for bidirectional light control,” lists Glauser, Vonk and colleagues as authors. A patent application has been filed and nominated for the institution’s Spark Award.
Coverage followed quickly. PetaPixel highlighted the 1-millimeter-tall colored ETH logo captured and reproduced by the prototype. Gizmodo noted the potential for screens that double as cameras. TechRadar captured the internet’s reflexive reaction—comparisons to the telescreens of “1984.” The jokes write themselves. But the physics points to practical gains.
Optical computing could accelerate. Metasurfaces already steer light for beam forming and sensing. This work folds analysis into the same footprint. Fiber-optic systems might embed diagnostics without extra taps. Medical imaging or augmented-reality glasses could shrink because one surface both projects and inspects the scene. Norris calls the short-term goal a working matrix. Scale that and the distinction between input and output hardware begins to dissolve.
But, challenges remain. Efficiency at scale, crosstalk between neighboring pixels, and integration with existing electronics all need attention. The current prototypes operate at 555 nanometers. Broadening the spectral range without losing fidelity will test the fabrication process. Still, the core insight holds. A surface shaped at the nanoscale can encode and decode the full optical field in one place.
So the pixel evolves. Once a passive dot on a grid, it now computes. Light arrives, transforms, and reports back using the same physical element. The line between emitter and detector vanishes. What arrives next is a generation of devices that see and show without switching roles. That shift carries consequences for displays, sensors, and the architectures that connect them.
Norris and his group have shown the principle works. The next papers will test how far the concept stretches when thousands of Fourier pixels operate together. Industry watchers will track power consumption, manufacturing yield, and speed. For now the demonstration stands. A pixel that both creates and understands its own image marks a quiet but substantial break from a century of one-way light control.
WebProNews is an iEntry Publication