A spotlight blazes straight into your eyes, and a few steps away a tiny firefly flickers. That's roughly what it's like hunting for an Earth-like planet around a neighboring star: the planet is a hundred million to ten billion times fainter than its star, and in the sky they almost merge into a single point. The star's light drowns out the faint planetary reflection, rendering ordinary telescopes helpless.

A team of researchers has proposed to bypass this obstacle with adaptive quantum measurement. The method doesn't try to directly photograph the star in hopes of spotting the planet. Instead, it selects a special "observation mode"—a spatial mode of light — in which every captured photon carries maximum information about the faint source. The process resembles tuning a musical instrument: the system iteratively shapes the light wave until it "hears" the planet's quiet voice amid the star's deafening noise.

The method works at contrasts up to \(10^{-8}\)—five orders of magnitude better than previous quantum schemes for finding exoplanets. It doesn't just detect a planet; it also precisely indicates how many are in the frame, where they are, and how bright they are. This is done using maximum likelihood estimation in a logarithmic brightness scale, and the number of sources is determined automatically via the BIC criterion — without any human-set thresholds. In tests on points separated by less than the diffraction limit and with an eight-order-of-magnitude brightness difference, the scene was recovered on average in 72.5% of cases.

Robustness to errors is another plus. Even with a camera shift of six pixels, success remains at 71.3%. Technical imperfections don't doom the observations. This approach offers the first opportunity to directly see Earth-like planets in the habitable zone — the region around a star where water can remain liquid. Previously, this range was out of reach for direct imaging.