The James Webb Space Telescope had already photographed the neighborhood of 51 Eridani—but its data appeared to show no planet. A new image-processing technique has now peeled away enough residual starlight to reveal 51 Eridani b, a known giant exoplanet that had been buried in Webb’s glare. The discovery demonstrates how much more Webb’s existing observations may still contain.
Called Physics-based Learning for Imaging Planets, or FLIP, the method was developed by Northwestern University astronomers led by graduate researcher Rodrigo Ferrer-Chavez, with assistant professor Jason Wang among the study’s co-authors. Their work, published in Nature on November 4, constructs a virtual “digital twin” of JWST’s Near-Infrared Camera. That model simulates how starlight travels through the telescope’s optics and lands on its detectors, allowing researchers to reproduce and subtract the unwanted glare more precisely. Northwestern says the approach improved detection sensitivity by between two and 15 times.
The challenge is not simply that stars are bright. Webb’s coronagraphs block much of a star’s light, but tiny thermal changes, vibrations from onboard hardware and minute optical shifts alter the remaining glare pattern. Traditional processing often uses another star as a reference image, then subtracts it from the target observation. Even an almost imperceptible mismatch can leave a bright residue that overwhelms a nearby planet.
FLIP instead combines a physical model of the telescope with optimization techniques commonly associated with machine learning. The researchers emphasize that it is not an artificial-intelligence system trained to recognize planets; the software uses known optical behavior to adjust millions of model parameters until its simulated glare matches the real observation. Applied to archival JWST observations of HIP 65426, 51 Eridani and TYC 5899, the method produced the first image of 51 Eridani b at 4.3 microns. Nature’s summary identifies that wavelength as the location of the new detection.
That infrared window is scientifically valuable because carbon dioxide absorbs strongly around 4.3 microns. Recovering 51 Eridani b there could help researchers study the planet’s atmosphere and composition, rather than merely confirming that the world exists. The planet itself was previously found using ground-based telescopes, but conventional processing of Webb’s observations could not separate it from the star’s residual light.
The larger payoff is the possibility of finding planets that current surveys systematically miss. Young, massive worlds are easier to image because they are comparatively bright, while smaller, colder planets resembling those in our own solar system can disappear against their host stars. A stronger software pipeline could therefore extract new science from observations already stored in the JWST archive, without changing the telescope’s hardware. Looking further ahead, techniques like FLIP may also help inform the design and data processing of future observatories intended to directly image Earth-sized planets around Sun-like stars.
