Skip to main content

Morning Overview

A hidden giant planet emerged from old Webb images that standard processing missed

A giant planet that was already known to astronomers has just been pulled out of Webb Space Telescope images that conventional processing could not see it in. The planet, 51 Eridani b, sat in data about two years old; a Northwestern University team recovered it above the 5-sigma detection threshold by rebuilding the telescope’s starlight on a computer and subtracting it.

The result, published Nov. 4 in Nature, is less about one planet than about a way to find fainter ones in images that already exist.

A planet a million times fainter than its star

Direct imaging of planets runs into a brightness problem. Rodrigo Ferrer-Chavez, a Northwestern astronomy Ph.D. student and lead author, put it in the university’s release this way: “Planets are a million times fainter than their stars.” The release adds that young giant planets can be about 100,000 times fainter than their hosts.

Webb’s coronagraph is built to fight that glare. NASA’s explanation of the NIRCam coronagraphs says masks physically block unwanted starlight so a planet’s faint light can reach the sensors, using five coronagraphic masks that can observe between 1.7 and 5 microns. Even so, some starlight leaks through, and astronomers measure the leftover pattern and subtract it. During commissioning, NIRCam reached contrasts better than 10 to the minus five at 1 arcsecond separation.

The standard way to subtract that pattern is to use a reference image of a different star. The reference works only if the telescope stays the same between the two observations, and shifts as small as a few nanometers can change the glare, the Northwestern release says.

FLIP, a digital twin of the telescope

The team’s method is called FLIP, for Physics-based Learning for Imaging Planets. Instead of leaning on a reference star, it builds what the researchers call a digital twin of Webb that simulates how starlight travels through the mirrors to the detectors. The model is adjusted, using millions of parameters, until it matches the real observation including the starlight, and the simulated glare is then subtracted. A Universe Today account describes it as a multistep optimization strategy with a flexible physical model informed by wavefront sensing data.

The release also notes that the same physics-based approach was benchmarked against KLIP, the method described in a 2012 paper by Soummer and colleagues. The tools resemble those used to train AI systems, but the team says no AI was involved because the physics was already known. Jason Wang, an assistant professor of physics and astronomy at Northwestern and a co-author, said the work showed that a planet people previously could not see in Webb data can in fact be recovered, a point that rests on the 51 Eri b detection.

The authors report 2 to 15 times better sensitivity to faint planets than conventional methods. The paper, “Imaging exoplanets below the telescope stability limit,” appears in Nature under DOI 10.1038/s41586-026-11213-z. Its author list also includes researchers at Caltech, MIT and the Space Telescope Science Institute, according to the Caltech authors record, which gives the citation as Nature 659, 58-62.

What came out of the old images

The test case was 51 Eridani A, whose planet 51 Eri b was first found with ground-based telescopes. Webb observed it with NIRCam’s F430M filter at 4.3 microns, a wavelength where carbon dioxide strongly absorbs infrared light. Conventional processing of the data, which the release cites as Balmer and colleagues in 2025, missed the planet, and the standard comparison method, KLIP, did not detect it. FLIP did, at greater than 5 sigma and at its expected location. The Caltech record calls it the first detection of 51 Eridani b at 4.3 microns.

The team also ran FLIP on two other stars. The release names them as HIP 65426 and TYC 5899, and says the latter has no known planetary companions. Neither the release nor the Universe Today account gives the planet’s distance, mass or age, and so those numbers are not included here.

Why stability matters for the next telescope

The larger motivation is a future space observatory. According to the release, NASA’s proposed Habitable Worlds Observatory may need stability at the picometer scale, roughly 100 to 1,000 times more stable than Webb. The authors suggest FLIP could let such a mission correct for small instabilities in software instead of requiring extreme hardware stability, and they propose studying planets at other wavelengths to learn more about their atmospheres.

The work was funded by the Space Telescope Science Institute, the Alfred P. Sloan Foundation and the U.S. National Science Foundation, according to the release, and both Ferrer-Chavez and Wang are members of Northwestern’s CIERA astrophysics center. The Caltech record shows the paper was submitted on March 2 and accepted on Sept. 25 before appearing Nov. 4. For now, FLIP’s gains are demonstrated on one planet that was already known; the test the field will watch is whether the method turns up a planet that no one has seen before.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.


More from Morning Overview


Morning Overview is reader-supported. Some links in our articles are affiliate links, and we may earn a commission at no extra cost to you. As an Amazon Associate I earn from qualifying purchases. Full disclosure.