A Northwestern-led team built FLIP (Physics-based Learning for Imaging Planets), a way to subtract a star’s glare from telescope images so faint planets show up. The Register covered the Nature paper.
What they did
- FLIP models how light moves through the telescope, corrects for tiny optical flaws, and simulates the glare to remove
- On Webb NIRCam data it got the first image of the giant planet 51 Eridani b at 4.3 micrometers, where normal processing failed
- Sensitivity improved 2 to 15 times over the usual techniques
My take
I love this kind of AI: it’s grounded in physics instead of guessing. Getting more out of telescopes we already have is how we get closer to imaging Earth-like worlds. “The heavens declare the glory of God” (Psalm 19:1).
God Bless, and Use the Source!
