Astronomers read the history of galaxies by their portraits. Previously, they used either lengthy analysis of light by colors or quick filtered snapshots, losing the details of shape. The new method combines images with brightness data: the algorithm learns to link each point in the image with physical properties — mass, age, amount of dust. This is the key to an old puzzle: dust can disguise an old galaxy as a young one, but our approach discerns the true age. It's amazing that Edwin Hubble classified galaxies by appearance, even though back then there were no tools for such deep reading of shape. Now, future sky surveys will be able to cover millions of objects, and instead of expensive spectroscopy, reliable estimates will be obtained directly from ordinary images. Moreover, the algorithm turned out to be so perceptive that it notices details that escape experienced eyes — a real breakthrough in cosmic demography.
Dust in galaxies doesn't just hinder observations — it tells its own story of star birth and death, because it consists of heavy elements ejected by old stars.