A new sky map brings together 5.6 trillion pixels and nearly four billion celestial objects in a single public database.
This gigantic panorama comes from the DESI Legacy Imaging Surveys.
Despite their name, these surveys were not conducted by the DESI instrument itself. They consist of several imaging campaigns carried out over 13 years to photograph the sky and prepare DESI’s future targets. More than 263,000 images were assembled and processed: they cover approximately three-quarters of the sky and record visible light as well as near-infrared light, a portion of radiation located just beyond the red detectable by our eyes.
A galaxy visible in the panorama
The map is described as two-dimensional because it first indicates where objects are located in the sky, without directly providing their distance. It brings together stars from our galaxy, as well as countless much more distant galaxies. Its 5.6 trillion pixels therefore do not correspond to a photograph taken all at once. They result from the methodical assembly of observations made with several telescopes.
DESI, for its part, is a spectroscopic instrument installed on the 4-meter Mayall Telescope in Arizona. The large image map serves as its target catalog: it precisely indicates where the galaxies and quasars to be observed are located. DESI then points its thousands of robotic optical fibers toward these objects and analyzes their light to measure their redshift.
To understand the difference, the Legacy Imaging Surveys essentially provide two coordinates, comparable to the position of a point on a road map. DESI then adds distance-related information by analyzing the light spectrum. Astronomers can then place the galaxies in space and reconstruct a three-dimensional map. DESI had already measured more than 47 million galaxies and quasars by April 2026.
The value of the new dataset nevertheless extends beyond the DESI mission. An astronomer studying an object can begin by consulting this map to check its surroundings and nearby sources.
The amount of information also provides a training ground for artificial intelligence tools: future observatories will produce volumes of images that are difficult to examine manually. Automated systems can thus learn to identify shapes, classify sources, or select unusual phenomena before a researcher examines the most interesting candidates.
In practice, this map is not reserved solely for laboratories. The data is public and comes with an online viewer. Professionals and amateurs alike can therefore browse the same images, zoom in on a galaxy, or examine a region of the sky.
Future analyses will be able to combine this panorama with new observations to search for rare phenomena still hidden among its billions of sources.