The first photo ever taken of a black hole looks a little sharper now. Originally released in 2019, the unprecedented historic image of the supermassive black hole at the center of the galaxy Messier 87 captured an essentially invisible celestial object using direct imaging. The image presented the first direct visual evidence that black holes exist, showcasing a central dark region encapsulated by a ring of light that looks brighter on one side. Astronomers nicknamed the object the “fuzzy, orange donut.” Now, scientists have used machine learning to give the image a cleaner upgrade that looks more like a “skinny” doughnut, researchers said. The central region is darker and larger, surrounded by a bright ring as hot gas falls into the black hole in the new image. A machine-learning technique was used to enhance the Event Horizon Telescope Collaboration’s image (left) of the supermassive black hole at the center of the galaxy Messier 87 and produce a sharper image. NOIRLab In 2017, astronomers set out to observe the invisible heart of the massive galaxy Messier 87, or M87, near the Virgo galaxy cluster 55 million light-years from Earth. The Event Horizon Telescope Collaboration, called EHT, is a global network of telescopes that captured the first photograph of a black hole. More than 200 researchers worked on the project for more than a decade. The project was named for the event horizon, the proposed boundary around a black hole that represents the point of no return where no light or radiation can escape. To capture an image of the black hole, scientists combined the power of seven radio telescopes around the world using Very-Long-Baseline-Interferometry, according to the European Southern Observatory, which is part of the EHT. This array effectively created a virtual telescope around the same size as Earth. ‘Maximum resolution’ achieved Data from the original 2017 observation was combined with a machine learning technique to capture the full resolution of what the telescopes saw for the first time. The new, more detailed image, along with a study, was released on Thursday in The Astrophysical Journal Letters. Dwarf starburst galaxy Henize 2-10 sparkles with young stars in this Hubble visible-light image With our new machine learning technique Astronom ultramass black hole using new technique. Computers using PRIMO analyzed more than 30,000 highresolution simulated images of blackholes to pick out common structural details. This allowed the machine learning — known as PRIMO — to fill in the gaps of the original image; the black hole’s shadow is represented by the dark central region. The new, more detailed image will help scientists make more accurate measurements of