Skip to content

Astronomy & Astrophysics Discoveries

CSIRO team independently validates landmark M87* black hole image

The CSIRO team's GenDIReCT method filtered atmospheric interference from the original Event Horizon Telescope data without prior knowledge of a black hole's appearance.

https://orbital-wire.com/astronomy-astrophysics/csiro-validates-m87-black-hole-image/

Key facts

  • An independent team of four researchers from Australia's CSIRO validated the 2019 Event Horizon Telescope image of supermassive black hole M87* using a new machine-learning imaging method.
  • The technique uses closure invariants, combining triplets of telescope measurements instead of standard pairs, to eliminate atmospheric and instrumental errors.
  • The CSIRO technique could improve medical imaging and earthquake mapping by combining data from multiple sensors to map unknown structures.
  • Albert Einstein published general relativity in 1915, predicting that mass curves spacetime and that a black hole's surroundings would appear as a ring of superheated glowing plasma with its shadow falling on it.

A four-researcher team from Australia’s Commonwealth Scientific and Industrial Research Organisation independently validated the Event Horizon Telescope’s landmark image of supermassive black hole M87* using a new machine-learning imaging method called GenDIReCT, publishing “Independent interferometric observable confirms black hole shadow in M87*” in Nature on Nov. 11, 2026, with DOI 10.1038/s41586-026-11241-9, to test whether atmospheric effects and image-processing methods shaped the historic picture.

The Event Horizon Telescope collaboration revealed the first picture of a black hole on April 10, 2019. The image shows M87* at the heart of galaxy Messier 87, about 55 million light-years away. The black hole has a mass around 6.5 billion times that of the sun. Hundreds of scientists produced the 2019 picture, capturing the first direct view of a black hole’s shadow outlined by light from swirling hot gas, and the image has become one of the most recognized scientific pictures of the modern era.

Four independent imaging teams within the collaboration participated in the original 2019 publication; all produced a ring-like structure but differed in fine-scale feature representation. Scientists have previously disagreed on whether the details and structure of the ring in the original Event Horizon Telescope publication were accurate, and one study fundamentally challenged the ring-like nature of the source. Those differences stem from human and algorithmic biases in imaging data from a sparse telescope array and from errors from individual telescopes and Earth’s atmosphere.

Very long baseline interferometry connects radio telescopes across the globe to mimic a planet-sized telescope, but this process can introduce errors and biases affecting image accuracy. Very large telescopes are not practically possible, so the Event Horizon Telescope operates as a linked network. The EHT combined signals from radio telescopes worldwide, enabling the sharpest imaging possible from Earth and ushering in a new era of imaging extreme environments around black holes.

Albert Einstein published general relativity in 1915, predicting that mass curves spacetime and that a black hole’s surroundings would appear as a ring of superheated glowing plasma with its shadow falling on it. Karl Schwarzschild derived a solution to the theory’s equations describing an infinitesimal region with infinite mass—a singularity—surrounded by a spherical event horizon within which not even light can escape. The original EHT image showed a black hole “shadow” outlined by light from swirling hot gas, just as predicted by general relativity, supporting the idea that galactic centers house supermassive black holes. When scientists first saw the M87* image, they were astounded at how closely it reflected general relativity’s predictions, confirming the theory once again.

The CSIRO team developed GenDIReCT to capture information about the black hole’s structure and filter out interference from viewing M87* through Earth’s atmosphere. The technique uses closure invariants, combining triplets of telescope measurements instead of standard pairs, to eliminate atmospheric and instrumental errors. The team trained GenDIReCT on terrestrial images of cats, dogs, and cars, so the model had no idea what a black hole looked like. Team leader Samuel Lai said the model had no prior knowledge of what a black hole looked like when applied to the M87 data.

The team validated GenDIReCT using a large suite of tests, including imaging non-ring structures and real datasets of other galaxies such as Centaurus A. On December 4, 2025, the CSIRO team allowed its fully validated imaging method to “look” at the EHT observations of M87 after freezing the training procedure. The method was then applied across all datasets, where the ring structure was found consistent.

“Our technique combines triplets rather than the standard pairs of telescope data to eliminate those errors,” said Nithyanandan Thyagarajan, CSIRO astronomer and project leader. Thyagarajan used experience from synchrotron experiments to help build the new imaging technique. Thyagarajan explained that creating an image through very long baseline interferometry is tricky and can introduce errors and biases. Although errors are normally estimated and corrected during analysis, leftover errors can produce wrong results, which is why the team developed closure invariants to combine triplets of telescope measurements.

Applying the CSIRO method called GenDIReCT to the original Event Horizon Telescope dataset produced a black hole almost identical in size and shape to the historic EHT image. The CSIRO result is consistent in diameter and overall orientation with Event Horizon Telescope collaboration images, though differences remain in fine details of ring width and surface brightness.

The ring’s appearance and diameter are consistent with a black hole of 6.5 billion solar masses observed nearly face-on, like looking straight down into the eye of a hurricane, and this consistency held across all datasets examined by the team. The result resolves previous discrepancies regarding the M87* ring structure while providing a fresh perspective on the iconic image.

Samuel Lai described the result as the first truly independent validation of the Event Horizon Telescope image, an important step in the scientific process. Samuel Lai, team leader and researcher with Australia’s Commonwealth Scientific and Industrial Research Organisation, said it compares to photographing a doughnut placed on the Moon, seen from a backyard. The analogy illustrates the extreme difficulty of imaging M87* from Earth.

Laurent Loinard, Director of the EHT Collaboration, said the independent method recovers the ring structure across several years in agreement with published EHT results. Studies like this strengthen confidence in what has been learned about black holes and enrich the toolkit available to the whole community. Loinard called the independent validation gratifying and valuable for novel results.

Laurent Loinard, Director of the EHT Collaboration, said the results strengthen confidence that the ring surrounding M87* is a genuine astrophysical feature rather than an artifact introduced during image reconstruction. The independent check adds support for general relativity in extreme gravity. The validation could settle questions about how much the final M87* image was influenced by atmospheric effects and processing methods used to reconstruct pictures from telescope data. Scientists have questioned the degree to which calibration errors and subjective decisions shaped the image.

A complementary imaging method based on traditional statistical techniques also validated the M87* structure using the new closure invariant data. The separate check demonstrated that the success did not depend solely on machine learning. The CSIRO technique could improve medical imaging and earthquake mapping by combining data from multiple sensors to map unknown structures.

The CSIRO method described in Nature cuts down on calibration errors and subjective image-processing choices that have troubled astronomers for many years. Scientists have questioned how much the final M87 image may have been influenced by processing methods used to reconstruct pictures from sparse telescope measurements. The CSIRO team built mathematical algorithms to eliminate these biases and calibration errors.

Share

Was this helpful?

Join the discussion

Please read our Comment Policy before commenting. First-time comments are held for moderation. By posting a comment you agree to our Terms of Use and Privacy Policy, which explain how your name, email address and comment are used.

Your email address will not be published. Required fields are marked *