Skip to main content
space-iconSpace and Physicsspace-iconAstronomy
clock-iconPUBLISHED36 minutes ago

Incredible Video Of Black Hole Jet Took 30 Years Of Data To Make And Is Highest Definition View Yet

Stitched together using a neural network, the work provides new insights into how the jet appears to move faster than light.

Dr. Alfredo Carpineti headshot

Dr. Alfredo Carpineti

Alfredo has a PhD in Astrophysics and a Master's in Quantum Fields and Fundamental Forces from Imperial College London.

Space & Physics Editor

Alfredo has a PhD in Astrophysics and a Master's in Quantum Fields and Fundamental Forces from Imperial College London.View full profile

Alfredo has a PhD in Astrophysics and a Master's in Quantum Fields and Fundamental Forces from Imperial College London.

View full profile
EditedbyKaty Evans
Katy Evans headshot

Katy Evans

Deputy Editor-In-Chief

Katy has a BA in Humanities and Philosophy, with over 20 years of experience in online and print publishing. She was named the Association of British Science Writers' Editor of the Year in 2023.

A still from the video showing the changing shape and structure of the jet

Something was going with this jet in 1998.

Image Credit: Foschi et al / Nature


It's taken nearly 30 years of data and a new machine-learning model, but we now have the highest-resolution video of a supermassive black hole's jet yet.

Galaxy 3C 345 is known as a blazar, an extremely active type of galaxy with a nucleus that is releasing a powerful jet of material, including gas and plasma, at nearly the speed of light. The source of that jet is a supermassive black hole, possibly orbited by a second one, creating a variable jet that has fascinated astronomers for decades. And now you can see it in motion. 

The team achieved this high-definition video using 116 images taken between 1995 and 2022 that were then stitched together by Kine, a machine learning algorithm based on neural networks that can sharpen these cosmic images. Several team members had also worked on the Event Horizon Telescope and used a similar one on the first-ever picture of a supermassive black hole.

three panels showing the changing shape and structure of the jet
The video shows three views of the jets: the total intensity, the polarization (a proxy for the magnetic field), and the optical flow (the projected velocity).
Image credit: Foschi et al / Nature

The high resolution has allowed the team to study the speed of the plasma in incredible detail. 

Something that seems absolutely impossible about this jet is that the material being ejected seems to move faster than the speed of light. This is impossible as far as we know. So what is going on here?

The culprit is geometry. Everything we see in space appears projected on the two-dimensional celestial vault. But things move in 3D. Due to the angle at which they appear, some of these jets (which are moving very close to the speed of light) appear to go many times faster than that.

In particular, the team measured that most of the gas is moving between 9 and 12 times the speed of light, but surprisingly the brightest regions are only moving 10 to 13 times the speed of light.

This is surprising because it was expected that the bright regions were shock fronts, so they should have been moving much faster than the surrounding material. This doesn’t seem to be the case at all.

"We found no evidence that traveling bright components are strongly shocked regions, as previously proposed, because the component speeds are of the same order as the average flow speed in the same regions. The shock interpretation is also disfavored by the absence of correlation between the bright features and the peaks in the fractional polarization," the authors wrote in the paper.

The shock front theory is not invalidated by this one case, but these observations will certainly raise eyebrows about exactly what goes on in the mysterious relativistic jets.

The key to this discovery is the reconstruction of the motion of the jet, using the neural network. This has shown the researchers how the plasma is moving throughout, rather than going photo by photo. It delivers about four times higher resolution than standard approaches and a whopping 140 times more contrast.

It's likely this machine-learning algorithm will be used when we finally get the first-ever video of a black hole's event horizon.  

The study is published in the journal Nature.


Add us as a Google preferred source to see more of our
trusted coverage in Search