A Stef and Oren Lewandowski production
Credits
BlackHoles.fyi is a project by Oren and Stef Lewandowski, a father and son team in the UK. Oren, who is 17, co-writes the words and checks the physics. Stef makes the pictures.
The site draws black holes from the physics of how light and gas behave near them. You can explore one on the homepage, look at ten real black holes in the gallery, each drawn from its published numbers, or read the Concepts pages, which take one idea at a time with a figure you can change.
Why we made it
It started with a book. Oren read Black Holes: The Key to Understanding the Universe by the physicists Brian Cox and Jeff Forshaw, and wanted to see two things in it. One was a Penrose diagram, a map that fits the whole of space and time onto one page, with someone on it you could move about. The other was what it looks like to fly into a black hole, from outside and from inside.
Stef went looking for pictures of those. He found plenty of beautiful computer-drawn pictures, but very few that explain what you are seeing. The best teacher we have found on the web is still Falling into a Black Hole, a site of real diagrams for a general reader that the astrophysicist Andrew Hamilton made around 2000. As far as we know it hasn't been rebuilt since, so we started drawing our own.
The site is ours: Brian Cox and Jeff Forshaw had no part in it.
Cast and crew
- Pictures
- Stef Lewandowski
- Words
- Oren Lewandowski
- Physics checking
- Oren Lewandowski
- Spacetime
- Albert Einstein, 1915
- The size of a black hole's dark silhouette
- James Bardeen, 1973
- First picture of the glowing gas round a black hole
- Jean-Pierre Luminet, 1979
- Render farm
- Freckle and Auburn, the two Macs that render the stills and films
- Typefaces
- Literata, by TypeTogether, and Public Sans, from the US Web Design System
How we made it
AI agents do the building
Stef directs a team of AI coding agents: programs that write and test code from a written brief. Each agent takes one piece of work, writes it, tests it and adds it to a private test version of the site once every check passes. Stef decides what looks right, and nothing goes public until Oren has signed it off.
Fibers keeps track of the work
We used Stef's AI engineering toolkit, Fibers, to build this project. Fibers coordinates many AI coding sessions working on one project at once. Each session claimed a piece of work on a shared board before starting it, logged its progress there as it went, and recorded each decision with who made it and why. That kept two agents from picking up the same job, and it means a decision made weeks ago can still be traced back to its reason.
Research comes first
Before we draw a part of the picture, it gets a research note tied to the papers it rests on. We also keep a note of what we tried and why it looked wrong.
Every number has a paper
Each number in the code records what it rests on, usually a paper, and how far that paper was read. Each part of the picture says how real it is: worked out from the physics, following a published model, or chosen for the picture.
Design choices are written down
Stef keeps a written record of the design choices he has accepted and rejected, each dated and in his own words. The agents read it before they design anything, so a look he has turned down is less likely to come back.
Sites we learned from
- Falling into a Black Hole, Andrew HamiltonReal diagrams of black holes for a general reader, made around 2000.
- Space Time Travel, Ute Kraus and Corvin ZahnPictures and films of relativity by two physicists. We took the idea of giving the reader a task with real data.
- Black hole shader, Eric BrunetonA shader, a program that runs on a computer's graphics chip, that draws a black hole live in the web browser, with a paper explaining how. The most honest picture of a black hole we know of on the web.
- Bartosz Ciechanowski's articlesLong, interactive articles on how things work. We took his habit of a small figure before the prose that needs it.
- Black hole, Oskar SeiskariA black hole drawn by following rays of light, with its code shared openly. It keeps a list of its own artefacts, the places where its picture is known to be wrong.
- Black hole simulation, blackholesimulation.web.appA black hole simulation in the browser, with a panel saying what is real and what is simplified. Our pictures carry the same kind of note.
- Quantum Country, Andy Matuschak and Michael NielsenAn essay on quantum computing that asks you questions as you read, then again days later, to check you still know it.
- DistillAn online journal of machine learning research written to be read, with bylines, references and corrections on every article.
Sources
The same parts, with a page for each area of the picture, are in Research.
Where each part of the picture comes from (13 parts)
Most of each picture is made by tracing light. For every pixel, the computer follows a ray of light back from your eye, bending it as the black hole's gravity would, to find where it came from. Some parts are worked out from the physics in full, some follow a model another team published, some are chosen for the picture where nothing more exact is known or practical, and many are partly worked out and partly chosen.
The dark patch at the centre
Worked out from the physics
Astronomers call it the black hole's shadow.
Each pixel is traced along the light ray that reaches it, so the dark patch is traced, not drawn. For a spinning hole, its outline matches the shape the physicist James Bardeen worked out in 1973 to within 1%, at spins up to 0.99, where 1 is the fastest a black hole can spin.
- Bardeen 1973, Timelike and null geodesics in the Kerr metric, in Black Holes (Les Houches lectures)
The thin bright ring round the dark patch
Worked out from the physics
This is the photon ring.
It is light that circled the hole at least once before reaching you. It comes out of the same traced rays as the shadow, and nothing is drawn on top of it.
The shape of the glowing disc
Partly worked out from the physics
Gas falling into a black hole circles it in a flat disc, called an accretion disc, and glows as it heats up.
The inner edge sits at the innermost stable orbit for the hole's spin: the closest that gas can circle without falling in, which moves inward the faster the hole spins. The gas circles at the rate for that orbit. M is a unit of distance set by the hole's mass, so the same drawing fits every black hole at its own scale. The disc is drawn out to 1 500 M, where the disc's own gravity would break it into clumps (Goodman 2003).
How hot the disc is
Follows a published model
The disc's colour and brightness come from its temperature.
The temperature follows the standard model of a thin disc, one much thinner than it is wide, corrected for the strong gravity near the hole (Novikov and Thorne 1973). The black hole on the site's front page is 3C 273, a real quasar: a feeding black hole at the heart of a distant galaxy. Its mass, 260 million suns, and how fast it feeds, 1.2 times the Eddington rate, are its published values. The Eddington rate is the rate at which the light from the falling gas starts to push the rest of it away.
- Shakura and Sunyaev 1973, Black holes in binary systems: observational appearance, A&A 24, 337
- Novikov and Thorne 1973, Astrophysics of black holes, in Black Holes (Les Houches lectures)
- Page and Thorne 1974, Disk-accretion onto a black hole, ApJ 191, 499
Why one side of the disc is brighter
Worked out from the physics
The disc's light changes in colour and brightness on its way to you.
For each point of the disc we work out how far its light is shifted towards blue or red: by the gas moving towards or away from you, by its light climbing out of the hole's gravity, and by the hole's spin. Gas coming towards you looks brighter and bluer, so one side of the disc outshines the other.
How thick the disc is
Partly worked out from the physics
The disc is much thinner than it is wide, but not flat.
The thickness follows the standard thin-disc model (Shakura and Sunyaev 1973). Above and below its middle the gas thins out in the shape that gas held up by its pressure takes.
The threads of gas in the disc
Partly worked out from the physics
Fine streaks of brighter and fainter gas that curve round the disc.
The threads move at the gas's real orbital speed, so the faster inner gas draws each one into a trailing arc, as it does in a real disc, where the inner gas outruns the outer. We drew their shape to match the gas in published simulations, rather than simulating it ourselves.
- Guan, Gammie, Simon and Johnson 2009, Locality of MHD turbulence in isothermal disks, ApJ 694, 1010
- Ripperda and others 2022, Black hole flares: ejection of accreted magnetic flux through 3D plasmoid-mediated reconnection, ApJL 924, L32
- Dexter and others 2020, Sgr A* near-infrared flares from reconnection events in a magnetically arrested disc, MNRAS 497, 4999
- Event Horizon Telescope Collaboration 2021, First M87 Event Horizon Telescope Results. VIII. Magnetic field structure near the event horizon, ApJL 910, L13
- Narayan and others 2022, Jets in magnetically arrested hot accretion flows onto rotating black holes, MNRAS 511, 3795
- Tremaine 2001, Slow modes in stellar systems, AJ 121, 1776 (lopsided, m = 1, discs)
- Dhruv and others 2025, A survey of GRMHD models for black hole accretion systems, ApJS 277, 16 (the Illinois v3 simulation library, open data)
The stars and the Milky Way behind the hole
Partly worked out from the physics
The hole's gravity bends their light, as a lens does.
The bending is traced, and each bright star sits at its catalogued place with its measured colour. The Milky Way's glow is NASA's map of the sky from Earth, stretched for contrast, and the stars' brightness is chosen so you can see them. Nine of the real black holes draw the sky from where they are instead: five in our Galaxy (Cygnus X-1, Gaia BH1, Gaia BH3, V404 Cygni and GRS 1915+105) from Gaia's stars, Sgr A* from the star cluster round it and the dust of the Galaxy's centre, and three outside it (M87*, TON 618 and 3C 273) from their host galaxy's glow and its own stars, modelled from measurements (for TON 618, whose host has never been seen, from scaling alone).
- NASA Goddard Scientific Visualization Studio, Deep Star Maps 2020
- van Leeuwen 2007, Validation of the new Hipparcos reduction, A&A 474, 653
- Høg and others 2000, The Tycho-2 catalogue of the 2.5 million brightest stars, A&A 355, L27
- NASA HEASARC, the Hipparcos and Tycho-2 tables
- ESA 1997, The Hipparcos and Tycho Catalogues, volume 1
- Ballesteros 2012, New insights into black bodies, EPL 97, 34008
The sky as seen from a real black hole
Partly worked out from the physics
Some of the real black holes show the sky from where they are, not from Earth.
Seen from the hole, the Milky Way is not Earth's view. The stars come from Gaia, the European Space Agency's star-mapping satellite, moved to this hole's place. The glow between them is a model fitted to Earth's sky. Over it we lay the dark bands of dust that cross the Milky Way, as they look from Earth. Where Gaia cannot see, we use Earth's sky itself. A panel beside the hole's picture lists which parts are measured, modelled, chosen or borrowed. The four holes outside our Galaxy have no Gaia stars to move; for now their sky is dark, which the panel says, and the glow of their own galaxy is a later step.
- Gaia Collaboration, Vallenari and others 2023, Gaia Data Release 3: Summary of the content and survey properties, A&A 674, A1
- Bailer-Jones and others 2021, Estimating distances from parallaxes. V. Geometric and photogeometric distances to 1.47 billion stars in Gaia Early Data Release 3, AJ 161, 147
- Miller-Jones and others 2021, Cygnus X-1 contains a 21-solar-mass black hole: implications for massive star winds, Science 371, 1046
- Stirling and others 2001, A relativistic jet from Cygnus X-1 in the low/hard X-ray state, MNRAS 327, 1273
- El-Badry and others 2023, A Sun-like star orbiting a black hole, MNRAS 518, 1057
- Gaia Collaboration, Panuzzo and others 2024, Discovery of a dormant 33 solar-mass black hole in pre-release Gaia astrometry, A&A 686, L2
- Reid and others 2014, A parallax distance to the microquasar GRS 1915+105 and a revised estimate of its black hole mass, ApJ 796, 2
- Kormendy and others 2009, Structure and formation of elliptical and spheroidal galaxies, ApJS 182, 216
- Mei and others 2007, The ACS Virgo Cluster Survey XIII: SBF distance catalog and the three-dimensional structure of the Virgo Cluster, ApJ 655, 144
- Oldham and Auger 2016, Evidence for a massive, extended globular cluster system around M87, from the Next Generation Virgo Survey, MNRAS
- Tamura and others 2006, Wide-field photometry of the globular clusters of M87 with Subaru, Papers I and II, MNRAS 373, 588 and 601
- Bahcall and others 1997, Hubble Space Telescope images of a sample of 20 nearby luminous quasars, ApJ 479, 642
- Ren and others 2024, HST/STIS coronagraph images of 3C 273's host galaxy, A&A 683, L5
- Loveday and others 2012, Galaxy And Mass Assembly (GAMA): ugriz galaxy luminosity functions, MNRAS 420, 1239
- Yasuda and others 2001, Galaxy number counts from the Sloan Digital Sky Survey commissioning data, AJ 122, 1104
- Shen and others 2003, The size distribution of galaxies in the Sloan Digital Sky Survey, MNRAS 343, 978
- Driver and others 2016, Measurements of extragalactic background light from the far UV to the far IR from deep ground and space-based galaxy counts, ApJ 827, 108
- Kormendy and Ho 2013, Coevolution (or not) of supermassive black holes and host galaxies, ARA&A 51, 511
- Vijaykumar and others 2024, Inferring host-galaxy properties of LIGO-Virgo-KAGRA's black holes, ApJ 972, 157
- Artale and others 2020, Mass and star formation rate of the host galaxies of compact binary mergers across cosmic time, MNRAS 491, 3419
- Cao, Lu and Zhao 2018, Host galaxy properties of mergers of stellar binary black holes and their implications for advanced LIGO gravitational wave sources, MNRAS 474, 4997
- Mapelli and others 2018, the host galaxies of double compact objects merging in the local Universe, MNRAS 481, 5324
- Bell and de Jong 2001, Stellar mass-to-light ratios and the Tully-Fisher relation, ApJ 550, 212
- Kregel, van der Kruit and de Grijs 2002, Flattening and truncation of stellar discs in edge-on spiral galaxies, MNRAS 334, 646
- Perna and others 2018, Binary black hole mergers within the LIGO horizon: statistical properties and prospects for detecting electromagnetic counterparts, MNRAS 477, 4228
- Varma, Isi and Biscoveanu 2020, Extracting the Gravitational Recoil from Black Hole Merger Signals, PRL 124, 101104
- Peng and others 2006, Probing the coevolution of supermassive black holes and galaxies using gravitationally lensed quasar hosts, ApJ 649, 616
- van der Wel and others 2014, 3D-HST+CANDELS: the evolution of the galaxy size-mass distribution since z = 3, ApJ 788, 28
- Schödel and others 2014, The nuclear cluster of the Milky Way: total mass and luminosity (long version), A&A 566, A47
- Schödel and others 2018, The distribution of stars around the Milky Way's central black hole II: diffuse light from sub-giants and dwarfs, A&A 609, A27
- Gallego-Cano and others 2020, New constraints on the structure of the nuclear stellar cluster of the Milky Way from star counts and MIR imaging, A&A 634, A71
- Fritz and others 2016, The nuclear cluster of the Milky Way: total mass and luminosity, ApJ 821, 44
- Sormani and others 2022, Self-consistent modelling of the Milky Way's nuclear stellar disc, MNRAS 512, 1857
- Launhardt, Zylka and Mezger 2002, The nuclear bulge of the Galaxy III: large-scale physical characteristics of stars and interstellar matter, A&A 384, 112
- Gillessen and others 2017, An update on monitoring stellar orbits in the Galactic Center, ApJ 837, 30
- Habibi and others 2017, Twelve years of spectroscopic monitoring in the Galactic Center: the closest look at S-stars near the black hole, ApJ 847, 120
- Paumard and others 2006, The two young star disks in the central parsec of the Galaxy: properties, dynamics, and formation, ApJ 643, 1011
- Schödel and others 2010, Peering through the veil: near-infrared photometry and extinction for the Galactic nuclear star cluster, A&A 511, A18
- NASA Goddard Scientific Visualization Studio, Deep Star Maps 2020
- ESA 1997, The Hipparcos and Tycho Catalogues, volume 1
The circles drawn round the hole
Partly worked out from the physics
They mark the horizon (the point of no return), the photon sphere (where light can circle the hole) and the innermost stable orbit (the closest that gas can circle without falling in).
Their sizes are 2 M, 3 M and 6 M, the ones the physics gives. M is a unit of distance set by the hole's mass, so the same drawing fits every black hole at its own scale. They are drawn as if light travelled straight, so they show where each sphere is, not how it would look.
The object you can drop in
Partly worked out from the physics
A small object falling towards the hole, so you can see how its fall looks from far away.
Its fall, the time on its own clock and on yours, the delay of its light and its fading are worked out. It is drawn as a dot whose size and colour are chosen, and the fainter images its light makes by circling the hole are left out, so a real one would fade more slowly.
- Ames and Thorne 1968, The optical appearance of a star that is collapsing through its gravitational radius, ApJ 151, 659
- Cardoso, Duque and Foschi 2021, The light ring and the appearance of matter accreted by black holes, PRD 103, 104044
- Yoshino, Takahashi and Nakao 2019, How does a collapsing star look?, PRD 100, 084062
- Chen and others 2022, Appearance of an infalling star in black holes with multiple photon spheres
- Podurets 1964, Astronomicheskii Zhurnal 41, 1090
- Cunningham 1975, Optical appearance of distant objects to observers near and inside a Schwarzschild black hole, PRD 12, 323
- Gralla, Lupsasca and Strominger 2018, Observational signature of high spin at the Event Horizon Telescope, MNRAS 475, 3829
- Johnson and others 2020, Universal interferometric signatures of a black hole's photon ring, Science Advances 6, eaaz1310
The paths of light
Partly worked out from the physics
Lines that show the routes light takes past the hole on its way to you.
Each path is found the same way the picture is: by following light back from your eye and bending it as the hole's gravity would. It is then drawn where it really runs in space, as if the hole's gravity did not bend your view of it. Paths stop at the middle of the glowing disc of gas that circles the hole. The picture gives that disc some thickness, so near its rim a path can end a little off what you see.
The black hole cut in half
Partly worked out from the physics
A view with the near half of the hole and its disc of gas cut away. Lines mark each region on the cut: the ergosphere, where nothing round a spinning hole can stay still; the corona, very hot gas just above the disc; the jet, a beam of gas fired out along the spin axis; the magnetic field threading the gas; and, for a hole that takes in only a little gas, the cloud of hot gas it has in place of a disc. Inside the horizon sit a second, inner horizon and the singularity, a ring where the maths of gravity gives no answer.
The gas on your side of the cut is left out, so the cut face is lit by the disc's gas, not painted on. The lines sit where each region is, as if light travelled straight. The horizon, the ergosphere and the disc's inner edge follow from the maths for the spin the hole is drawn with. The disc's thickness, the cloud's width, the corona's reach and the jet's edge come from the models the rest of the picture is drawn with, and the magnetic field is a pattern we drew, not one worked out. The inner horizon and the ring come from the idealised maths.
- Tchekhovskoy, McKinney and Narayan 2008, Simulations of ultrarelativistic magnetodynamic jets from gamma-ray burst engines, MNRAS 388, 551
- Narayan and others 2022, Jets in magnetically arrested hot accretion flows onto rotating black holes, MNRAS 511, 3795
- Porth and others 2019, The Event Horizon General Relativistic Magnetohydrodynamic Code Comparison Project, ApJS 243, 26
- Ripperda and others 2022, Black hole flares: ejection of accreted magnetic flux through 3D plasmoid-mediated reconnection, ApJL 924, L32
- Begelman, Scepi and Dexter 2022, What really makes an accretion disc MAD, MNRAS 511, 2040
- Tchekhovskoy, Narayan and McKinney 2011, Efficient generation of jets from magnetically arrested accretion on a rapidly spinning black hole, MNRAS 418, L79
- McKinney, Tchekhovskoy and Blandford 2012, General relativistic magnetohydrodynamic simulations of magnetically choked accretion flows around black holes, MNRAS 423, 3083
- Hamilton and Avelino 2010, The physics of the relativistic counter-streaming instability that drives mass inflation inside black holes, Physics Reports 495, 1
More parts of the pictures are added here as each is checked against the code that draws it.
The real black holes (10 black holes)
Black holes that astronomers have measured, each drawn from its published numbers, with the count of papers those numbers come from.
- Gaia BH1, 2 papers
- Gaia BH3, 1 paper
- GW150914 remnant, the black hole left behind by the first merger of two black holes ever detected, 3 papers
- TON 618, 6 papers
- 3C 273, 5 papers
- Sagittarius A*, 7 papers
- M87*, 8 papers
- Cygnus X-1, 19 papers
- GRS 1915+105, 10 papers
- V404 Cygni, 9 papers
Data (4 sources)
The star maps and catalogues the pictures use, credited as each source asks.
- Deep Star Maps 2020. NASA/Goddard Space Flight Center Scientific Visualization Studio; Gaia DR2: ESA/Gaia/DPAC. Used for the Milky Way behind the hole.
- The Hipparcos and Tycho-2 star catalogues. van Leeuwen 2007 and Høg and others 2000, from NASA's HEASARC archive. Used for the bright stars, placed one by one.
- The SXS catalogue of black hole mergers. The SXS Collaboration (Simulating eXtreme Spacetimes), under the CC BY 4.0 licence, which lets anyone reuse it with credit. Used for the Concepts page where two black holes spiral together and merge.
- SIMBAD. The reference database of named objects in the sky, run by CDS, Strasbourg. Used for the real holes' names and places in the sky.
Software (12 projects)
The open-source code the site is built with, and the licence each is shared under, which says how others may reuse it.
- three.js, MIT
- React Three Fiber and drei, MIT
- React, MIT
- React Router, MIT
- MDX, MIT
- Vite, MIT
- Tailwind CSS, MIT
- Motion, MIT
- Zustand, MIT
- hls.js, Apache 2.0
- Bun, MIT
- Vitest, MIT
Further reading (2 works)
The book this site began with, and the site that showed us black holes can be explained with real diagrams.
- Black Holes: The Key to Understanding the Universe, Brian Cox and Jeff Forshaw
- Falling into a Black Hole, Andrew Hamilton
After the credits
Read next: Why a black hole looks black