Black Hole Vision: Imaging and Analysis

• Black Hole Vision is defined as the ensemble of observational, geometric, and numerical practices that render black holes visible through null geodesics and gravitational lensing effects.
• Recent advances with EHT and BHEX demonstrate horizon-scale imaging that reveals detailed structures like photon rings, shadows, and event-horizon silhouettes.
• Comprehensive ray-tracing simulations and analytic models offer quantitative tools to distinguish Kerr black holes from alternative spacetime metrics.

Black Hole Vision is the ensemble of observational, geometric, numerical, and pedagogical practices through which black holes are rendered legible by the behavior of light in strong gravity. In current literature, the expression spans the Black Hole Explorer mission concept centered on the photon ring, theoretical work on the shadow, inner shadow, and lensed event-horizon image, full general-relativistic ray-tracing of accretion flows and all-sky environments, and an interactive iOS application that synthesizes black-hole-lensed views from live camera feeds. Across these contexts, the common premise is that black holes are not seen as luminous surfaces, but through null geodesics, photon capture, repeated lensing, Doppler and redshift effects, and the geometry-dependent image structures these processes generate near the horizon (Galison et al., 2024, Johnson et al., 2024, Berens et al., 6 Mar 2026).

1. From indirect visibility to event-horizon-scale imaging

Black-hole vision begins from a standard relativistic constraint: the event horizon is not a material surface and cannot be observed directly as an emitting boundary. What becomes observable is the way curved spacetime bends, captures, delays, and redshifts radiation from accretion flows, jets, background sources, or infalling matter. This is why black-hole imaging has always been as much a problem in general-relativistic optics as in instrumental astronomy. Early numerical simulations already showed the now-canonical crescent-like appearance of a black hole with an accretion disk, including a central dark region magnified by gravitational lensing and strong brightness asymmetry from gravitational redshift and Doppler boosting (Luminet, 2018).

The observational transition from theory to measurement occurred through very-long-baseline interferometry at millimeter wavelengths. The Event Horizon Telescope does not produce a direct photograph in the ordinary sense; it records signals at widely separated antennas, synchronizes them with atomic clocks to sub-nanosecond precision, and reconstructs an image from correlated data. One night can generate about 2 petabytes of data. In this framework, black-hole vision is a reconstructed view of horizon-scale structure rather than an optical image in the classical sense (Luminet, 2018).

Recent observations of M87* and Sgr A* established that event-horizon-scale structure can be imaged, revealing a dark central region and a bright, asymmetric ring at millimeter wavelengths. Complementary measurements, such as GRAVITY’s center-of-light astrometry of infrared flares near Sgr A*, reinforce the interpretation that the observed emission originates very close to the black hole. The field has therefore shifted from asking whether black holes can be imaged at all to asking which geometric structure a given image is tracing, and with what precision it can constrain the underlying spacetime (Narayan et al., 2023).

2. Shadow, photon ring, inner shadow, and event-horizon image

The main interpretive difficulty in black-hole vision is that several distinct image features can produce superficially similar dark or bright structures. The literature therefore separates geometric capture boundaries, lensed horizon projections, and higher-order lensing features rather than treating every central depression as “the shadow.”

Feature	Geometric definition	Typical visibility regime
Classical shadow	Capture cross-section on the observer’s sky, bounded by rays that asymptotically approach photon spheres	Distant luminous background beyond the photon spheres
Photon ring	Very thin, bright, highly lensed image formed by photons that orbit the black hole one or more times before escaping	Optically thin near-horizon emission
Inner shadow	Region whose rays hit the horizon without crossing the equatorial plane even once	Equatorial emission extending to the horizon
Event-horizon silhouette / image	Gravitationally lensed image or projection of the event horizon, nested inside the classical shadow	Inner accretion flow or plunging matter near the horizon

In Schwarzschild spacetime, the classical shadow radius is $r_{\rm sh}=3\sqrt{3}\simeq 5.2$, while the event-horizon silhouette is smaller, with $r_{\rm eh}\simeq 4.457$. A distinct projected radius $r_{\rm EW}\simeq 2.848$ corresponds to the nearest hemisphere of the horizon. These values underpin a central claim of the silhouette literature: the horizon-associated dark region is not identical to the full photon-capture shadow (Dokuchaev et al., 2020, Dokuchaev, 2024).

This distinction becomes observationally consequential in the interpretation of M87*. One line of analysis argues that the first EHT image does not show the classical shadow in a straightforward sense, but rather a lensed dark image of the southern hemisphere of the event-horizon globe, with a size compatible with a thin-disk silhouette for $a\gtrsim 0.75$; in that reading, the classical shadow is “invisible at all” in the image (Dokuchaev et al., 2020). A different line of work separates the photon ring from a smaller inner shadow and shows, using semi-analytic models and MAD GRMHD simulations, that both may be present simultaneously in submillimeter images of M87*, while current observations remain limited primarily by dynamic range rather than by the existence of the structures themselves (Chael et al., 2021).

A more radical horizon-centered interpretation treats the black hole image as the gravitationally lensed image of the event horizon itself, projected onto the observer’s sky and located inside the classical shadow. In that framework, what is seen is not merely a region where light is absent, but a general-relativistic projection of the horizon globe reconstructed from photons emitted very near it (Dokuchaev, 2024, Dokuchaev et al., 2018). A separate perceptual analysis reaches a related but distinct conclusion: for a semi-transparent spherical shell approaching the Schwarzschild radius, affine and binocular distance compress the apparent depth so strongly that the object is thought of as a two-dimensional membrane by outside observers restricted to those distance measures (Lin, 2019).

3. Null geodesics, unstable photon orbits, and the geometry of vision

The geometric substrate of black-hole vision is the null geodesic structure of the spacetime. In Schwarzschild, the relevant unstable light orbit lies at

$r \equiv 3GM/c^2,$

the photon sphere. In Kerr, the corresponding set is not a single spherical surface but a photon shell: a family of unstable spherical photon orbits whose image-plane projection depends on spin and inclination. This difference is responsible for the transition from a perfectly circular ring in the non-rotating case to asymmetric, thickness-dependent, and brightness-dependent structures in the rotating case (Galison et al., 2024).

The Kerr metric is frequently written in Boyer–Lindquist coordinates $(t,r,\theta,\phi)$ with

$$\Delta = r^2 - 2Mr + a^2, \qquad \Sigma = r^2 + a^2 \cos^2\theta.$$

In Kerr–Newman form, the same imaging program is extended by adding charge through

$\Delta=r^2-2Mr+a^2+e^2,$

with horizon radii

$r_{\pm}=1\pm\sqrt{1-a^2-q^2}$

in dimensionless units $GM/c^2=1$. In these formulations, Carter separability reduces the geodesic problem to the radial and polar potentials $r_{\rm eh}\simeq 4.457$0 and $r_{\rm eh}\simeq 4.457$1, and the observer’s image is parameterized by the impact parameters $r_{\rm eh}\simeq 4.457$2 and $r_{\rm eh}\simeq 4.457$3 or $r_{\rm eh}\simeq 4.457$4, depending on convention (Dokuchaev, 2024).

The BHEX visualization program makes an additional conceptual move by emphasizing instability itself. In its treatment, nearby photon trajectories around a spinning black hole can diverge rapidly, with their separation controlled by the Lyapunov exponent. Photons are divided into critical, subcritical, and supercritical classes: critical photons asymptotically remain in orbit, subcritical photons fall into the black hole, and supercritical photons escape to infinity. The photon ring is therefore the boundary between capture and escape, and its self-similar subrings, indexed by $r_{\rm eh}\simeq 4.457$5, encode the instability of near-horizon null geodesics. This is the basis of the paper’s description of the photon ring as a probe of “spacetime chaos” (Galison et al., 2024).

The same work links ray dynamics to visual geometry through embedding diagrams. In the Schwarzschild equatorial slice, the embedded surface in Euclidean $r_{\rm eh}\simeq 4.457$6 satisfies

$r_{\rm eh}\simeq 4.457$7

while the Kerr case must be obtained numerically. The conceptual role of such embeddings is not calculational efficiency but visual mediation: they turn abstract curvature into a shape that can be followed alongside the lensed trajectories of photons (Galison et al., 2024).

4. Interferometric black-hole vision: EHT, BHEX, and space VLBI

Present black-hole vision is fundamentally interferometric. On the ground, the EHT reaches a longest baseline of about $r_{\rm eh}\simeq 4.457$8, giving a resolution of roughly $r_{\rm eh}\simeq 4.457$9 at 230 GHz,

$r_{\rm EW}\simeq 2.848$0

That is sufficient to resolve the broad shadow-scale structure of M87* and Sgr A*, but not the narrowest substructure associated with the photon ring (Johnson et al., 2024).

The Black Hole Explorer is proposed as a hybrid space-ground extension of submillimeter VLBI. Its core concept is an orbiting millimeter telescope at about 20,000 km altitude operating together with terrestrial millimeter radio telescopes, including the Event Horizon Telescope and its future extensions. By extending the baseline beyond Earth’s diameter, BHEX is designed to discover and measure the bright and narrow photon ring around M87* and Sgr A*, and to use its detailed shape and thickness as direct tracers of near-horizon geometry rather than of the more complex accretion flow (Galison et al., 2024).

In the BHEX science case, the photon ring is a visibility-domain observable as well as an image-domain one. In the optically thin regime appropriate for M87* and Sgr A* at submillimeter wavelengths, $r_{\rm EW}\simeq 2.848$1 of the flux density comes from the photon ring, and this flux falls as $r_{\rm EW}\simeq 2.848$2 on long baselines as the ring is increasingly resolved. Because the ring shape depends on Kerr geometry, BHEX is positioned as a route to direct spin measurement, with simulated Bayesian fits suggesting constraints at the level of $r_{\rm EW}\simeq 2.848$3 accuracy even accounting for astrophysical systematics. The mission is also intended to extend horizon-scale imaging to dozens of additional supermassive black holes and to connect black holes to their relativistic jets (Johnson et al., 2024).

The mission architecture is correspondingly specialized. The hardware concept includes a rigid 3.5 m antenna with $r_{\rm EW}\simeq 2.848$4 surface accuracy, dual-frequency receivers spanning 240–320 GHz and 80–106 GHz, frequency phase transfer between low and high bands, cryogenic cooling near the quantum noise limit, and a 100 Gb/s optical downlink. The Japanese mission study adds a baseline fringe-sensitivity goal of $r_{\rm EW}\simeq 2.848$5 mJy, observing windows centered on M87* and Sgr A*, and a potential single-dish mode in the 50–70 GHz band for molecular-universe science, including molecular oxygen (Akiyama et al., 2024).

5. Numerical, analytic, and immersive rendering

The computational realization of black-hole vision ranges from full radiative-transfer pipelines to compact parametric image models. At the most complete end, general-relativistic magnetohydrodynamical simulations and covariant radiative transfer are combined to render the accretion flow, jet, shadow, and lensed sky self-consistently. A notable example is a full $r_{\rm EW}\simeq 2.848$6 steradian virtual-reality movie built from BHAC simulations and post-processed with RAPTOR. In that framework, an observer is placed inside the flow, rays are launched over the full celestial sphere using an orthonormal tetrad, and images are generated at $r_{\rm EW}\simeq 2.848$7 GHz, $r_{\rm EW}\simeq 2.848$8 GHz, $r_{\rm EW}\simeq 2.848$9 GHz, and $a\gtrsim 0.75$0 GHz. The resulting movie has $a\gtrsim 0.75$1 frames, and at closest approach the observer encounters a local luminosity of about $a\gtrsim 0.75$2. Because some rays orbit the black hole one or more times before reaching the observer, multiple repeated colored patches appear directly in the rendered sky (Davelaar et al., 2018).

At a more abstract level, sparse-VLBI image reconstruction motivates analytic models with closed visibility-domain representations. One such approach replaces a full pixel grid with an eccentric ring carrying a brightness gradient plus a two-dimensional Gaussian. The nine-parameter “xringaus” model is fit to visibility amplitudes and closure phases using a Metropolis–Hastings MCMC sampler with replica exchange. Its purpose is not to reproduce plasma microphysics in detail but to estimate shadow size, crescent asymmetry, and image orientation directly from limited $a\gtrsim 0.75$3-plane coverage (Benkevitch et al., 2016).

These two approaches define complementary poles of black-hole vision. Full GRMHD plus radiative transfer emphasizes physical completeness, time dependence, and observer-state dependence; analytic geometric modeling emphasizes computational efficiency, direct comparison with interferometric observables, and uncertainty quantification when data are sparse. The field routinely uses both modes, often in the same interpretive pipeline (Davelaar et al., 2018, Benkevitch et al., 2016).

6. Black Hole Vision as interactive software and quantitative laboratory

Black Hole Vision also denotes a specific interactive platform: an open-source iOS application that synthesizes black-hole-lensed views from live camera inputs. The app combines video feeds from the front- and rear-facing iPhone cameras, projects them onto a notional source sphere at radius $a\gtrsim 0.75$4, maps screen pixels to image-plane coordinates $a\gtrsim 0.75$5 through

$a\gtrsim 0.75$6

and then traces null geodesics backward in Schwarzschild or Kerr spacetime to determine where each ray intersects the source sphere. In the Kerr implementation, the conserved quantities are

$a\gtrsim 0.75$7

with

$a\gtrsim 0.75$8

The app is explicitly designed to highlight the shadow, repeated subimages, and photon-ring-like distortions expected in BHEX-scale black-hole imaging (Berens et al., 6 Mar 2026).

A subsequent pedagogical paper converts the same platform into a quantitative laboratory for Schwarzschild spacetime. There, the phone is mounted on a tripod, pointed at a calibrated meter stick or a 2D grid, and used to recover the simulated mass from four independent probes: weak-field lensing slope, shadow / critical-curve mass, Einstein-ring mass, and shadow-capture / vanishing-point mass. The measurements are combined through inverse-variance weighting,

$a\gtrsim 0.75$9

yielding a best-estimate mass of about $r \equiv 3GM/c^2,$0 mm in geometrized units. The same study analyzes the lensing Jacobian, integrated coordinate length, exponential demagnification of higher-order echoes, and shadow circularity, reporting agreement in integrated coordinate length at about $r \equiv 3GM/c^2,$1 and an axial ratio of

$r \equiv 3GM/c^2,$2

It also isolates non-physical artifacts, identifying the “4-leaf clover” shadow in high-magnification mode as a Cartesian grid aliasing artifact and tracing small asymmetries to a measured camera tilt of about $r \equiv 3GM/c^2,$3 (Burko, 21 May 2026).

This pedagogical development is significant because it reframes black-hole vision from qualitative visualization to measurement practice. The app-generated data are not astrophysical observations, but the workflow reproduces many of the same metrological distinctions that arise in real imaging: critical curves versus horizon-associated structures, physical signatures versus sampling artifacts, and geodesic instability versus rendering bias (Burko, 21 May 2026).

7. Metric discrimination, alternative spacetimes, and unresolved interpretation

Because black-hole vision links image morphology to geodesic structure, it is routinely used to compare Kerr with non-Kerr metrics. For regular black holes illuminated by a thin Novikov–Thorne disk, the Schwarzschild, Bardeen, and Hayward cases all produce direct and secondary disk images, but the Hayward spacetime is distinguished by a characteristic minimal distance between the innermost region of the direct image and the outermost region of the secondary image. The same study reports that increasing inclination angle separates higher-order images more strongly, making image-order structure itself a discriminant (Guo et al., 2023).

When a spherically symmetric black hole is embedded in a Hernquist-type dark matter halo, the geodesic structure can differ substantially from isolated Schwarzschild, yet the observable consequences in the astrophysically relevant low-compactness regime remain modest. The primary and secondary tracks of isotropic orbiting sources, and the width, location, and relative luminosity of the corresponding photon rings, undergo only minor modifications. The authors conclude that this “troubles distinguishing between both geometries using present observations of very-long baseline interferometry” (Macedo et al., 2024).

In 4D Einstein–Gauss–Bonnet gravity, by contrast, the visual trends are more systematic. Under a celestial light sphere, increasing the coupling parameter $r \equiv 3GM/c^2,$4 decreases the shadow radius and the corresponding photon ring, while increasing the spin parameter $r \equiv 3GM/c^2,$5 enhances asymmetry and frame dragging. Under a thin accretion disk with the inner edge extended to the event horizon, the inner shadow gradually decreases with $r \equiv 3GM/c^2,$6 and becomes progressively asymmetric with $r \equiv 3GM/c^2,$7. The study further reports consistency of the considered rotating EGB model with current M87* and Sgr A* constraints for the explored parameter range (Aslam et al., 18 Apr 2026).

A plausible implication is that black-hole vision is becoming a comparative science of metrics rather than a single-image enterprise. The same image plane now hosts several nested questions: which emitting geometry is dominant, which dark feature is being identified, how much of the brightness pattern is universal versus plasma-dependent, and whether current angular resolution and dynamic range are sufficient to discriminate nearby alternatives to Kerr. The literature does not collapse these questions into one definition, but it increasingly treats them as aspects of a common program: using lensed light to make strong-field spacetime empirically legible.