Every other method here is indirect — an inference from a dip, a wobble, a delay. This one collects photons that came off the planet.
The obstacle is contrast. A Jupiter next to a Sun-like star is something like a billion times fainter and sits a fraction of an arcsecond away: a firefly beside a lighthouse, seen from another city.
Two tricks make it possible. A coronagraph masks the star inside the instrument, and adaptive optics reshapes a deformable mirror hundreds of times a second to undo the atmosphere's blurring. What survives is a faint dot beside a black disc.
The planets that show up are young — tens of millions of years old — still glowing from the heat of their own formation, and far from their star. An old, cool planet like Jupiter today is beyond reach.
The arithmetic
Contrast required
C = Fp / F*
~10-9 for a mature Jupiter in reflected light; ~10-6 for a young, self-luminous one in the infrared. That difference is the whole reason the method finds only young planets.
Smallest separation resolvable
θ ≈ 1.22 λ / D
The diffraction limit. An 8 m telescope at 1.6 µm reaches ~0.05 arcsec — about 5 AU at a distance of 100 light years.
Temperature from the light
L = 4πR²σT4
The planet's own thermal glow gives its temperature directly, and its spectrum gives the atmosphere — no starlight needed as a backdrop.
Why so few, and which ones
HR 8799 system
4 planets, 15–70 AU, ~30 Myr old
all four imaged, orbits tracked since 2008
Contrast achieved
~10-6 at 0.5 arcsec
enough for young giants, not for Earths
Typical host
young, nearby, often a bright A-type star
the population is tiny and very unrepresentative
Under a hundred planets in the catalogue came this way, but each one can be studied in a depth the other methods cannot approach — you can watch it move along its orbit and read its atmosphere directly.
What it does well
The planet's own light: temperature, weather and composition become measurable
Works at wide separations, where transits and wobbles are hopeless
The orbit can be watched unfolding over years
Where it cannot help
Only young, hot, massive planets far from their star
Needs the largest telescopes with adaptive optics
Orbits take decades to trace, so masses are poorly constrained
Fewer than a hundred planets found in total
Landmark discovery
HR 8799 (2008) — four giant planets photographed around one star, the first multi-planet system ever imaged.
Instruments
VLT/SPHERE, Gemini/GPI, Subaru/SCExAO, JWST; the Extremely Large Telescope next
Archive name
Imaging — the value in NASA's discoverymethod field