This site computes colours. It does not measure anything. Every number it starts from, every model spectrum, every real photograph and every line of the colour maths comes from someone else's published work, listed here in full. Where a source asks to be acknowledged in particular words, those words appear below exactly as asked.
Data, model grids and code. Between them these produce every swatch on this site; without them there would be nothing to compute.
Every number about a planet and its star — how big, how hot, how far out, what kind of star it orbits — comes from here. It is the catalogue the whole site is built on: without it there is nothing to compute a colour for.
CiteChristiansen et al. (2025), PSJ, doi:10.3847/PSJ/ade3c2 — doi.org/10.3847/PSJ/ade3c2
This research has made use of the NASA Exoplanet Archive, which is operated by the California Institute of Technology, under contract with the National Aeronautics and Space Administration under the Exoplanet Exploration Program.
A set of precomputed model atmospheres: what fraction of light a Jupiter- or Neptune-like planet reflects at each wavelength, worked out for a range of distances from the star and cloud states. One of the two engines behind the modelled spectra, and the reference marks under the model-space slider.
CiteCahoy, Marley & Fortney (2010), ApJ 724, 189, doi:10.1088/0004-637X/724/1/189 — doi.org/10.1088/0004-637X/724/1/189
Deriving colours from it is uncontroversial; redistributing the grid files is a separate act we have not cleared. Excluded from any dataset deposit.
Real measurements — Jupiter, Saturn, Uranus and Neptune, recorded through a telescope at ESO in 1995. Four of the five solar-system anchors that let you check this site's method against planets whose colours we already know.
CiteKarkoschka (1998), Icarus 133, 134-146, doi:10.1006/icar.1998.5913 — doi.org/10.1006/icar.1998.5913
The measured spectrum of Earth, assembled from spacecraft and earthshine observations. It is the fifth anchor, and the strictest test on the site: if our pipeline got Earth's colour wrong, nothing else here would be worth reading.
CitePayne, Villanueva, Kofman et al. (2026), PSJ, doi:10.3847/PSJ/ae2feb — doi.org/10.3847/PSJ/ae2feb
Redistributed verbatim under CC BY 4.0, (c) the authors, without warranties of any kind (see section 5 of the licence). Interpolated onto our 5 nm grid; otherwise unmodified.
NASA's open-source radiative-transfer code. Given a planet's gravity, temperature and atmosphere it computes the reflected-light spectrum properly, from physics rather than from a lookup table. Used for selected targets.
CiteBatalha et al. (2019), ApJ 878, 70, doi:10.3847/1538-4357/ab1b51 — doi.org/10.3847/1538-4357/ab1b51
PICASO 4.0.1, with the opacity database from Zenodo record 14861730 — both are needed to reproduce the committed spectra, and the opacity database carries its own citation. See docs/picaso-runbook.md.
The finding that smaller giant planets hold more metal-rich atmospheres than bigger ones. It is what stops every modelled planet at the same temperature coming out the same colour — for most of the catalogue, this paper is quietly setting the chemistry.
CiteThorngren, Fortney, Murray-Clay & Lopez (2016), ApJ 831, 64, doi:10.3847/0004-637X/831/1/64 — doi.org/10.3847/0004-637X/831/1/64
Smaller planets get more metal-rich atmospheres, which is why a Neptune-mass world and a Jupiter-mass world at the same temperature do not come out the same colour here.
The work that showed clouds do not simply burn away as a planet gets hotter. Different substances condense at different temperatures, so a planet's sky can clear and then cloud over again further up the thermometer — around 1,600 K it is hot enough for rock itself to condense into a bright haze. It is why some of the hottest worlds here are pale rather than dark.
CiteParmentier, Fortney, Showman, Morley & Marley (2016), ApJ 828, 22, doi:10.3847/0004-637X/828/1/22 — doi.org/10.3847/0004-637X/828/1/22
Clouds are not simply burned off as a planet gets hotter: each condensate is stable over its own narrow temperature range, so the sky clears and then clouds over again. This paper predicts the silicate-cloud brightening between roughly 1,600 and 1,900 K that gives planets like Kepler-7 b their high measured albedo.
The measurement of how much light Kepler-7 b actually reflects — about a third of it, far more than a scorched planet has any right to. It is the real number our model of hot, cloudy planets is set against, and the reason we know the old version of that model was wrong.
CiteDemory, Seager, Madhusudhan et al. (2011), ApJL 735, L12, doi:10.1088/2041-8205/735/1/L12; Demory, de Wit, Lewis et al. (2013), ApJL 776, L25, doi:10.1088/2041-8205/776/2/L25 — doi.org/10.1088/2041-8205/735/1/L12
Kepler-7 b's geometric albedo of 0.32 +/- 0.03 is the one measurement the hot-cloud model here is fitted to; the 2013 paper resolves the cloud as high-altitude, off-centre and probably silicate.
The description of how cloud droplets fall. A cloud is only visible if it stays up, so on a planet with weak gravity the deck floats high and bright, while stronger gravity pulls it down out of sight. It is why two equally hot planets here are not equally cloudy.
CiteAckerman & Marley (2001), ApJ 556, 872, doi:10.1086/321540 — doi.org/10.1086/321540
Whether a cloud deck reaches the visible atmosphere or rains out below it is a contest between settling and mixing, which is why low-gravity planets here end up cloudier than high-gravity ones at the same temperature.
The climate-model calculation of where a star's habitable zone begins and ends. It is the arithmetic behind the 'could there be liquid water' filter — which is about orbital distance only, and never about an atmosphere anyone has measured.
CiteKopparapu, Ramirez, SchottelKotte, Kasting, Domagal-Goldman & Eymet (2014), ApJL 787, L29, doi:10.1088/2041-8205/787/2/L29 — doi.org/10.1088/2041-8205/787/2/L29
Table 1 coefficients, for a 1 Earth-mass planet. The zone is orbital distance only: no atmosphere has been measured for any planet we mark.
The study that worked out which known exoplanets Roman's coronagraph could actually catch in reflected light. Its Table 4 is the target board: the shortlist, and the counts quoted on that page.
CiteCarrión-González, García Muñoz, Santos, Cabrera, Csizmadia & Rauer (2021), A&A 651, A7, doi:10.1051/0004-6361/202039993 — doi.org/10.1051/0004-6361/202039993
Table 4 — planets whose reflected light the Roman coronagraph could reach. The board quotes their scenario counts directly rather than recomputing them.
NASA's own specification of the Roman coronagraph's filters — which colours of light each one lets through. Every 'as Roman would see it' swatch on this site is the model spectrum pushed through those three filters and reassembled.
CiteRoman Coronagraph Instrument Primer, Community Participation Program, 8 January 2025, p. 5
We model the three flight bands — 575 nm, 730 nm and 825 nm — as top-hat filters at their nominal design widths (10%, 15%, 10%). That is a CONVENTION, and it matters: real filter profiles have sloped shoulders, and the as-built widths measured on the ground run one to two points wider than the nominal figures. Band 2 (660 nm) is on the filter wheel but was never characterised as a supported observing mode, so it is not modelled here.
The reference numbers for the solar system's own planets — size, orbit, distance from the Sun. The five anchors are not in the exoplanet catalogue, so theirs come here.
The anchors are not in the Exoplanet Archive, so their orbital numbers come from here instead.
The library that turns a spectrum into a colour: it carries the CIE 1931 colour-matching functions — the measured response of human vision — and the standard transform into sRGB. We use it rather than hand-rolled maths so the colour step is a published standard, not our arithmetic.
The official map of which patch of sky belongs to which constellation. It is why a planet page can tell you its host star is in Lyra, and why the sky chart can point you at the right part of the night.
CiteRoman (1987), PASP 99, 695 — CDS catalogue VI/42; VizieR: doi:10.26093/cds/vizier, 2000, A&AS 143, 23 — doi.org/10.26093/cds/vizier
Catalogue VI/42 is redistributed verbatim as pipeline/data/constellation_boundaries.dat. VizieR's terms require the original catalogue author and publication to be cited explicitly alongside VizieR itself, which is why Roman (1987) is named first here.
Two of our sources ask to be acknowledged in particular words. Here they are, unedited. Anything built on this site's data inherits them, because we inherit them:
NASA Exoplanet ArchiveThis research has made use of the NASA Exoplanet Archive, which is operated by the California Institute of Technology, under contract with the National Aeronautics and Space Administration under the Exoplanet Exploration Program.
VizieR / CDS, StrasbourgThis research has made use of the VizieR catalogue access tool, CDS, Strasbourg, France (DOI : 10.26093/cds/vizier). The original description of the VizieR service was published in 2000, A&AS 143, 23.
Two kinds of real image appear on this site: the surface maps that give the five solar-system worlds their real geography, and the telescope images of the handful of planets anyone has actually photographed. Neither sets a colour. Every map is rescaled so its average matches the colour the physics produced, and the telescope images are infrared and false-coloured — they are shown beside the swatch, never as it.
| Credit | Licence | Used on |
|---|---|---|
| NASA Earth Observatory (Reto Stöckli, Robert Simmon) surface map | Public domain | Earth |
| Solar System Scope (maps from NASA elevation and imagery data) surface map | CC BY 4.0 | Jupiter, Neptune, Saturn, Uranus |
| ESO / J. Rameau telescope image | CC BY 4.0 | HD 95086 b |
| ESO / Lagrange / SPHERE consortium telescope image | CC BY 4.0 | beta Pictoris b |
| NAOJ (National Astronomical Observatory of Japan) telescope image | NAOJ terms, credit required | GJ 504 b |
| NASA / Apollo 17 crew telescope image | Public domain (NASA) | Earth |
| NASA, ESA, CSA, STScI, W. Balmer (JHU), L. Pueyo & M. Perrin (STScI) telescope image | CC BY 4.0 | 51 Eridani b, HR 8799 b, HR 8799 c, HR 8799 e |
| NASA/JPL telescope image | Public domain (NASA) | Neptune |
| NASA/JPL-Caltech telescope image | Public domain (NASA) | Uranus |
| NASA/JPL/Space Science Institute telescope image | Public domain (NASA) | Jupiter, Saturn |
Not science, but not ours either — third-party files this site ships to your browser.
The pixel face this site's labels, readouts and captions are set in — most of what makes it look like an instrument rather than a web page.
OFL.txt ships beside both copies of the font, as section 2 of that licence requires.
The small JavaScript library behind the menus, toggles and sliders. Everything important still renders without it.
Copyright and permission notice retained inside the minified file.
The colours, palettes, spectra plots, Roman-view reconstructions, planet renders,
share cards and sky charts — everything computed here — are published under
CC BY 4.0.
In the data file that means the fields
true_colour · spectrum · palette · instrument_views · phase_colours · sun_swap · habitability · meta.
In the downloadable dataset: This file has two rights layers. The derived_fields are ours and are CC BY 4.0. The republished_fields are upstream facts we pass through and cannot license to you; use them under their own source's terms, listed in sources. Those are the
fields
params · host_star · sky · discovery.
“Exoplanet Palette (jocarino), CC BY 4.0”, with a link back — and, more important to us than the credit, the word . Every colour in this file is MODELLED, not photographed. Reproducing these values as observations misrepresents them. The five solar-system planets are the exception and are derived from measured albedo spectra; each record's provenance field says which it is.
Copyright (c) 2026 jocarino. Provided without warranties of any kind; see section 5 of the CC BY 4.0 licence.
Full terms, including the one source whose redistribution rights are still unresolved: LICENSE-DATA. If you publish anything built on these colours, please cite the sources above as well as this site — the Archive acknowledgement in particular travels with the data.