AptifoldContact
The shadow observatory
An interactive field trip · v3.34

Worlds beyond the photograph

How do you find a world you can't see?

When a planet crosses its star, less starlight reaches the telescope. Drag the planet to see the reading change. In 3D, drag the star or sky to turn your view.

With the scene focused, arrow keys move the planet; W, A, S and D rotate your viewpoint. Shift makes smaller camera moves.
THE TELESCOPE'S VIEW
● Even star · starlight reaching the telescope100.000%Move the planet to make a shadow.
Sun & Earth artwork · invented orbit
0° · far side180° · near side360°
Smaller shadowBigger shadow
Across the middleMisses the star
Gold keeps an evenly bright star.

Follow the light

Drag either graph. The shadow beside it shows how much of the star is covered at that moment.

The shadow and both markers show the same moment.

Model from the telescope

Projected model in the telescope's fixed direction, not a resolved observation. The gold disc is the evenly bright star.light total
97.903%of the even star’s light arrives
◆ Dimmer edges: 100.000%
Why this reading?

2.097% blocked + 97.903% remains

Only part of the planet covers the star here: 2.097% is blocked and 97.903% reaches the telescope. The gold dot marks the light that still arrives.

Compare a centre crossing, an edge crossing and a miss. The controls stay live. Return to my view restores your previous settings and pauses motion.
One complete orbitNow · 165.0°
Crossing detail7.2× zoom

Circle and diamond show this moment. The pale band marks the close-up; both graphs keep the same brightness scale. ● Gold: even star · ◆ Dashed: dimmer edges

Drag either light curve to move the same planet. Left and right arrows move one degree, Shift moves five degrees, Home goes to zero and End goes to 360.

The close-up shows 155 to 205 degrees, the pale band on the whole orbit. Both graphs keep the same 85 to 100 percent brightness scale. The solid line and circle show an evenly bright star; the dashed line and diamond show a star with dimmer edges. Each starts at 100 percent of its own unblocked light. Outside this window, its markers disappear and Show transit returns the planet to 180 degrees.

Drag planet: move it. Drag sky: look around. Arrows move planet; WASD changes view.

The planet blocks light only on the near side, where its silhouette overlaps the star.

Would the dips line up in real light?

An invented planet makes this model's dip. Now look for repeating dips in the brightness TESS measured from a real star, WASP-126.

Inside this model: geometry, brightness and what is invented

This is an invented circular orbit with two optional star-brightness models. It is not a reconstruction of WASP-126 b. The orbital radius, planet size and orientation are teaching choices, not estimates from the telescope readings in the third instrument.

Drag the sky to inspect the scene from another angle. The telescope and its light reading stay fixed. With the scene focused, W/A/S/D turns the view; the arrow keys move the planet. Whole orbit fits the whole path; Telescope returns to the observing view.

A distant star is usually too small on the sky for a transit telescope to make out the planet crossing its face. Instead it measures the star’s total light. The small telescope-direction diagram makes that alignment visible; the light budget adds up the whole star, and the moving gold dot puts the current reading on the curve. A crossing seen from that telescope is called a transit. NASA: how transits reveal planets.

The solid gold curve uses an evenly bright star. Once the planet is fully inside its disc, the blocked area stays the same and the bottom is flat. The dashed blue curve uses a star with a brighter centre and dimmer edges, so the same shadow removes a changing share of the light. This effect is called limb darkening. Use Star profile to change the linear coefficient. At u=0 the two curves overlap. The coefficient is chosen for explanation, not measured from WASP-126. The two small star icons illustrate surface brightness; the textured scene remains artwork.

The close-up enlarges the pale band from 155° to 205°. Both graphs keep the same fixed 85–100% brightness scale, so a shallow dip stays shallow. Drag either graph to move the same planet. If the planet is elsewhere in its orbit, the close-up hides both markers; Show transit brings it to the near-side midpoint. The wider horizontal scale helps reveal the curved entry and exit without changing the model.

Star radius = 1; orbital radius a = 3.4 stellar radii
θ = (position in degrees − 180) × π / 180
x = a sin θ; y = b cos θ; z = √(a² − b²) cos θ
b is the projected separation at the near-side midpoint.

The telescope is on the positive z side, looking toward the origin. For the evenly bright star, a foreground planet (z > 0) removes the fraction of the star's disk covered by its projected disk. The calculation uses the analytic area of intersection of two circles, divided by π. Otherwise brightness is 100%. For a completely overlapping planet of radius r, the evenly bright star loses the fraction r². For the dimmer-edge example, covered light is integrated over the whole finite planet silhouette using the linear law below. Each curve is divided by its own star’s total unblocked light, so both start at 100%; their intrinsic luminosities are not being compared. No planetary emission, reflected light, starspots, noise or orbital dynamics are inferred.

Linear limb darkening: I(μ) = I₀ [1 − u (1 − μ)]
μ = √(1 − ρ²), with ρ the distance from the disc centre in stellar radii
Adjustable u = 0 to 1 (default 0.6); edge intensity = (1 − u) × centre intensity
Unblocked disc light = π I₀ (1 − u/3)
Remaining light = 1 − covered light / unblocked disc light

This radial profile is a simple teaching choice. Real profiles depend on the star and the wavelength observed; a suitable fitted law can be more complex. The geometry and the finite size of the planet are retained in both calculations. The added integral is numerical, with separate accuracy checks; the original uniform-star calculation is unchanged.

Model sources: batman author documentation: brightness laws and normalization · Mandel & Agol (2002): transit light curves. These sources explain the model family; this page uses its own checked numerical integration, not the batman package.

The three guided paths keep the planet radius at 0.16 stellar radii and show the near-side midpoint at 180°. The closest projected separations are b=0, b=1 and b=1.25. The middle path has a full-overlap interval; the edge path never fits the whole planet inside the stellar disc; the last path misses even at its closest point. The guide uses u=0.6 for the blue profile and keeps the telescope fixed. The guide widens the scene once so every example remains visible at the same scale. The sliders and camera stay available; size, path or profile edits are labelled Your variation. Return to my view restores the settings from before the guide, with playback paused.

The glow is artwork. The planet uses Earth’s surface and a static cloud map as illustration; this is not an Earth orbit or a picture of WASP-126’s planet. The textured surface, corona and blue rim are illustrations. They do not generate the charts: the same orbit state drives the scene, curves and readouts. A spectator camera changes only the view. The light curve always belongs to the original telescope. Play advances one model lap at an illustrative speed; there is no fitted period or physical time scale here.

A closer look

Behind the exhibit

Can we find a rhythm in starlight?

No photograph of this planet, just a diary of its star's brightness. A repeating dip might hide in those dates. Let's try a different way to line them up.

A star's diary, one dot at a time.

Each dot records a date and a brightness. Lower dots mean less light reached the telescope. Reading the diary in date order keeps the story straight, but little dips spread across weeks can be hard to compare.

These are 993 preserved telescope readings of WASP-126, a star with an already-known planet. This is an investigation of a viewing method, not a new planet search.

Cut the calendar. Keep the readings.

Start with an invented diary: dips on days 1, 4 and 7. Cut its timeline into three-day pieces and put each piece beneath the first. Every dip sits one day into its piece. The dates are different; their positions within a piece agree. That's the useful trick called folding.

Try four-day pieces instead. Day 1 sits one day in, day 4 sits at the next piece's start, and day 7 sits three days in. The dips have not changed. The ruler you are using to compare their timing has.

Now give the telescope diary a repeat time.

Use the same idea on all 993 readings. The supplied reference is 3.28883 days per piece. A second trial, 3.51883 days, makes the pieces a little longer. Watch where readings from different dates meet. Both views keep every original date and brightness; neither is a cleaner version of the data.

Native reading 62 is our starting example. Its date and brightness stay fixed while its position within a trial cycle changes. The reference period is supplied, not fitted by this page.

A better question for the same evidence.

Folding asks whether readings from different dates share a position in a proposed repeating cycle. A sensible repeat time can make a recurring feature easier to inspect. A mismatched one can spread it out. The operation changes how the evidence is arranged, not what the telescope measured.

A dip is a clue, not a planetary passport. Eclipsing stars, stellar activity and instrument effects can imitate one. This display alone neither establishes a planet nor measures its properties.

  • A reading has an original date and brightness.
  • Folding changes the cycle coordinate, not the original measurement.
  • The trial period defines the length of each piece.
  • The invented three-day and four-day views use the same three dips.
  • All 993 real readings remain in each view.
  • The two reference settings are supplied, not fitted.
  • Background is a source column, not a flux error bar; row numbers are added here.
  • A pattern alone does not establish a planet.
Software & data credits

Aptifold’s AI agent researched the sources, implemented the models, designed the visuals, wrote the code and ran the tests. Aptifold’s founder set the goals and reviewed the results. Software license MIT License. Third-party material retains the licenses below. See model and source notes.

External observation input

MIT TESS Science Office QLP, Huang et al. (2020a,b), via MAST.
Dataset DOI: https://doi.org/10.17909/t9-r086-e880
Archive and documentation: https://archive.stsci.edu/hlsp/qlp
License recorded by the owning atlas: Creative Commons Attribution 4.0.
License link: https://creativecommons.org/licenses/by/4.0/

Frozen source filename:
hlsp_qlp_tess_ffi_s0001-0000000025155310_tess_v01_llc.txt
Native SHA-256:
db6e6e546884e90184ed8d02101b40b567999b7fc91da3bd105684859124586d

The owning atlas qualified these exact native bytes on 17 October 2026.
This package retains all 993 numeric rows and adds no new observational
clipping, detrending, normalization, fitting or error estimates. The displayed
cycle coordinates and original diagrams are new representations, not newly
measured data. The original has three columns: time, brightness and background.
Display row numbers are added; background is not a flux uncertainty.

No institutional endorsement or image of the observed exoplanet is implied.
The credited star and planet artwork illustrates an invented system; it is not
WASP-126 data or an input to the photometry calculation.

Three.js r180, version 0.180.0, MIT License.
Reused from the owning atlas; source: https://github.com/mrdoob/three.js/tree/r180
Exact local source: src/vendor/three-r180.cjs
SHA-256: 2980c5831824f46d79a39a93feb70dd851538ce5b5480a6463883b319022c51f
The emitted engine changes two comment em dashes to hyphens, removes one space
before an indentation tab on the material lookup line, and wraps exports for
lazy local initialization. The exact original vendor bytes remain preserved.
The complete Three.js copyright and license follow below. The library is
bundled locally; this page loads no CDN or remote asset.

The MIT License

Copyright © 2010-2025 three.js authors

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.

Solar artwork: Solar System Scope / INOVE, https://www.solarsystemscope.com/textures/ . Sun texture licensed CC BY 4.0, https://creativecommons.org/licenses/by/4.0/ . Original JPEG unchanged; mapped onto a sphere with rendered limb-brightness, tonal-grade and color-space adjustments. Illustrative artwork, not WASP-126 data or the photometry model. No endorsement implied.
Planet artwork: Solar System Scope / INOVE, https://www.solarsystemscope.com/textures/ . Earth day map and cloud map licensed CC BY 4.0, https://creativecommons.org/licenses/by/4.0/ . Original JPEGs unchanged; rendered together with standard lighting and illustrative material adjustments. Earth imagery is used as artwork in an invented system, not as an observed exoplanet surface or a photometry input. No endorsement implied.

APTIFOLD / AGENTIC AI

What would you like to work through?

Aptifold develops its own agentic AI to investigate questions, test ideas and build working software. This experience is one example of that work.

Explore Aptifold Discuss a question

Company links open in a new tab, so you can keep this experiment open.