Nobody left a webcam on the Moon. There is no CAM 1, no CAM 2, and no website from 1997.
The picture is calculated in your browser, right now, from NASA data: the shape of the ground, where each piece of hardware stands, where the Sun and Earth are in the sky for the date you picked, and which stars are up. Give it a date, and it draws that date. That is why it can rewind 57 years.
The hardware is real, and it is still there. This page is about the machines people left behind at one spot on the Moon, and what they found out.
| REAL (measured or documented) | MODELLED (our reconstruction) | INVENTED (fiction) |
|---|---|---|
|
The shape of the ground (LROC height model, 2 m) Its brightness pattern (LROC photo, 0.5 m) Where each object stands (LROC coordinates, mission maps) Mission dates and times to the second The Sun's and Earth's position for any date Earth's phase and which side faces the Moon Star positions and brightness Sizes of the hardware |
The lander and Surveyor 3 as you see them: cut-outs of the crews' own photographs (a panorama taken beside the TV camera, and AS12-48-7100), turned to face the camera and relit for the date. Simple 3D models cast their shadows. The smaller hardware (flag, dish, instruments) as simple 3D shapes Shadows and lighting Rocks and small bumps below 2 m (generated from Surveyor 3's own rock counts) Camera exposure, noise, scan lines, glare and glitches (an imitation of 1960s TV and 1990s webcams) Autofocus blur Earth's surface picture (a 2002 Blue Marble image, not the clouds of that day) Flag and antenna orientation (approximate) |
The webcams Their tripods The 1997 website The visitor counter |
Things this page does not show: the astronauts themselves (objects appear at the minute the crew set them up), and the footpaths at the right time. The ground picture is from 2010, so its faint tracks are visible at every date, even 1967.
The picture looks quiet. Behind it, every frame is a small physics calculation. Here is what goes into it (counted from this page's own code and data):
| 1,048,576 | measured terrain heights, one every 2 metres, from NASA's Lunar Reconnaissance Orbiter (LROC stereo model). Another 65,536 heights carry the ground out to the horizon, 12 km away. |
| 0.5 m | size of one pixel of the orbital photo that gives the ground its brightness. Its 2010 shadows were calculated away so that the Sun of any date can light it again. |
| 5,080 | stars, every one visible to the naked eye (Yale Bright Star Catalogue), each with its own position, brightness and colour, turned with the Moon's real rotation. |
| 7,707 | rocks near the two cameras, placed with the size distribution Surveyor 3 itself measured in this crater in 1967: many pebbles, few boulders (218 bigger than 24 cm). |
| 1.9 billion | seconds between Surveyor 3's landing and today. The Sun, Earth, stars and shadows are calculated fresh for whichever one you pick. Nothing is a recorded video. |
| 33 | objects and events, each switched on or off at its documented time, many to the second (touchdown 06:54:35 UTC). |
| up to 50 million | terrain-height lookups in every frame, only for the shadows: each ground pixel looks up to 56 steps toward the Sun to find out whether a crater rim is in the way. |
| 16.8 million | depth samples in the shadow map, so the lander, flag and rocks cast shadows of the right length. |
| 1 arcminute | accuracy of the Sun and Earth positions (Astronomy Engine). At touchdown the page gives a Sun 5.15° above the horizon; the mission documents say 5.1°. |
| 3 | Apollo 12 crew photographs give the picture its realism: the lander (from the crew's panorama), Surveyor 3 (AS12-48-7100) and the close-up soil grain (AS12-49-7284) are the astronauts' own pictures, relit for each moment. |
| 31 | sources you can open and check, listed below. |
| about 2,800 | lines of our own code: 1,900 of JavaScript and graphics shaders, 500 of HTML and CSS, 400 of Python to prepare the NASA data. On top of that come the open libraries three.js and Astronomy Engine. |
| 0 | servers doing the work. Everything above is calculated in your own browser, many times a second. |
Try it: tap FF three times (one day per second) and watch the shadows swing round, the stars wheel past and Earth change its phase, all from those numbers. [start at the Apollo 12 landing]
For every frame the page asks where the Sun and Earth are, using the same kind of calculation NASA's planetary ephemerides use (the Astronomy Engine library). It also needs to know exactly how the Moon is turned, because the Moon rotates once a month and wobbles slightly. A day here lasts about two weeks, then a night of two weeks.
Check: at Apollo 12's touchdown the documents say the Sun was 5.1 degrees above the horizon. This page calculates 5.15. [show me]
Second check: the 2010 orbital photo has shading baked in. Fitting that shading against the height model gives the Sun at azimuth 272 degrees when the photo was taken. The calculation gives 273.
Shadows are cast from the real Sun direction over the real ground shape. For each point of ground the page marches toward the Sun across the height map and checks whether a hill or crater rim is in the way. Near landing time the Sun is so low that the crater fills with shadow.
Try the Shadows layer. [shadows off at touchdown]
Moon dust does not scatter light like paint or sand. It throws light back toward the Sun, so ground seen with the Sun behind you looks bright and flat, and ground seen toward the Sun looks dark. Honest note: this camera does not show that effect yet. Its ground is lit like ordinary dull paint (Lambert's law). We wrote the real lunar formula (McEwen's Lunar-Lambert model with an opposition surge), but switching it on made the down-sun ground 3 to 4 times brighter than the rest of the picture was tuned for, and there was no time to retune it before the deadline. The ground's brightness pattern is the real one, from the orbiter photo, with an average albedo of 0.11.
There is no air to scatter sunlight, so the sky stays black even at noon. You still cannot see stars in daylight: the ground is so bright that the camera's exposure is far too short for them. At night the camera turns its gain up, the stars come out, and the picture gets grainier.
The Moon always keeps the same side toward Earth, so from this spot Earth hangs about 60 degrees up in the eastern sky and hardly moves. It goes through phases, opposite to the Moon's: when it is night here, Earth is nearly full and lights the ground with blue-white Earthlight. It also turns: watch the continents go by. [show me]
When Earth sees a lunar eclipse, the Moon sees Earth slide in front of the Sun. Nothing is coded for it: the Sun and Earth positions simply line up. Sunlight bent through Earth's atmosphere makes a thin red ring, and the ground turns dark copper. On 24 April 1967 Surveyor 3 photographed exactly this from its crater. The page finds that eclipse at the time the NASA eclipse catalogue gives (greatest eclipse 12:06 UTC). [show me] [the copper ground]
Solar panels would be useless for two weeks out of four. The station used a SNAP-27 nuclear generator: 3.8 kg of plutonium-238 that stays hot on its own. Its heat halves every 87.7 years, so even today it would still give off about 800 W of heat. Its spec sheet shows that number for whatever date you pick.
CAM 1 and CAM 2 can see each other. Switch on Camera markers: the page draws where the other camera should be, using only the coordinates and the height map. If the tripod is not under its marker, something is wrong. The page also tests the line of sight over the crater rim. CAM 1 to CAM 2: 187 m, bearing 145 degrees one way and 325 the other, and the ground does not block the view.
This page was built on 14-15 November 2026: 57 years to the day after Apollo 12 launched (14 November 1969, 16:22 UTC, struck by lightning twice on the way up).
Every date, position and number on this site comes from a document you can open. The spec sheets and the "did you know" lines link to the exact chapter. Here is the full list:
Event times are converted from mission elapsed time (launch 14 Nov 1969, 16:22:00 UTC) using the journal transcripts. Positions in metres from the lander are in data/positions/positions.csv in the source code, with the source for each row.
NASA data and resources:
Other: Yale Bright Star Catalogue, 5th ed. (Hoffleit and Warren, via CDS). Astronomy Engine by Don Cross (MIT licence). three.js (MIT licence).
No NASA logos or insignia are used. The US flag on the Moon is shown as the physical object it is.
One person, NASA Space Apps Challenge 2026, challenge Abandoned but not Forgotten. The idea, the research, the choice of facts and every decision are the author's; the data was gathered before the weekend, nothing on the site was built before it.
Use of AI: the code (JavaScript, shaders, the Python data-preparation script) and drafts of the page text were written with an AI coding assistant, Claude Code (Anthropic), working from the author's plans and fact sheets. Each code file says so in its header. No image, video or audio on this site is AI-generated: the ground and Earth pictures are NASA images, the models are NASA's or built from simple shapes.
Source code: see the project repository (link on the project page).