In 1969 a FORTRAN program on a UNIVAC 1108 at NASA's Manned Spacecraft Center drew what the Apollo crews would see out their windows and through their optics: star fields for navigation, the Earth and Moon, the lunar surface through the LM's landing window. The frames went onto microfilm, and views were "incorporated into the Apollo flight-plan documents" (NASA TN D-6853, p. 10). We know of no surviving source code for it.
VIEW-1108 is a what-if: how it might have felt to operate VIEW if that 1108 could run the simulation in real time.
In 1969 a view was a batch job. The program integrated the trajectory, drew "at any nth value of the integration step", and a camera photographed the cathode-ray tube onto microfilm. Users could request "crude printer-plot images" for "a quick-look evaluation before the microfilm frames are received", and for most uses the frames were "printed on standard sheets of paper" (TN D-6853, p. 3). You asked a question and got the answer back later, on paper.
Here the answer comes back while you move. Drag, and the window turns; type a g.e.t., and the sky, the Earth and the Moon move to that moment of the Apollo 11 mission. The picture above is a frame from our kernel, not from the film. On the page every frame is computed live by FORTRAN 66/77-style code compiled to WebAssembly (or its JavaScript translation where WebAssembly is unavailable), and drawn as white lines on black, like the film.
Where it's going: from one film's scenes to whole Apollo missions (Apollo 8 and Apollo 11 first), phase by phase, with windows, cabins and spacecraft, and every modern addition labelled as one. See docs/vision.md.
Every statement here comes from the two NASA reports in reference/.
- "The view program operates on the UNIVAC 1108 computer. The program is written in the FORTRAN V language ..." Its output "consists of microfilm frames produced by a camera that photographs an image constructed on the surface of a cathode-ray tube." (NASA TN D-6853, Hyle & Lunde, 1972, p. 3)
- It began as an aid "in the early Gemini rendezvous and docking studies" and was revived for Apollo. (TN D-6853, p. 2)
- On Apollo 13, with the guidance system powered down, "the views of the Earth were used as the sole attitude-reference source for a midcourse correction." (TN D-6853, p. 10)
- "The major analytical tool used to produce this report was developed by Mr. G. B. Roush of the Computation and Analysis Division." (MSC Internal Note 69-FM-197, Revision 1 to Views from the CM and LM During the Flight of Apollo 11 (Mission G), A. N. Lunde, 3 July 1969; a 321-page scan, 303 numbered pages, mostly VIEW plots)
A short clip believed to be VIEW film output (shared by NB, @Noahbolanowski; its archive source is unknown) has four shots: an Earthrise, the Earth approaching, the LM turning, and the LM's descent window. On load, the page replays those four shots live from the kernel, then settles into a slower tour of the same scenes.
The page is the console. It starts by replaying the film, then hands you the controls. docs/modes.md lists every mode, scene, toggle and link parameter, with example links to share.
| Mode | What it does |
|---|---|
| Attract | The film's four shots, at the film's pace (36 s). Plays once on load. |
| Tour | A slow loop through Earthrise, Earth approach, Earth limb, the LM pirouette, the LM descent, and a slow spin of the whole Moon. |
| Live | The Apollo 11 mission clock at 1× (or 10×, 60×). The scene follows the mission phase from the g.e.t. you type or scrub to. |
| Beam | Traces each frame vector by vector, in the kernel's output order, on a phosphor that fades. Speeds run from an estimated 1108-plus-recorder pace down to a slow trace you can watch, up to a persistence-of-vision blur. The recorder rates are our estimates (see docs/univac-1108.md). |
| Free-look | Any drag, wheel or key. Look around (yaw, pitch, roll, field of view 1°–170°) at the current moment. |
Keys: arrows look, Q/E roll, +/- field of view, space pause, [ ] speed, R reset, 1–6 scenes, L copy link. The "Fortran listing" link shows the kernel source the page is running.
The [ LINK ] button (key L) copies a URL that reproduces the current view. You can also write one by hand. Bad values are ignored, parameters override stored preferences for that visit only (nothing is written to storage), and with no mode the page starts in Attract.
| Parameter | Values | Example |
|---|---|---|
mode |
attract, tour, live, free, beam |
mode=live |
scene |
1..6 (Free-look, Beam; in Live only the LM windows 4 and 5 apply) |
scene=5 |
get |
g.e.t. as h:mm:ss or seconds |
get=102:45:40 |
utc |
UTC as YYYY-MM-DDTHH:MM:SS |
utc=1969-07-20T20:17:40 |
fov, yaw, pitch, roll |
degrees (fov 1 to 170) |
fov=100&pitch=-10 |
rate |
1, 10, 60, 300, 1000 (Live, Free-look) |
rate=60 |
bspeed |
1..4: 1108 + recorder, recorder only, slow trace, persistence (Beam) |
bspeed=3 |
labels, frame, hidden |
0 or 1 (names, plot frame, hidden lines) |
labels=0 |
bloom, jitter, dust, fps |
0 or 1 (film effects; fps=1 is the 16 fps film rate) |
bloom=1 |
catalog |
nav (391 stars) or full |
catalog=full |
listing |
dark or light (Fortran listing) |
listing=light |
bare, still=earthrise, film=N |
chrome hidden; frozen Earthrise; Attract at film second N (for screenshots) | still=earthrise |
Example: https://aaronsb.github.io/view1108/?mode=live&get=102:45:40&fov=100 opens Live at the landing, with a 100 degree field of view.
src/vdrive.f the kernel's lead element: scene setup and the frame driver
src/*.f the other kernel elements (ephemeris, trajectories, pen, models, one file per
drawing layer): fixed-form FORTRAN, DOUBLE PRECISION, COMMON, DO/CONTINUE
src/viewdata.f star, coastline and crater tables as BLOCK DATA
src/shell.f90 a thin modern-Fortran shell that exports the kernel to WebAssembly
web/ the "film recorder": draws the kernel's vectors, stars and labels
The kernel takes a time and a look direction and fills a display list of lines, stars and labels: the job we assume the 1969 program did for its recorder. The rules it follows, checked by make lint:
- FORTRAN V limits: identifiers of at most 6 characters (NASA-CR-150010, 1976) and no
IMPLICIT NONE(UNIVAC UP-4046 Rev 3 §10.4.1); see docs/univac-1108.md. - Functional first. Code we believe a 1969 FORTRAN V programmer could not have written stays in, fenced with
C RESTOMOD:comments that give the reason and, where known, the year. Examples: fileINCLUDE, real-valuedPARAMETERs, blockIF(FORTRAN 77, 1978), Liang–Barsky clipping (1984), Park–Miller random numbers (1988), the IAU lunar orientation model, post-1969 map and crater data, and a display buffer larger than the 1108's 262,144-word core.
Where the reports are silent, the choice is ours and the fortran code has been commented to indicate this. One example: the projection. Neither report states it. The film's straight horizons, and a straight horizon in a 100° docking-window plot of MSC IN 69-FM-197 (PDF p.170), fit a gnomonic (true-perspective) plot; the same page's 170° panel shows a fisheye dome, which fits stereographic. The kernel blends from one to the other as the field widens (comment at PROJ in src/pen.f).
Probably, with changes: the fenced items above rewritten in FORTRAN V terms, and the display buffer streamed out instead of held in core. docs/univac-1108.md works through the 1108's word formats, core size and instruction times from the UNIVAC manuals. Its rough, unmeasured estimate is on the order of a second of 1108 compute per frame for an assumed workload; we have not yet counted our kernel's operations. The 1108's double precision has a 60-bit fraction against IEEE's 53 bits, so the original format was the more precise of the two. We should also consider the batch job nature of the code - it likely wasn't one large monolith but several interoperating codes that were loaded for different sequences of batch jobs - control tapes and data tapes. A sibling MSC report describes exactly that pattern for other Apollo analysis programs: an integrator wrote a "trajectory ephemeris tape" that programs "that contained no integrator also used ... for input", all "run in a batch-processing mode" (NASA TN D-6855, Allday, 1972, pp. 7-8). docs/batch-pipeline.md collects the evidence and a labelled conjecture of VIEW's job sequence; a batch mode for this page is proposed in #1.
A UNIVAC 1100-series computer at the U.S. Census Bureau, 1970s. Public domain, Wikimedia Commons.
- Stars: the 37 Apollo navigation stars from the Apollo Guidance Computer's star table (Comanche055, via chrislgarry/Apollo-11), plus a catalog to visual magnitude 4.5 (XHIP via d3-celestial). TN D-6853 (p. 12) describes two alternative catalogs, 391 navigation stars or 1,078 stars to magnitude 4.5; we draw the 37 named stars and the magnitude-4.5 catalog together.
- Earth: Natural Earth 1:110m coastlines.
- Moon: craters from the IAU/USGS Gazetteer of Planetary Nomenclature, plus seeded small craters where the gazetteer is sparse.
- Trajectories: low-precision Sun and Moon ephemerides and simple Kepler and circular orbits keyed to Apollo 11 event times. Not a precision tool.
Needs LFortran 0.66, LLVM/clang 23 and binaryen 121 (conda-forge), plus gfortran, node and python3.
micromamba create -p ~/lf -c conda-forge lfortran=0.66.0 llvm-tools=23.1.2 lld=23.1.2 clang=23.1.2 binaryen=121
make # list targets
make build # FORTRAN -> wasm -> web/view1108.html, then the self-test
make serve # http://localhost:8108/view1108.html (make stop to end)
# page.template.html?mock: the page modules from web/src/, with a toy kernel
make check # render every scene natively with gfortran to build/check/
make lint # dialect check and compiler warnings
make sheet # regenerate the film comparison sheet (headless Chromium)
Every push to main rebuilds the page from source in GitHub Actions and publishes it to GitHub Pages.
- G. B. Roush, credited with developing the major analytical tool behind the Apollo 11 views (MSC IN 69-FM-197); A. N. Lunde and C. T. Hyle, who documented the program (TN D-6853).
- NB (@Noahbolanowski), whose thread on VIEW started this project.
- Data sources and their terms: THIRD_PARTY.md.
MIT, see LICENSE. Third-party data keeps its own terms.