Record a sighting.

Summary#

State what this recording is about: an ID, a description, tags. The ID is this account's own, the day and then who saw it (1964-04-24-ZamoraLonnie); a recording does not name its case, it is the case that names its accounts and gives a page its observer picker.

Witness#

Who gave the account — and, just as importantly, what they observed it through. An eye is not a lens: naked-eye vision maps an angle to an angle, a camera maps it to f·tan θ, and the two draw genuinely different frames. Pick a camera and its focal length, aperture, exposure and focus distance become available — each disabled where the device fixed it, because the owner of a fixed-focus snapshot camera had nothing to choose. Instruments outside the observation's own date are left out of the list, except the one the recording already names, which is kept and flagged.

A camera's frame is letterboxed inside the element at its own proportions, and its field becomes the default one; the eye, or a camera of unknown make, is drawn 16:9 with a 60° vertical field. Changing the instrument re-projects the shapes: they keep their angles, so they move as well as change size. The exposure stays within what the device allowed (1/1000 to 8 s for an unknown camera, 10 s for a phone); only the 35 mm SLRs have a B setting, up to an hour. A long pose is drawn at once as the viewfinder showed it, then fills in into the photograph as the scene settles.

Roll sits here rather than with the place, because it is how the device was held — a camera askew, a head leaned over — not where the observer stood.

Location#

Account names a place, it does not give coordinates. So type the name and press Locate: latitude and longitude are filled from OpenStreetMap's own geocoder, every candidate stays listed, and picking another moves the observer. What is stored is the qualified name that was resolved, so a later reader lands on the same spot. Move a coordinate by hand and a name that came from a search is re-derived, or cleared — a name describing somewhere the sighting is no longer at would be a written false statement. A name you typed yourself is kept as typed and never replaced, even when no lookup finds it.

Heading is the direction faced, Tilt how far up or down, and Altitude is above sea level, floored by the ground's own height at that location: an observer in the Alps is not at 0 m. Relief and imagery sources are chosen right here, under the coordinates whose ground they describe.

Position, heading and tilt are written at the playhead, as a keyframe of the observer's track: set them at another moment and the observer moves or turns between the two.

The map over the render does the same for the position: click bare ground and the observer stands there at the playhead, exactly as if both coordinates had been typed. The wheel zooms around the pointer, a drag moves the ground, and +, − and ⤢ (back to the whole path) are there for whoever has neither.

Date#

A start, an end, a duration — and a time zone, which is the rule, not the number. Pick the observer's own zone and the offset is derived from that zone's rules at the observation's date: Valensole in July 1965 resolves to UTC+1, not today's UTC+2, because France only reintroduced summer time in 1976.

The zone is filled in from the coordinates, and never replaces one you chose. Zone boundaries are coarse (Montgomery, Alabama falls in America/Chicago, which kept summer time in 1948 when Alabama did not), so the two are one thing said both ways, kept in step: pick a zone and the offset follows; type an offset and a zone that reads it that day is picked (the place's own when it fits). The only other choice is unknown, which has no offset, and a number no zone reads that day is refused. An offset that no clock at that longitude could have kept is flagged, the meridian's solar time in its tooltip. Editing either date drops an explicit duration once the two dates give an exact length of their own.

Each date is either precise — a picker, for an exact full instant — or said in EDTF, a text, for everything a picker cannot say: a bare year, a month, a time with no date, or any date in doubt, which ends with ? (uncertain), ~ (approximate) or % (both). The button beside it says which it is and swaps the two. Hovering the text shows, in its tooltip, what was understood (“→ July 1965 — uncertain”), and a (?) in the field lists the syntax. Most archives need EDTF — of 241 case files on rr0.org, 43% state a bare year and only 17% a date with a time. A case's own date is the same field. Duration is said the same way: precise, a number of seconds, or as it was stated — an ISO 8601 duration with a doubt or a range (PT10M~, PT5M/PT10M) — and, in brackets after it, the one length the simulation is played at.

Moments names the instant at the playhead (A, B, C… as case sketches do), with a sentence of what happens then. Picking one in the list goes there; 🎯 goes back to the one shown, which the list cannot do once playback has moved on past it.

Environment#

What stood around the observer, at a real distance east and north: buildings with their floors and windows, trees, streetlights, vehicles, aircraft — and other observers. This is the only thing that can put a number on a distance: if the object passed behind that hangar it was at least that far, if in front of that tree, at most. Each crossing narrows the object's real width from one side for the whole recording; the result appears under the apparent size, and reads “unknown” when nothing crosses its line of sight, which is the honest answer for most sightings.

The scenery comes by itself, where a place has been surveyed. When you give the recording a place (by name, by coordinates, or by moving the observer), the editor asks the survey that covers it for what stands within a hundred metres, and puts each tree, building, bridge and span of electric line in the list, at its measured place, height and width: in France, the IGN's LiDAR HD (a scan of the whole country from the air) for the trees and buildings, and its map for the bridges and the overhead lines, with their voltage. Nothing to press. A line under the list says how many were found, and a choice beside it says which survey it comes from: IGN LiDAR HD + BD TOPO (the default, which knows the most), the laser alone, the map alone, or File, which keeps the scenery the recording says and asks no survey any more, even as the observer moves (the choice is the recording's, and links to its item of the Files tab), or None, which leaves the scenery to you and takes away what a survey had brought (Undo puts it back; what you placed stays; your browser remembers the choice: asking sends the place to that service). Choosing another survey asks it at once for the place the recording stands at, and replaces what the first brought. Elsewhere, and wherever no survey covers the place, nothing is asked and nothing is said. What is taken is of the day of the survey, which is not the day of the recording: keep it, move it, resize it or delete it, like anything else. Of each kind it brings nothing where you have placed some of your own (those you placed stay), and it is never added to a recording that is merely loaded. Moving the observer to another place replaces what a survey brought for the first, and nothing else.

The survey also names what it knows the name of. A building takes the name of the zone the map gives it (Mairie de Saint-Lô, a post office, a belfry), the parts of one numbered; a bridge is named for the road it carries (Bridge (Rue de Carentan)), a span of line for its voltage; the others are numbered like anything you add. In France the roads of the scene come from the same map, with their names, rather than from OpenStreetMap, which too often did not answer. Buildings, like trees, stand on the lowest ground under them, as they are built into a slope; and a span of line meets the ground at both its poles.

The radius next to the choice says how far round the observer the survey is asked, from 20 m to 3 km (100 m by default, and the browser remembers it): a wider one brings the farther buildings, bridges and lines from the map. The laser, which gives the trees, is asked within 150 m only. Outside France, where the laser has not flown, the buildings come from OpenStreetMap (anything its volunteers wrote of a building is drawn: height, roof, colours; the rest is an ordinary building), without trees; choose it in the list of surveys to have it anywhere.

The street lamps, traffic lights, pedestrian crossings and road signs come with any survey, from OpenStreetMap; in Paris the lamps come from the city's own inventory (Paris Data), every lamp at its place with the colour and the flux of its light, a sodium lamp's orange as an LED's white. OpenStreetMap is often slow or down: the line under the survey then says the lighting is awaited, and plants it when it comes, without you doing anything (one step of the history), or says it is unavailable for now with a Retry button.

While the scenery is being looked up, a spinner stands over the picture; choosing "None" in the list of surveys stops the wait (what the survey answers afterwards is dropped). What was taken is dated: the status line gives the day of the laser flight, of the aerial photograph and of the map's last edition, and each object keeps the newest day it was drawn from (surveyedOn). Where that day is later than the observation, a warning says so, and goes on saying it if the date of the observation is changed afterwards: the scenery is then a landmark, not what stood there.

A building's roof is read too: from the laser, a roof with two or four slopes (and which way its ridge runs) or a flat one, with the building then as high as its ridge; from the aerial photograph, its colour (tile, slate, zinc), taken without the green of the trees over it. A roof the survey cannot read stays a plain box, not a guess.

The walls are drawn as walls: the colour of the material the map states for them (stone, brick, concrete, breeze block, wood; a pale plaster, a little different from one house to the next, where it states none, which is two buildings in five), with windows on every floor and a door on the ground floor of the front face. They are the typical windows of a facade, not the windows of that house, and some of them are lit once the Sun is well below the horizon: about half in the evening, practically none from three to five in the morning, a few more at dawn, by the hour on the clock (always the same windows, so those lit at three are still lit at nine). The hours follow the French time-use survey (INSEE, Enquête Emploi du temps 2009-2010: half of couples asleep at 22:40 and half up at 7:00, nine in ten asleep from 1:00 to 6:00); the curve between those points, the two in a hundred always up and the half of the windows lit when everybody is up are assumptions. Everything about the walls and the roof (their colour, plain walls, the roof's shape, rise and ridge) can be edited afterwards in the building's properties, like any other decor.

Click an object in the picture to select it: the Environment group opens on it, framed, ready to move, size or turn. Drag the selected object to carry it over the ground; hold Shift to carry it east-west only, Alt north-south only, Ctrl up and down only. Hold Cmd (or two of those keys) with the press and it is the view that turns, as over bare ground — the way to turn it from over a forest, where every press lands on a tree.

Several observers, and their points of view. An observer placed here can carry the URL of their own recording, and right-clicking them in the scene offers to view it — which loads their account and puts you where they stood, looking the way they looked. Two people a hundred metres apart did not see the same thing, and being able to step from one to the other is how that stops being an assertion and becomes something to check. The 🎯 button does the smaller version of the same thing: it turns the current observer to face whatever is selected.

Published together, those recordings become the observer picker a reader gets — see the case that lists them.

Beyond placement and size, an object has the properties its kind has, drawn from what the kind says it lets you set: floors, the floor the observer is on and a window on each side for a building; the windows of a car, whose windscreen and rear window never open; the span, deck and railing of a bridge; the poles and strand of an overhead line. A field left empty states nothing and shows, greyed, the value the scene uses. A model carries its own colours, so the single Colour is offered for the built-in shapes only; a model can instead offer the colour of a part of itself — the costume of a figure, the leaves of a tree — or a switch of one of its parts — an airliner's retracted landing gear — and any other setting of its own, which appear here the same way.

Decor can also move (an aircraft crossing the sky, a car driving past) and carry lights with real, regulated flash rates — anticollision beacons at 40–100 a minute, hazard flashers at 60–120. On a long exposure that rate is drawn: steady lamps leave lines, flashing ones leave dots at regular intervals, which is exactly how a photograph of an airliner is told from a photograph of something that does not blink.

The ready-made sets of lamps offered for an object are those that suit the model it is drawn as, and are scaled to the size of the machine: an airliner is not offered a helicopter's, a patrol car gets its emergency beacons. The model also says what noise an aircraft makes in flight (an A350 is heard as a wide-body, a stand-in for something whose noise nobody reported is silent) and gives a vehicle its engine.

A vehicle's lights are a setting and not a switch, because not everybody puts them on, nor as they should: off, position lights (which show the vehicle and light nothing), dipped beam, main beam (far, narrow, dazzling) or fog lamps (low and wide). Choose it in the object's State group at the instant you want it to change: it is written as a keyframe, as the lamp switch is, and holds until the next. A beam really lights the road, the verges, the trees and the walls in front of the vehicle. The vehicle the observer rides (the recording's vehicle) puts its dipped beam on by itself from the dusk on unless the file says otherwise.

Sky#

The one group that is not account. Weather is a measurable fact about a place at an instant, and the two groups above already state both — so it is looked up from ERA5, the ECMWF reanalysis, and shown read-only above a line naming the dataset and the exact UTC instant described. A wrong time zone shows up there before it shows up in the sky.

Choosing File rather than Open-Meteo (in Data sources, behind the gear at the top right of the group, Weather, in the Sky group) reads the weather from the recording's own file (inside it, or in a file beside it: see the Files tab) and asks nobody for it any more: the values stay what they were, with the record they came from, and the fields become editable; once changed, the weather is the observer's and the record is no longer named. A recording that names a source is replayed exactly as authored and never looked up again, so a published case file reads identically offline.

A lookup writes a weather keyframe at the start, at each whole hour and at the end of the recording, each at wherever the observer stands at that moment; the track is laid out again when the recording's length changes.

The group has three handles of its own, one part open at a time like the groups themselves: Precipitation — its type and intensity, and the Storm box, which is where a weather code reports a thunderstorm, beside rain and hail, and which turns on the lightning and the thunder at its real delay — Clouds, and Wind, the direction it blows toward and its speed, which is what carries every cloud that has no wind of its own.

The clouds are the Clouds part of this group. A looked-up sky arrives as three of them, the low, middle and high bands of the record; a sky the observer described is as many as they saw. Every number in the panel is written into the recording the moment it is valid, and the first edit pauses playback, so nothing is committed at a moving playhead.

Edit scope decides what an edit is about. Current time writes a weather keyframe at the playhead, or updates the one that is there: this is how a sky changes during the observation — set the deck at the start, seek, set it again. Whole observation applies the one property you change to every keyframe there is, leaving the others as they were, which is what you want for “it was cumulus, not stratus”.

Layer lists them by rank, type and base. Add layer puts a half-covered cumulus deck at 1 500 m for you to reshape; Delete layer removes the one selected. A layer keeps its identity across keyframes, so adding one at 40 seconds fades it in from the previous keyframe rather than popping it.

Then the layer's own fields: Cloud type (cumulus, stratus, stratocumulus, cirrus, unknown — it shapes the tops, and a cirrus is the deck that makes haloes), Base and Thickness in metres above the reference ground, Coverage as the percentage of sky it really covers, Cloud size as the width of one cloud in metres — separate from coverage, so the same percentage can be many small clouds or a few large ones — Density (0 transparent to 2) and Darkness (0 white to 1). Layer wind direction and speed are for a deck that moves differently from the ground wind, which the high one usually does; left empty they mean the general wind. Pattern seed picks another arrangement of the same numbers, when the one drawn puts a cloud where the account says there was none.

Individual clouds is for the one the account places: the one the phenomenon went behind, the one that was there and nowhere else. Add individual cloud puts one in the layer, ahead of where you are looking, at the layer's own base. It is drawn as one of the layer's own — the same texture, the same edges — and differs from its neighbours only in standing exactly where you say, even with the layer's coverage at 0 %. Point at cloud turns the observer to it. Delete cloud removes it.

Tick Select and drag clouds in the sky and the picture itself becomes the control: click a cloud to select it, drag to move it — in the plane facing you, so it keeps its distance while its bearing and its altitude follow the pointer, and it never goes below the ground. Untick it and clicks are the player's again. The numeric fields do the rest, and do it exactly: East and North position in metres from where the observer started, its own base, thickness, width, depth and rotation, its density and a darkness that, left empty, is the layer's. Keyframe it twice and it drifts, grows or darkens between the two; on top of that it rides the layer's wind like every other cloud.

Any cloud edit takes the weather away from the record: the source is dropped and no later lookup may overwrite what you set. Choose Open-Meteo again as the source and the record's three layers come back, replacing yours. The one exception is Crystal alignment, offered on a cirrus layer only: no weather record measures it, so it stays editable with ERA5 selected. Tumbling crystals give a bare ring; level plates and rolling columns give sundogs, arcs and a pillar.

Below sits the “Sky:” line — read-only, and not a lookup at all. A meteor shower is a position in Earth's orbit and a comet's orbit is a solved problem, so the date and the place alone decide both. It states what else was in that patch of sky: the shower and its rate over the sporadic background, the comet and its magnitude, any nova or supernova whose recorded light curve covers that night (the 🌟 button turns the observer towards it), whether low orbit was still sunlit, whether the Milky Way or the zodiacal light could have been seen at all. From February 2021 it also names the satellites that really crossed that sky bright enough to be seen, the brightest with the time of its pass on the observer's clock, and says when some of them were a Starlink train; the 🛰 button moves the playhead to each pass in turn, brightest first, and turns the observer towards it. 🌠 and ☄ do the same for the next meteor and for the comet. Where the instrument reaches deeper than magnitude 9, the line says the stars are drawn only as far as this project's catalogue goes; and a rainbow is mentioned only when rain was reported. Whether any of it explains anything is the reader's conclusion, never the file's claim.

Phenomenon#

The object itself. Oval or polygon, colour, transparency, halo, brilliance (how dazzling it was — a light you cannot look at washes out the field around it, throws the spikes its aperture makes, and clips to white, which no halo does) and blur (how indistinct its edges looked). Several shapes can share one timeline — a craft, a trailing flame, a second object — each with its own name, and grouped, reordered or deleted from the canvas's own context menu.

Apparent width, real width and distance are three readings of one relation, kept in step: edit any one and one of the other two follows — the one Hold pins never moves. Only the angle is the recording's; the real width is derived, and the distance is where the scene draws the shape, a hypothesis to try rather than a statement. Through an eye at 60° across a 360-pixel canvas, one degree is exactly 6 pixels and the full Moon about 3 — so an object 3.5 m wide at 90 m is 13 pixels across, not the 90 an author reaches for unaided. Getting this wrong is the single most common way a reconstruction ends up false.

That distance is never saved with the recording, and Withdraw the distance hypothesis takes it back. A shape drawn inside a larger visible one stands at that one's distance, just in front of it, so a light on a craft stays on the craft.

Sampling rate is how often the pointer is read while recording.

The selector above the render, Account, says how the account is drawn and edited: 2D · shapes, the angles the observer drew, or 3D · bodies, what they said those were. Choosing one opens the matching part of this group, and opening a part sets the selector; a reading of an account has no shapes, and no selector. Those are the group's Shapes: what the observer drew. Its Bodies are what the observer said those shapes were, in 3D: each body names the shapes it stands for and the model it is drawn as, a built-in shape, a model of the catalogue, or a glTF file at an address. The address can be relative to the recording's own file, so a sighting.json and the craft.gltf beside it work together from any page; its credit (name and licence) is required before it is drawn. Interpretation title, at the top of this part, names what the recording claims — “Craft on its legs”, “April 2013: a model on a thread” — and is always offered: typing one creates the interpretation before any body does. + adds a body (and the interpretation, when the recording has none): it stands for the selected shape, in its direction, at the distance the scene draws it and as big as its apparent width makes it there, in a single keyframe at the playhead; on the ground at that distance when the line of sight goes into it first; with no shape, where the observer is looking. 🎯 turns the observer towards the body. A recording may hold only bodies: its last shape can be deleted. At the playhead, the body's position (from the observer, or in the world), size and attitude are shown as they are at that instant; editing one writes a keyframe there, which is how a body is given its movement: move the playhead, edit again. Such a keyframe states nothing else, so the body's light, flame and moving parts go on as before. With this part open, a body can also be taken on the picture itself: dragging it moves it by as much as the pointer turns (over the relief when it stands on the ground, at the same distance when it flies), and the wheel over it takes it nearer or further. The body on show is framed with handles: a corner sizes it as a whole, a side stretches it level, the top or bottom raises it, and the stem above turns it as the pointer goes left or right, and with Shift held pitches it as the pointer goes up or down and rolls it as it goes sideways. All of these write the same keyframe as the fields, which show where it went. A body's movement is stated in the file, and shown here. The picture draws the bodies, with the shapes beside them as outlines to read them against, whichever group is open, except in the Shapes part, where the shapes are drawn alone and in full.

What happened in the sky while the recording runs, as facts (a laser shot, a rocket launch), is not part of the Phenomenon group but of Sky, beside the state the sky was in (precipitation, clouds, wind): it is neither the interpretation, which reads the account, nor decor, which outlasts the observation. Its Beams tab lists the beams of light (a laser, a searchlight): the points of each one's path (where it leaves the emitter, any mirror, where it ends), its wavelength, power, divergence and when it is on; how bright it shows is what the air scatters of it, computed. Its Launches tab lists the rocket launches, their path, engine and plume. An empty field shows, greyed, the value in use. These are drawn whatever group is open.