# Line of sight is the whole protocol

> A rooftop plan draws sight lines between antennas wherever a roof doesn't block them. Two buildings end up joined to nothing, and the map can't say why.

August 19, 2026 · https://justog.club/stories/mesh-antenna-rooftop-map

A rooftop plan, seen straight down, printed flat in one high-vis yellow. Nine building outlines,
drawn to their relative size against each other.

Seven of them carry an antenna mark. Straight lines join antenna to antenna wherever a sightline
clears every roof standing between them.

Two buildings sit under the same yellow ink as the rest and connect to nothing at all.

The map isn't a network diagram in the usual sense. It's a geometry problem, solved once, drawn
directly onto the roofscape it was solved against.

## What is a mesh antenna rooftop map?

A plan of a district from above, showing which rooftops carry a radio link and which of those
links can actually see each other.

Every straight line on the sheet represents a working path — two antennas with nothing tall enough
between them to block the signal. Every missing line represents a path that was checked and
failed. The two unconnected buildings aren't drawn differently from the rest; they're just the
rooftops where every line that was tried ran into a roof.

## Why does a straight line decide whether two antennas can talk?

Because the radio link this kind of mesh depends on travels in a straight path, and a roof in the
way stops it the same way a wall stops a torch beam.

Unlike a wired connection, which can run around an obstacle inside a duct or up a stairwell, this
link only exists along the direct line between two points — [the same constraint a warning light
on a mast exists to announce from a distance](/stories/radio-mast-warning-lights), because
anything tall enough to interrupt that line is also tall enough to matter to low-flying aircraft.
A taller building erected between two rooftops doesn't degrade the link. It ends it.

## What makes the two isolated buildings the actual point of the design?

That the map can't tell you why they failed, only that they did.

A line that's missing could mean the roof between them grew a floor since the plan was last
walked, or that the owner of the intervening roof never agreed to host anything on it, or that the
angle was simply wrong from the start. The drawing records the outcome and stays silent on the
cause, the way a rooftop version of
[a tenant board only lists who's inside, never who was asked and said no](/stories/apartment-block-directory).
That gap is where the genre's real subject lives: the map shows geometry, but geometry was never
the only thing deciding who gets connected.

## Would a rooftop plan like this need anyone's permission?

An invented skyline needs nobody's permission, because no real property is being mapped.

Nine building outlines drawn for this design are shapes, not addresses — nothing about the print
identifies an actual roof, an actual owner or an actual mesh network operating anywhere. A plan of
a real, identifiable district, especially one showing which real rooftops carry equipment and
which real owners refused, is a different kind of document and a much more sensitive one to
publish. The invented version does the whole job: the geometry is the point, and geometry doesn't
need an address to make its argument.

## Does the same drawing work for a mesh that isn't wireless?

Anywhere a network's shape is decided by which points can physically reach which other points
rather than by any central plan.

A chain of relay beacons across open country, plotted against the land that blocks or clears each
sightline, uses the identical logic — [the same discipline that reduces a location to a precise
six-figure square on a map](/stories/grid-reference-six-figure) is what a relay survey runs
on before a single antenna goes up. Whatever the medium, if reach depends on an unobstructed line
between two fixed points, the resulting map looks like this one: mostly connections, a few
conspicuous gaps.

## Isn't line of sight just a limitation, something a better antenna fixes?

No amount of transmitter power gets a signal around a building that's physically in the way — this
isn't a limitation the hardware can be upgraded out of.

A more sensitive receiver extends how far a link can reach once it has a clear path. It does
nothing for a link that has no path at all, because the obstruction isn't attenuating the signal,
it's blocking it entirely. That's why the map matters more than the equipment spec sheet: knowing
which roofs can see which other roofs decides the shape of the whole network before anyone chooses
what to bolt to the mast.

## Does it print well?

Yes, and the restraint is what sells it.

One ink against a dark ground — high-vis yellow reads as exactly the colour a rooftop survey would
actually use, since it's the tone chosen for anything meant to be seen from a distance in poor
light. Keep the building outlines as simple flat shapes and the sightlines as single hard-edged
strokes; anything softer turns a precise survey into a loose sketch. No text is needed and none
should be added — the two gaps argue the whole point on their own.

## How do I get one made?

Describe the rooftops at [JustOG](/) — how many buildings, which carry an antenna, which sightlines
you want left unresolved. Pick a direction, drag the crop frame, see it composited on the real
garment, and it's made to order and shipped.

Designs other people have published are in the [shop](/shop).

Seven working links, two buildings under the same roof survey that never joined the rest.
