Routster

Layers, tiles and presentation maps

One event can carry several maps. Which kind you pick changes what runners see and what the analysis computes.

Every event has one primary map. That is the analytical source of truth: the georeference, the terrain classification, the passability model and the ISOM course overprint all read it and nothing else. Extra layers change what people look at; they never change what is computed.

Understanding that separation is what makes the rest of this page make sense. You can hide the primary map completely and the analysis still works, because none of the pipelines consult the layer list.

The three roles

RoleWhat it isHow it is drawn
OverlayA second georeferenced image on top of the primary mapDrawn over the primary at an opacity you choose
PresentationA runner-facing rendition of the same terrain — contours only, paths hidden, a scanned hand-drawn editionReplaces the primary as the base map in analysis views; inherits the primary's georeference
TileOne panel of a map that comes in several pieces — a page of a multi-page training PDF, or one loop cut out of a sheetTiles from the same upload form a group and are shown in order

Plan

Presentation maps and multi-tile groups are a Club-plan feature — they are event-organiser tools rather than runner tools. Ordinary overlay layers are not gated. During the current beta every premium feature is unlocked for everyone, so you can build and test all of this now.

Overlay layers

The simplest case: a second image over the first. Useful for an old edition of the map, a planner's sketch, or a satellite crop of a section people got lost in.

Each layer carries its own georeference and its own opacity. You can georeference it by:

  • Dropping a world file — a .wld, .jgw, .pgw or .tfw beside the image. Instant and exact, when the projection is one we recognise.
  • Clicking matching points — pick the same feature on the layer and on the reference, three or four times, and the transform is solved from them.

Whichever route you take, preview the result — against the primary map or against satellite imagery — before you move on. Layers are ordered, and the order is what decides which sits on top.

Presentation maps

A presentation map is the map the runners actually held, when that is not the map the analysis needs. The classic case is a training where the OCAD render exists purely to serve the machinery — georeferencing, terrain, passability, the overprint — while the artefact people ran with was a PDF with the paths removed.

It inherits the primary map's georeference and dimensions, so there is nothing to align: the two are renditions of the same ground.

SettingDefaultDescription
Show only this map to viewersOffRemoves the primary map from the layer switcher, so the presentation map is the only map on offer. The primary is still used for georeferencing, terrain analysis and the course overprint — this changes what is offered, not what is computed.
The course is already printed on this mapOffStops Routster drawing its own ISOM circles on top. The controls stay in the data, so splits, leg splitting and every analysis still work — only the drawing is suppressed.

The printed-course flag is per layer, not per event

A runner-facing presentation map very often ships with the course on it while the primary OCAD render does not, and the viewer switches between the two. Setting the flag on the event rather than on the layer would give you a doubled course on one of them whichever way you chose.

Tiles

A tile group is one map that arrives in pieces. Two shapes of upload produce one:

Split modeInputResult
One tile per PDF pageA multi-page PDF, or several images dropped togetherEach page or image becomes a tile, in order
Regions on one sheetA single page with several maps printed on itYou draw a rectangle or a circle around each one and each becomes a tile

The second is the common case for a slingor training, where four or five loops are printed on one A3 sheet. Cropping each loop out gives the analysis a proper map per loop — which is what georeferencing, anchor alignment and terrain classification all want.

Keep the whole page

Cropping loses something, though: the sheet was arranged. Which loop is which, and where each sat on the page, is information a runner recognises and a bare crop throws away.

Keep the whole page stores the uncropped sheet alongside the tiles. Picking a loop then still shows only that loop's course, but drawn on the whole sheet with the rest faded — so the runner sees which loop it was and where it sat.

The sheet is uploaded once per group, not once per tile, and needs no georeference of its own: a crop is a pure translation of its sheet, so each tile's own transform is enough to place the sheet behind it. Storing it also costs nothing if you change your mind — the setting can be turned off later without re-uploading.

Georeferencing a tile group

This is where the real time goes, and there are two strategies:

ModeHow it worksWhen to use it
Per tileGeoreference every tile individually against satellite imageryTiles covering genuinely different ground, or few enough that it is not worth the setup
Anchor to baseGeoreference one tile properly, then align the rest by clicking the same feature on both imagesTiles cut from the same sheet or covering overlapping ground — which is nearly always

Anchor alignment solves a similarity transform — uniform scale, rotation and translation — from your matched points, then composes it with the anchor's own pixel-to-world transform. Two matches are the minimum; three or four spread across the image are much better, because a pair clustered in one corner leaves the rotation poorly determined.

The anchor can be another tile or the primary map. Choosing the primary is usually right when it exists: it has the most trustworthy georeference in the event.

Do the anchor tile properly

Every other tile inherits its error. Spend the extra minute placing four well-separated control points on the anchor, check it against satellite imagery, and the rest of the group will fall into place from a couple of clicks each.

Linking a tile to a class

A tile can be linked to a class by name. In a loop event that is what lets the viewer crop a runner's route to just that loop when the layer is active — otherwise you would see the whole race drawn across a map of one loop.

Which map people actually see

For viewer-facing surfaces — event cards, the event page, the analysis base map — the map is picked in this order:

  1. The presentation map, if there is one
  2. Otherwise the first tile of the first tile group
  3. Otherwise the primary map

The primary map remains the analytical source of truth throughout. This order only decides which picture is shown.

Worked example: a slingor training

  1. Upload the OCAD file as the primary map. It georeferences itself and gives the terrain and passability pipelines something to read.
  2. Add the runner's A3 sheet as a tile group, split by regions on one sheet — one region per loop.
  3. Turn on keep the whole page, so each loop is shown in its place on the sheet.
  4. Georeference with anchor to base: do the first loop against the primary map, then align the rest by matching features.
  5. Tick the course is already printed on this map on the tiles, since the loops came off the printer with their courses on them.
  6. Link each tile to its class, so routes are cropped to the right loop.

The result: runners see the sheet they held, with their own loop highlighted and their route on it, while every number in the analysis comes from the OCAD render underneath. See Georeferencing for how to check the alignment before you publish.