Shortbread does not aim to replace the current raster rendering and provide the same level of details.
In fact, most vector tile maps have much less details compared to many rasters.
Before creating issues, make sure they match the goal of the project (shortbread or spirit).
It actually does show them but only point-like ones, not areas which is at least a common way to map them. Looking at Vector Tile Schema 1.0 â Shortbread it should âjust workâ from a schema and style pov as amenity areas should be include as points, I expect that it is being missed during tile generation so you should probably look at GitHub - pnorman/tilekiln
charging_station are not listed in shortbread schema (or I missed them), they are not in the pois theme definition which is here spirit/themes/shortbread/topics/pois.lua at main · pnorman/spirit · GitHub
GitHub - pnorman/spirit is a better place for such an issue to at least have the data in the vector tiles, tilekiln is independent from the schema/style.
Then to have them render is another step furtherâŠ
I was just getting confused by âtoo many layersâ the only points for chargers was/is an issue with one of the raster layers, not with shortbread which doesnât render them at all as you point out.
I wrote a blog post on where to report problems that should help people.
This could be an issue in
charging_station,charging_station when it shouldIn this case, searching the shortbread docs shows no mention of charging_station, so the right place to open an issue to get the schema changed is in the shortbread-docs issue tracker. There is no issue mentioning charging_station, so itâs appropriate to add one.
Some good arguments for including would be how itâs similar to other street-level amenities like phones, vending machines, recycling bins, post boxes, drinking water, and others. Itâs also worth mentioning that amenity=fuel isnât in the POI list.
There is some related discussion on what kind of POIs should be included.
Would it be possible to make vector tiles show the same area as the raster tiles when switching between them? Currently this seems to depend on zoom level, and not the same at that with OMT vs. Shortbread.
Do you mean the zoom level in the tile URL, or the zoom level in the URL of the osm.org homepage? By convention, zoom levels for MVT are offset by one.
I mean, when I look at OSM.org, then I switch mapstyle, then âwindowâ or âcropâ (what I see, depends on zoom of course) sometimes changes, sometimes it doesnât. Raster always stays same, vector behaves a bit inconsistently. Noticeable very good in close zooms. (Was comparing a mapping of mine, how others may see it.)
PS: Vector tiles always create a pause here, perhaps computer too old?
Each of the layers has a different maximum zoom level:
| Layer | Maximum zoom level (raster) | Maximum zoom level (vector) |
|---|---|---|
| Standard | 19 | |
| CyclOSM | 20 | |
| Cycle Map | 21 | |
| Transport Map | 21 | |
| TracesTrack Topo | 19 | |
| Humanitarian | 20 | |
| Shortbread | 23 | 24 |
| MapTiler OMT | 23 | 24 |
If you switch to a layer that doesnât support the current zoom level, the map will automatically zoom out to the maximum zoom level that is supported. If youâre seeing otherwise, please file a bug report against openstreetmap-website.
Rendering is done by the browser and uses GPU.
With old builtin GPU it may take some time⊠it really depends on your definition of âtoo oldâ.
About zoom levels, be aware that at higher zooms (above 14), vector tiles are just âoverzoomedâ and will show artifacts due to coordinates resolution.
A (considerable) amount of panning, zooming and switching layers later I come to the conclusion:
When switching from a vector layer to a raster layer and then panning or zooming the raster layer, the cursor (current position of the where in the map the current window is in) does not change in the vector layers. Depending on previous position, I get to see a misaligned crop (or even nothing at all when panned too far away) when switching back to the vector layer. Starting to pan immediately fixes this.
(This has nothing to do with different zoom numbers or overzooming. Is above useful for an issue?)
If you included the URLs at each stage and screenshots of what you see, yes. My guess is that (notwithstanding @pnormanâs post above) the actual issue may be within the Leaflet plugin used for Maplibre or the way it is deployed on osm.org.
And yes, I see the same thing going from vector here to raster here, panning in raster to here and then changing to vector.
Thank you for the pointer, this has already been reported here Switching layers and panning can cause vector tiles to be misaligned · Issue #6226 · openstreetmap/openstreetmap-website · GitHub
Currently doesnât work when WebGL is disabled for anti-fingerprinting reasons.
Rendering is done using GPU thru WebGL⊠which is thus mandatory for vector tiles.
Iâve spotted a small little issue on the Shortbread layer but Iâm not sure where exactly the fault lies.
It looks like the location of road name labels isnât being re-calculated when the geometry of the road is altered (or at least not right away, 4 hours later and it still hasnât). This leads to the road name labels sometimes appearing off the road.
Link: OpenStreetMap
Changeset: Way History: âȘRaven Drive⏠(âȘ54756008âŹ) | OpenStreetMap
Doesnât seem like this is a style or renderer issue, is it the tile generator?
The stylesheet gets the streets and street names from the same set of vector tiles. If some elements of the map are more stale than others, it would likely be an issue with the tile generator, which the OWG maintains.
For completeness a different map style looking at the same tiles shows the same problem. Those elements are in different layers, so likely one has not updated.
Thanks both. Iâve reported it on the OWG repo (couldnât add a label though).
