Open dktue opened 6 years ago
Perhaps a problem with the data here, but dots at the walls looks strange:
Yes, there is a problem with the data ("light points") as it includes not only street lamps but also spotlight in monuments and other kinds of lights operated by the City Council. It has attributes to indicate what model of light point it is, but as there are hundreds of types I haven't bothered yet to try to separate the street lights.
Please ignore where there are many dots together as those are not really street lamps.
But wall mounted streetlamps would render exactly the same. Rendering them not would be confusing
But wall mounted streetlamps would render exactly the same. Rendering them not would be confusing
That example you provided can give us an idea on how wall mounted streetlamps would render, but only if we look at the bottom left side of the building. The reason is that while there are quite a few wall mounted streetlamps on that particular building, there are even more than are not really streetlamps, as the purpose of those is to illuminate the building or monument itself, not the street.
BTW, that building was the City Hall from 1696 to 2007. No idea what are those "light points" inside the building.
As said above, in general I would be in favour of rendering this starting with z20, with an appropriate icon. If you could come up with a PR?
@sommerluk, I have a branch open with a rendering like that of natural=tree
that is based off of meased's icon at zoom 19+ (although this could easily be changed to an svg, moved to 20+, or something else based on feedback). I can submit a pull request if you want.
You can find the branch here.
The comparison can be viewed here.
Here are some previews all at z19:
Tucson, United States:
Edmonton, Canada:
Paris, France:
Currently there is no consensus among the maintainers that rendering highway=street_lamp
is desirable. The main arguments against it are:
lit=yes
on roads in mapping and selectively rendering just one variant, but not the other could be considered nudging the mappers.Of course counter-arguments can be given against all of these:
lit=yes
. Since we render bollards the display of (free standing) street lamps could also be considered useful simply as obstacles/orientation points (i.e. a tall bollard).highway=street_lamp
and many mapping tools encouraging use of lit=yes
mapping of highway=street_lamp
remains popular without indications that it is becoming less so (https://taghistory.raifer.tech/?#***/highway/street_lamp&***/lit/yes)highway=street_lamp
does not rule out rendering lit=yes
on roads (in a similar way as rendering of urban land uses and buildings is not mutually exclusive).What @BubbaJuice has shown is an implementation of the non-blocking rendering similar to the french style. This is done above the POI and label layers - hence the light symbols cover these in some cases, which is not a good idea. Non-blocking symbols would have to be rendered before the symbol and label layers.
Also the dark yellow color is not very intuitive - as previous comments have already mentioned. It has very low lightness contrast with urban landuses and pedestrian roads - making the lighting aspect fairly non-obvious. The french style coloring works much better in that regard.
I have some time ago further compacted the french design in my style and use that starting at z19 - which IMO works reasonably even at dense low latitude settings:
This would match our approach to rendering trees where we also start early (z16) with a very subtle design and move to a more distinctive symbols as you zoom in.
See also a real world example at z19.
I am re-opening this to encourage renewed discussion but note that does not mean we have consensus among maintainers that this is desirable to render. So opening a PR at this time might be premature, further tuning the design concept shown and seeing if that convinces the maintainers this is a viable approach might be the more prudent approach.
Streetlamps don't show up on the map, yet but I think it would encourage mappers to add them to OSM if they could see that they are actually being rendered.
At zoomlevel 19 this should cause no harm.