I'm don't know that much cartography terms and techniques, I only know
what I know from using maps. I have noted though that traditional maps
simplify the geographic shapes depending on scale. Sometimes in
interesting ways, instead of removing small islands they are sometimes
drawn bigger instead. It's actually quite elegant as it makes the map
pack more information. It's probably not the right thing to do for an
automatic generalization though, as it would be impossible to know which
islands that has importance enough to be enlarged and kept rather than
just removed.
Although OSM-Carto doesn't do any generalization of the shapes when
zooming out I think it works quite fine on a computer screen, so I
personally don't see it as a problem, but I guess a true cartographer
will :-). In any case, when going to vector shapes it will obviously be
necessary from technical reasons.
Maybe this is self-evident to anyone that knows more about this than I
do, but I have to ask, are you saying that when we have to implement
generalization to be able to serve vector tiles, it's also natural to
include generalization of names? Meaning that we could see more names
than we do now when we zoom out, so perhaps rural areas don't get the
empty-map-syndrome? That would be awesome.
In addition I still want some method to name features in the landscape
though, that supports automatic generalization. I thought named areas
was an elegant way to do this, but it seems some have very strong
opinions against it. Named points of natural features of different sizes
would also work (like isolated dwelling < hamlet < village < town) but I
don't think it is as elegant and provides less information as one
actually can map out an area even if the borders are fuzzy. But, a point
with size does the job so I think that is acceptable if one could get
consensus on that. Points all with the same prominence which is offered
today for some of the features I need to name does not work though as
the generalization will then not be representative of the area.
On 2020-11-07 20:44, Tomas Straupis wrote:
One more thing to consider: generalisation is one of the most
important things for cartography, but it is also extremely important
for vector tiles. 2-3 years ago we've played with government data and
it produced huge (up to 4MB) vector tiles (pbf) for middle scales
(zoom 8-12). Browsers (especially mobile ones) were struggling with
that amount of data. Even moderate generalisation reduced the amount
of data to 0,5M. It is something which is currently brushed under the
carpet as using raster will never produce a tile that large. What I'm
saying here is that generalisation (the real one, not DP) will have to
be done anyways as OSM community is starting to see the disadvantages
of legacy raster maps and is getting used to the idea of vector maps
(for the client, not between servers).
2020-11-07, št, 21:23 Anders Torger rašė:
(I had to run it in Chrome, it didn't render properly in my Firefox,
but
this vector stuff is new tech and Linux Firefox seems to have some
issues with that.)
Strange. Firefox on linux is my primary browser, it is the way I
always use/test *.openmap.lt...
Okay, change that to "Firefox on *my* Linux computer" :-)... it's not
the first time it seems like I am having issues with that which no-one
else has... hmm... maybe I have been toying too much with the GPU
settings.
_______________________________________________
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging
