Renderer limits means overview maps not useful in rural areas

I’ve been a casual contributing mapper for a number of years here in Sweden, and recently tried to up my game to make really high quality maps in the areas I live. I have a lot of local knowledge combined with very high quality government maps (already preloaded into the editor) together with Bing which today has much better alignment than a few years ago (still government maps are best on that). So good reference is there too, I have all I need to make a good job.

My areas are bit more rural, more nature. Villages and hamlets, small cities. Nature is prominent and naming nature is important, many old names but still in active use by forestry, wildlife goers, hunters and locals. I immediately run into basic issues that I’m surprised that they aren’t solved already, but I assume that’s because OSM is focused on densely populated places and perhaps mostly roads rather than nature.

My observations:

** Tagging bays and straits as areas work great, as the renderer gets and idea how prominent it is and it can make proper text sizing and they can be seen even if zoomed out if the area is large. Lots of our lakes, even quite small ones have sub-naming, and with these features I can make really good mapping of this.

** Tagging and naming areas on ground does not seem to be developed much at all, unfortunately.

** There is natural=peninsula so one can tag and name an area of varying size, but it doesn’t seem to render (unless I’ve made some mistake…)

** I can’t make an area to name hills or slopes, which is very common around here (natural=hill would be nice). There’s peak, but that’s only for point for the highest peak with elevation, so it doesn’t fit my purpose.

** Valleys can only be tagged as ways, but here it would make much more sense to make an area, as sizes of these valleys vary a lot, and the renderer need to know which one is more prominent to make sane overview renders.

** Due to limitations in area-based name tagging the map look empty just when zoomed out a little, as names disappear almost directly, so despite detailed mapping and tagging the overview map is just not useful.

** I can’t make a group name. It’s here very common that several smaller islands are named as a group, smaller ponds are named as a group etc. There are tags for that (group/cluster), but not rendered. Alternatives to make it a named multipolygon means that the renderer put the name only in the biggest sub-area, and if naming each area the same the name gets repeated in each area, which looks ugly and clutters the map.

** No good tag for anonymous gravel yards, which there are a lot of. Abandoned quarry is the closest, but still not right, as only some actually were gravel/sand pits to start with. Those gravel yards are often leftovers from construction work or forestry often even locals don’t exactly know when or why they were made.

I’ve tried filing issues on openstreetmap-carto for this, but they always get immediately closed as they don’t really look into adding new features or even recognize these things as a problem, and urge me to discuss these things on some better suited forum, so here I am.

I think these issues are major problems for mapping in Sweden, and for me personally as it takes away from the joy of mapping with quality if the default render that most get to see still is, well, bad. The government maps are easily available here and the typical Swedish online services get their maps from there, so the competition is stiff. International services still often use OSM (facebook, garmin, strava, various generic route plotting services, small independent services etc), so we Swedes still have a need for OSM to be good, and for it to be good it needs to have features that make the map work well in rural areas and nature, as we have a lot of that.

In detailed areas such as major cities the map looks fairly good also here in Sweden, but as soon as it gets more nature with wider spaced features and larger named areas the map doesn’t look good and isn’t useful for getting an overview. The rural areas also have special features that not appear in cities (like these anonymous gravel yards), and it would be nice if these could be properly tagged and rendered.

There are issues in the cities too, I just took a look in Stockholm and there are nature areas there, which naturally have named hills and such too, names which appears properly sized in the government maps (as they are area-defined), and they are not available at all in OSM.

It seems to me that it would be quite simple and non-intrusive to fix most of these issues.

I think the bay/strait type of feature is the way to go (ie fuzzy areas that can be drawn on top, area is not rendered, it’s just a lead where to place and how large to make the text), naming forest etc as an alternative would be extremely messy and in many cases not work as these areas can be split up in say one part forest and one part wetland.

Hi. I’ll discuss one aspect I’m somewhat familiar with, multiple related objects, eg a group of lakes with a common name like Moss Lakes. The best approach seems to be a multipolygon, with the lakes (or islands, etc) being outer members, with the name etc. being recorded only once, as part of the relation. This simplifies maintenance, and lets better renderers (hopefully) deal with it better.

We’re discouraged of course from “mapping for the renderer”, but more specifically, against mapping something incorrectly for the renderer. I’ll mention that in the grouped-lakes-as-multipolygon example, I’ve seen three different behaviours from three different renderers, all reasonable in my opinion:

  • Name on each
  • Name on the largest
  • Name at the centroid of the group

In the latter case, it seems the renderer (openandromaps) also recognizes the larger size of the group, and may show the name at lower zoom levels. Well handled, IMHO.

But keep in mind, there are many different renderers for OSM, so the best approach ultimately is, map it as correctly as possible, and let the renderers worry about their job of presentation. :slight_smile:

Thanks, I’ll go the multipolygon route, seems to be most correct and best chance of being rendered properly.

I don’t think it’s a good strategy for the OSM project that mappers should “blindly” map and hope that renderers sometime in the future will actual show all their hard work. Far from all mappers are deeply experienced and committed long-term, so if a basic feature is not rendered or documented properly today, it will not be mapped. I think understandable results directly and clear documentation on the wiki would be much better. I think it’s of key importance that the default style, the one displayed on, has solid support for all kinds of naming that is required for generating a high quality general-purpose map. And today I think there are some glaring things missing, at least for the mapping needs we have here locally. But I would be surprised if those needs are only local to Sweden.

That these basic features can be missing without that much complaints I think it’s more about that in places with high density and where the map is mostly used very much zoomed in they are not missed as much, but as said when I made a brief look in Stockholm I find places also there where it’s needed, it’s just not as obvious that it’s missing if you don’t know it should be there.

As I understand now (still new to OSM community inner workings), traditionally tagging happens first and rendering later, but I don’t think that is viable today. Someone needs to take the lead or we will be stuck with basic features missing as with the growing size of the community it gets increasingly harder to get features in.

To me the things I listed in the initial post are basic things that is required for making high quality cartography, so it should be uncontroversial to have such information represented in the database. But for some reason it’s not there still today in 2020. Without these features maps rendered from OSM data can never reach similar quality like the general-purpose maps the government provides here locally (for free these days).

Below is an image with examples from renderings of our Swedish government general-purpose maps, which highlight common mapping needs I have that cannot be fulfilled with the current default style and tagging guidelines. Personally I find these to belong to basic cartography and should be uncontroversial and I think can be easily added without disturbing existing data.

The big thing which doesn’t seem to have been much used so far in OSM is to have name tags bound to areas, which gives the renderer possibility to know how large to render the text, which text that are more prominent than others when cluttering needs to be reduced. Where this is used on straits and bays it seems to work very well already with the default style.

Currently OSM is unable to make good overview maps especially in rural areas, but I see problems also in high density areas, as it cannot sort between name label prominence, simply because the database lacks the required information.

The right way to do this, which is not “tagging for the renderer”, but rather the opposite, is to connect tags to areas that are drawn to cover the named area. Some of these will indeed be fuzzy as bays and straits are today (there’s no exact start or end on a hill or valley), but I don’t think that is a problem. Maybe we want to add some extra tag to indicate that it is a fuzzy area.

For hamlets and cities I suppose the information can be inside the point (ie how large it is), I haven’t really got the knowledge how those points are mapped today as I’m mostly mapping rural stuff. Cities prominence is about number of inhabitants, not surface area like the natural features.

I’m no expert on OSM so I may not be able to suggest the best solution to these problems that fit in an OSM context (I’m guessing named areas is the best way and it is already in use with bays/straits, but I’m not sure), but I am quite experienced with using maps and I think these basic naming problems are certainly worth looking into.

It seems that I’m not alone in seeing these issues, but they are rarely highlighted (at least I haven’t found much discussion when searching). I’m afraid that is because it’s just really really tough to introduce the tiniest of new features. As said I’ve tried to take up these things as issues on github, but it leads nowhere. Arguments are not met on point, but diverted in the style “here’s the design goal document read that”, which I have read, and have concluded that fixing these issues are 100% in line with the documented design goals. Also popular evasive action is to point at the commercial styles that exist “use them instead”, but of course these have the same issues for many of these things as the required tagging (and guidelines of how to use them) is not there in OSM.

As I don’t see how it could be any controversy in supporting these to me obvious naming features, I’m just confused.