Bicycle routes at high zoom hide the road infrastructure

[Phyks]

What road is "D10"? https://www.cyclosm.org/#map=15/43.3889/-1.2789/cyclosm

Looks like a small road with little traffic, due to the presence of the bicycle route:

It is actually a secondary https://www.openstreetmap.org/way/74170035!!!

[Florimondable]

link to #480

[Phyks]

The more I think about this, the more I'm inclined to remove completely the bicycle routes, and provide the Waymarked Trails layer.

It really seems to be meant as a dynamic feature at high zoom levels. From my own experience in France, I would completely remove them from Z12/Z13.

It seems perfectly fine for both cities with dense and structured bicycle network https://www.cyclosm.org/#map=13/48.5906/7.7442/cyclosm (current labelling is unusable + network is already obvious from the cycle infra) as well as in the countryside (https://www.cyclosm.org/#map=12/48.6359/4.9226/cyclosm).

My only concern is how this would work with cycle junctions. @hiddewie as you worked on this feature and are living (I think) in an area where usage might differ, would it make sense to have cycle junctions rendered but not the bicycle routes themselves at zoom 12/13 and beyond? https://www.cyclosm.org/#map=13/52.3449/6.5533/cyclosm

[rkflx]

The more I think about this, the more I'm inclined to remove completely the bicycle routes, and provide the Waymarked Trails layer.
It really seems to be meant as a dynamic feature at high zoom levels.

It sounds like you've already made up your mind, so it's probably pointless to argue against removal from low zoom levels. I'm trying anyway, since cycle routes are one of the main reasons users are attracted to CyclOSM (at least IMO, unless you have hard data to prove the contrary?):

I guess we can agree that on high zoom levels routes are a perfect way to get inspired to plan a multi-day trip, i.e. a more touristic use case. Still, after zooming in to see more details, users would now experience routes disappearing, which would be a bit odd, wouldn't it?

However, my main point is that cycle routes can be incredibly useful on a local scale and for low zoom levels too:

• Often there is a good reason why a cycling route is going a particular way instead using another road nearby (note this is a low-level zoom routing choice, i.e. cycle routes only on high zoom levels would not help at that level of detail!). While equal in visualization on the map (either because of all attributes being the same, of some features not being mapped yet), one road might provide way worse conditions (i.e. beyond the existence of cycleways, a speed limit and a particular surface quality) in real life compared to a manually curated cycle route. Even if not planning a long distance tour along the whole route, e.g. for a simple visit across town, it would be beneficial to prefer cycle routes for some segments where it makes sense over alternatives.
• A cycle route does not necessarily require the existence of cycling infrastructure as displayed on the map. It could simply mean a route suitable for cycling, recommended over other less suitable routes in a particular area. Those hints are useful information on a local scale too, which would not be available if removed. Note that the actual name of the route is less relevant here, the shields are mainly to help users understand what the coloring is all about.
• If something is part of a cycle route, there might be other cyclists around, there might be sign posts, the infrastructure might have gotten at least a little bit more fixes for cycling than elsewhere (flat curbs, winter plowing, less bumpy surface, ...), the route selection hopefully was done by someone with some cycling experience etc. See also https://en.wikipedia.org/wiki/Bike_freeway and similar more local projects to build dedicated cycling-only infrastructure: On the map they might appear as just another cycling path, but in reality they could be important, high-capacity and well-built connections explicitly for cyclists, for which cycle routes are an appropriate map visualization.
• In addition, cycle routes allow users to more intuitively understand routing choices of BRouter (with "ignore_cycleroutes" unchecked) by cross-checking with cycle routes on the map. They also help to increase confidence in manually chosen routes.

Waymarked Trails as a replacement are unsuitable: They are probably fine for an initial draft of a long distance route, but for everyday use of the map they are visually distracting, hide much of the underlying map, are cumbersome to add and remove etc. You'd lose all advantages of the integration of features at rendering time: No overlapping labels, coordinated color scheme, well-chosen opacity of lines without having to make labels transparent too, cannot get out of sync due to different database update schedules, and more. In the specific case of CyclOSM, Waymarked Trails would hide cycling infrastructure, road types, road labels and road conditions as well.

Granted, the visualization of cycle routes is not always perfect yet. But instead of removing routes altogether, why not keep them and try to implement improvements instead? Some ideas:

• "Start and end of routes not visible": This could be mitigated by drawing additional shields at the beginning and the end, e.g. styled as a circle or lollipop (vs. the rectangular shape of the normal route shields). This way there would also be a shield at a predictable location at all times (currently when zooming shields jump around a lot, making it hard to determine start and end).
• "Bicycle routes at high zoom hide the road infrastructure": This is not only a problem at high zoom levels, but at low zooms too, and is mainly caused by too much opacity and too much color saturation.
  • By increasing the width a lot, we could afford to decrease opacity and saturation. See OpenCycleMap as an example.
  • Not sure if it would look weird, but we could try rendering routes below the road network (or rather as an additional casing), to create a wide "glow" effect (still somewhat transparent though, to not obstruct nearby features like sidewalks too much).
• Multiple routes on top of each other are confusing, commonly national vs. regional routes: They are hard to distinguish because of their similar color. Again, see OpenCycleMap: It has more distinct colors between categories (which do not pose an issue vs. other map features because they use less opacity and saturation, and more width instead). It is hard to tell if CyclOSM is already doing this (I guess it is?), but not adding multiple transparent routes on top of each other is also a good idea, i.e. only use the color of the most important route, and indicate less important routes by shield only (see OpenCycleMap). Better grouping of shields would also help, so they are visible at a glance (ATM, in some cases you have to zoom and pan to eventually read all shields of a particular location).
• Route colors (and styling in general) are too similar to cycleways, paths and speed limit indicators: See above for ideas. In general it should be fine if explicit cycling infrastructure like a cycleway is rendered more prominently than "virtual" infrastructure such as a cycle route (completely hiding cycle routes in those cases is probably not a good idea, though).
• Road colors different from white (e.g. primary/secondary roads) are hard to see: This is not ideal, but not that bad either: Often the type of road can additionally be inferred from its width as well by its numbering scheme. Of course, not drawing cycle routes directly on top would be better.
• (Rendering routes names at Z17 should be kept, they are much better search terms than the abbreviated shield labels to find out more.)

[Phyks]

Hi @rkflx, thanks for the detailed answer. My mind is not completely made up, and I'm very happy to hear feedbacks.

First, I think we are not talking about the same thing for high and low zoom levels. OSM zoom levels go from 0 (world map) to 20 (very local area). Therefore, for me low zoom level are zoom <= 12 (regional views) and high zoom levels are zoom from about 12 and above (city / neighborhood views).

I guess we can agree that on high zoom levels routes are a perfect way to get inspired to plan a multi-day trip, i.e. a more touristic use case. Still, after zooming in to see more details, users would now experience routes disappearing, which would be a bit odd, wouldn't it?

Definitely 👍, I'm anyway planning to keep them at (my) low zoom levels, https://www.cyclosm.org/#map=10/45.2836/-0.0989/cyclosm for instance. It is an issue at (my) high zoom levels where:

• It tends to mask the underlying road type, which is a pity https://www.cyclosm.org/#map=15/43.3320/-1.4531/cyclosm
• It clutters the map in cities with information which is basically a duplicate of cycle network https://www.cyclosm.org/#map=12/48.8455/2.3521/cyclosm.

Often there is a good reason why a cycling route is going a particular way instead using another road nearby (note this is a low-level zoom routing choice, i.e. cycle routes only on high zoom levels would not help at that level of detail!). While equal in visualization on the map

This is, sadly, not my experience here in France. Hence, why I ask for feedbacks from different parts of the world or different experiences.

Some salient examples I came across lately being:

• http://brouter.de/brouter-web/#map=14/43.3792/-1.7220/cyclosm&lonlats=-1.664343,43.384779;-1.782446,43.372239 this is supposed to be part of a EuroVelo route (so, definitely good for cycling + diversity of profiles including families). Actually, this is a very crowdy (with car traffic) and hilly secondary https://www.openstreetmap.org/#map=14/43.3824/-1.7303, with absolutely no bicycle infrastructure nor usable roadside. Definitely not something you want to be riding when back-country offers you a very cycle-friendly (very low car traffic, no axial marking road) similarly hilly route http://brouter.de/brouter-web/#map=14/43.3802/-1.7116/cyclosm&lonlats=-1.664343,43.384779;-1.711636,43.377667;-1.782446,43.372239 with a very nice (if not better) view as well!
• http://brouter.de/brouter-web/#map=13/43.3435/-1.6263/cyclosm&lonlats=-1.663227,43.386708;-1.600571,43.32493 this is exactly the same (although this one is actually a local route, advertised for sport cycling, which is not possible to emphasize from OSM cycle route tagging) : a very car-crowdy secondary with no bicycle infra. If you are not into sport cycling, but like to ride in quiet places, once again, you can find a cosy route nearby http://brouter.de/brouter-web/#map=13/43.3583/-1.6292/cyclosm&lonlats=-1.663227,43.386708;-1.644773,43.370931;-1.608853,43.331203;-1.600571,43.32493.
• Situation is the same further away: http://brouter.de/brouter-web/#map=15/43.3503/-1.4214/cyclosm&lonlats=-1.400414,43.360247;-1.446419,43.337783 vs http://brouter.de/brouter-web/#map=15/43.3472/-1.4226/cyclosm&lonlats=-1.400414,43.360247;-1.40213,43.357419;-1.414275,43.353738;-1.428566,43.346282;-1.446419,43.337783

Note my examples are very local here (same region), but this is something I encountered widely around France. This is not so much about this specific area, but rather about France having many "cycle routes" which are actually designed for people willing to ride "Tour de France"-like sport cycling. This is not something which is currently possible to infer from OSM tagging, see #283

Waymarked Trails as a replacement are unsuitable

I definitely hear your argument here. Brouter-web has the ability to change the opacity of the waymarked trails layer, which is a nice and must-have option. This might not be the case of other softwares :/

"Bicycle routes at high zoom hide the road infrastructure": This is not only a problem at high zoom levels, but at low zooms too, and is mainly caused by too much opacity and too much color saturation.

At https://www.cyclosm.org/#map=11/43.3716/-1.3705/cyclosm, I'm mainly looking at planning my multi-day trip, drawing general directions. Therefore, I would not mind too much about the underlying cycle infrastructure. Once my main stops are set, I'd zoom in (probably using BRouter as well) and draw the detail of my route. At https://www.cyclosm.org/#map=15/43.3470/-1.4313/cyclosm, I want to get a feel of whether the "6" route, which offers no bicycle infra, is using a primary / secondary / tertiary, and how much car traffic I should expect.

By increasing the width a lot, we could afford to decrease opacity and saturation. See OpenCycleMap as an example.

Indeed, but this is also putting a lot (too much, in my opinion) emphasis on cycle routes taking secondary roads. See:

I can imagine as a newcomer and not putting too much thoughts into this thinking: "wow, this is blue, so it is bicycle friendly, this is large, so it's a really nice cycle route", while the message is actually the opposite.

Not sure if it would look weird, but we could try rendering routes below the road network (or rather as an additional casing), to create a wide "glow" effect (still somewhat transparent though, to not obstruct nearby features like sidewalks too much).

I really like this idea! I'll try to come up with a PoC, this could actually be a very nice option to keep cycle routes as a hint and find some middle ground between relying purely on Waymarked Trails and keeping as it currently is.

[Florimondable]

I agree with @rkflx cycle route is also very usefull at high zoom, and I'm very against removing it (I prefer switching to another map to check the highway tag of a road than switching to waymarked trail).

We must try to improve the current cycle route rendering (link to #480 for path/track cycle route issue)
Ideas :

• render cycle route as thin line in the middle of the road at high zooms (don't solve MTB visibility at high and low zooms #480)
• render cycle route in dash line at high zoom
• render cycle route under road/path (solve MTB visibility at high and low zooms #480)
• ...

[hiddewie]

@hiddewie as you worked on this feature and are living (I think) in an area where usage might differ, would it make sense to have cycle junctions rendered but not the bicycle routes themselves at zoom 12/13 and beyond?

Having the junctions rendered without the routes would be pretty strange, and I think maybe useless.

The junctions are connected together by (many) local routes, and if you consult a map to find the way to a junction, it is very useful to see the routes connecting the junctions together. On the junctions are maps like https://images.fietsnetwerk.nl/img/news/de-beste-manier-om-een-fietsroute-te-plannen-langs-knooppunten/fietsroutenetwerkknooppunten.jpg, and on the routes between the junctions are many (small) signs like http://www.fietsknooppunten.eu/wp-content/uploads/2011/01/bewegwijzering-200x200.jpg

In The Netherlands, if there is a cycling route going over a road, this is always a road you can perfectly cycle on (even with children). Local routes, regional routes and national routes. Because there is a cycling route on the road, in The Netherlands it does not really matter what type of road it is.

I have no experience with MTB routes, they are different and I don't know how people consult MTB routes on a (digital) map. MTB routes are shorter and usually local (in a forest, or in a small region). If I can, I filter MTB routes from my maps.

---

A small overview of the zoom levels and how I would personally read the map for cycling routes:

• zoom 9-11: national routes, get an overview of a region where there are many cycling routes
• zoom 12-13: during cycling, finding the way
• zoom 14-15: navigating through city

Even for zoom 14-15 it is useful to see which roads have a cycling route, which I might follow.

Also, for me personally the routes are more important than the roads with cycling lanes, because I want to follow the route and not the cycling lanes. https://www.cyclosm.org/#map=13/51.8640/5.7783/cyclosm