Render foot/cyclepaths without regard to surface

[pnorman]

Rendering with a different dash pattern to indicate paved, unpaved, or unknown was a good idea, but in practice the difference between them is too subtle to be seen and understood.

The existing rendering also constrained us cartographically, making it near-impossible to balance the patterns against a variety of backgrounds.

I expect this idea to be somewhat controversial, and I like the theory of showing the surface with different dash patterns. It just doesn't work in practice with our wide assortment of backgrounds.

Fixes #1765
Fixes #1793

[jeisenbe]

I agree, the current rendering is not clear. I've always forgotten that there is a difference, because the rendering seems too subtle to notice.

It would be great to have a clear way of distinguishing paved and unpaved paths and cycleways. I would rather show this information instead of access=.

But I would support this change for the moment, until we can develop an effective rendering.

[jeisenbe]

I am in favor of showing footwayand cycleway surface, even if this means removing the access=private/no rendering. However, I approve of this change, because the current rendering is not clear.

It would be ideal to show the paved footways and cycleways with a solid line, but this will require changing the cycleway color since a solid blue line could be confused with a waterway. This can be considered in a separate PR.

Test images:

Before z14

after z14

Before z15 (access=private is shown above)

after z15

Before z16

after z16

Edited: I would like to see if we can improve this first

[jeisenbe]

I'm not sure that #1793 will be fixed by this? Also see previous discussion in PR #1788

[imagico]

I think this would strategically point us in the opposite direction of where we need to be going. And i think this neither fixes #1765 nor #1793 in a convincing and long term viable fashion and it does not fix or promises a path to fix #1748.

This style has moved with past changes (in particular #747) very much in direction towards emphasis of automobile transport and de-emphasis of other modes of transports. This trend should be reversed because one of the key unique selling points of OSM compared to other map data sources is the extensive and differentiated mapping of non-road infrastructure.

There have been plenty of demonstrations already been made how well readable and balanced differentiation of narrow line signatures can be designed - assuming you accepts these being more prominently visible than they are right now - @SomeoneElseOSM has shown that (https://map.atownsend.org.uk/maps/map/map.html#zoom=15&lat=-24.9959&lon=135.05924) - i have also demonstrated it specifically for the problem discussed here (http://blog.imagico.de/drawing-the-lines/).

Frankly if the desire is to reduce differentiation of non-automobile path rendering i would rather remove the difference between footway and cycleway (which is in parts based on an irritating ruleset anyway) and maintain the paved/unpaved difference.

[jeisenbe]

I find your arguments pretty convincing. After looking back at the attempts to solve this previously, I'm not sure if I could quickly replace this with a better design.

@imagico, would you be willing to submit a PR to implement the ideas in (http://blog.imagico.de/drawing-the-lines/) or something similar?

Solid purple, red, brown (and green for bridleways?) lines would not be mistaken for the admin boundaries, but we can work on changing the colors of those if needed.

Let's see if we can fix the rendering this month instead

[imagico]

@imagico, would you be willing to submit a PR to implement the ideas in (http://blog.imagico.de/drawing-the-lines/) or something similar?

I don't think this would work well with the purple boundaries, in particular for access restricted paths. And it would mean a partial revert of #3467.

[jeisenbe]

access restricted paths

I'm willing to lose the distinction of access=no if necessary to render surface= in a sensible way. But the purple boundaries will be changing shortly, I think. Even the slightly desaturated purple (e.g. Lch(47,30,327) is sufficiently different that there will not be a problem implementing the purple cycleways and magenta motorways, based on test images and delta E calculations. I'm also happy to see #3467 reverted partially or entirely.