Talk:Proposed features/crossing:markings

From OpenStreetMap Wiki
Jump to navigation Jump to search

Looks good

Implied values will not burden mappers in countries with consistently applied markings: e.g. here in Austria, it is all Zebra plus sign or signal; it lets mappers from other countries where details vary a lot give all the details. May be especially useful in places, where the implications of markings on legal status of pedestrians or drivers cannot be read from a single base tag. --Hungerburg (talk) 22:23, 10 July 2022 (UTC)

PS: Can this considered a starting point, as work on an ontology, that can capture all the different customs there are worldwide? So crossing=traffic_signals would become a shorthand, much like an editor preset, that could be elaborated in a multiple-choice fashion into a list of crossing:*=* subkeys? In my area, it would come out crossing:markings=zebra + crossing:signals=vertical_redgreen. A mechanical edit, even if performed by an editor in its validation stage, should keep the legacy crossing key in place though. --Hungerburg (talk) 19:06, 11 July 2022 (UTC)

Surface vs. paint

Resolved: cleared up confusion, but also see discussion on surface vs markings below

--Popball (talk) 12:46, 31 July 2022 (UTC)

Example 4, Fireman's crossing, is using surface to mimic paint. Two rules apply! --Hungerburg (talk) 22:23, 10 July 2022 (UTC)

In my definition, I noted that the markings are often done with paint or bright bricks. I said this because something like this:
Cobblestone zebra crossing crosswalk (27813061567).jpg
is clearly marked. This is because road markings in general are not necessarily paint. See
Certainly these:
Rails de trams (33819684188).jpg
are road markings, even if they are made of bricks and not paint.--Popball (talk) 00:02, 11 July 2022 (UTC)
I see, it is only the bright bricks that count; The picture of Fireman's mislead me into thinking, it actually showed a "solid" marking, due to the red bricks. In any case, bricks are much more durable than paint. --Hungerburg (talk) 19:15, 11 July 2022 (UTC)

Marked, marking, markings

Resolved: added reason for naming scheme to proposal

--Popball (talk)

crossing:marked=* is already used 240 times, and crossing:marking=* is documented for all of 22 occurrences, but crossing:markings=* has yet to be used or documented. This proposal should state why crossing:markings=* is considered the superior key name of the three or how it would coexist with the other two. – Minh Nguyễn 💬 07:51, 11 July 2022 (UTC)

The page for crossing:marking=* appears to have been created two days ago, and I was not aware of it. I chose it over crossing:marked=* because it is not a binary value, and over crossing:marking=* as it is multiple road markings in a pattern (at least usually).Ultimately I'm mostly indifferent to the name, and if the community prefers a different key name, I'm all for it.--Popball (talk) 08:13, 11 July 2022 (UTC)
I don't have any problem with crossing:markings=*; it just seemed like a question for the proposal to get out of the way. – Minh Nguyễn 💬 03:37, 12 July 2022 (UTC)


Resolved: I updated the rationale with the feedback given here

--Popball (talk) 10:47, 3 August 2022 (UTC)

Don't pedestrian crossing markings usually have only one design per country? Adding one or two additional tags (crossing:markings=* + crossing:marking:colour=*) to every single pedestrian crossing if they all look the same seems like a waste of time and memory space and brings no additional information. If you really think that the design of pedestrian crossing markings is worth mapping (IMHO there is much more important information that is currently missing on OSM), then it would make more sense to store such information as country-wide defaults. --Dafadllyn (talk) 16:01, 11 July 2022 (UTC)

They can vary within a country. The biggest example of this is the US, where the types of crosswalks are often installed based on a risk assessment. (Higher risk -> more visible crosswalk designs) The US MUTCD (i.e. the US road design standards bible) defines several designs to be used based on speed and traffic volumes. In any case, I'll add this bit of background to the proposal to make it more understandable to more people --Popball (talk) 16:32, 11 July 2022 (UTC)
Thanks for the clarification. Still, i don't see a value of adding this tag in countries which only have one type of pedestrian crossing markings. --Dafadllyn (talk) 19:37, 11 July 2022 (UTC)
crossing:markings=yes would be an option for any mapper who doesn't want to go into detail about the style of marking, just as crossing_ref=* is largely optional today. – Minh Nguyễn 💬 03:36, 12 July 2022 (UTC)
crossing:markings=yes/no still gives zero additional information in countries where all crossing=uncontrolled/crossing=marked and/or crossing=traffic_signals look the same. If this tag won't be kind of banned in these countries, it will result in a waste of time, waste of memory space and unnecessary new object versions. --Dafadllyn (talk) 16:21, 18 July 2022 (UTC)
IMO we shouldn't be too concerned about the memory usage as much as an accurate data model; the entire world in OSM is only 64GB compressed. In terms of time, well, if a mapper thinks something is a waste of time, they won't map it. Perhaps to another its something useful and they will map it. --Popball (talk) 17:05, 18 July 2022 (UTC)
Yeah, memory space isn't the primary concern. But expanding "German zebra crossing" to half a dozen tags degrades usability for every mapper who needs to edit or review the object later. Presets do not solve that problem either, or rather, they only solve it for the first mapper, not everyone who comes after. That doesn't mean this proposal is bad – it's quite useful! But it does mean that we, in a separate effort, should find a good way to map standard crossing types, maybe a cleaned-up crossing_ref=*, and should then consider discouraging the use of crossing:markings=* if the standard markings for that particular crossing type are used. --Tordanik 11:53, 23 July 2022 (UTC)
The situation may be similar to surface:colour=*. Certainly we can assume that surface=asphalt will be black or gray, but we are able to capture the times where it does deviate (red asphalt for in this case) --Popball (talk) 12:21, 23 July 2022 (UTC)

@Tordanik: Thanks for pointing out that presets only solve the usability problem when first tagged. I suppose this problem would be most prominent for JOSM users, which mostly only uses presets in that initial phase; by contrast, the presets in other editors show up for everyone who comes after as well. On the other hand, railway=crossing has had a variety of companion subkeys for a long time, without much in the way of implied tags, and it doesn't seem to have posed a significant problem. I think it would be great to ensure consistency between the two kinds of pedestrian crossings.

crossing_ref=* badly needs a cleanup. Part of the problem with that key is that it's being used for a variety of purposes. This proposal would chip away at the problem by partially deconflicting the key. For instance, I don't have to choose between indicating the marking style or signal style of a HAWK crossing, which could have any of several marking styles. In that light, it reminds me of the successful push to replace crossing=island with crossing:island=yes. I have no idea if crossing_ref=zebra implies crossing_ref=zebra in the UK, as it happens to on the ground there, but if the key's values were namespaced by country, it would be more practical to expect that inference.

 – Minh Nguyễn 💬 06:13, 28 July 2022 (UTC)

crossing:island vs double_zebra

Resolved: Updated example to reduce confusion

Is there a significant difference between crossing:markings=double_zebra and crossing:island=yes + crossing:markings=zebra? --CjMalone (talk) 10:06, 13 July 2022 (UTC)

@CjMalone: I think crosing:markings=double_zebra is intended to represent zebra stripes that have a line down the middle, like this: ¦¦¦¦¦¦¦¦¦. The example is slightly confusing because it focuses on the island that happens to interrupt that striped pattern, but I don't think that's what the tag would be about. – Minh Nguyễn 💬 01:57, 14 July 2022 (UTC)
Thank you, that makes sense. --CjMalone (talk) 07:05, 15 July 2022 (UTC)
I replaced the photo with one that doesn't have a traffic island to avoid confusion. – Minh Nguyễn 💬 18:50, 15 July 2022 (UTC)

New U.S. terminology for high-visibility crosswalks

Resolved: Added new designs

--Popball (talk) 12:49, 31 July 2022 (UTC)

The U.S. Federal Highway Administration is funding a new standard for high-visibility crosswalk markings that's due out soon. It seems to be adopting new terminology for crosswalk marking styles, for example "longitudinal bar" for what had been called "continental" (zebra). [1] Basic "transverse" markings will be deprecated in favor of high-visibility crosswalks. [2]

There's also a new "double-paired" style, which has already made its way into some state standards [3][4] and is being used in Seattle. [5] Apparently the purpose of this style is to offset the bars away from typical wheel paths where they would get worn away. [6]

 – Minh Nguyễn 💬 19:11, 15 July 2022 (UTC)

Thanks for looking into this! I'll add "paired_zebra" and "paired_ladder" as they appear to be in the books as standard designs. In terms of terminology, I don't believe there should be any change other than noting the terminology in the tag description. --Popball (talk) 10:17, 18 July 2022 (UTC)

Missing tagging possibilities for markings in Germany / Austria

Resolved: New schemes added to the proposal, as well as examples

--Popball (talk) 11:42, 4 August 2022 (UTC)

I miss the ability to define bike lane crossings at intersections in the proposals. Spontaneously, I can name two markings that are frequently used in my area, but currently can't really be covered well with crossing=*.

There are bicycle and/or pedestrian crossings that consist of only a dashed line on the left and right. See photo here: At least in my region, these crossings are mostly for bike lanes only. So far I tag this with crossing=marked and bicycle=yes. But with this it can't be distinguished with a zebra-crosswalk. Especially since in Austria there are also zebra stripes with offset dashed line on both sides, which are released for bicyclists. See:

Furthermore, there are explicit bicycle crossings at intersections, which are exclusively for cyclists and consist of a red lane marking and white dashed lines. These can, but do not have to be in combination with a bicycle protection lane. Example:

—Preceding unsigned comment added by Mcliquid (talkcontribs) 08:41, 18 July 2022‎ (UTC)

@Mcliquid: Would your first example be adequately described by crossing:markings=dashed_transverse? The article illustrates that tag with a pelican crossing in the UK, but should it be distinct from the thicker style you see in Germany, which apparently is used for a different purpose?

This proposal seems to call for your second example to be tagged as crossing:markings=solid;dashed_transverse, perhaps with colour=red.

 – Minh Nguyễn 💬 08:48, 18 July 2022 (UTC)

@Minh Nguyen: Yes, the only difference would be the traffic light, which then does not exist in my examples.
I was initially put off by the wording "transverse" as I had understood this to mean "diagonal". But as already shown in dashed_transverse, it should fit. —Preceding unsigned comment added by Mcliquid (talkcontribs) 09:06, 18 July 2022‎ (UTC)
Picture is from Germany. It shows a cycle-ford (that is what they call that). The dashed lines mark the continuation of the bidirectional cycleway over the carriageway of the lower priority street at the intersection. The cycleway further down to the background is doubling as a footway, non-segregated, while at the intersection it looks like pedestrian and cyclist paths separate. The other dashed line marks the pedestrian-ford (as they call it.) I'd say, the dashed_transverse fits both. I would not map a solid marking, because in my mind the red paint does not mark the crossing, cycleways sometimes are red too where there is no crossing. --Hungerburg (talk) 21:09, 23 July 2022 (UTC)
Picture was shot in Austria. If it was not for the extra squares besides the zebra, the young lady would not be allowed to cycle there in direction of the crosswalk. Corresponding signage can be seen too. This is a fairly new feature. It combines into a single feature, what can be seen here as accompanying, rather than separate feature. The legislation calls that a cycle-crossing, if on its own, it is also a a case for dashed_transverse - in a picture from Czechia. The combinations with a zebra, when mapped as a single node, if there is consensus on using semicolons in the value, might be called zebra;dashed_transverse. Tagging bicycle=yes and segregated=yes|no remains due of course, even tells a bit more about the arrangement of paint there. --Hungerburg (talk) 19:49, 24 July 2022 (UTC)
I reformatted the names, so now it is easier to create refinements of marking types ypes. In terms of the Austrian combined zebra crossing, my first instinct is to call it one feature crossing:markings=zebra:block_edge (it seems the markings are called Blockmarkierungen) however I'm not sure how descriptive of a name this is. But maybe you can think of something nicer! Alternatively crossing:markings=zebra;transverse:dashed might be better. In the case of separated crossings, there is already footway:crossing:markings=* and cycleway:crossing:markings=* (along with segregated=*)--Popball (talk) 14:08, 26 July 2022 (UTC)
From my understanding, this can be marked "solved". That is how I read Mcliquid. For completeness, a drawing of the markings.
Markings of Pedestrian and Cyclist Crossings in Austria (and other countries)
I'd call the pattern transverse:dotted, from my background of map reading, that is; but I rather not complicate the issue beyond the reasonable. --Hungerburg (talk) 22:17, 27 July 2022 (UTC)

Dot vs dash

Resolved: Dots and dashes added to proposal

--Popball (talk) 11:43, 4 August 2022 (UTC)

Not sure this is directly relevant Your example photo for dashed (whatever the tagging will be) looks more like "dotted" (rectangular strips vs square-like spots). Seems this ambiguity comes from the US terminology. (in fact, dashed line is called "broken" line in UK officially; but I don't think that should be used). Painted dots (or studs) are used on zebra and pelican etc signalized crosswalks. Problem is dashed line is used on zebra crossings as a give-way line. Often there can be dashed line and white zebra stripes, but without studs or painted dots. This risks mistaking the give-way line as the crosswalk border (if the user is not familiar with markings). Furthermore, as this sections points to, on parallel crossings, both dashes (give-way) and dots ("elephant's foot" for bikes) are used. Therefore I concur with System-users-3.svgHungerburg in renaming the currently illustrated "dashed" to "dotted". --- Kovposch (talk) 12:08, 29 July 2022 (UTC)

A great leap in the right direction for crossing tagging

Tagging pedestrian crossing can be a bit of a mess in the edge cases, and this proposal neatly addresses one dimension. It is clearly defined which physical aspects of the road are covered, and it is simple to understand exactly how the crossing looks. And we get to leave crossing_ref to do its specific job.

The harmony with other crossing: tags is a great vote of confidence and future uses of :markings tags look interesting. The documentation that is provided in the proposal itself sets a good standard too.

I am all-around in favour of this proposal and look forward to seeing it succeed. --BudgieInWA (talk) 04:44, 21 July 2022 (UTC)


Resolved: The proposal in its current state is now much more intelligible for non-technical users. --Hungerburg (talk) 21:19, 31 July 2022 (UTC)

Only now I consulted the dictionary, how transverse will be translated. So I see, why this caused problems, see above, where it was understood to mean oblique. In a strictly mathematical (see wikipedia) meaning, it is the line from one kerb to the other across the carriageway? To the native speakers, is there a way to express that, not taking the view of the motorists? As pedestrian/cyclist, the lines for me are longitudinal.--Hungerburg (talk) 22:51, 27 July 2022 (UTC)

@Hungerburg: I assume this proposal has adopted the term "transverse" from U.S. transportation engineering parlance, since European standards classify markings by purpose or zoo animals rather than visual style. In a typical reflection of the car culture here, Americans refer to road markings that cross the road as "transverse", not only for crosswalks but also for other markings like stop lines. Here, "longitudinal" would present the opposite problem, because the zebra pattern is described as "longitudinal", even among pedestrian and cycling advocates. Ultimately, crosswalk markings represent guidance to all road users – pedestrians, cyclists, and motorists – and even the term "crossing" is understood to mean that nonmotorized traffic crosses the road rather than vice versa. Rather than invert the terminology, maybe we could find a concise term that's less ambiguous. – Minh Nguyễn 💬 05:48, 28 July 2022 (UTC)
First of all, I have to say I'm opposed to *=*:*. It's too cumbersome. surface=concrete:* is not good, compared to surface=paving_stones + paving_stones:*=*
This depends on the reference. highway=crossing is more defined on the road, while highway=*way + *way=crossing is transverse to it. It's difficult to make them work together
In OSM, *:across=* is proposed in incline:across=*. Should be useful to generalize it. However, I'm doubtful whether they can be defined consistently.
--- Kovposch (talk) 07:04, 28 July 2022 (UTC)
@Minh Nguyen: You're right about it being from US engineering jargon; I'm not aware of any British term for it (as OSM prefers British English terms), so I chose the most widespread term anyway. I don't think it causes any actual confusion, but I'm open to suggestions. --Popball (talk) 23:10, 28 July 2022 (UTC)
It's used on lane and junction traffic control lines, but not for crosswalks markings. Kovposch (talk) 07:59, 29 July 2022 (UTC)
@Popball: Please consider "lines:solid|dashed|dotted". I do not think anybody will have problems differentiating lines (that line the crosswalk) from ladders and zebras. The dictionary says about : "in a position or direction that is at an angle of 90° to something else." It is synonymous with orthogonal and perpendicular. Here I read, "transverse lines on either side" @Minh Nguyen:, or "transverse lines where diagonal crossing …". This is nonense anywhere except maybe some engineering circles. The use of such jargon may sound sophisticated, but will make translations really hard. For the sake of simplicity, let us call lines lines and move the term transverse to the description. --Hungerburg (talk) 08:41, 29 July 2022 (UTC)
@Hungerburg: I agree, this is probably the right way to go. The only thought I had was whether the namespacing is necessary in this case at all, as I'd like to avoid it if not ambigious for these common values. So perhaps simply crossing:markings=lines crossing:markings=dashed crossing:markings=dotted wouldn't cause any confusion and would be easier to type out. --Popball (talk) 09:27, 29 July 2022 (UTC)
@Popball: I came here mostly, because I see this proposal as a chance to get rid of the crossing-ref=zebra, that are plain wrong in my area. I have no idea how consumers are helped or hindered by namespacing a.k.a. hierarchy building. I'd welcome solid vs. dashed vs. dotted as top level members of the collection of values. They have quite different uses. I happen to look at raw tags too, and consider those first citizens, however potent the wizards in some editors are, so easy reading always welcome, especially where editors need time to catch up with developments.
crossing:markings=lined|dashed|dotted, which are adjectives, still is an outlier from zebra|ladder, which are nouns, but this only due to grammar, and grammar is negligible - Note, I wrote "lined", not "lines", I do care about grammar, to some extent :) crossing:markings=lines|dashes|dots should also be fine, grammatically. --Hungerburg (talk) 23:20, 29 July 2022 (UTC)
The Austrian "Bodenmarkierungsverordnung" (regulation of road markings) knows "Längsmarkierungen" (longitudinal markings) and "Quermarkierungen" (transverse markings). Among the latter are "Haltelinien" (stop-lines), "Schutzwege" (cross-walks) and Radfahrerüberfahrten" (cyclist-crossings). No animals were harmed in this production ;)
From that perspective, all of the markings in this proposal are transverse. Even though some (ladder) or all (zebra) the lines run longitudinal to the carriageway. Transverse is a bit an outlier in the naming convention, it is not about shape, but direction, based on a different categorisation.
That leads me to crossing:markings=solid_lines|dashed_lines|dotted_lines. Or lines:solid, lines:dashed, if you want the extra colon. --Hungerburg (talk) 09:28, 28 July 2022 (UTC)
@Hungerburg: For completeness, the Vienna Convention refers to the components of a pedestrian crossing marking as "stripes" (as in zebra stripes) and the components of a cyclist crossing as "lines". In general, markings that span the road are called "transverse"; these markings are perpendicular to "longitudinal" markings. [7] – Minh Nguyễn 💬 18:22, 29 July 2022 (UTC)

Pedestrian rail crossings

Resolved: Added to the proposal that this can be used with railways as well

--Popball (talk) 11:08, 3 August 2022 (UTC)

I think crossing:markings=* should also be allowed in combination with railway=crossing. Where I live, most pedestrian rail crossings have no markings, but a nontrivial number have transverse markings, often but not exclusively at tram crossings. Usually it's just standard (transverse) markings, but I've seen continental (zebra) and ladder (ladder) markings as well. Some are also controlled by traffic lights and pedestrian signals, so there's a conflict with crossing=traffic_signals. – Minh Nguyễn 💬 06:43, 28 July 2022 (UTC)

Agreed, the scheme is already consistent with the other crossing:*=* tags already in use with railroads, I see no reason it shouldn't be used here as well. --Popball (talk) 22:49, 28 July 2022 (UTC)

Marking vs pavement

Resolved: Clarified "markings" in the proposal as well as add examples of what it is not

I find *=solid as mixing up the fill pattern for a marking and surfacing. *=zebra may be described as solid bars, while *=zebra:paried is a pair of bars. *=transverse and *=ladder are solid lines. If the *=no example is changed to a paving material of different colour, it's suddenly *=solid, which doesn't make sense to me. What if an concrete road is repaved with asphalt at some locations or vice versa? Obviously it should not be considered a colored crosswalk. Moreover, there is a risk of conflating the coloring of humps and tables (if done in some countries), or bus and bike lanes. Kovposch (talk) 07:13, 28 July 2022 (UTC)

This crosswalk is painted in various colors, but it isn't a solid marking; the white transverse lines are still required for it to be a legal crosswalk.
@Kovposch: I suspect the solid value comes from U.S. standards, despite the example image from Germany. It's actually one of the oldest styles of painted crosswalk markings in the U.S. but somewhat rare, unrelated to aesthetic treatments or the recent trend of painting lanes in various colors. I've typically seen them around train stations or on traffic_calming=table. The FHWA makes a clear distinction: only high-contrast white (or in some states, yellow) solid paint can count on its own as a legal crosswalk. Otherwise, there must be high-contrast transverse lines on either side for it to be a legal crosswalk. Concrete and brick pavers don't count as solid. The combination of transverse lines and solid paint between the lines is also prohibited. [8] I think it's worth keeping a distinction between crossing:markings=solid and crossing:markings=transverse surface=paving_stones as long as there's a jurisdiction where this distinction matters in reality. – Minh Nguyễn 💬 22:10, 28 July 2022 (UTC)
I do think it'd be nice for this proposal to at least suggest a tagging for crosswalks with murals or decorative features, whether they be painted or created using colored paving stones. FHWA-endorsed or not, these treatments are becoming increasingly common (in the U.S., at least), and certainly are interpreted by all but the bureaucrats as part of the markings for the crosswalk, enhancing its visibility. I typically see them in pedestrian-oriented areas. I suspect people would be interested in mapping them in detail too. I'd be tempted to coin something like crossing:markings=art or mural for the rainbow crosswalk you post here, but if you really desire to keep the crossing:markings=transverse for the legal reasons, maybe crossing:markings:mural=yes? Though that's a lot of modifiers... Maybe adding tourism=artwork+artwork_type=mural could work too, though people might not like how that renders. In any case, something to consider. --Willkmis (talk)

@Willkmis: I agree that it's desirable to tag art crosswalks explicitly. artwork_type=mural seems perfectly logical to me. It's even possible to dual-tag with area:highway=footway on an area. Reusing the artwork tagging scheme allows us to indicate a number of other details, such as the artist, without having to derive new keys specific to art crosswalks. It's true that openstreetmap-carto renders murals awkwardly, but crossing:markings=art won't solve that for most murals anyways.

As for colored paving stones, surface=paving_stones is already commonly used in these situations. I think that should be adequate, though it raises the question of whether colour=* should refer to the markings or the surface in that situation. colour=* is often used to distinguish white markings from yellow markings etc.

 – Minh Nguyễn 💬 10:45, 29 July 2022 (UTC)

Thanks, I forgot to mention rainbow and other coloring. As for pavers, in Proposed_features/crossing:markings#Examples photo 4, surface=paving_stones is visibly colored different to the road. More uniformly, Red bricks are commonly used.
Here are US examples without solid white lines.
(can't find the European cases that are cheap substituents to a raised crosswalks in my mind)
Can that be considered *=transverse;solid? That's why I don't like *=solid.
Besides, someone may be interested in paving_stones:*=* patterns. Such as this herringbone example.
So I'm of the position "solid" colored patch should be reserved for surface=* tagging, independent of its legal meaning.
--- Kovposch (talk) 12:17, 29 July 2022 (UTC)
@Kovposch: For what it's worth, the U.S. examples you cited are considered to be unmarked, at least from a legal perspective. I'm unsure about the law in Virginia, where those photos come from, but some other states have relatively strict laws on when pedestrians have the right of way, requiring the crosswalk to be properly marked. A low-contrast change in pavement quality wouldn't suffice. In practice, pedestrians may consider the crosswalks to be marked, but motorists may not; this is the safety problem FHWA cites when requiring supplementary transverse markings. – Minh Nguyễn 💬 18:17, 29 July 2022 (UTC)
For bikes, this is what I'm think of:
That's solid blue, but it's an indication for bikes, not exactly the crosswalk. See also red ones in Netherlands. Again, there can be no other crossing markings.
They make another reason for leaving "solid" to surfacing to me.
--- Kovposch (talk) 12:38, 29 July 2022 (UTC)
@Kovposch: To add to what you've shown, here's one that I found out in the wild in Germany:
Cycleway crossing driveway.jpg
The cycleway was painted red where the driveway crosses the segregated cycle and foot path. Somewhat interestingly, the footway was left as is. The color of the paint matches what is used for cycle lanes in the area. Cars must yield to bicycle and pedestrians here, but not because its a crosswalk, but rather because turning traffic must yield to pedestrians/cyclists on the cycleway/sidewalk, and traffic entering the road (coming from the right) must yield to traffic on the road (it's a driveway after all).
In any case, I think that all the examples you have shown are not "marked" for the purpaces of crossing:markings=* and that surface=* or surface:colour=* should be used in these cases.
I do remain convinced that a crossing done in all high visibility white paint should be counted in the scheme, but as Mihn has pointed out, these seem to have gone out of fashion (probably because they are less safe/more expensive than the alternatives). I think that perhaps the best solution is to simply describe this in the key description. This may also help clear up confusion that already exists, as I'm certain that many crossings similar to the ones you've shown are currently tagged as crossing=marked. --Popball (talk) 14:30, 30 July 2022 (UTC)

To me, the bigger issue about solid is that it's very difficult to find an example of it being used as a high-visibility crosswalk marking per se, as opposed to a bike lane marking or speed table marking. The train stations I'm thinking of have since been remodeled, removing these markings. They may still be allowed, technically, but maybe they've fallen out of use. Maybe they doesn't merit inclusion in the tagging scheme right off the bat, creating confusion, until we find examples of them in the wild. – Minh Nguyễn 💬 18:17, 29 July 2022 (UTC)

Indeed, the picture is from Germany. The regulations there do not mention neither the marking nor the sign, that tries to mimic the official one. While official regulations apply also to private parking spots like this one, this "crossing" is not legally binding for motorists or pedestrians. Yet, this is openstreetmap after all, not an outlet of governmental data, but fed by user contributions. So how to handle such cases? No idea, but certainly, from my point of view, it is not good enough to make "solid" worth a top level entry in the proposed scheme. --Hungerburg (talk) 22:41, 29 July 2022 (UTC)

@Kovposch, Hungerburg, and Minh Nguyen: I updated the definition and the tag description for crossing:markings=solid. Feel free to let me know how well this clears things up and whether I should change/add anything else. --Popball (talk) 23:57, 2 August 2022 (UTC)

Documenting the availability of the value Solid will not make me disapprove of the proposal. Maybe one day, it can even serve as a means to detect unmarked crossings :) --Hungerburg (talk) 20:41, 4 August 2022 (UTC)
@Hungerburg: by detect unarked crossings you mean that because true solid markings are rare, it would be a good indication that it may be crossing:markings=no? In this case I agree, and anyway I don't think that leaving out solid from the documentation would stop mappers from tagging coloured but unmarked crossings as crossing:markings=yes or something else. In any case, I think this issue is resolved and I think the proposal is more clear because of it :) --Popball (talk) 11:36, 5 August 2022 (UTC)

Markings vs studs

Resolved: Added that studs can be used to mark the crossing

--Popball (talk) 08:55, 3 August 2022 (UTC)

In Commonwealth jurisdictions, silvery metallic road studs were commonly installed along a crosswalk. They are now often replaced by white short-dashed/dotted lines. How do we draw the line between markings and other devices? Studs is asked on tactile_paving=* in Talk:Key:tactile_paving#Should_pedestrian_crossing_studs_be_tagged_as_tactile_paving_on_the_way?. --- Kovposch (talk) 07:17, 28 July 2022 (UTC)

Given that in these countries they mean the same thing and are often replaced with a regular marking (i.e. equivelent in the eyes of the road authority), I'm fairly certain they should be included. Wikipedia includes them in the article on road surface markings if thats any further confirmation. --Popball (talk) 23:28, 28 July 2022 (UTC)

Text visbility and roadway vs sidewalk marking

Resolved: Clarified this in the proposal

--Popball (talk) 11:03, 3 August 2022 (UTC)

Is eg look left/right on the roadway, alone or together, a crosswalk marking? For wider roads, their size makes them only seen by pedestrians, or it may be blocked by (illegal) parking. On the contrary, on narrower roads they can take up half the lane.
Sometimes there are very long warning messages that take up the entire lane.
Furthermore, in some places there are even text marked on the sidewalk at the kerb or before it. Do they count? Technically, that section would not be footway=crossing. But it is relevant to the crosswalk.
All these situations can be found on railway=crossing as well. Kovposch (talk) 07:29, 28 July 2022 (UTC)

@Kovposch: I updated the proposal so that this does not refer to look left/look right. They serve a diffrent purpace and can exist with or without a normal marking. --Popball (talk) 23:14, 28 July 2022 (UTC)

Edge cases: crossings with multiple markings

Resolved: added this tagging to the proposal
A few zebra markings indicate that diagonal crossing is allowed, but they disappear in the middle of the intersection.
Zebra markings near the curbs indicate diagonal crossing is allowed, but they switch to transverse through the middle of the intersection, perhaps to avoid visual clutter

I think this proposal is great, and would advance the possibilities for detailed tagging of crossings, which are currently a bit of a mess IMO. I wonder how this proposal would accommodate crosswalks which change markings partway through an intersection. Here are a couple examples near me. Both of them are "scramble" intersections where diagonal crossing is permitted during a dedicated pedestrian signal phase. In both, there are a couple of zebra lines near the curbs which hint to pedestrians that diagonal movement is possible, but they switch mid-intersection to either a transverse treatment or nothing at all. Would splitting the ways be ideal, or should they be tagged in a special way? Or should they be tagged just by their mid-intersection portion, which makes up most of their length? --Willkmis (talk)

Just split the highway=footway there? Kovposch (talk) 07:54, 29 July 2022 (UTC)
Yeah, using highway=footway footway=crossing you can capture this easily. On the highway=crossing node itself you can add the most prominent type or simply crossing:markings=yes. --Popball (talk) 17:06, 29 July 2022 (UTC)

Double lines

Resolved: Added this pattern

--Popball (talk) 11:44, 4 August 2022 (UTC)

Crossover in Toronto

In Ontario, one of the standard markings for a crossover (signalized pedestrian crossing) has two parallel double lines instead of two parallel single lines. [9] – Minh Nguyễn 💬 10:42, 3 August 2022 (UTC)

Thanks for finding this one, I added it to the table under crossing:markings=lines:paired. Let me know if you think crossing:markings=lines:double would be better. --Popball (talk) 11:01, 3 August 2022 (UTC)

The photo makes it painfully obvious, that crossing:signals=* needs some love. This one is as much a signalled crossing as a zebra crossing in the original sense is one. --Hungerburg (talk) 20:51, 4 August 2022 (UTC)

@Hungerburg: This is just flashing_lights=button, not crossing:signals=yes. The difference is that cars must stop and wait when someone is already in the crosswalk, but they can proceed after the pedestrian clears the crosswalk, even if the lights are still on. The same rules would apply if there were no flashing lights, but the lights are there as a warning. But I agree that there needs to be a formal push for crossing:signals=* adoption as well. This proposal gives us a blueprint for putting together a more successful crossing:signals=* proposal than the one from a few years ago. – Minh Nguyễn 💬 21:16, 6 August 2022 (UTC)

Please find a better picture for lines:paired. It looks like lines (withoud paired) without zooming in --Nielkrokodil (talk) 12:57, 6 August 2022 (UTC)

I added diagrams for each value, so now the examples aren't quite as cruitial :) --Popball (talk) 20:15, 6 August 2022 (UTC)

Behold, a ladder with double lines: Toronto – Minh Nguyễn 💬 20:50, 6 August 2022 (UTC)

I kinda just want to call it crossing:markings=zebra;lines:paired just to keep things simple. Of course, a normal ladder could be defined in the same way, but I think it deserves a top level level due to its ubiquity in the US anyway. --Popball (talk)
Yes, I think that's reasonable. I haven't found any indication that this particular combination is standard anyways. On the other hand, it is very common to leave a gap between the rungs and lines of a ladder marking. I've been ignoring that detail, but I guess there's an argument to be made that it's actually a combination of zebra and lines. – Minh Nguyễn 💬 21:12, 6 August 2022 (UTC)

Why don't deprecate zebra and marked?

Resolved: add deprecation for crossing:marking=*, leave crossing=zebra as is

--Popball (talk) 09:36, 19 August 2022 (UTC)

I am not familiar with the discussion so far, but let me ask a question. If crossing:marking=* represents road marking, does crossing=* come to represent the function of a pedestrian crossing? If so, is there any reason not to deprecate crossing=marked and crossing=zebra? marked is a duplicate of uncontrolled and the value does not represent a function. zebra should be crossing_ref=zebra according to the proposal. --Gyotoku810 (talk) 14:05, 6 August 2022 (UTC)

@Gyotoku810: According to all the tags approved for crossing=*, yes, crossing=* is meant to represent the function of a pedestrian crossing. A proposal suggested to use crossing=* to represent the presence of crossings with marked and unmarked and to use crossing:signals=* to represent the presence of a pedestrian signal. The proposal never seemed to get anywhere (probably because it attemped to depricate existing values of crossing=*), however crossing=marked now has tons of usage, probably because of the iD preset. As such, it seems that it (de facto) is a duplicate of crossing=uncontrolled, at least judging by the relative low usage of crossing:signals=*.
On the other hand, crossing=zebra seems to be a duplicate of crossing=uncontrolled + crossing_ref=zebra. Given that at one point it was a iD preset for "crosswalk", this tag has seen a lot of misuse (all the usages in my city were distinctly not zebra). As such, given the lack of documentation on this tag, it seems it also is (de facto) a duplicate of crossing=uncontrolled.
Ultimately, however, these tags simply have too much usage to deprecate and lots of software already supports these values, despite their ambiguity. --Popball (talk) 15:12, 6 August 2022 (UTC)
@Popball: Thank you for your answer. I understood that the problem is that those tags are already heavily used. The proposal is so great but I hope one day those tags will be deprecated. Thanks. --Gyotoku810 (talk) 15:32, 6 August 2022 (UTC)
Oops, I found that the proposal is crossing:markings=*, not crossing:marking=*. That's no problem. --Gyotoku810 (talk) 15:40, 6 August 2022 (UTC)
Should crossing:marking=* at least be deprecated, while there's still a chance to head off the confusion seen above? Or is it unnecessary to do that as part of the formal proposal process because that value is still quite new and rare? – Minh Nguyễn 💬 20:53, 6 August 2022 (UTC)
Well we certainly shouldn't have both! It's quite confusing indeed. I think given the low usage, (it's only rainbow and zebra at this point) and poor definition, I think we can solve it on its page to migrate it to surface:colour=* and crossing:markings=*. I see you've already taken the first steps. I'm not against formally deprecating it, but it seems a bit silly to do so when that page was created after this proposal was first drafted :D --Popball (talk) 21:29, 6 August 2022 (UTC)

Marking vs surface (Recap)

Resolved: The proposed new key is not immune to known problems in the field. Providing more details though will only help consumers in assessing the implications of the mapping on what they infer from it. --Hungerburg (talk) 21:02, 11 August 2022 (UTC)

The few here, in this talk, we care about legal status. That made me shy of solid, and likely others too, see Markings vs pavement section. While everybody with a driving licence should be able to follow suit, we cannot take it for granted in the mapping community at large, to look at the subject from same point of view. So I learned to see the goal of this proposal to fill in details about physical properties, something that can be universally expected to be gotten right; And legal status out of scope here, however useful the supplementary information can be, in determining that.

How about: Pass the legal cup to the crossing key! Then the German parking lot example might become crossing=unmarked, crossing:markings=solid; Perhaps the Azalea-Trail one too? Or is that too much illogical? Is the contradiction in words only, or is it in subject matter? Gonna prepare a foto, later today. --Hungerburg (talk) 10:16, 7 August 2022 (UTC)

I think trying to redefine existing keys with lots of usage should probably be avoided; mappers have been using crossing=* for a long time, and redefining these keys would mean that there would suddenly be a lot of mistagged crossings. --Popball (talk) 10:30, 7 August 2022 (UTC)
Not a crossing

Here the promised foto: I suppose, lots of people would tag a "crossing=marked + crossing:markings=solid". From my reading of local legalese, there is not even a "highway=crossing" there at all. If I'd tag "crossing=uncontrolled + crossing:markings=solid", then I would not even notice the contradiction, I mentioned above. --Hungerburg (talk) 22:56, 7 August 2022 (UTC)

Key name

Resolved: adding crossing:markings=surface makes the key less confusing to users who will see a change in surface as a marking

--Popball (talk) 10:09, 29 August 2022 (UTC)

As System-users-3.svgMateusz Konieczny (on osm, edits, contrib, heatmap, chngset com.) has pointed out on the mailing list, the name crossing:markings=* may be confusing to mappers who may be expecting that a change in surface should be considered a marked crosswalk. It is worth looking into seeing if a better name is possible. Maybe something like crossing:road_markings=*? I don't want the name to get too out of hand, as many keys require the user to read the wiki anyway (or for whichever application they are using to give them easy to follow presets; I think this proposal could be a good StreetComplete quest anyway), and I think that if a mapper is inputting the key=value pair themself, then they have read the wiki.

The other option is to allow something like crossing:markings=surface to indicate that the crossing is only differentiated by the change in surface, and a user can then fill in the advanced surface info under surface=*, surface:colour=*, etc. I think it's best to avoid something like crossing:markings=surface;zebra in any case.

One other option is to make something like crossing:differentianted_surface=yes/no (or maybe crossing:diff_surface=* is less unwieldy) to tag whether the surface of the crossing differs from the road surface. I don't know if this actually adds any useful information, but it might be nice to have a different key for this in general.

Currently I'm leaning towards crossing:markings=surface but making it clear that it should be used when the change in surface is the only feature "marking" the crossing on the roadway, and to note that not all juristictions recogize these as proper crosswalks. --Popball (talk) 14:47, 7 August 2022 (UTC)

The key name is perfect. There is no reason to change it to something contrived. The easiest way out would be to just drop "solid" form the list of proposed values. I tried to argue in favour of keeping solid, see previous section. --Hungerburg (talk) 23:06, 7 August 2022 (UTC)

@Popball: There is something to be said for consistency with road_marking=* (which was largely popularized through a stealth import), but I think crossing:markings=* is reasonably intuitive. crossing:road_markings=* would be incompatible with railway=crossing or possibly highway=crossing (such as where a footpath formally crosses a bikeway at a marked crossing).

I'm not sure of the value of the other suggestions versus a common-sense highway=crossing surface=*, footway=crossing surface=*, or area:highway=crossing surface=* – together, these combinations already have hundreds of thousands of occurrences. Differentiation is just a matter of noticing that the surface=* differs between the crossing and the intersecting highway=residential.

crossing:markings=solid does seem to be a bit of a troublemaker. I appreciate the clarification that it only applies when the color is a normal crossing marking color, but this nuance doesn't seem very durable because of the broader-sounding tag value. obnoxiously_thick_line would probably communicate the distinction more effectively. Incidentally, a friend pointed out that any kind of solid-painted surface, including for bike lanes, can be an effective surface for sport=hydroplaning unless grit is added to the paint.

 – Minh Nguyễn 💬 01:23, 8 August 2022 (UTC)

Given the opposition in the talk thread, I think it might make sense to move crossing:markings=solid to a footnote in crossing:markings=surface given it rarity, andour inability to find a proper example, despite its seeming existsnce in the design guidelines. --Popball (talk) 23:07, 12 August 2022 (UTC)
Sounds good to me. There's already an escape hatch below the table so that you can use any tag you like in response to a city painting any marking they like, which should be enough for however many formal solid crosswalks there are in reality. – Minh Nguyễn 💬 03:10, 13 August 2022 (UTC)
Curiously, though not really off-topic, Mateusz' objection, luckily not in vote phase, highlights a deficiency of the Crossing=Marked tag. I'd say, our host Popball, did a great job, as several times before, in adjusting the proposal, still I hesitate to mark the issue resolved, while myselves considering that a matter of fact. --Hungerburg (talk) 22:27, 10 August 2022 (UTC)
Thank you @Hungerburg: :D I think this is precisely why I began thinking that crossing:markings=surface makes sense: to many it may seem marked but to an agency it is not. I do think is is a lot more intuitive for new mappers, although slightly less consistent, although easily filtered to get "legal road markings" by excluding surface.--Popball (talk) 23:07, 12 August 2022 (UTC)
If this proposal succeeds, it could be a good blueprint for reviving the crossing:signals=* proposal, this time perhaps with an opportunity to distinguish between common kinds of pedestrian signals. Less conflicts between tagging schemes and more opportunities to capture details that weren't expressible before: a win-win scenario. – Minh Nguyễn 💬 23:19, 10 August 2022 (UTC)

Lines longitudinal to the road

Resolved: I took care of this one, the lines truly are parallel/longitudinal to the cross-walk, and transverse/orthogonal/perpendicular to the carriageway --Hungerburg (talk) 22:16, 10 August 2022 (UTC)

Just a small observation: In the table below "Tagging" it reads twice "Lines longitudinal to the road". This looks to me like a typo. Lines are orthogonal to the carriageway, by the definition given in the intro. --Hungerburg (talk) 09:35, 9 August 2022 (UTC)

More missing marking varieties

Resolved: Added zebra:bicolour

Here I've listed a number of additional pedestrian crossing marking varieties that I am aware of. The tags in the tables are just a quick possible proposal - it's up for the author how to include those into the final proposal text.

And a bit of off-topic concerning crossing:markings=no. Author should keep in mind that there exist pedestrian crossings with only a sign but without markings and crossings with only markings but without a sign.

Example Tagging Comment
Podil, Kiev, Ukraine, 04070 - panoramio (263).jpg High-risk zebra crossing (Ukraine)
Пешеходный переход в Судалице.jpg High-risk zebra crossing (Russia)
Medzilaborce .jpg High-risk ladder crossing (Slovakia)
40-е корпуса.jpg Directional crossing (Russia)
External image Conventional zebra crossing with a lane for visually-impaired people (Ukraine)
External image Unconventional zebra crossing with a lane for visually-impaired people (Ukraine)

--VileGecko (talk) 14:38, 29 August 2022 (UTC)

Hi, thanks for looking into these. I know the bicolor zebras are very common in some regions, so it may make sense to include this as a top level entry (crossing:markings=zebra:bicolour seems to be straigtforward enough). The ones for the visually impaired I'm thinking make sense as the existing crossing:markings=zebra:double, and the additional guidance might make more sense in a different tag (a more universal system of tagging high contrast pavement for the visually impaired). As for the directional one, I'm not convinced that the arrows are prominent enough to differentiate from crossing:markings=zebra:double. (not that you suggested it, but bicycle pictograms in bicycle crossings shouldn't be tagged here either for the same reason).
As for crossings with only a sign, my idea would be something like crossing:traffic_sign=*, but that belongs in its own proposal IMO. --Popball (talk) 07:15, 30 August 2022 (UTC)
Mapping pedestrian crossing lanes for visually impaired people similarly to how cycleway crossings are mapped indeed seems reasonable. However this probably deserves and additional proposal to extend current tactile_paving=* key to include non-tactile high contrast safety lanes for visually impaired people both on sidewalks and on pedestrian crossings. As of now the scheme seems to be lacking in regard to non-tactile safety lanes. --VileGecko (talk) 11:02, 30 August 2022 (UTC)

Where it is really useful to survey?

Are there countries where there are some strong differences between crossings captured by new tags listed here? If yes - which countries and why it is useful to collect? (asking as it may be viable StreetComplete quest if this info is likely to be used for something) Mateusz Konieczny (talk) 14:12, 7 September 2022 (UTC)

In the US it is quite useful, I'd say especially so at anything with crossing=traffic_signals, as it is quite common to find signalized crossings without markings (and up until this proposal there wasn't a real way to capture markings for signalized crossings. I think Canada is a similar situation.
In Germany capturing it is less useful, as usually they are so consistant. It might make sence, however, for the "What kind of pedestrian crossing is this?" to have the following four options:
1. signalized
2. zebra crossing (most commonly crossing=uncontrolled, crossing_ref=zebra)
3. unmarked
4. dashes (crossing:markings=dashes) (this is typically only used for bicycle crossings, although I don't believe it has any legal implications, it is merely for safety; Germany wikipedia has an article about this: Fußgängerfurt on Wikipedia).
Perhaps also useful would be an surface option with crossing:markings=surface but with those 4-5 options you can cover the vast majority of pedestrian crossings in Germany. —Preceding unsigned comment added by Popball (talkcontribs) 17:23, 7 September 2022 (UTC)
But is this data at least potentially useful for something beyond 3D rendering? Mateusz Konieczny (talk) 18:03, 7 September 2022 (UTC)

A third MUTCD-influenced country, New Zealand, allows both zebra and lines for essentially the same kind of crossing. There seems to be significant variation in Czechia as well, specifically between zebra and zebra:double. Apart from these patterns, the values surface and no seem to be applicable all over the world.

It's increasingly common for 2D renderers to display high detail like crossings at high zoom levels; crossing markings could be useful there without getting into 3D. A router for wheelchair and visually impaired users might prefer high-visibility crossing markings over nearby low-visibility ones that motorists are less likely to notice. That said, it may be less important for StreetComplete to have a quest for something that's normally visible in aerial imagery. A future crossing:signals=yes/no/shared/separate proposal would be a better candidate for a quest. It would be more straightforwardly applicable to routing and not as easy to determine via aerial imagery.

 – Minh Nguyễn 💬 18:48, 7 September 2022 (UTC)