Worldwide routable Garmin maps: Missing/incorrect feature requests

As written here I’m downloading Thailand data from time to time directly to check my mapping.

I wouldn’t display all unpaved roads as tracks. There are also paved tracks. I think doing it like Rendering of MTB maps on Garmin units would be better.

Thanks again. I have no idea how the mtb people get their maps to look the way they do. I don’t think having multi-coloured roads is a good idea, the lines will get too thick and clutter up the map. Minko somehow managed to make bridges appear thicker, which looks good at some zoom levels, but as you zoom in they disappear and as you zoom out they ‘detach’ from the road they mark.

To me as a motorcycle traveller it is important that I can see whether a road is paved or not. In fact, this info is probably more important than any other. For a car driver this may be less important. That, and the fact that Garmin maps (non-NT) don’t distinguish between cars and bikes signals to me that we need a separate motorcycle map. Or perhaps we could abuse the taxi restriction that is not used by anybody?

Peter,
I dont see any sea on your screenshot. Wanmil has recently improved the sea generation.
See http://forum.openstreetmap.org/viewtopic.php?id=17343

About address search, this uses administrative districts, I dont know how this in Thailand works. See --location-autofill and --bounds=directory in the mkgmap options

Regarding bridges: yes I use only one linetype with fixed pixel distance which has those issues. Maybe you can make more linetypes for every zoomlevel: resolution 24 a wider one, resolution 22 a more narrow bridge. Besides, I dont know if Garmin has more zoomlevels than mkgmap can show.

Taxi restrictions doesnt work on most Garmin units afaik

Hi,

Thanks for joining in. :slight_smile:

This map is more of a proof that it works in the details that we worked on, so the probs with addresses and sea don’t bother me at this stage. When Lambertus incorporates it I’m sure this will all be sorted.

A separate bike map it will have to be, then.

Regards,
Peter.

Do you mean motorbike map?

Yes, of course. Cyclists already seem well catered for.

I have reported this issue again and they are willing to have a look onto it but can’t promise anything:
https://forums.garmin.com/showthread.php?p=142590#post142590

Good on you.

I did a little testing and so far it seems that the ‘no cars/motorcycles’ restriction in Garmin maps doesn’t work for anything other than walkways. As I couldn’t find any roads on our NZOpenGPS maps with car restrictions I used the ferries from Auckland to Waiheke Island. The results:

In MS everything gets routed via the passenger ferry.

In BC cars go on the passenger ferry, bicycles get sent on a long detour via the car ferry. (Unless car pool avoidance is turned off.)

What we need now is a location on an official Garmin NT map, where we could test these restrictions. I.e., for a road where cars are not allowed, will bikes get routed through there? (I only have US maps and I don’t know where to find such a location.)

I couldnt tested this on the latest maps, I only have an older CN Europe. As far as I can see, for Garmin bicycles DON’T exist so they are treated the same as cars :rage:

You have either pedestrians or cars, nothing in between. I dont even think that they have a specific pedestrian routing, its only a switch car=yes / no. If it is no, only pedestrians can route on them, not even bicycles. Bicycles are routed the same as cars.

Now the challenge is, can we find the code what blocks car traffic from certain roads? It is clearly not motorcar=no or access=no because this tag doesnt work in BC (unless you switch carpool avoidance on)

It doesn’t matter if the map is old. You just need to know a street where cars are not allowed to test the restriction. As far as I can tell so far in non-NT maps car restrictions only work on walkways. In our NZ maps we disconnect roads where cars are not allowed. Malls have to be mapped as walkways.

Ok, I found the map now (didn’t look on the front page). Seems you used the wrong typ and style files, this doesnt look good to me.
It was some time ago but we decided to put the changes in the world-test map and not in “world”. So I either have to copy the test versions to the world map or you can try another spin of these files:
http://code.google.com/p/mkgmap-style-sheets/source/browse/#svn%2Ftrunk%2Ftyp%2Fworld-test
http://code.google.com/p/mkgmap-style-sheets/source/browse/#svn%2Ftrunk%2Fstyles%2Fworld-test%253Fstate%253Dclosed

Ah, yes, I assumed the ‘world’ would be the style I needed. I’ll change to ‘world-test’ asap. Unfortunately an update is already running and I cannot run two updates at the same time.

No problem, it’s still a test anyway (and there are some routing bugs in the latest mkgmap too, so dont worry).

I wasn’t aware of that, either and have used world.

My understanding was that we use the test style and typ for development and the world style and typ for production. Otherwise, what is world for?

If you agree then let’s copy test over world and just keep Lambertus’ cycle running.

I have not seen the output from test…

This discussion should really be moved to the topic we started for that purpose.

In the test style were the latest changes made, we started first with ‘world’ but you suggested it was better to use this style for the production (which never made it until now).

Sure, i can copy them to the “production” styles.

Those test versions were put here: http://code.google.com/p/mkgmap-style-sheets/downloads/list

Lambertus, can you post the mkgmap parameters here?
I think that something with make-cycleways is not needed and in the test styles I also use a land-tag=natural=background

Tile rendering:

<?php
$name = "generic_new";
$description = "OSM Generic Routable";
$family_id = "2000";
$product_id = "6";
$mapid = 63240000;
$style_dir="/home/lambertus/garmin/utils/styles/";
$style = "generic_new";
$bounds = "/home/lambertus/garmin/utils/bounds.zip";
$max_size = 24*1024*1024;
$initial_nodes = 1500000;
$max_nodes = 1200000;
$type = "o5m";
//$type = "osm.gz";
$split_dir = "/home/lambertus/garmin/split-o5m";

$script_dir = "/home/lambertus/garmin/$name";
$cities = "/home/lambertus/garmin/utils/cities15000.zip";

$mkgmapOptions = array(
		" --reduce-point-density=4",
		" --reduce-point-density-polygon=8",
		" --precomp-sea='/home/lambertus/garmin/utils/sea'",
		" --show-profiles=1",
		" --add-pois-to-lines",
		" --index",
		" --location-autofill=bounds,is_in,nearest",
		" --latin1",	// implies code-page=1252
		" --ignore-maxspeeds",
		" --remove-short-arcs",
		" --min-size-polygon=10",
		" --merge-lines",
		" --add-pois-to-areas",
		" --preserve-element-order",
		" --route",
		" --name-tag-list=name:en,int_name,name:zh_py,name:engels,name"
	);
?>

Shall I add --generate-sea=land-tag=natural=background?

Combining:

"generic_new" => array(
		"description" => "OSM generic routable (new style)",
		"underscored" => 'osm_generic_new',
		"family_id" => 2000,
		"product_id" => 6,
		"overview_name" => "63240000",
		"typ_dir" => "generic_new",
		"copyright" => "'Map data © openstreetmap.org and contributors'"
	)

Edit: added the land-tag and updated the style+typ in time for the already running map update to pick-up the changes.

Great! :slight_smile:

Thank you both. Looks like all of a sudden things are moving fast! :smiley:

I updated my files and compiled a quick test map using old osm data. Looks almost the same as before. Downloaded N Thailand with JOSM, but now I get only POIs, no lines or polygons. I don’t know whether this is a problem with my osm file or an mkgmap bug. I have previously made an even bigger map without problems.

Incidentally, when it does work mkgmap complains about:

WARNING: Style uses tag 'display_name' which is deprecated and will be removed soon. Please use the new tag 'mkgmap:display_name' instead.

Could it be that the lines and polygons are covered by a land background polygon (0x27?)
You need to use --generate-sea=land-tag=natural=background and the typ file in order to get a decent map.

About display_name, I replaced it now with mkgmap:display_name in the lines style.