Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Toby, On 07/07/2014 08:57 PM, Toby Murray wrote: Is this a tagging problem or a rendering problem? As Christoph has already spotted, the server on which I made these tiles had a slightly older style sheet and ocean shape. I'll re-do the tiles based on the current Carto style and we'll have a look then! Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Hi, On 06/30/2013 01:32 AM, Frederik Ramm wrote: On 12.03.2012 08:56, Frederik Ramm wrote: There's nothing keeping one from applying the Tiles@Home lowzoom process to a slight variation of our standard Mapnik style however, and out comes this (for zoom levels 0-8; from z9 on, the standard Mapnik style looks fine): http://fred.dev.openstreetmap.org/lowzoom/ I have updated these tiles with current data. (I had somehow lost the code that produced the tiles and had to reverse engineer my way back from the old tiles... the year-old tiles are still available from the layer switcher.) A new year, a new update. I like it how this map reveals some things that otherwise remain unseen, like for example a several 100km long secondary road in Antarctica (near New Zealand) that seems to have been there for a year (I've deleted it now)... Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
On Monday 07 July 2014, Frederik Ramm wrote: A new year, a new update. I like it how this map reveals some things that otherwise remain unseen, like for example a several 100km long secondary road in Antarctica (near New Zealand) that seems to have been there for a year (I've deleted it now)... That was actually an existing feature - a transport route from the coast to the French Concordia station - you can even marginally see it on the LIMA mosaic if you look closely: http://mc.bbbike.org/mc/?lon=123.43508lat=-75.07765zoom=12num=2mt0=bing-satellitemt1=mapnik See also: http://en.wikipedia.org/wiki/Concordia_Station The tagging might of course be considered wrong - it could be more like highway=track and surface=snow but for the standards of the region highway=secondary might even be an understatement. The map could use an update of the coastline file. -- Christoph Hormann http://www.imagico.de/ ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Hi, On 07/07/2014 01:41 PM, Christoph Hormann wrote: That was actually an existing feature - a transport route from the coast to the French Concordia station - you can even marginally see it on the LIMA mosaic if you look closely: Oh dear. I automatically assumed that it must be some kind of doodle because of being long and straight and practically nodeless ;) The tagging might of course be considered wrong - it could be more like highway=track and surface=snow but for the standards of the region highway=secondary might even be an understatement. What would you recommend? Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
2014-07-07 13:41 GMT+02:00 Christoph Hormann chris_horm...@gmx.de: See also: http://en.wikipedia.org/wiki/Concordia_Station The tagging might of course be considered wrong - it could be more like highway=track and surface=snow but for the standards of the region highway=secondary might even be an understatement. I wouldn't use track, service might be appropriate in other settings (with alternatives), but in this environment it is probably a primary (or at least secondary as you put it). I'd go for primary + surface cheers, Martin ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Am 07.07.2014 16:13, schrieb Frederik Ramm: Hi, On 07/07/2014 01:41 PM, Christoph Hormann wrote: That was actually an existing feature - a transport route from the coast to the French Concordia station - you can even marginally see it on the LIMA mosaic if you look closely: Oh dear. I automatically assumed that it must be some kind of doodle because of being long and straight and practically nodeless ;) The tagging might of course be considered wrong - it could be more like highway=track and surface=snow but for the standards of the region highway=secondary might even be an understatement. What would you recommend? Seems to be a winter_road only usable in summer ! Cheers colliar signature.asc Description: OpenPGP digital signature ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
On Monday 07 July 2014, Frederik Ramm wrote: The tagging might of course be considered wrong - it could be more like highway=track and surface=snow but for the standards of the region highway=secondary might even be an understatement. What would you recommend? Well - there is not really a frame of reference here - going strictly by the wiki it would be highway=trunk - there are certainly no more than a handful of such long distance roads on the whole continent so this certainly qualifies as 'one of the most important'. On the other hand it is not used frequently so highway=secondary might not be such a bad choice. The question is of course if this qualifies as road at all since for the most part it is not really a built, permanent structure, on the plateau essentially these are just the tracks of the transports that went this route. The other end of the road is tagged highway=road by the way: http://www.openstreetmap.org/way/187792384 -- Christoph Hormann http://www.imagico.de/ ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
This rendering has highlighted inconsistent tagging in county borders in the U.S. However the inconsistency is only on member ways. The county boundary relations are very consistent but don't seem to be used in rendering. Should they be? I do like that counties are a little more prominent than on osm.org. Right now rendering seems to be based on the admin_level=6 tag on ways only. Look at the difference between Colorado and Kansas at zooms 7-10: http://tile.openstreetmap.fr/?zoom=7lat=38.0209lon=-102.2259layers=B000FFF Whoever originally did the county border relations in Kansas deleted the admin_level=6 tag from the ways. I did the ones in Colorado and left it there. At z11+ it seems that the relations are used. Or perhaps it is just a generic rendering of the boundary=administrative tag which I think is still on most of the member ways. And actually I just noticed that this affects all admin boundaries. If you go to z5, there are a lot of state borders missing as well as the US/Canada border: http://tile.openstreetmap.fr/?zoom=5lat=42.64191lon=-84.04284layers=B000FFF Is this a tagging problem or a rendering problem? Toby On Sat, Jul 6, 2013 at 4:02 AM, Christian Quest cqu...@openstreetmap.fr wrote: I slightly modified the boundary rendering I also added islands and archipelago rendering, which makes seas a bit less empty... Example: http://tile.openstreetmap.fr/?zoom=7lat=58.13383lon=-2.9452layers=B0FFF Work still in progress... and feedback (with permalinks) welcome ! 2013/7/4 Michael Kugelmann michaelk_...@gmx.de: Am 04.07.2013 11:18, schrieb Christian Quest: Here is zoom 7 (others are in the render_list queue): to me this looks really nice and a huge improvement to the current OSM default lowzoom rendering! Best regards, Michael. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk -- Christian Quest - OpenStreetMap France Un nouveau serveur pour OSM... http://donate.osm.org/server2013/ ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
I slightly modified the boundary rendering I also added islands and archipelago rendering, which makes seas a bit less empty... Example: http://tile.openstreetmap.fr/?zoom=7lat=58.13383lon=-2.9452layers=B0FFF Work still in progress... and feedback (with permalinks) welcome ! 2013/7/4 Michael Kugelmann michaelk_...@gmx.de: Am 04.07.2013 11:18, schrieb Christian Quest: Here is zoom 7 (others are in the render_list queue): to me this looks really nice and a huge improvement to the current OSM default lowzoom rendering! Best regards, Michael. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk -- Christian Quest - OpenStreetMap France Un nouveau serveur pour OSM... http://donate.osm.org/server2013/ ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
On Saturday 06 July 2013, Christian Quest wrote: I slightly modified the boundary rendering I also added islands and archipelago rendering, which makes seas a bit less empty... Example: http://tile.openstreetmap.fr/?zoom=7lat=58.13383lon=-2.9452layers= B0FFF I very much like the island labels - this would be great to have in the standard map (at the moment i think islands are not labeled there until z=12 which is kind of pointless for any but the smallest islands - especially in polar regions). And of course it well points out missing names (and in this case missing french names). I guess the lack of labels in the north of http://tile.openstreetmap.fr/?zoom=6lat=-56.31392lon=-34.0843layers=B0FFF is due to incomplete rendering. The style also produces some funny stuff like: http://tile.openstreetmap.fr/?zoom=11lat=-53.14353lon=73.053layers=B0FFF Greetings, -- Christoph Hormann http://www.imagico.de/ ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Am 06.07.2013 11:02, schrieb Christian Quest: I slightly modified the boundary rendering As in the OSM Mapnik style, rendering of admin_level=3 boundaries seems to be identical to the one for admin_level=2 boundaries, which may lead to misundertandings for countries where level 3 is used, such as Slovenia and Slovakia. http://tile.openstreetmap.fr/?zoom=7lat=45.84969lon=14.82642layers=B0FFF Best regards Rainer ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Hi, Frederik Ramm frede...@remote.org wrote: http://fred.dev.openstreetmap.org/lowzoom/ I have updated these tiles with current data. (I had somehow lost the code that produced the tiles and had to reverse engineer my way back from the old tiles... the year-old tiles are still available from the layer switcher.) On 07/04/2013 02:00 AM, Tirkon wrote: I am not sure: Do you want to replace rhe lowzoom levels at osm.org? I think the lowzoom levels at osm.org look very bland and my approach attempts to fix that, trying not to be a cartographically solid map (for that, MapQuest has done an excellent job I think) but instead highlighting where OSM has data and where it hasn't. I don't have a particular agenda about replacing the lowzoom tiles on osm.org; if someone wants me to install my process on the OSM tile server I can do that but if most people are happy with what we've got then I have no problem with that. At present the shown names seem to be taken from the place tag. [...] The first priority for rendering could be the admin level. Certainly an idea worth pursuing, however some magic would have to be applied to match place tags (usually nodes - they will have a population or capital tag) with admininistrative boundaries (that have an admin level but neither capital nor population info). Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
On Thursday 04 July 2013, Tirkon wrote: [...] The first priority for rendering could be the admin level. Within the admin level at first the towns with a capital tag are rendered wirh a star next to the name. If this is space-kompetitiv at a given zoom level, the winner will be decided on the population. Then all other towns of the given admin level are rendered. Space-competition is decided first on the place tag (which could indicate independent cities) and then on population. The problem about this is not how to do it in principle but that label placement optimization is a highly non-local problem so doing it well requires doing it for the whole globe at once and you would have to redo it every time something changes anywhere. Even the current very simple approach suffers from this problem resulting in occasional cut off labels at tile edges. The thing is the standard osm.org map is meant to allow near real time updates and compromises rendering quality for that. But there is very little actual real time data in the rendering at the lowest zoom levels anyway - none in 0 and 1, only labels in 2 and 3. Frederik's approach would bring the real time osm data to the lower zoom levels but in the end only few of the individual changes in the database have a visible effect on the map at this scale. So it would make sense to take a split approach to the lower zoom levels - create a real time version that allows reviewing edits made to the database and a second high quality version that is pre-rendered periodically using techniques that can be more expensive. Greetings, -- Christoph Hormann http://www.imagico.de/ ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
2013/7/4 Frederik Ramm frede...@remote.org On 07/04/2013 02:00 AM, Tirkon wrote: I am not sure: Do you want to replace rhe lowzoom levels at osm.org? I think the lowzoom levels at osm.org look very bland and my approach attempts to fix that I partly agree (have thought that for a long time, but somehow got used to the status quo in the meantime). I believe they come from an era in OSM when we were still proud to have some country boundaries and the coastline (what we now take mostly for granted). , trying not to be a cartographically solid map (for that, MapQuest has done an excellent job I think) but instead highlighting where OSM has data and where it hasn't. yes, that's what makes them interesting, it is somehow osmarender lowzoom 2.0 ;-) (btw.: MapQuest seems to use a handmade map for really low zooms: 1-3) I don't have a particular agenda about replacing the lowzoom tiles on osm.org; if someone wants me to install my process on the OSM tile server I can do that but if most people are happy with what we've got then I have no problem with that. maybe we could have both? In the end low zoom tiles don't require so much disk space and won't have to be updated very often. cheers, Martin ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
On 04/07/13 09:39, Christoph Hormann wrote: The thing is the standard osm.org map is meant to allow near real time updates and compromises rendering quality for that. But there is very little actual real time data in the rendering at the lowest zoom levels anyway - none in 0 and 1, only labels in 2 and 3. Low zooms (0 to 12) are not actually updated in real time anyway - they are only updated periodically by a batch job. Tom -- Tom Hughes (t...@compton.nu) http://compton.nu/ ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Based on Frederik lowzoom idea, I started improving the low zooms on our OSM-FR style... Here is zoom 7 (others are in the render_list queue): http://tile.openstreetmap.fr/?zoom=7lat=46.88069lon=3.13697layers=B0FFF Here is how I did it: - modify my OSM-FR to remove all labels and boundary at zoom 8 (which was already using landcover/landuse), - generate 4 large PNG using nik2img - combine them and generate one large tiled z8.tif using gdal_translate (133MB, if you want to play with it I can share the result) - create a very simple stylesheet to test using this z8.tif as raster, then add placenames and boundaries It's a work in progress. I'm currently merging this lowzoom stylesheet to my main OSM-FR style sheet. I've not played with imagemagick, it is just a lanczos downscaling made by mapnik. Label ordering is handled by using the place=* tag, then the population=* tag so large cities are placed first. The placename density is limited using text-min-distance. -- Christian Quest - OpenStreetMap France Un nouveau serveur pour OSM... http://donate.osm.org/server2013/ ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
On Thursday 04 July 2013, Tom Hughes wrote: Low zooms (0 to 12) are not actually updated in real time anyway - they are only updated periodically by a batch job. But you can still trigger a rerender using /dirty. Greetings, -- Christoph Hormann http://www.imagico.de/ ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
On 04/07/13 10:48, Christoph Hormann wrote: On Thursday 04 July 2013, Tom Hughes wrote: Low zooms (0 to 12) are not actually updated in real time anyway - they are only updated periodically by a batch job. But you can still trigger a rerender using /dirty. At the moment you probably can, yes. I have suggested a feature in mod_tile to tell it to never render tiles below a certain zoom though so that may not be true forever ;-) Tom -- Tom Hughes (t...@compton.nu) http://compton.nu/ ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Frederik Ramm frede...@remote.org wrote: The first priority for rendering could be the admin level. Certainly an idea worth pursuing, however some magic would have to be applied to match place tags (usually nodes - they will have a population or capital tag) with admininistrative boundaries (that have an admin level but neither capital nor population info). ÖK! So the idea could be slimmed down: No consideration of admin_level. Rendering as today decided on the place tag. But if there is much empty space the next lower value of the place tag will be used as well. No berücksichtigung Here is an example of empty space: http://www.openstreetmap.org/?lat=53.12lon=11.7zoom=8layers=M The other features could be added (possibly partly), because population and capital are now available. In particular preferred rendering of capital-names and its stars could be added for comparatively little costs. This alone gives a better orientation cause these cities are known better in the most cases. The magic in the extended version could be provided by adding the correspondent place-node with admin_centre-role as a member to the boundary-relation. But this has to be done worldwide and thus it is a problem. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Am 04.07.2013 11:18, schrieb Christian Quest: Based on Frederik lowzoom idea, I started improving the low zooms on our OSM-FR style... Here is zoom 7 (others are in the render_list queue): http://tile.openstreetmap.fr/?zoom=7lat=46.88069lon=3.13697layers=B0FFF Here is how I did it: - modify my OSM-FR to remove all labels and boundary at zoom 8 (which was already using landcover/landuse), - generate 4 large PNG using nik2img - combine them and generate one large tiled z8.tif using gdal_translate (133MB, if you want to play with it I can share the result) - create a very simple stylesheet to test using this z8.tif as raster, then add placenames and boundaries It's a work in progress. I'm currently merging this lowzoom stylesheet to my main OSM-FR style sheet. I've not played with imagemagick, it is just a lanczos downscaling made by mapnik. Label ordering is handled by using the place=* tag, then the population=* tag so large cities are placed first. The placename density is limited using text-min-distance. Nice One minor problem: Some city names are not rendered but the city nodes. Probably it makes sense to allow to move label a bit if they are to close. Cheers colliar signature.asc Description: OpenPGP digital signature ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Am 04.07.2013 11:18, schrieb Christian Quest: Here is zoom 7 (others are in the render_list queue): to me this looks really nice and a huge improvement to the current OSM default lowzoom rendering! Best regards, Michael. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Am 01.07.2013 22:32, schrieb Joseph Reeves: Hi Frederik, I really liked this when I saw it last year; am pleased to see it back on the mailing list! To be honest, I prefer the last year's version, but only because of one feature: country borders. I often use the MapQuest Open tiles, for example, at low zooms because they show borders much more clearly than the standard Mapnik tiles do. Likewise, last year's low zoom tiles are easier to understand when it comes to borders (and is therefore more useful to people looking to identify the general location of an entire country, something OSM isn't the best resource for). +1 Would love to see this on osm.org http://osm.org! +1 Yeah, mapnik's low zoom level are quite empty and ugly. colliar signature.asc Description: OpenPGP digital signature ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
On Sat, Jun 29, 2013 at 4:32 PM, Frederik Ramm frede...@remote.org wrote: On 12.03.2012 08:56, Frederik Ramm wrote: There's nothing keeping one from applying the Tiles@Home lowzoom proces http://fred.dev.openstreetmap.**org/lowzoom/http://fred.dev.openstreetmap.org/lowzoom/ Nice. Those strike a pretty reasonable balance between seeing activity and clutter. They give a nice sense of human activity across the planet. Only the ferry routes seemed odd: in some cases they are prominent enough to look a bit like landform outlines or rendering mistakes. --- For the opposite end of the spectrum (high zoom) I'd love to fill in blank areas of the map with whatever meager data is available: https://github.com/mapnik/mapnik/issues/1906 Here the goal is to show more of the features when zoomed into a sparse area. In the extreme example, show the only oasis in the middle of an otherwise featureless desert tile :-). ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Hi Frederik, I really liked this when I saw it last year; am pleased to see it back on the mailing list! To be honest, I prefer the last year's version, but only because of one feature: country borders. I often use the MapQuest Open tiles, for example, at low zooms because they show borders much more clearly than the standard Mapnik tiles do. Likewise, last year's low zoom tiles are easier to understand when it comes to borders (and is therefore more useful to people looking to identify the general location of an entire country, something OSM isn't the best resource for). Would love to see this on osm.org! I think it would also be a requirement for an open replacement to Google Earth (I know Marble from the KDE project does this, but I think it could do it much better). Cheers, Joseph On 1 July 2013 21:08, Bryce Nesbitt bry...@obviously.com wrote: On Sat, Jun 29, 2013 at 4:32 PM, Frederik Ramm frede...@remote.orgwrote: On 12.03.2012 08:56, Frederik Ramm wrote: There's nothing keeping one from applying the Tiles@Home lowzoom proces http://fred.dev.openstreetmap.**org/lowzoom/http://fred.dev.openstreetmap.org/lowzoom/ Nice. Those strike a pretty reasonable balance between seeing activity and clutter. They give a nice sense of human activity across the planet. Only the ferry routes seemed odd: in some cases they are prominent enough to look a bit like landform outlines or rendering mistakes. --- For the opposite end of the spectrum (high zoom) I'd love to fill in blank areas of the map with whatever meager data is available: https://github.com/mapnik/mapnik/issues/1906 Here the goal is to show more of the features when zoomed into a sparse area. In the extreme example, show the only oasis in the middle of an otherwise featureless desert tile :-). ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
On Sunday 30 June 2013, Frederik Ramm wrote: I have updated these tiles with current data. (I had somehow lost the code that produced the tiles and had to reverse engineer my way back from the old tiles... the year-old tiles are still available from the layer switcher.) Hello Fred, what method did you use for resizing the tiles? It seems to be different in the new and the old tiles emphasizing the line features now. Since the tiles darken as you zoom out you are probably resizing in gamma corrected color space which is not such a good idea (for example Labrador is dissolving into the ocean as you zoom out). A good starting point for resizing techniques in general is: http://www.imagemagick.org/Usage/resize/ And when rendering the various area features at z=9 it would probably be best to turn off the area patterns like on the glaciers which just result in artefacts when scaling down. Greetings, -- Christoph Hormann http://www.imagico.de/ ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Hi, On 30.06.2013 10:27, Christoph Hormann wrote: what method did you use for resizing the tiles? It seems to be different in the new and the old tiles emphasizing the line features now. Yes, I simply didn't manage to reproduce last year's version exactly. What I do is precisely the following: * Start from a modified z9 rendering (with thicker linework, no rivers, and lighter area colouring) - these are my raw tiles for z9. * For tiles on zoom z-1, use tiles from z and ImageMagick's Scale method to downscale them to 50%; merge labels onto these tiles; the result is the cooked tiles on z-1. * To produce the raw tiles for z-1 (which are never visible but only used as an input for z-2 tiles later), I not only downscale the tiles on z, but after that, use ImageMagick's Compose function with a mode of Darken to repeat the placement of downscaled tiles three times, with an offset of (0,1), (1,0) and (1,1). This step is crucial because otherwise the roads would thin out to a hint of road after 2 scale-down steps, but at the same time this is also resposible for the darkening of tiles as you get to lower zooms. (The Darken method chooses the darker of the two inputs.) I would be happy to hear ideas about how to do this differently, and try them out. And when rendering the various area features at z=9 it would probably be best to turn off the area patterns like on the glaciers which just result in artefacts when scaling down. Yes, I've currently commented out a large amount of areas but should maybe look more closely. Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
On Sunday 30 June 2013, Frederik Ramm wrote: * To produce the raw tiles for z-1 (which are never visible but only used as an input for z-2 tiles later), I not only downscale the tiles on z, but after that, use ImageMagick's Compose function with a mode of Darken to repeat the placement of downscaled tiles three times, with an offset of (0,1), (1,0) and (1,1). This step is crucial because otherwise the roads would thin out to a hint of road after 2 scale-down steps, but at the same time this is also resposible for the darkening of tiles as you get to lower zooms. (The Darken method chooses the darker of the two inputs.) I see - when you do this just one zoom level above you will likely get significant aliasing (the offset operations introduce 2x2 pixel structures which are then resized to half the size). In my experience such morphological tricks work best with at least two zoom levels difference between processing and target grid: convert z9.png \( +clone -morphology Erode Disk:2.5 \) \ -compose Darken -composite -scale 12.5% z6.png I would be happy to hear ideas about how to do this differently, and try them out. If you want to emphasize the roads it would probably be best to render them separately at z=9 to be able to process them in a different way. This would double the ressources required of course. You can also try something like: convert -sigmoidal-contrast 15,50% -scale 12.5% \ +sigmoidal-contrast 15,50% z9.png z6.png this will however somewhat distort the colors when you do it in RGB. Greetings, -- Christoph Hormann http://www.imagico.de/ ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Hi, On 12.03.2012 08:56, Frederik Ramm wrote: There's nothing keeping one from applying the Tiles@Home lowzoom process to a slight variation of our standard Mapnik style however, and out comes this (for zoom levels 0-8; from z9 on, the standard Mapnik style looks fine): http://fred.dev.openstreetmap.org/lowzoom/ I have updated these tiles with current data. (I had somehow lost the code that produced the tiles and had to reverse engineer my way back from the old tiles... the year-old tiles are still available from the layer switcher.) Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
* Frederik Ramm (frede...@remote.org) wrote: There's nothing keeping one from applying the Tiles@Home lowzoom process to a slight variation of our standard Mapnik style however, and out comes this (for zoom levels 0-8; from z9 on, the standard Mapnik style looks fine): http://fred.dev.openstreetmap.org/lowzoom/ It so happened that I was experimenting with the same thing in Feb, and this week I've finished the utility which can combine highzoom tiles into lowzoom ones and apply overlays in a fast and easy way, and I though you (and others) may be interested in it. Wiki: http://wiki.openstreetmap.org/wiki/Tiletool Source docs: https://github.com/AMDmi3/tiletool Example rendering from z9 mapnik tiles (plain, no captionless tiles and overlays): http://lowzoom.osm.rambler.ru/ -- Dmitry Marakasov . 55B5 0596 FF1E 8D84 5F56 9510 D35A 80DD F9D2 F77D amd...@amdmi3.ru ..: jabber: amd...@jabber.ruhttp://www.amdmi3.ru ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Hi, On 03/12/2012 08:56 AM, Frederik Ramm wrote: There's nothing keeping one from applying the Tiles@Home lowzoom process to a slight variation of our standard Mapnik style however, and out comes this (for zoom levels 0-8; from z9 on, the standard Mapnik style looks fine): http://fred.dev.openstreetmap.org/lowzoom/ I have uploaded a new version of these tiles, exaggerating the linework so that now even areas with no landcover import but a dense road network (e.g. South America) look busy on the map. I think this comes pretty close to the Tiles@Home lowzoom tiles now. I might still have to do something about the country borders and labels. I would like to repeat that this map is not intended to be particularly pretty; it is intended to be something like an index to OpenStreetMap data, something that tells you that there's something to discover if you zoom in on a particular location. Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
2012/3/17 Frederik Ramm frede...@remote.org Hi, On 03/12/2012 08:56 AM, Frederik Ramm wrote: There's nothing keeping one from applying the Tiles@Home lowzoom process to a slight variation of our standard Mapnik style however, and out comes this (for zoom levels 0-8; from z9 on, the standard Mapnik style looks fine): http://fred.dev.openstreetmap.**org/lowzoom/http://fred.dev.openstreetmap.org/lowzoom/ I have uploaded a new version of these tiles, exaggerating the linework so that now even areas with no landcover import but a dense road network (e.g. South America) look busy on the map. I think this comes pretty close to the Tiles@Home lowzoom tiles now. I might still have to do something about the country borders and labels. I would like to repeat that this map is not intended to be particularly pretty; it is intended to be something like an index to OpenStreetMap data, something that tells you that there's something to discover if you zoom in on a particular location. I think it's beautiful! The only thing that I don't understand is the jump of administrative borders rendering around zoom 4-5, on lowest zooms seems a physical map (borders are nearly invisibile) then suddenly become thicker. I have this impression more on Europe where's a lot of landuse. Also when appear the city labels for the first zoom, the font is the same size as the countries labels.. Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 __**_ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.**org/listinfo/talkhttp://lists.openstreetmap.org/listinfo/talk ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
There's nothing keeping one from applying the Tiles@Home lowzoom process to a slight variation of our standard Mapnik style however, and out comes this (for zoom levels 0-8; from z9 on, the standard Mapnik style looks fine): http://fred.dev.openstreetmap.org/lowzoom/ I love it. Australia is the sort of place where European defaults often don't work well, because our population density is so low. At zoom 7, the state of Victoria looks really empty: http://www.openstreetmap.org/?lat=-37.25lon=145.42zoom=7layers=M The lowzoom mode fills in the space better, but what it really needs is more place names (for our case, anyway). At that zoom, the whole state only shows 8 place names. Google Maps has dozens at a similar level, maybe more than 50 Steve. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Am 13. März 2012 00:20 schrieb Stephan Knauss o...@stephans-server.de: The strength of OSM is not in displaying landuse which we typically import from somewhere. This rendering is emphasizing the wrong aspect of OSM. around here we typically map the landuses by ourselves (aerial imagery and local knowledge), so it is genuine OSM-data. cheers, Martin ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
On 03/12/2012 10:21 AM, Erik Johansson wrote: On Mon, Mar 12, 2012 at 08:56, Frederik Rammfrede...@remote.org wrote: There's nothing keeping one from applying the Tiles@Home lowzoom process to a slight variation of our standard Mapnik style however, and out comes this (for zoom levels 0-8; from z9 on, the standard Mapnik style looks fine): http://fred.dev.openstreetmap.org/lowzoom/ That looks great, it would be nice to see it on the map. It's much nicer to see that map style than the default. +0 (I've always wanted this myself) I'm guessing most of the terrain pixels are Corine Land cover? The only problem I have with this is that it will make the Corine import go faster in all those white areas. Maybe most, but not all, we mainly used Landsat, Yahoo, Bing GPS for landuse. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
2012/3/12 Frederik Ramm frede...@remote.org http://fred.dev.openstreetmap.org/lowzoom/ The land cover looks great. I would maybe add country borders and remove region labels. If not remove, then make different than country labels. It's a bit of a mess like this. Janko ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
[OSM-talk] Mapnik Lowzoom TIles - proof of concept
Hi, one thing I particularly liked about the now-deceased Tiles@Home style was the lowzoom tiles which, in contrast to our rather bland Mapnik tiles http://www.openstreetmap.org/?lat=52.3lon=7.8zoom=5layers=M or those beefed up by terrain colouring http://www.openstreetmap.org/?lat=52.3lon=7.8zoom=5layers=Q actually showed real OpenStreetMap data. There's nothing keeping one from applying the Tiles@Home lowzoom process to a slight variation of our standard Mapnik style however, and out comes this (for zoom levels 0-8; from z9 on, the standard Mapnik style looks fine): http://fred.dev.openstreetmap.org/lowzoom/ I guess it's always a question of what audience a map is for, and my lowzoom tiles above look funny in areas where landcover has partially been imported; but for the mapper to get an idea of where there is data and where not, I quite like it. I think that if one applied a little more bitmap processing and maybe included, like t@h did, the boundaries on the lower zoom renderings, that could be a nice alternative to what we have. I'd like to have the roads a bit more prominent against the landcover perhaps. This is just a proof of concept and I'm not updating these tiles at the moment, but it would be quite easy to set this up on our main tileserver and keep it updated, and it wouldn't use a lot of resources. (Technical details: Took normal OSM Mapnik style and separated it in one style drawing only labels for z0-z8 (transparent background), and another style drawing everything but labels on z9; rendered world-wide set of no-label tiles on z9; produced z0-z8 tiles by bitmap-scaling/compositing appropriate number of z9 tiles and overlaying the label-only tile. All sources - hacky, Perl-y - available on request.) Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
On Mon, Mar 12, 2012 at 08:56, Frederik Ramm frede...@remote.org wrote: There's nothing keeping one from applying the Tiles@Home lowzoom process to a slight variation of our standard Mapnik style however, and out comes this (for zoom levels 0-8; from z9 on, the standard Mapnik style looks fine): http://fred.dev.openstreetmap.org/lowzoom/ +0 (I've always wanted this myself) I'm guessing most of the terrain pixels are Corine Land cover? The only problem I have with this is that it will make the Corine import go faster in all those white areas. -- /emj ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Hi, On 03/12/12 10:21, Erik Johansson wrote: I'm guessing most of the terrain pixels are Corine Land cover? The only problem I have with this is that it will make the Corine import go faster in all those white areas. Yes, that must not be encouraged. I think that if we make roads more prominent, then the non-landcover areas will look better. Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
On 12 mar 2012, at 11:22, Frederik Ramm wrote: Hi, On 03/12/12 10:21, Erik Johansson wrote: I'm guessing most of the terrain pixels are Corine Land cover? The only problem I have with this is that it will make the Corine import go faster in all those white areas. Yes, that must not be encouraged. I think that if we make roads more prominent, then the non-landcover areas will look better. It would be nice if landuse (forest etc) were _much_ less prominent on the default Mapnik layer. Really hard to see even major roads at low zoom if they go through a forest area as it is now. It's even a bit difficult at high zooms. :) Ex: http://www.openstreetmap.org/?lat=63.412lon=22.659zoom=10layers=M I think Hike Bike Map has better, more muted, colors for landuse: http://hikebikemap.de/?zoom=10lat=63.412lon=22.659layers=BF /Joakim ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
On 12.03.2012 11:22, Frederik Ramm wrote: Yes, that must not be encouraged. I think that if we make roads more prominent, then the non-landcover areas will look better. I also think it does focus to much on natural=* How would it look if adding highway=motorway,trunk,primary to the rendering? Maybe in light shades of grey? Is transparency an option? So areas with higher mapping density would come up more prominent? Was the query speed a problem for the mapnik setup? Stephan ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
that looks great fred, and yes Erik, I recognize the corine import that i did there in southeast europe. mike On Mon, Mar 12, 2012 at 10:21 AM, Erik Johansson e...@kth.se wrote: I'm guessing most of the terrain pixels are Corine Land cover? The -- James Michael DuPont Member of Free Libre Open Source Software Kosova http://flossk.org ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Hi, On 03/12/2012 09:18 PM, Stephan Knauss wrote: How would it look if adding highway=motorway,trunk,primary to the rendering? Maybe in light shades of grey? All these *are* already on the rendering; the rendering is based on plain z9 tiles which are just bitmap reduced. They're just not prominent enough. Using a few simple tricks a la http://www.remote.org/frederik/tmp/tiles/ the roads could perhaps be brought out more. Of course I could also simply not render landcover at all at z9. Is transparency an option? So areas with higher mapping density would come up more prominent? I don't think that would work with the bitmap approach. Was the query speed a problem for the mapnik setup? No, it was relatively benign, took my tile server a few hours to make the base tiles on z9 and then the rest was easy. Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
On 12.03.2012 22:05, Frederik Ramm wrote: On 03/12/2012 09:18 PM, Stephan Knauss wrote: How would it look if adding highway=motorway,trunk,primary to the rendering? Maybe in light shades of grey? All these *are* already on the rendering; the rendering is based on plain z9 tiles which are just bitmap reduced. They're just not prominent enough. Ah, you are right. I was looking at level 5. Americas looks as empty as south-east asia. If you zoom in far enough you can see the streets. I feel that this makes us look too much like any random topo map. The strength of OSM is not in displaying landuse which we typically import from somewhere. This rendering is emphasizing the wrong aspect of OSM. Did you create the style by yourself or was it some readymade you could reuse? Stephan ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Mapnik Lowzoom TIles - proof of concept
Stephan Knauss writes: ... displaying landuse which we typically import from somewhere. Careful with that we Eugene[1]! Some of us typically enter landuse[2] from our knowledge of the use of the land. [1] http://en.wikipedia.org/wiki/Careful_with_That_Axe,_Eugene [2] http://www.openstreetmap.org/?lat=44.6987lon=-74.9774zoom=13 -- --my blog is athttp://blog.russnelson.com Crynwr supports open source software 521 Pleasant Valley Rd. | +1 315-600-8815 Potsdam, NY 13676-3213 | Sheepdog ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk