HDM-CartoCSS icon indicating copy to clipboard operation
HDM-CartoCSS copied to clipboard

Tiles render with transparent lines between them at global scale

Open timlinux opened this issue 10 years ago • 9 comments

We are using the HOT OSM tiles for the InaSAFE user map at http://users.inasafe.org The map looks great but at global scale there are strange transparent lines between tiles as per the image below:

screen shot 2014-02-11 at 10 19 30 am

We can duplicate this using chrome and firefox (test on OSX and Linux Mint).

It would be great if these tiles could be made seamless.

timlinux avatar Feb 11 '14 03:02 timlinux

I'm just an observer, but... Are you talking about the lines on Antarctica? Those are not inter-tile lines, but lines marking territorial claims in Antarctica (i.e. they're essentially the "same" as national boundaries).

danstowell avatar Feb 11 '14 08:02 danstowell

Hi. Yes I am talking about those. The problem is that with the current styling they look like transparency rather than actual lines.

On Tue, Feb 11, 2014 at 3:36 PM, danstowell [email protected]:

I'm just an observer, but... Are you talking about the lines on Antarctica? Those are not inter-tile lines, but lines marking territorial claims in Antarcticahttps://en.wikipedia.org/wiki/Territorial_claims_in_Antarctica(i.e. they're essentially the "same" as national boundaries).

Reply to this email directly or view it on GitHubhttps://github.com/hotosm/HDM-CartoCSS/issues/216#issuecomment-34734963 .

Tim Sutton

Visit http://linfiniti.com to find out about:

  • QGIS programming services
  • GeoDjango web development
  • QGIS Training
  • FOSS Consulting Services Skype: timlinux Irc: timlinux on #qgis at freenode.net Tim is a member of the QGIS Project Steering Committee

timlinux avatar Feb 11 '14 08:02 timlinux

hey @timlinux :) Thanks for the report. @danstowell is right, those are administrative boundaries. Now, the rendering is complex because each way is part of many boundaries, and we in general want to render all of them; I'm not sure I can make a dedicated rule for Antarctica. But I agree that the result is not smart, I will see what's doable. And as always I'm open to suggestion and pull requests ;)

yohanboniface avatar Feb 11 '14 11:02 yohanboniface

Maybe one option is not to render maritime boundaries below zoom 5 or something like this.

yohanboniface avatar Feb 11 '14 11:02 yohanboniface

Yeah that might work - thanks for looking into this!

On Tue, Feb 11, 2014 at 6:33 PM, Yohan Boniface [email protected]:

Maybe one option is not to render maritime boundaries below zoom 5 or something like this.

Reply to this email directly or view it on GitHubhttps://github.com/hotosm/HDM-CartoCSS/issues/216#issuecomment-34746222 .

Tim Sutton

Visit http://linfiniti.com to find out about:

  • QGIS programming services
  • GeoDjango web development
  • QGIS Training
  • FOSS Consulting Services Skype: timlinux Irc: timlinux on #qgis at freenode.net Tim is a member of the QGIS Project Steering Committee

timlinux avatar Feb 11 '14 11:02 timlinux

@yohanboniface Thanks! Do you know when these changes will propogate to the published map tiles?

timlinux avatar Mar 15 '14 18:03 timlinux

@timlinux code is already deployed, but low zoom levels are only generated by cron once a week.

The bad news is that I've tested some tiles manually and it seems that some boundaries are still visible. I need to investigate. I'm afraid it may be linked to some conflict between way tags and relation tags…

yohanboniface avatar Mar 15 '14 21:03 yohanboniface

@yohanboniface OK thanks for the update.

timlinux avatar Mar 15 '14 23:03 timlinux

OSM-carto uses a shapefile under zoom 4, I think it's the only viable option, given how the relations are imported by osm2pgsql at this time (one geometry for all the objects of the relation, with one set of tags, and so only one value for each tag, even if not all the relation elements have the same value; but it's certainly a hard problem to solve and keep optimizated). Each of the elements of the relation is also imported separately (with of course the original tag values), so I've tried not to select relations (osm_id < 0) but for now without success: the resulting dataset is not the one I'm expecting. To be continued…

yohanboniface avatar Mar 16 '14 17:03 yohanboniface