search

AndrewAnnex / planetcantile

tile matrix sets for other planets
BSD 3-Clause "New" or "Revised" License
9 stars 1 forks source link

Vector Tiles #4

Open jlaura opened 1 year ago

jlaura commented 1 year ago

A natural follow on the raster tiler work might be to look at vector tiles. timvt is a vector tile setup (data in PostGIS). This uses morecantile in the background, so there should be a fair bit of overlap. One thing I do not fully understand is how this might (or might now) work with elastic search.

Why ES? Because pygeoapi has support for CQL only on ES providers. Other providers are working this I'm sure since this is part 3 (I believe) of the OGC features API. Right now, if one wanted to serve vector data for CQL searches and the same data with a vector tile provider they would need to run ES and PostGres.

This opens a whole can of worms then about styling and symbology. Here is a link to Alessandro's repo with SVG-SLD work.

AndrewAnnex commented 1 year ago

@jlaura there are a lot of backends for vector data sources that do and don't support CQL (geoserver does for example). I was also looking at https://github.com/kylebarron/dem-tiler which does something similar to timvt by serving mapbox vector tiles for contours using gdal_contour and tippecanoe.

One could imagine using the postgis driver in ogr2ogr to do something similar, or be semi agnostic to the specific backend provided there was gdal driver for the backends... But I wonder if the conversation about pygeoapi is really appropriate to this repo, as the moment these tilesets work it should be possible to use whichever solution you want.

I mean I do see that these vector tile formats also assume web mercator projections by default, maybe there's something to figure out there?

regarding the SVG-SLD work, how much of that is solved by Daven Quinn's project: https://davenquinn.com/projects/geologic-patterns/

jlaura commented 1 year ago

@jlaura there are a lot of backends for vector data sources that do and don't support CQL (geoserver does for example). I was also looking at https://github.com/kylebarron/dem-tiler which does something similar to timvt by serving mapbox vector tiles for contours using gdal_contour and tippecanoe.

For sure, but not also OGC Features 1.0 compliant. And I agree that these should be possible to use, but I am not sure that they will be. Not all of the solutions are then fully compliant with the OGC requirements. Sure, maybe not appropriate for this repo, but we have upstream stuff to take care of to be able to use this work across the ecosystem.

So much of the ecosystem assumes web mercator, I'm not sure how to get around that honestly.

One the SVG - SLD side, thanks for linking Daven's work. There is a little bit of overlap in the patterns. Daven has more patterns supproted than Alessandro. Alessandro has more of the line / point symbologies (they are noted as future for Daven.) So, this still looks to be a wide open problem.

AndrewAnnex commented 1 year ago

Hmm not so sure about that if you were saying geoserver didn’t support it https://docs.geoserver.org/latest/en/user/community/ogc-api/features/index.html but if not no worriesThis is a fairly new spec and something to keep an eye on for sure as more clients and servers come along, the geojson/gml part is something to look into more, maybe we can ship geojsons using the native crs coordinates and not worry about the whole “geojson is 4326” thing given the use of the tilematrixset…. - Dr. Andrew AnnexOn Dec 15, 2022, at 10:30 AM, jlaura @.***> wrote:

@jlaura there are a lot of backends for vector data sources that do and don't support CQL (geoserver does for example). I was also looking at https://github.com/kylebarron/dem-tiler which does something similar to timvt by serving mapbox vector tiles for contours using gdal_contour and tippecanoe.

For sure, but not also OGC Features 1.0 compliant. And I agree that these should be possible to use, but I am not sure that they will be. Not all of the solutions are then fully compliant with the OGC requirements. Sure, maybe not appropriate for this repo, but we have upstream stuff to take care of to be able to use this work across the ecosystem. So much of the ecosystem assumes web mercator, I'm not sure how to get around that honestly. One the SVG - SLD side, thanks for linking Daven's work. There is a little bit of overlap in the patterns. Daven has more patterns supproted than Alessandro. Alessandro has more of the line / point symbologies (they are noted as future for Daven.) So, this still looks to be a wide open problem.

—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you commented.Message ID: @.***>