natcap / urban-online-workflow

This repository hosts the beta implementation of the Urban Online ES Workflow. The project is intended to give urban planners the ability to create and assess scenarios using InVEST Urban models.
1 stars 5 forks source link

Serve vector tiles #5

Closed phargogh closed 1 year ago

phargogh commented 2 years ago

We'll need the ability to serve standard map tiles based off of the wallpapered LULC, so it'd be helpful to try out some existing tile servers to see if there's one that better suits our needs. A quick google search turned up:

dcdenu4 commented 2 years ago

Related this this, it would be good to have a discussion about the zoom levels we'll need as this can directly effect the size and computation time taken for tiling. For instance Mapbox has a tiling tool called 'tippecanoe' for vector tiling that might be useful. https://github.com/mapbox/tippecanoe#zoom-levels

dcdenu4 commented 2 years ago

This other mapbox reference might also be useful for thinking about precision: https://docs.mapbox.com/mapbox-tiling-service/guides/#precision-levels-and-square-kilometers

emlys commented 2 years ago

I tried out the Mapbox Tiling Service on the San Antonio parcel data with zoom level 16 and it seems good.

davemfish commented 2 years ago

@phargogh It's looking more and more like we'll use REGRID's Tileserver for parcel vectors. So unless we have other vector data that we need on a map, we may not need to setup our own tileserver.

dcdenu4 commented 2 years ago

Depending on scale and usability we still might want to serve our raster outputs (for visualization) via tiles or via a OCG format.

phargogh commented 2 years ago

It's looking more and more like we'll use REGRID's Tileserver for parcel vectors

@davemfish great! I'm really glad that those tiles will be able to be served through REGRID.

Depending on scale and usability we still might want to serve our raster outputs (for visualization) via tiles or via a OCG format.

@dcdenu4 and if I remember correctly, it sounded a bit like we might still be needing to convert wallpapered LULCs to vectors in order to serve those to the frontend, right? If so, a vector tile service might still be relevant.

davemfish commented 2 years ago

@dcdenu4 and if I remember correctly, it sounded a bit like we might still be needing to convert wallpapered LULCs to vectors in order to serve those to the frontend, right? If so, a vector tile service might still be relevant.

I'm not sure why we would want to convert to vector instead of serving the LULC as raster tiles?

phargogh commented 2 years ago

Sorry, I'm still catching back up on things here coming back to the office. I thought I remembered from a previous conversation that it would be easier or better (I forget why) for us to render vectors in OpenLayers than it would be to render rasters. If that's not the case, great!

davemfish commented 2 years ago

We're all still learning but I think rendering rasters is just fine for any data where we don't want to interact with features & geometries.

dcdenu4 commented 2 years ago

Ah! Sorry, I was just thinking of serving raster tiles specifically. I see this issue is "Serve vector tiles".

davemfish commented 1 year ago

Now that we're deeper into the project, I don't think we have a need for serving vector tiles at the moment.