Comment by ssl232 | Hacker News Reader

Comment by ssl232 | Hacker News Reader

I dabbled in web-based mapping a couple years ago and found the tech stack to be really quite polished. Starting from a database of features from e.g. OpenStreetMap hosted in a Postgres-based PostGIS map server, and geographical boundaries via special shape files, you can write XML files representing the features you wish to see on each layer of the map (using SQL queries intended for PostGIS), and CSS-like style files representing how the features should look. You then set up a tile server (Apache or nginx or whatever) that generates tiles on demand and performs caching. Finally you use a JavaScript library to provide the "slippy map" that users can drag around.

All of these layers of the tech stack have FOSS implementations; indeed most of the most widely used ones are FOSS. And then there's QGIS (also FOSS) to do most of this in a similar fashion, but offline.

As merely a dabbler having a bit of fun, it's rather excellent I must say. It would be interesting to hear from anyone working in this field professionally as to how the FOSS tools measure up.

QGIS is enormously valuable to those of us who need to do what I would non-authoritatively term "intermediate level geospatial analysis" in jobs that might not have geospatial analysis as part of the official job description, but don't have thousands of dollars available for the ludicrously expensive ArcGIS license.

I find the analysts still need ArcGIS, but on the geospatial software development side I've never run into something I couldn't use QGIS for

My current job has pretty low standards for maps in reports, and unfortunately QGIS has a dealbreaker issue: I'd like to just import a basemap, display my data on top, and then export as an image; however, in QGIS the resolution of the basemap scales with the resolution of the export. So, for example, if you want a 300 dpi image, all the city and and street labels (from the basemap) will be microscopic. It's baffling that the software does this, and as a result I have to use ArcMap for these simple maps.

QGIS will, for raster basemaps, fetch the detailed version to get you your high DPI detail but this of course involves stitching together lots of tiles from the basemap's lower zoom levels - which then have small labels. For raster maps which are not labelled, it makes complete sense and produces a much better result than the alternative (picking the "scale based" zoom level and interpolating).

This is kind of the only "right" answer when dealing with raster basemaps. It's either pixellated or going to be rendered "too small" for the zoom level.

Vector basemaps don't have this problem, and QGIS supports them, so that's the way to go if you can get data. QGIS can then render at the required DPI in full clarity but with elements scaled/positioned appropriately.

> Vector basemaps don't have this problem

I was about to firmly disagree, but perhaps my issue is that I've been assuming XYZ tiles were vector. Maybe _some_ are vector, based on the connection? If I'm remembering right, I imported a load of XYZ sources, and then never thought twice about them. I should have been more discerning. I'll look deeper into getting some proper vector basemaps.

> This is kind of the only "right" answer when dealing with raster basemaps.

I believe ArcMap uses rasters (when you choose the simple "add data" feature and one of their pre-selected basemaps), but they still render the labels appropriately when you export.

Here are a few workarounds for raster tiles [0]. You can either:

- limit the maximum zoom for raster tiles layers

- use higher resolution tiles (if they exist @2x or @4x)

- lower the DPI in Layout Settings -> Export settings (it should not impact vector layers)

- user vector tiles

[0] https://gis.stackexchange.com/questions/286167/qgis-gives-di...

We do a huge amount of very complex analysis and entirely run QGIS. Looked at ArcGIS and decided it wasn't anywhere near worth the money, and getting into their ecosystem at all makes it borderline impossible to interop with e.g. QGIS sensibly.

That is such an absurdly specific but accurate description. I used to manage a site that had a lot of geospatial elements, including various custom maps and obscure (outside of the GIS world) data formats like geotiffs or KML or DTED or the WMM. Coming from a non-GIS background and without the need for an expensive solution like ArcGIS, I have no idea how I would have sanity checked any of it without QGIS

I've recently started working in this space. Loads of applications in research, Ag tech and environmental science and it's mostly using this tech stack (as well as GeoServer.) Definitely a nice change from my typical ecommerce/saas work. Working with some long term geo/surveyor types and they seem happy with the open source offerings.

Reply on news.ycombinator.com