MalaGIS

Sharing GIS Technologies, Resources and News.

Georeferencing in the AI Era: From Manual Struggles to One-Click Solutions

As a GIS professional, who hasn't been driven mad by the manual georeferencing of historical imagery? When rushing to meet project deadlines, the author often had to hunch over the screen, scrutinizing scanned maps for control points — they had to be "timeless" features like road intersections, bridges, or landmark buildings. If the old map was blurry or the landmark had been demolished, it meant digging through archives all over again. Then, switching to ArcGIS, we would painstakingly align points one by one against a reference basemap, manually inputting latitude and longitude coordinates. Each point had to be checked three times, fearing a single entry error could throw the entire map out of alignment. What's even more torturous is that too few control points compromise accuracy, while too many can introduce distortion. And when batch-processing aerial photos? That meant burning the midnight oil in front of the computer, repeating the tedious operations until your fingers trembled.

The scene described by the author is sure to send shivers down the spines of many, a true nightmare. However, in 2026, AI is finally starting to save (or take over the jobs of) GIS drafters. In a previous article, "QGIS Map Georeferencing Tutorial (with AI)", the author introduced an AI georeferencing plugin for QGIS, but its performance was somewhat lacking. Recently, @yaoyao, author of Kongtian Perception, recommended a tool to the author: georeferencer.ai. Let's take a look at it today.

Read More >>

A GIS Practitioner's Guide: The Quest for POI Data on "Free Noodle Refill" Restaurants

A few days ago, news about the "noodle refill" incident in Jiajiang County, Leshan, Sichuan Province, became a hot search topic, and someone also shared this news in the "Malá Là GIS" group chat. While debates about cultural differences and other issues were still raging on Weibo, a breath of fresh air appeared in our GIS group when someone asked: How can we obtain POI data for restaurants that offer "free noodle refills"? (See? Our GIS group truly has a professional spirit.)

As you know, I usually don't slack off, but I couldn't resist such an interesting GIS question. I consulted with various experts and did some hands-on experimenting. Although the final result wasn't entirely satisfactory, I'll still summarize the process for everyone today.

Traditional POI Databases

I've shared many POI datasets before, the most famous being the POI dataset from OSM. However, I found that the data fields provided by such sources are currently too simplistic, as shown below:

Read More >>

A Practical Guide to Patching Node Modules in GIS Frontend Development

As a GIS frontend developer, I work extensively with various large-scale graphics libraries: OpenLayers, Leaflet, Mapbox GL JS, Cesium, Turf.js... These libraries often contain tens of thousands of lines of code. I wonder if you've encountered situations during development where you needed to modify the source code within node_modules. For example, in my old projects using OpenLayers 6.x, some third-party plugins were no longer updated. In such cases, upgrading to newer plugin versions would require upgrading the OpenLayers version, which is often an impossible workload for a project delivered years ago.

Read More >>

MapLibre Tile (MLT): A Next-Generation Vector Tile Standard

Just a few days ago, the MapLibre community officially released the MapLibre Tile (MLT) format. This could very well be the most fundamental and hardcore technological innovation in the WebGIS domain since Mapbox defined the MVT standard a decade ago. The official announcement claims that MLT can achieve compression rates up to 6 times better than MVT and decoding speeds 3 times faster! So what exactly is MLT? Why is it positioned to challenge MVT? Let's delve into the details today.

Read More >>

A Fresh Contender in IP Geolocation for GIS: public-ip-address

In previous articles such as "Revisiting Locating User Positions Using IP Addresses (ip2region)" and "Summary of Methods for Locating User Positions Using IP Addresses", we introduced many methods and tools for IP geolocation. These include GeoIP, ipip, ip2region, as well as numerous third-party service providers like Amap and Baidu. Among them, some are online, some are offline, some are paid, and some are free. A few days ago, the editor came across another interesting open-source IP geolocation library: public-ip-address.

Read More >>

Offline Map Tile Downloader

During GIS project delivery, situations with weak or no network connectivity are frequently encountered. Typically, we need to manually download map tiles from various providers. Currently, there are numerous map downloaders available on the market, both free and paid. Upon closer thought, this process isn't particularly complex; it essentially breaks down into three steps: defining the area of interest, downloading all tiles within that area, and exporting. Therefore, while working on a project, the author searched for a simple, small-scale downloader and ultimately found one on GitHub: OfflineMapDownloader.

Project Introduction

Original Project Repository:

https://github.com/0015/OfflineMapDownloader

After testing, the author identified several significant issues, primarily in the following three areas:

  1. It only supports OpenStreetMap and ArcGIS Online.
  2. It does not support downloading tiles for regions in China.
  3. The downloaded .mbtiles files cannot be opened in QGIS.

The issue with China regions was particularly critical, rendering the tool unusable for the author. Consequently, the author made some simple fixes and created a new version. The author's version repository:

https://github.com/sailor103/OfflineMapDownloader

This version primarily addresses the three aforementioned problems. It can now correctly export maps for Chinese regions and supports opening the files in QGIS, as shown below:

Read More >>

Wish3D: A Lightweight Open-Source Engine for 3D Reality Mesh Publishing

In today's era of booming real-world 3D technology, OSGB format models are widely used. However, "how to publish efficiently and browse smoothly" has always been a pain point for developers: slow loading, large file sizes, laggy interactions on mobile devices... Even star engines like Cesium struggle with issues such as "bloated architecture and subpar mobile experience." In mid-January this year, Zhongke Tuxin open-sourced a lightweight real-world 3D model publishing engine — Wish3D. It claims to be free and open-source, accessible for both individuals and enterprises without barriers. The editor saw it in the MalagiGIS group yesterday and immediately downloaded it for a test drive. Overall, it's indeed fast, but there are several minor issues. Let's share the experience today.

Read More >>

MBTiles: An Efficient Map Tile Storage Format

Recently, while working on an offline map deployment project, I found that the transfer efficiency was relatively low and offline deployment cumbersome when dealing with massive amounts of scattered tile files. Initially, I considered using a ZIP package for transfer and extraction, but later discovered that a format already existed which implements this concept and does so even better: MBTiles. MBTiles is a map tile storage specification based on an SQLite database. It packages thousands of independent map tiles (PNG, JPG, or vector PBF) into a single .mbtiles file.

Read More >>

Mapbox Launches 3D Lanes for Enhanced Navigation at Complex Intersections

On January 5, 2026, Mapbox, a leading global location services platform, officially announced the launch of its Mapbox 3D Lanes feature, designed to address the common problem of taking wrong turns at complex intersections. The Mapbox 3D Lanes feature includes lane geometry, lane markings, and 3D models of overpasses and tunnels, helping drivers follow their routes more easily.

How Effective Is It?

Frankly, our team was immediately captivated by the released preview images. After all, when it comes to aesthetic appeal, Mapbox has never been one to shy away from a challenge.

Read More >>

Comparative Overview of Two Global Building Footprint Datasets

In a previous article titled "Introducing GlobalBuildingAtlas: A Global Building Height Dataset ", I shared a global building dataset that includes height information. However, some readers noted concerns about its data quality. Therefore, I would like to share two other global building footprint datasets I discovered while working on CIM-related projects: one from Microsoft (open-source) and another from Google. Both are products of major tech companies, and their usefulness can be evaluated through testing.

Microsoft GlobalMLBuildingFootprints

Microsoft's GlobalMLBuildingFootprints open-source repository provides a global building footprint dataset. It detects approximately 1.4 billion buildings from multi-source Bing Maps imagery (2014–2024) and is freely available under the ODbL license. The data is provided in row-separated GeoJSON format (with a .csv.gz extension) using the EPSG:4326 coordinate system and includes attributes such as estimated height and confidence scores.

Read More >>

Copyright © 2020-2026 MalaGIS Drive by Typecho & Lingonberry Sitemap

Back to top