L7 is an open-source, WebGL-powered geospatial data visualization and analysis engine from Ant Group's AntV team, written in TypeScript and aimed at developers who need to render and explore large-scale location data inside web maps, BI dashboards, and GIS-style applications.

What it is

L7 is a browser-side rendering engine for geospatial data. The name encodes its scope: L stands for Location and 7 stands for the seven continents, signalling an intent to provide visual analysis for location data worldwide. It is published under the MIT licence as the npm package @antv/l7, with map integrations shipped separately in @antv/l7-maps. Rendering runs on WebGL, which is what lets it draw large volumes of spatial data in real time and in motion rather than as static images. The project sits inside the AntV visualization ecosystem maintained by the Ant Group data visualization team, alongside sibling tools such as L7Draw, L7Plot, LarkMap, L7VP, and L7Editor, and its source and issue tracker live on GitHub with roughly 4,000 stars, 656 forks, and 216 open issues.

The concrete problem it solves is the gap between raw tabular spatial data and a legible map. Teams that want symbols, paths, filled regions, or heatmaps on a basemap would otherwise have to write their own WebGL rendering layer and a spatial data transformation pipeline. L7 removes both: it accepts CSV, JSON, and GeoJSON directly, or custom formats, without requiring complex spatial data conversion, and it exposes layer construction through a small chained API. The README positions it for common map charts, visual analysis inside BI systems, and application development in GIS, transportation, electric power, land, agriculture, and urban domains.

Key capabilities

• A Scene object initializes the map canvas, combined with a basemap provider such as Mapbox imported from @antv/l7-maps.
• Layers are built by chaining data-driven visual mappings, for example new PointLayer().source(data).shape('circle').size('mag', [1, 25]).color('mag', ['#5B8FF9', '#5CCEA1']), so visual variables such as colour, size, texture, direction, and volume carry the data meaning.
• Data ingestion supports CSV, JSON, and GeoJSON, plus user-defined formats, without complex spatial data transformation.
• Point layers cover bubble charts, scatter plots, symbol maps, 3D column maps, aggregation maps, composite chart maps, and custom markers.
• Line layers cover path maps, arcs in 2D and 3D, great-circle routes, and contour lines.
• Polygon layers cover fill maps and 3D fill maps, while heatmaps come in classic, honeycomb, and grid forms, and raster layers handle image and raster data.
• Basemaps are abstracted behind one interface, with Amap used for domestic Chinese compliance and Mapbox for international business, and offline intranet deployment is supported.

Who uses it and how

• Product teams building map charts and visual analysis into BI systems, where the engine handles the rendering while the surrounding application supplies the query layer.
• Spatial information management and analysis systems in GIS, transportation, electric power, land, agriculture, and urban planning, where 2D and 3D views of the same dataset are both required.
• Organizations running international and domestic operations side by side, switching between Mapbox and Amap basemaps without changing the data layer code.
• Deployments on isolated intranets, which the project supports explicitly and which rules out basemap services that require public network access.
• JavaScript and TypeScript front-end developers already working in the AntV ecosystem, who can pair the engine with L7Plot, LarkMap, L7VP, or L7Editor rather than building those layers themselves.

Getting started

Install with npm install @antv/l7, add @antv/l7-maps for basemap support, then construct a Scene with a Mapbox basemap and add layers such as a PointLayer. Documentation, tutorials, and examples are hosted at https://l7.antv.antgroup.com.

How it compares

The facts do not name a set of paid products that L7 substitutes for, so the honest comparison is positional. Mapbox and Amap appear here as basemap providers that L7 consumes rather than as alternatives to it, and the closest named neighbours are the AntV siblings L7Plot, L7Draw, LarkMap, L7VP, and L7Editor, which extend the same ecosystem at different levels of abstraction. Anyone choosing between L7 and another rendering stack is choosing a layer engine, not a map provider.

When to use it — and when not to

L7 is a client-side rendering library, so there is no database, object storage, or SMTP service to operate; the work is bundling it into a front-end build and supplying a basemap. Anyone who needs a server-side geospatial query engine, a tile server, or a no-code mapping product should look elsewhere, because L7 is a rendering and visual-analysis layer, not a data platform. The main frictions visible in the facts are documentation that is Chinese-first with an English translation at ./README.en-US.md, and 216 open issues, so teams without Chinese-language capacity or a tolerance for tracking upstream fixes should weigh that before adopting.

npm install @antv/l7

import { Scene } from '@antv/l7';
import { Mapbox } from '@antv/l7-maps';

const scene = new Scene({
  id: 'map',
  map: new Mapbox({
    style: 'light',
    pitch: 0,
    center: [107.054293, 35.246265],
    zoom: 4.056,
  }),
});

import { PointLayer } from '@antv/l7';

const pointLayer = new PointLayer()
  .source(data)
  .shape('circle')
  .size('mag', [1, 25])
  .color('mag', ['#5B8FF9', '#5CCEA1'])
  .style({
    opacity: 0.3,
    strokeWidth: 1,
  });

scene.addLayer(pointLayer);