This talk explores how map styles and underlying data can be automatically optimized together to improve performance and resource efficiency. Using practical examples, it demonstrates how data- and style-driven techniques can reduce tile sizes, shorten loading times, and make client-side rendering more efficient.

With the introduction of the OpenStreetMap Foundation’s vector tiles, the barrier to using high-resolution, client-rendered vector maps has never been lower. However, these maps often struggle with performance compared to server-rendered alternatives.

How can we close this gap?

One overlooked opportunity is the joint optimization of visual styles and underlying data. Existing systems - and even much of the research - treat these separately, leading to unnecessary data transfer, higher server load, and reduced client performance.

In this talk, I present my master’s thesis, which approaches data- and style-driven optimization of client-rendered vector maps as a spatial query problem using database techniques. The result: reduced data size and complexity, and improved performance - without compromising visual quality, but at the cost of worse debuggability and flexibility.

The talk focuses on the real-world impact of these optimizations on network usage, rendering performance, and energy consumption across actual map styles and datasets.