Compressing vector graphics

A vector graphic isn't a grid of pixels. It's a drawing list of drawing objects and their properties: "circle here, this big, this colour".

So RLE on pixels and lowering the resolution have nothing to work on. Instead, you compress the drawing list itself.

Most vector files are already small. But a detailed map, a technical drawing or a complex illustration can hold thousands of objects and hundreds of thousands of coordinates, and they're sent across the web all the time.

A drawing list is data, often stored as text, so the lossless ideas you've met all work on it:

• Store repeated properties once. If 300 objects share the same fill colour, line colour and line thickness, store that set once as a named style and give each object a short reference to it.
• Store repeated objects once. A star drawn 50 times can be defined once and reused at 50 positions.
• Compress the file's text with dictionary-based compression, so repeated words such as circle or line colour become short indexes, or with RLE wherever the data contains runs.

The decompressor can expand every reference back, so no data is lost.