README

TinySDF

TinySDF is a tiny and fast JavaScript library for generating SDF (signed distance field) from system fonts on the browser using Canvas 2D and Felzenszwalb/Huttenlocher distance transform. This is very useful for rendering text with WebGL.

Demo

Usage

Create a TinySDF for drawing glyph SDFs based on font parameters:

const tinySdf = new TinySDF({
    fontSize: 24,             // Font size in pixels
    fontFamily: 'sans-serif', // CSS font-family
    fontWeight: 'normal',     // CSS font-weight
    fontStyle: 'normal',      // CSS font-style
    buffer: 3,                // Whitespace buffer around a glyph in pixels
    radius: 8,                // How many pixels around the glyph shape to use for encoding distance
    cutoff: 0.25              // How much of the radius (relative) is used for the inside part of the glyph
});

const glyph = tinySdf.draw('泽'); // draw a single character

Returns an object with the following properties:

data is a Uint8ClampedArray array of alpha values (0–255) for a width x height grid.
width: Width of the returned bitmap.
height: Height of the returned bitmap.
glyphTop: Maximum ascent of the glyph from alphabetic baseline.
glyphLeft: Currently hardwired to 0 (actual glyph differences are encoded in the rasterization).
glyphWidth: Width of the rasterized portion of the glyph.
glyphHeight Height of the rasterized portion of the glyph.
glyphAdvance: Layout advance.

TinySDF is provided as a ES module, so it's only supported on modern browsers, excluding IE.

<script type="module">
import TinySDF from 'https://cdn.skypack.dev/@mapbox/tiny-sdf';
...
</script>

In Node, you can't use require — only import in ESM-capable versions (v12.15+):

import TinySDF from '@mapbox/tiny-sdf';

Development

npm test  # run tests
npm start # start server for the demo page

License

This implementation is licensed under the BSD 2-Clause license. It's based directly on the algorithm published in the Felzenszwalb/Huttenlocher paper, and is not a port of the existing C++ implementation provided by the paper's authors.

Usage no npm install needed!