How Vector Graphics Are Rendered on the GPU (and Why It Matters)
How vector graphics get from paths to pixels: CPU vs GPU rendering, tessellation, anti-aliasing, renderers like Skia, and what it means for animation.
Core Takeaway
To draw vector graphics, a renderer turns paths into pixels: it works out which pixels each shape covers, smooths the edges and fills them with colour. CPUs can do this, but GPUs do it in parallel and much faster for complex scenes. Modern renderers such as Skia can use the GPU, which is why rich vector animation can run smoothly on phones.
From path to pixels
A vector shape is a set of curves. To display it, the renderer must decide coverage (which pixels are inside the shape), anti-alias the edges so they do not look jagged, and blend the colour with what is behind it. Doing that for many shapes, 60 or more times a second, is the core challenge.
CPU vs GPU
CPU rendering is flexible and simple but slows down with complexity. GPU rendering processes many pixels in parallel.
- Tessellation: break curves into triangles the GPU can draw
- Stencil-and-cover: mark covered pixels, then fill them
- Compute approaches: newer techniques that use GPU compute for coverage
Renderers you already use
Skia, the graphics library behind Chrome, Android and Flutter, can render on CPU or GPU. On the web, WebGL and the newer WebGPU API give access to the GPU from the browser.
What it means for your animations
GPU rendering makes many shapes cheap, but some things still cost: very large blurs, many overlapping transparent layers, constantly changing paths with many points, and uploading large images. Keep effects modest, simplify paths, and test on low-end devices, where GPUs are much weaker.
Key Principles to Remember
- Rendering means coverage, anti-aliasing and blending
- GPUs parallelise this and handle complex scenes better
- Skia, WebGL and WebGPU are common paths
- Blur, transparency and dense paths still cost
Frequently Asked Questions
Sources
- Skia — skia.org
- WebGPU API — MDN Web Docs