Rasterization
Also known as: rasterisation, raster rendering, rasterized rendering
Rasterization is the technique graphics cards use to turn 3D objects, made of triangles, into the pixels you see on screen.
It projects each triangle onto the screen, works out which pixels it covers, and colours them using shaders. While ray tracing gets the headlines, rasterization still does most of the work in almost every game, because it's dramatically faster.
Rasterization, explained at three levels
At home
Everything in a 3D game is built from tiny triangles. Rasterization is how the computer flattens those triangles onto your screen and fills in the right colours, many times every second.
At work
Rasterization makes real-time 3D graphics possible on phones, consoles, and PCs. It's why interactive product viewers, simulations, and games run smoothly.
At home
The GPU takes each triangle, figures out where it appears on screen, and fills in the pixels it covers. A 'depth buffer' keeps track of which object is closest, so things in front hide things behind them.
At work
The rasterization pipeline includes vertex processing, primitive assembly, rasterization, pixel (fragment) shading, and output merging. Effects such as shadows, reflections, and ambient occlusion are approximated with techniques like shadow maps and screen-space methods.
At home
Rasterization only knows what's on screen, so screen-space reflections disappear when the reflected object moves out of view. In my opinion, that's the clearest example of why ray tracing feels different: it can see what the camera can't.
At work
Modern engines use deferred rendering, mesh shaders, virtualized geometry (such as Unreal Engine's Nanite), and temporal techniques to scale rasterization. Performance tuning focuses on draw calls, overdraw, memory bandwidth, and shader complexity, especially on mobile and standalone VR hardware.
Real-world examples
At home
Fatima noticed that in an older game, reflections in a lake vanished when she tilted the camera down. A friend explained it was a rasterization shortcut that only reflects what's already on screen.
At work
Sophie builds an online furniture configurator for a Montreal retailer. Because it uses rasterization through WebGL, customers can rotate sofas smoothly on ordinary laptops and phones.
How it compares
| feature | Rasterization | Ray tracing |
|---|---|---|
| Speed | Very fast | Slower |
| Reflections and shadows | Approximated | Physically simulated |
| Hardware support | Every GPU | Newer GPUs |
| Best for | Real-time graphics everywhere | Realistic lighting effects |
Rasterization is fast approximation. Ray tracing is slower simulation.
For analysts
For interactive 3D on customer devices, rasterization is the safe default. I'd reserve ray-traced features for audiences with high-end hardware, since most users won't have it.
History & origin
Rasterization grew from 1960s and 1970s computer graphics research, including work at the University of Utah, where Ed Catmull and others developed key techniques such as the depth buffer (Z-buffer), and Henri Gouraud and Bui Tuong Phong created shading methods. Silicon Graphics built rasterization hardware for workstations in the 1980s and 1990s. Consumer 3D graphics cards, such as 3dfx's Voodoo in 1996 and Nvidia's GeForce 256 in 1999, brought it to home computers. Programmable shaders in the 2000s made rasterized graphics far more realistic. In my view, rasterization's decades of clever tricks are why games looked so good long before ray tracing was practical.