The Evolution of Game Graphics: From Pixels to Ray Tracing
How game graphics evolved, from vector displays and pixel art to 3D polygons, GPUs and shaders, PBR, ray tracing and AI upscaling.
Over a few decades, game graphics have gone from a handful of dots to near-cinematic visuals. But this change can't simply be summed up as "it kept getting better." Every era had its own hardware limits, and developers devised ways to express themselves within those limits. Those inventions sometimes became the technology of the next era.
In this post we look in order at the major stages game graphics have gone through, starting with vector displays and pixel art, then 3D polygons, graphics accelerators and GPUs, shaders, physically based rendering, and finally ray tracing and AI upscaling.
The Era of Lines and Dots: Vector and Raster
Early arcade games used two kinds of display. One was the vector display, which drew lines directly with an electron beam; games like Asteroids and Tempest used it. Lines were extremely crisp, but filling in surfaces or using rich color was difficult. A home vector-based console called the Vectrex was also released.
The other was the raster method, which, like a television, scans the screen line by line to plot dots (pixels). Raster ultimately won the market, and vector displays all but disappeared after the mid-1980s. On raster screens, the approach of moving small images called sprites over a background took hold.
Art Born of Constraints: 8-Bit and 16-Bit Pixel Art
Consoles of the 8-bit and 16-bit era had strict limits on the number of colors per screen, the size and number of sprites, and memory. So designers had to decide the position and color of every single dot by hand. Images created by deliberately placing pixels like this are what we now call pixel art.
Techniques used in this era are still widely used today:
- Palette swapping: Changing only the colors of the same image to represent different enemies or players.
- Dithering: Alternating dots of two colors to make them look like an in-between shade.
- Parallax scrolling: Moving background layers at different speeds to create a sense of depth.
Pixel art survived as an art style even after the technical constraints disappeared. Today it's especially popular in indie games because it lets small teams create a consistent look.
The Arrival of 3D Polygons
In the early 1990s, 3D graphics—building solid shapes by joining polygons such as triangles and quadrilaterals—entered games in earnest.
| Year | Game / Hardware | Significance |
|---|---|---|
| 1993 | Star Fox (SNES) | Put a Super FX chip in the cartridge to achieve polygon 3D on a home console |
| 1993 | Virtua Fighter (arcade) | First arcade fighting game with fully 3D polygon graphics |
| 1994 | PlayStation (Japan launch) | Home console designed around 3D polygon graphics |
Sega's Virtua Fighter built its characters from polygons with flat-shaded faces, and even that was a huge shock at the time. Seeing the game's success, Sony set the PlayStation's direction around 3D polygons. As texture mapping—applying images onto polygons—became widespread, 3D visuals quickly grew more realistic.
Graphics Accelerators and the GPU
On PCs, it was too much for the CPU to handle all the 3D calculations. Dedicated 3D accelerator cards solved this problem.
Founded in 1994, 3dfx opened up the PC 3D acceleration market with its Voodoo Graphics chip. In the late 1990s, it was almost a given that 3D games supported 3dfx's Glide API. But as competition intensified, 3dfx ran into trouble, and in late 2000 Nvidia acquired most of its assets.
Nvidia released the GeForce 256 in 1999, calling it "the world's first GPU." The key was that the graphics chip itself handled transform and lighting (T&L) calculations previously done by the CPU. From then on, the graphics card was no longer just an auxiliary device but a core component that determines game performance.
Shaders: Programmable Graphics
Early graphics chips used a fixed-function pipeline that could only draw the screen in predetermined ways. That changed in 2001, when Nvidia's GeForce 3 became the first to support programmable pixel shaders. Shaders are small programs written by developers that determine how each vertex or each pixel on the screen is processed.
Shaders made the following effects possible in real-time games:
- Normal mapping: A technique that applies a texture containing surface bump information to a low-polygon model to make it look far more detailed. It began appearing in early-2000s PC games and titles like Virtua Fighter 4 (2001), and became nearly standard by the Xbox 360/PS3 generation.
- Dynamic shadows, reflections and water: Lighting effects that change with the scene could now be calculated.
- Post-processing effects: Applying effects such as bloom, depth of field and motion blur to the entire screen.
Graphics hardware later evolved into a unified shader architecture, in which the same compute units share all the work instead of having separate vertex and pixel shaders. Thanks to this versatility, GPUs would later be widely used for AI computation as well.
Physically Based Rendering: Standardizing How Light Is Simulated
In the 2010s, the way materials are represented was largely standardized. Previously, artists often eyeballed how shiny or reflective things should be scene by scene. Physically based rendering (PBR) defines materials by physical properties such as metalness and roughness, and calculates how light reflects off surfaces with consistent formulas.
The advantage of PBR is that materials look natural even when the lighting changes. In 2013, Remember Me partially adopted it using a modified Unreal Engine 3, followed the same year by Ryse: Son of Rome and Killzone Shadow Fall. As major engines including Unreal Engine 4 adopted physically based materials by default, it became the industry standard.
Ray Tracing: An Old Theory Goes Real-Time
Until now, most game graphics have used rasterization—a method of quickly drawing 3D shapes by projecting them onto the screen's pixels. Ray tracing, by contrast, traces the paths of light ray by ray to calculate reflections, refractions and shadows.
The idea of ray tracing is old. In 1968, Arthur Appel published a method for creating shaded images using rays, and in 1979 Turner Whitted of Bell Labs demonstrated recursively tracing reflections and refractions. However, because the computation was so heavy, it was long used only in fields like film CG, where a single frame could take a long time to render.
The turning point came in 2018:
- In September 2018, Nvidia released the GeForce RTX 20 series, which included dedicated ray-processing units (RT cores).
- In November of the same year, Battlefield V became the first game to use this feature through Microsoft's DirectX Raytracing (DXR).
- In 2020, AMD unveiled the Radeon RX 6000 series with hardware ray tracing support, and the PlayStation 5 and Xbox Series X|S also shipped with ray tracing hardware.
Most games today use a hybrid approach, applying ray tracing only to certain effects such as shadows and reflections and rasterizing the rest. But games that require a ray tracing-capable graphics card have started to appear, such as Doom: The Dark Ages, released in 2025.
AI Upscaling and the Road Ahead
Because ray tracing is still demanding, upscaling technology—rendering at a lower resolution and then reconstructing a higher one—has developed alongside it. Nvidia's DLSS was introduced with the RTX 20 series in 2018, and from DLSS 2 in 2020 it no longer needed to be trained separately for each game, which helped it spread widely. Frame generation, which uses AI to create and insert in-between frames, was added later. The Nintendo Switch 2, released in 2025, is the first console to use DLSS.
This trend shows that the direction of game graphics is gradually shifting. Where the goal used to be more polygons and higher resolutions, the key challenges now are how realistically light can be calculated, and how efficiently that calculation can be reduced.
Meanwhile, as technology advances, more games are deliberately choosing pixel art or simple 3D styles. In every era, good graphics have depended less on high or low specs than on how well they convey a game's mood and gameplay.