Pre-rendering
Pre-rendering is the technique of creating images before they are ever played back, often on an entirely different computer system than the one that will display them. At its core, the idea is simple: because rendering images demands far more computing power than simply playing them back, you can offload that heavy work to powerful machines, or let it run over extended periods, and deliver polished results to hardware that could never have produced them on its own.
This distinction between pre-rendered and real-time images has quietly shaped decades of film, television, and gaming. When you watch a cutscene in a video game or a scene from an early computer-animated film, the question of whether those images were calculated in advance or on the fly determines almost everything about how they look, how they behave, and what trade-offs the creators had to accept.
The story of pre-rendering runs from arcade halls in September 1983 through blockbuster PC titles, console franchises, and Hollywood animation, right up to a technique called Machinima that emerged in the mid-2000s as real-time graphics finally caught up.
Real-time rendering operates under a strict contract: images must appear fast enough to create the illusion of motion, which means a computer must calculate and display each frame within a tight time window. Pre-rendering dissolves that constraint entirely. A studio or developer can use multiple computers working over days or weeks, producing results whose visual complexity would be impossible to achieve live.
The contrast is illustrated sharply by two rail-shooter arcade games. Maximum Force used pre-rendered 3D levels alongside 2D sprites for enemies. Virtua Cop rendered everything in 3D polygons in real-time. Maximum Force looked more realistic by the standards of the era, but Virtua Cop could portray enemies at varying distances, register body-specific hits, and respond to multiple impacts. The pre-rendered game won on visual fidelity; the real-time game won on depth of interaction.
This trade-off is not merely aesthetic. A game using pre-rendered backgrounds must commit to fixed camera angles. Pre-rendered video cannot reflect changes a character undergoes during gameplay, such as wounds or customized clothing, without storing an alternate version of that video. Storing high-quality pre-rendered assets at multiple variants is generally not feasible given the storage space required.
Two arcade laserdisc games introduced in late 1983 mark the earliest documented use of pre-rendered 3D computer graphics in video sequences. Interstellar, introduced by Funai at the AM Show in September of that year, and Star Rider, introduced by Williams Electronics at the AMOA show in October, both used this approach before it had any widespread name.
The X68000 enhanced remake of Ys I: Ancient Ys Vanished, released in 1991, used 3D pre-rendered graphics for the boss sprites. The result was described as creating "a bizarre contrast" with the game's otherwise 2D graphics. One of the first games to use pre-rendered graphics and full motion video extensively was The 7th Guest. Released in 1993, it was among the first PC games sold exclusively on CD-ROM, and it proved hugely popular, though critics gave it mixed reviews. Its pre-rendered video sequences ran at 640x320 at 15 frames per second, a feat that had been considered impossible on personal computers at the time.
Donkey Kong Country, released on the SNES in 1994, became one of the first significant console games built around pre-rendered graphics. The technique also took hold on the original PlayStation through two major franchises: Resident Evil and Final Fantasy. Both used pre-rendered backgrounds and movies to deliver visual quality far beyond what the console could produce in real-time, while still embedding real-time characters and items for interactivity.
Shortly after The 7th Guest, Myst arrived in 1993 and pushed the popularity of pre-rendered graphics and CD-ROM distribution even further. Most of the rendered work created for Myst was later reused as the foundation for the remake realMyst: Interactive 3D Edition, which replaced the static pre-rendered images with free-roaming real-time 3D graphics.
The high point of entirely pre-rendered visuals in games is often cited as Myst IV: Revelation, released in 2004. By that point, the series had stretched the technique as far as it could go within the constraints of fixed images and scripted sequences.
One notable exception to the limitations of pre-rendered video appeared in Final Fantasy VIII, where real-time assets were composited directly with pre-rendered video. This allowed the game to show dynamic backgrounds and shifting camera angles, working around the usual problem that pre-rendered lighting cannot change state convincingly once it has been baked in.
Games built on pre-rendered backgrounds did not have to sacrifice all responsiveness. By rendering only the static environment in advance and keeping characters, items, and effects as real-time elements, developers could direct spare processing power toward the interactive parts of the scene. This sometimes produced a level of detail in those elements greater than what was typical for the host platform, even when the gap between the pre-rendered background and the real-time characters remained visually obvious.
Games such as Warcraft III: Reign of Chaos found a middle path. That game used pre-rendered cutscenes for the beginning and end of a campaign, and switched to the in-game engine for level briefings and character dialogue during a mission. The choice matched the visual budget to the narrative weight of each moment.
Some games also used 16-bit pre-rendered skyboxes. Half-Life (in its GoldSrc version), Re-Volt, and Quake II all employed this technique for their backgrounds while rendering the rest of the game in real-time.
While making Doom 3, id Software used pre-rendered high-detail models as the source material for generating normal maps, specular maps, and diffuse lighting maps. Those maps were then applied to lower-polygon real-time models, allowing the game engine to simulate the surface detail of the original complex model without the processing cost of rendering it live. The engine used, id Tech 4, supported bump mapping and dynamic per-pixel lighting, capabilities that had previously appeared only in pre-rendered video.
Pre-rendered lighting more broadly is a technique that has declined in popularity. During a game's production, processor-intensive ray tracing algorithms can be run to generate light textures, which are applied on top of ordinary hand-drawn textures in the final game. This offloads the most expensive lighting calculations to development time rather than runtime.
The generation of texture sets from pre-rendered models became an increasingly common workflow for 3D games, pairing complex pre-rendered detail with real-time algorithms to simulate levels of surface fidelity that live rendering could not yet produce on its own.
Entirely pre-rendered computer graphics reached mainstream audiences through CG films including Toy Story, Shrek, and Final Fantasy: The Spirits Within. Each frame of these productions was calculated in advance, with no constraint on rendering time per image.
By the mid-2000s, advances in consumer PC and video game console hardware allowed real-time game engines to produce the photorealism that had previously required pre-rendering. The growth of Machinima reflected this shift: filmmakers began using game engines to render sequences in real-time, rather than relying on offline rendering pipelines. Developers increasingly turned to their own engines to render cinematics, rather than commissioning separate pre-rendered sequences. The id Tech 4 engine in Doom 3 is one specific example of a real-time engine whose lighting capabilities had previously been the exclusive territory of pre-rendered work, pointing toward a future where the boundary between the two techniques continues to narrow.
Up Next
Common questions
What is pre-rendering in computer graphics?
Pre-rendering, also called offline rendering, is the technique of calculating images before playback, often using more powerful computers or longer processing times than would be available in real-time. It contrasts with real-time rendering, where images must be generated within a tight time window to create the illusion of motion. The pre-rendered images or video are then played back on the target system, which may be less powerful than the system that rendered them.
What were the first video games to use pre-rendered 3D graphics?
Two arcade laserdisc games introduced in late 1983 are the earliest documented examples: Interstellar, introduced by Funai at the AM Show in September, and Star Rider, introduced by Williams Electronics at the AMOA show in October. The X68000 remake of Ys I: Ancient Ys Vanished, released in 1991, also used 3D pre-rendered graphics for its boss sprites.
What were the main disadvantages of pre-rendered backgrounds in video games?
Pre-rendered backgrounds forced games to use fixed camera angles and could not reflect changes characters underwent during gameplay, such as wounds or customized clothing, without storing multiple alternate versions of the video. Storing high-quality pre-rendered assets at scale was generally not feasible due to storage requirements. Pre-rendered lighting also could not change state convincingly during play.
Which video games are cited as the most advanced examples of pre-rendered graphics?
Myst IV: Revelation, released in 2004, is often cited as the most graphically advanced use of entirely pre-rendered graphics in games. The 7th Guest, released in 1993, was notable for running pre-rendered video at 640x320 at 15 frames per second, a feat previously considered impossible on personal computers. Donkey Kong Country, released on the SNES in 1994, was one of the first significant console games built around pre-rendered graphics.
What CG films are entirely pre-rendered?
Toy Story, Shrek, and Final Fantasy: The Spirits Within are all entirely pre-rendered CG films, meaning every frame was calculated in advance rather than generated in real-time.
How did id Software use pre-rendering for Doom 3?
While making Doom 3, id Software used pre-rendered high-detail models to generate normal maps, specular maps, and diffuse lighting maps. These maps were applied to lower-polygon real-time models in the game, allowing the id Tech 4 engine to simulate the surface detail of the original complex models without the processing cost of rendering them live. The engine also supported bump mapping and dynamic per-pixel lighting, capabilities previously found only in pre-rendered video.