Sprite (computer graphics)

History

The use of sprites originated with arcade games. The first video game to represent player characters as human player images was Taito's Basketball, which was licensed in February 1974 to Midway, releasing it as TV Basketball in North America.

Signetics devised the first video/graphics processors capable of generating sprite graphics for home systems. The Signetics 2636 video processors were first used in the 1976 Radofin 1292 Advanced Programmable Video System.

The Atari VCS, released in 1977, features a hardware sprite implementation where five graphical objects can be moved independently of the game playfield. The term sprite was not in use at the time. The VCS's sprites are called movable objects in the programming manual, further identified as two players, two missiles, and one ball. These each consist of a single row of pixels that are displayed on a scan line. To produce a two-dimensional shape, the sprite's single-row bitmap is altered by software from one scanline to the next.

The Atari 400 and 800 home computers of 1979 feature similar, but more elaborate, circuitry capable of moving eight single-color objects per scan line: four 8-bit wide players and four 2-bit wide missiles. Each is the full height of the display—a long, thin strip. DMA from a table in memory automatically sets the graphics pattern registers for each scan line. Hardware registers control the horizontal position of each player and missile. Vertical motion is achieved by moving the bitmap data within a player or missile's strip. The feature was called "player/missile graphics" by Atari.

The Namco Galaxian arcade system board, for the 1979 arcade game Galaxian, featured animated, multi-colored sprites. It pioneered a sprite system that animated pre-loaded sprites over a scrolling background, which became the basis for Nintendo's Radar Scope and Donkey Kong arcade hardware and home consoles such as the Nintendo Entertainment System. According to Steve Golson from General Computer Corporation, the term "stamp" was used instead of "sprite" at the time.

The Elektor TV Games Computer was an early microcomputer capable of generating sprite graphics, which Signetics referred to as "objects".

The term sprite was first used in the graphic sense by one of the definers of the Texas Instruments 9918(A) video display processor (VDP). The term was derived from the fact that sprites, rather than being part of the bitmap data in the framebuffer, instead "floated" around on top without affecting the data in the framebuffer below, much like a ghost or "sprite". By this time, sprites had advanced to the point where complete two-dimensional shapes could be moved around the screen horizontally and vertically with minimal software overhead.

The CPU would instruct the external chips to fetch source images and integrate them into the main screen using direct memory access channels. Calling up external hardware, instead of using the processor alone, greatly improved graphics performance. Because the processor was not occupied by the simple task of transferring data from one place to another, software could run faster; and because the hardware provided certain innate abilities, programs were also smaller.

Hardware sprites

Many early arcade games and home video game systems had chips with sprite support, in which small images were integrated into the screen, often with priority control with respect to the background graphics, at the time the video signal was being generated by the graphics chip.

These contrasted with software and blitter methods of 2D animation which modify a framebuffer held in RAM, which required more memory cycles to load and store the pixels, sometimes with an additional mask, and refresh backgrounds behind moving objects. These methods frequently required double buffering to avoid flickering and tearing, but placed fewer restrictions on the size and number of moving objects.

The sprite engine is a hardware implementation of scanline rendering. For each scanline the appropriate scanlines of the sprites are first copied (the number of pixels is limited by the memory bandwidth and the length of the horizontal retrace) into very fast, small, multiple (limiting the number of sprites on a line), and costly caches (the size of which limit the horizontal width) and as the pixels are sent to the screen, these caches are combined with each other and the background. It may be larger than the screen and is usually tiled, where the tile map is cached, but the tile set is not. For every pixel, every sprite unit signals its presence onto its line on a bus, so every other unit can notice a collision with it. Some sprite engines can automatically reload their "sprite units" from a display list. The sprite engine has synergy with the palette. To save registers, the height of the sprite, the location of the texture, and the zoom factors are often limited. On systems where the word size is the same as the texel there is no penalty for doing unaligned reads needed for rotation. This leads to the limitations of the known implementations, in the following table.

Table of Sprite Hardware Features

These are base hardware specs and don't include additional programming techniques, such as using raster interrupts to repurpose sprites mid-frame.

Use in 3D rendering

2D images with alpha channels constrained to face the camera may be used in 3D graphics. They are common for rendering vegetation, to approximate distant objects, or for particle effects. These are sometimes called "billboards" or "Z-sprites." The technique was most heavily used in Sega game machines in the late 1990s, prior to the era of polygon rendering. If rendered on the fly to cache an approximate view of an underlying 3D model, such sprites are called impostors. Modern hardware may have a specific mode for rendering such point sprites without needing each corner to be defined, or these may be generated by vertex shaders.

Synonyms

Some hardware makers used terms other than sprite, and other terms exist to describe various forms of software-rendering of sprites: