Our industry has seen three phases with respect to the quality of the graphics we use, and a fourth will be coming soon.
ROM and Casette Graphics
During the first phase, in 1979-82, most games were sold on ROM or on casette; in either case, you just couldn’t put a lot of stuff in there (4K to 16K was typical) so the graphics were necessarily quite limited. Of course, the display resolutions were also very low, so you could get a screen or two out of 8K.
The second phase started around 1982 when a significant fraction of the installed base had floppy disk drives. This permitted game designers to use floppies as their primary delivery medium, and the amount of graphics that we could deliver jumped up to the neighborhood of 80K - 140K, a ten-fold increase. This was especially fortuitous, because the screen resolutions did not increase concomitantly; thus, we were able to pack several dozen screens’ worth of graphics onto a floppy, using primitive compaction techniques. This phase lasted until about 1987.
Hard Disk Graphics
By 1987 two new developments had changed things dramatically. First, the 8-bit machines had been supplanted by the 16-bitters. This meant that we had more processing power. It also meant that we had more display space. The simple black-and-white display on the original Macintosh, for example, required 21K worth of image.
Another big change was the triumph of the hard disk. By 1987 this was standard equipment on home computers. Initially this created a problem for game developers: we couldn’t require the user to boot from our floppy, which in turn shot down most of our best copy-protection schemes. The advent of the hard disk spelled the doom of on-disk copy-protected software.
The positive value of the hard disk did not become evident until perhaps 1989, when the size of computer games, as measured by the number of floppies in the package, began to rise sharply. Prior to 1989, most games were one-disk affairs. But then we started to see more multi-disk games. The user was expected to install the files from the floppies onto his hard disk, and then play the game from the hard disk.
The constraining factor was the price of floppy disk space. It had been about $2/megabyte in 1988; by 1990 it had fallen to $1/megabyte. Because most other expenses are fixed, a game publisher can afford to spend about $2 on floppies for a $40 game, $4 for a $50 game, and $6 for a $60 game. The combination of falling disk prices and rising game prices made possible an explosion in the sizes of games.
Almost all of this growth has been in graphics and sound. Consider the following table of my own products:
Year Product Object Code Size Total Product Size
1985 BOP 100K 181K
1986 PVR 81K 304K
1987 SBT 80K 273K
1989 G&B 153K 522K
1990 BotP 122K 1371K
1991 PSB (est) 150K 4000K
In six years the object code has fallen from 55% of total product size to 4%.
There is another factor to consider: hard disk accesses are fast enough to permit substantial use of graphics that are pulled in from the hard disk only as needed. We keep an animation file on disk, with the first frame of the animation in RAM; when the animation is called for, we pop up the first frame and run to the hard disk. By the time the user’s eye has registered on the first frame, the animation is ready to begin.
The Next Phase: CD-ROM Graphics
CD-ROM and the various graphics compression standards (JPEG et al) will change things again, making possible another big leap in graphics. But that remains several years in the future...