Jeff,

M.Predko's video generation technique is "character mapped" not "bit-mapped" (which is typical of many micro-generated video approaches). The displaying pixel field of a video line is the result of reading an character value, using this as an index into ROM'd character patterns, and the shifting of these patterns to the Display via the Video DAC. The algorithm keeps track of what row of 8 for a character is being generated and indexes into the correct pattern page for that line.


From previous post's you know the platform video is bit-mapped, TV monochrome. When a character is "written" to the display, what occurs is that a "display routine" does the character pattern lookup and then just stores these patterns into a "Frame Buffer". So the pattern/data entry into the Frame Buffer occurs outside of Video Generation proper.

The video generation process itself is just the reading of the Frame Buffer byte by byte and shifting these patterns out to the Video DAC. As you can imagine reading/shifting is faster than reading/lookup/shifting.

The Predko experiment is a wonderful tutorial and introduction, he left it as a problem for the reader to implement a usable system which would not incorporate "hard coded" message content, and greater resolution.

Regarding, "extra patterns":
Character Patterns greater than 127(dec) are the IBM PC ISO character patterns (the borders/lines/non-english characters) familiar to PC owners. However, Mr. Predko's program to get them from the VGA ROM can scramble these patterns and render them useless. By sticking which 256 possible characters, a technique known as "tiling" can be used to generate "bit-mapped" looking displays with minimum data.

Regarding other Formats:

TV Color and Mono/Color VGA are definitely targets for future works. This is sort of a "walk before you run" effort.

regards,
p