# HG changeset patch
# User Paul Boddie <paul@boddie.org.uk>
# Date 1605054796 -3600
# Node ID 6e4fcde513009a29e55c688a58c1c6c124e758a2
# Parent  0ddf105800424bdc6cf7df23c0f4a5aa5ad65dcd
Added character generator support enhancement.

diff -r 0ddf10580042 -r 6e4fcde51300 ULA.txt
--- a/ULA.txt	Sun Aug 23 22:26:14 2020 +0200
+++ b/ULA.txt	Wed Nov 11 01:33:16 2020 +0100
@@ -1382,6 +1382,41 @@
 character. When no attribute information is provided, the colours are defined
 globally.
 
+Enhancement: Character Generator Support and Vertical Scaling
+-------------------------------------------------------------
+
+When generating a picture, the ULA traverses screen memory, obtaining 40 or 80
+bytes of pixel data for each scanline. It then proceeds to the next row of
+pixel data for each successive scanline, with the exception of the text modes
+where scanlines may be blank (for which the row address does not advance).
+This arrangement provides a conventional bitmapped graphics display.
+
+However, the ULA could instead facilitate the use of character generators. The
+principles involved can be demonstrated by the Jafa Mode 7 Mark 2 Display Unit
+expansion for the Electron which feeds the pixel data from a MODE 4 screen to
+a SAA5050 character generator to create a MODE 7 display. The solution adopted
+involves the replication of 40 bytes of character data across as many pixel
+rows as is necessary for the character generator to receive the appropriate
+character data for all scanlines in any given character row. If only a single
+40-byte row of character data were to be present for the first scanline of a
+character row, the character generator would only produce the first scanline
+(or the uppermost pixels of the characters) correctly, with the rest of the
+character shapes being ill-defined.
+
+Here, the ULA could facilitate the use of memory-efficient character mode
+representations (such as MODE 7) by holding the row address for a number of
+scanlines, thus providing the same row of screen data for those scanlines,
+then advancing to the next row. Visualised in terms of pixel data, it would be
+like providing a display with a very low vertical resolution. Indeed, being
+able to reduce the vertical resolution of a display mode by a factor of eight
+or ten would be equivalent to the above character generation technique in
+terms of the ULA's screen reading activities.
+
+By combining this vertical scaling or scanline replication with a circuit
+switchable between bitmapped graphics output and character graphics output,
+MODE 7 support could be made available, potentially as a hardware option
+separate from the ULA.
+
 Enhancement: Compressed Character Data
 --------------------------------------