Digital Typesetter

The present invention relates to a "digital" typesetter for imaging graphics quality characters of a specified font which is stored in the machine in digital form.

Digital typesetters image typographical characters which have been coded in digital form and stored on some kind of digital storage medium such as a magnetic tape, drum or rigid or floppy disk. Such digital typesetters are normally provided with a cathode ray tube (CRT) or laser beam imaging system for writing characters onto photographic film or paper. In the case of digital CRT typesetters, successive characters are each completely imaged, one at a time, by the CRT beam which is position-controlled in two dimensions by the CRT X-Y deflection voltages. Normally, each character is built up from a series of vertical strokes which commence from the left side bearing of the character and proceed successively to the right side bearing. Once a character has been completely stroked, the CRT beam is moved to the left side bearing position of the next successive character, and so on.

When a laser beam rather than a CRT imaging system is used to image typographical characters on film or paper, a modulated one dimensional scan has generally been used. In this case, the character-writing beam is swept across the entire width of the output film or paper in a single line and this print medium is moved in a direction transverse to the direction of the scan line. While such a laser-based imaging system is substantially less expensive than a cathode ray tube system, such a system requires computation of the beam switch points to turn the scanning beam on and off with sufficient speed to keep up with the scanner on a real time basis. In order to overcome this problem, the characters of each font have been digitized as a separate digital master for each type-size to be set. In this way, the encoded character information may be used directly without scaling, for example, which requires extensive computation. The disadvantage of this arrangement is that an entire font of characters must be stored in the machine for each of the desired type sizes. This results in excessive use of storage space within the machine or requires that the machine be repeatedly "dressed" with a different digitally-encoded font each time a different type size, or a different font are to be used.

It is a general object of the present invention to provide a digital typesetter which combines the advantages of the CRT and laser typesetters described above.

It is a more specific object of the present invention to provide a digital typesetter having an imaging system of the type producing a one-dimensional scan line across the entire width of the output print medium.

It is a further specific object of the present invention to provide a digital typesetter which is capable of imaging typographical characters of virtually any size from a single digitized master font.

These objects, as well as further objects which will become apparent from the discussion that follows, are achieved, according to the present invention, by providing an electronic data processing system which receives first digital data defining the identity, form, size and placement of characters to be typeset; receives second digital data defining the contour of each character to be typeset with respect to a normalized encoding set of first and second coordinates; and produces third digital data defining the character boundaries intersecting a raster (scan) line. This third digital data, for at least a portion of the raster line extending the width of several characters, is temporarily stored in a raster line storage buffer. A character imaging device, connected to the line storage buffer, is provided to image the successive raster lines on a print medium. Drive means are also provided for moving the print medium in a direction transverse to the direction of the raster (scan) line.

In a preferred embodiment of the invention, the raster line extends substantially the width of the print medium which may be at least the size of a conventional typewritten page. The print medium may comprise paper which can be used in a conventional way for paste-up or to make office copies, or the print medium may comprise a printing plate so that printed copies may be made directly from the typesetter output.

Preferably, the raster line extends horizontally on the print medium; i.e., parallel to the lines of type. An image of this raster line may be created by a scanning device, such as a laser recorder, having means (e.g., a laser) for generating a scanning beam and a means (e.g., a mirror) for moving the scanning beam across the print medium in a scan line. Suitably, the scanning beam is switched on and off in response to the line storage buffer, as the beam is scanned across the print medium, thus forming the raster line image.

The line storage buffer preferably includes two line buffers, each of which may be formed of a plurality of memory elements each storing a single binary digit corresponding to a respective, unique raster point along the raster line. The binary digits in the first line buffer may thus represent positions on the scan line at which the scanning beam is to be switched on and the binary digits in the second line buffer may represent the positions on the scan line in which the scanning beam is to be switched off. The line storage buffer may also include a third line buffer, which may be identical to the other two. The binary digits in this third line buffer may thus represent the positions of the scan line at which the beam on/off switching is to be reversed, thus effecting a color reversal for either an entire page or a rectilinear section thereof.

The first digital data defining the identity, form, size and placement of the characters to be typeset may originate from a computer system (with the typesetter on line) or from a storage medium such as a floppy disk. The second digital data is preferably stored on a magnetic record, such as a floppy disk.

The manner in which the second digital data defines the contour of each character of a font will be described in detail hereinbelow. Essentially, this second digital data comprises digital numbers defining and X and Y coordinates of the start points of character outlines and digital numbers defining the length and direction of a plurality of straight line vectors extending successively along the character outlines from the start points. The length and direction of each vector is represented by the first coordinate distance .DELTA.X and the second coordinate distance .DELTA.Y from one end of the vector to the other.

In accordance with a preferred feature of the present invention, the digital numbers defining the vectors are arranged such that the vectors of an entire string are successively defined before defining the vectors of another string. In addition, the second digital data includes further digital numbers, associated with the digital numbers defining the coordinates of each start point, which constitute the starting address of the digital numbers defining the vectors of the associated string. In this way, a single vector string may be addressed from a plurality of start points within a font. Finally, it is preferable if the digital numbers defining the coordinates of a start point further specify the quadrant (either right or left) of at least the first vector of the associated, addressed vector string.

In general, character designers (persons who design character fonts) tend to create a few basic character shapes which are repeated throughout the font, either directly or in mirror image. Consistency dictates that a few shapes be repeated throughout the font while symmetry dictates that mirror images be used. The vector strings utilized in the digital definition of characters permit the quantity of data required to define an entire font to be substantially reduced because a single vector string may be addressed from the start points of various characters and may be directed to extend either toward the left or toward the right, thus forming mirror images.

The electronic data processing system used in the typesetter preferably includes a random access memory for storing fourth digital data, a data management subsystem for receiving and storing the first and second digital data and producing and storing the fourth digital data in the random access memory, and an outline converter subsystem for receiving the fourth digital data from the random access memory and computing from this fourth digital data the third digital data defining the character boundaries intersecting the raster line. The manner in which the fourth digital data is processed and stored and thereafter converted into the third digital data will be described in detail hereinbelow.

For a better understanding of the invention, together with other and further objects, reference is made to the following description, taken in conjunction with the accompanying drawings.

FIG. 1 is a block diagram of the overall typesetting system according to the present invention.

FIG. 2 is a schematic diagram of a one-dimensional scanner (laser recorder) which may be used in the system of FIG. 1.

FIG. 3 is a block diagram of the Output Data Processing System in the system of FIG. 1.

FIG. 4 is a block diagram of the Data Management Subsystem in the Output Data Processing System of FIG. 3.

FIG. 5 is a block diagram of the Outline Converter Subsystem in the Output Data Processing System of FIG. 3.

FIG. 6 is a diagram showing the structure of the Font File contained in the font storage unit in the system of FIG. 1.

FIG. 7 is a diagram showing the structure of the Font Control File in the Font File of FIG. 6.

FIG. 8 is a diagram showing the structure of the Character Width File in the Font File of FIG. 6.

FIG. 9 is a diagram showing how the character data contained in the Font File is scaled.

FIG. 10 is a diagram showing the structure of the outline data words contained in the Outline Data File in the Font File of FIG. 6.

FIG. 11 is a diagram showing how the Outline Data File is arranged on a floppy disk.

FIG. 12 is a diagram showing the structure of the header and outline data contained in the Outline Data File of FIG. 11.

FIG. 13 is a diagram showing the input code structure for those terminal elements which are to be sent from the Input System for a font data transfer.

FIG. 14 contains flow diagrams showing the basic operation of the Data Management Subsystem and the Outline Converter Subsystem.

FIG. 15 is a flow diagram for the building of data in the Data RAM in the Data Management Subsystem.

FIG. 16 is a diagram of the layout of data in the Data RAM.

FIG. 17 is a diagram showing the structure of the outline data in the Data RAM.

FIG. 18 is a diagram showing the parameters used in the basic calculation carried out by the Output Data Processing System.

FIG. 19A and 19B are diagrams showing the data and control inputs for the hardwired processor in the Outline Converter Subsystem.

FIGS. 20A and 20B are functional block diagrams of the hardwired processor including the output data (raster line) storage buffers.

FIG. 21 is a block diagram showing the principal sections of the hardwired processor.

FIG. 22 is a block diagram of the data and control buffers and storage section of the hardwired processor of FIG. 21.

FIG. 23 is a block diagram of the output data buffer address computation section of the hardwired processor of FIG. 21.

FIG. 24 is a block diagram of the output data buffers and video logic section of the hardwired processor of FIG. 21.

The preferred embodiment of the present invention will now be described with reference to FIGS. 1-24 of the drawings. Identical elements shown in the various figures are labeled with the same reference numerals.

The overall system according to the present invention is shown, in block form, in FIG. 1. This general system is divided into an Input System which supplies instructions, character information and font data from separate sources 30 and 32, and an Output Data Processing System 34 which drives a Character Imaging System 36.

The input device 30 may be a paper tape or magnetic tape reader, a separate computer, an input terminal with a keyboard and CRT screen, or a data transmission channel such as a telephone line. This input device 30 supplies to the processing system 34 digital data defining the identity, form, size and placement of characters to be typeset. As used herein, the term "identity" of characters is intended to mean the name of each particular character chosen, such as upper case "A", lower case "a", upper case "B", numeral "5"; semi-colon ";" and the like. This identity is given by an ASCII or TTS code. As used herein, the term "form" is intended to designate the shape of each character; i.e., the particular font and the amount and direction of slant. The term "size", as used herein, is intended to designate the size in both the X direction ("set width") and in the Y direction ("point size") of each character. Finally, the term "placement", as used herein, is intended to mean the coordinate (X,Y) position of the character on the page to be typeset. In this particular embodiment, the input device 30 designates the X position and Y position of the upper left corner of the "em" square of at least the first character on the page. Subsequent characters are positioned in relation to the immediately preceding character if no placement information is given.

In addition to the identity, form, size and placement of characters, the input device 30 may also supply page variant information; that is, "global commands" which apply to all or a group of characters on a page. Examples of such commands are "wrong reading", which effects a left-right mirror image on the page by flipping the X positions of all characters, and "reverse video" which effects a color reversal for an entire page. For example, with reverse video a page may be imaged as white on black, rather than black on white.

Commands from the input device 30 may also effect a color reversal for a section of a page, such that only a rectilinear portion of the page is white on black rather than black on white.

The font storage unit 32 is essentially a floppy disk reader which may be a part of the input device 30. This font storage unit supplies to the Output Data Processing System 34 digital data defining the font of characters previously selected by the input device 30. This "second" digital data (as distinguished from the "first" digital data supplied by the input device 30) defines the contour of each character of a font with respect to a normalized encoding set of first and second coordinates. In particular, this second digital data defines the profiles or black-white boundaries of each character. If a "profile" is considered to be simply one boundary of a character, it will be seen that any "dark" portion of a character (if the character is dark on a light background) must lie between two profiles (outer boundaries or edges) of the character. By defining all the profiles of the character, with respect to a coordinate set, the "contour", outline or shape of the character is completely defined.

One aspect of this second digital data which defines the contour of each character of a font is that the character contours are defined in terms of a normalized set of coordinates, such as the X-Y coordinates of a Cartesian coordinate set. The term "normalized", as used herein, is intended to mean that the definition of a character in terms of the coordinate set is only related to any given absolute size or to the final size of the character when it is imaged. Thus, the digital values defining a character in this normalized set of coordinates are the values from which the character is scaled, up or down, to the final output resolution. Unless the scale factor just happens to equal 1 (a unique situation), the character will be defined with a different resolution than the final output.

As an example, the output data processing system 34 which will be described in detail hereinbelow is capable of scaling characters with point sizes in the range of 3-130, an expansion factor of 43 to 1. Notwithstanding this range of point sizes, the contour of each character is defined only once with respect to the normalized encoding set of coordinates.

The Output Data Processing System 34 receives the first digital data defining the identity, form, size and placement of characters to be typeset and the second digital data defining the contour of each character of the chosen font and produces third digital data defining the character boundaries intersecting a raster line. This third digital data is stored in one or more raster line buffers, also located within the Output Data Processing system, in readiness for the Character Imaging System 36. The raster line storage buffer(s) are preferably formed of a plurality of binary memory elements, each storing a single binary digit corresponding to a respective, unique raster point along the raster line. The line buffer(s) store sufficient raster (third digital data) for a portion of the raster line extending the width of at least several characters. In fact, the line buffer(s) preferably store sufficient data to define an entire raster line extending the complete width of the output print medium which may, for example, be at least the size of a conventional type-written page.

The information stored in the raster line storage buffer(s) is translated into a raster line image by a Character Imaging System 36 connected to the Output Data Processing System 34. This Character Imaging System creates an image on a print medium for the particular raster line defined by the information stored in the raster line storage buffer(s). A drive mechanism is also provided in the Character Imaging System for moving the print medium in a direction transverse to the direction of the imaged raster line.

The Character Imaging System may comprise some means for creating an image for an entire raster line at once. However, the Character Imaging System preferably includes a device, such as a CRT or laser source, for generating a scanning beam and some means, such as beam deflection circuits or a movable mirror, for moving the scanning beam across the print medium in a scan line. One such one-dimensional scanner is illustrated in FIG. 2.

FIG. 2 shows a laser source 38 which produces an optical beam 40. The beam passed through a light modulator 42, such as an acousto-optic modulator, which switches the beam on and off by a video control signal 44. This signal is generated in the Output Data Processing System 34 as a time function of the third digital data stored in the raster line storage buffer(s). The intermittent beam 40 which emerges from the shutter is passed to a movable mirror 46 which repetitively tilts back and forth about an axis 48 parallel to the plane of the mirror and perpendicular to the plane of the drawing. The mirror 46 is driven by a galvanometer motor which moves the mirror at an essentially constant angular speed in one direction, rapidly returns the mirror to its initial position, and repeats this scanning operation upon receipt of the next trigger pulse on line 50.

The beam is reflected from the mirror 46 and passes through a field-flattening optical system 52 which focuses it on a suitable recording or print medium 54. The purpose of the field-flattening optical system, and a specific system which may be used with the laser recorder shown in FIG. 2, is described in the U.S. Pat. No. 3,881,801 issued May 6, 1975 to Edwin W. Bechtold and entitled "Optical Scanning System". The print medium 54 used with the laser recorder of FIG. 2 may be a photographic or electrophotographic film or paper which is sensitive to the particular light frequency of the laser beam. This print medium is moved in a direction transverse to the direction of the beam scan line, preferably during the retrace movement of the mirror 46. The print medium is moved by an advance roller 56 and drive motor 58 in response to a trigger pulse on line 60. This trigger pulse is similarly generated by the output data processing system 34 as a time-shifted version of the trigger pulse on line 50.

Generally, then, it may be seen that the function of the Output Processing System is to convert the first and second digital data received from the input device 30 and the font storage unit 32 into third digital data which may be conveniently used to control the Character Imaging System 36. The format of the first and second digital data received by the Output Data Processing System, as well as the structure of the Output Data Processing System will now be described in detail with reference to FIGS. 3-23 of the drawings. This description has been organized into separate sections for convenience of the reader as set forth in the Index below.

1.1 General: The Output Data Processing System is responsible for computing the horizontal coordinates, on the page to be typeset, at which the laser scanning beam must be turned on or off for each and every raster line on the page. Its computation is based upon the particular raster line which is required (depth down the page); on the particular characters (i.e., identity) which are to be set at that point on the page; and on the form and size as well as the shape of these characters as defined by the Input System.

Since the conversion from the "second" digital data, defining the contour of the characters to be set, into raster data is complex, and since the raster output form requires repeated, multiple character data access, the time required for computation of each raster line becomes a significant factor in the system architecture. In an effort to minimize the computation time, the Output Data Processing System has been divided into two major subsystems:

(1) The Data Management Subsystem (DMS) and,

(2) The Outline Converter Subsystem (OCS). A Z80A microprocessor is used in the former and an 8.times.300 (or "SMS 300") microcontroller with a hardwired processor is used in the latter.

FIG. 3 shows the Output Data Processing System in block form. This system receives the first digital data defining the identity, form, size and placement of the characters to be typeset as well as the second digital data defining the contour of each character from a common Input System. The Input System operates with a programmed 8080 microcomputer 62 supported by a RAM 64 of suitable size. The microcomputer and memory are arranged on a 8080 bus 66 as are two floppy disk read/write units comprising floppy disk controllers 68 and the disks 70 themselves. One disk 70 contains the text information or "first" digital data, while the other disk contains the font information or "second" digital data. The bus terminates in an IOP80 interface 72 which communicates with an interface 74 in the Output Data Processing System. This latter interface is arranged on a Z80A bus 76 as are the Z80A microprocessor 78 and four memory units 80, 82, 84 and 86 of the Data Management Subsystem.

The memory unit 80 serves to store the program for the Z80A microprocessor 78 and is a workspace for the microprocessor computations. The memory 82, called a "font RAM", stores the second digital data defining the characters of the chosen font. This data is processed and supplied in a convenient form, which will be described in detail below, to two memories 84 and 86 called "Data RAM's".

The Data RAM's 84 and 86 are "shared" by the Data Management Subsystem and the Outline Converter Subsystem. Basically, the Z80A microprocessor supplies data to these RAM's and the 8.times.300 microcontroller 88 receives and analyzes this data, under control of a program stored in another memory 90, and supplies pertinent data to a hardwired processor 92. This hardwired processor converts the data into the so-called "third" digital data which is stored in three raster line buffers. The information contained in these buffers is then converted into a video control signal by an interface 94 and supplied to the laser recorder in synchronism with the movement of the scanning beam.

1.2 Data Management Subystem: The circuit blocks and their interconnections employed in the Data Management Subsystem are shown in FIG. 4. In general, the responsibility of the Data Management Subsystem is to organize and supply data to the memory shared with the Outline Converter Subsystem so as to facilitate rapid processing by the Outline Converter Subsystem. More specifically, the Data Management Subsystem executes the following process steps:

(a) When ready, read the next required typographical "line segments" into a Data RAM memory file from the text floppy disk.

(b) Transfer the font data from the font floppy disk to one of the font RAM memories for a "font data file".

(c) Set-up an "outline file" in the Data RAM for the Outline Converter Subsystem. This file contains the X and Y start points of each outline of each character required, as well as "vector" data defining the contour of each character.

(d) Revise the "line segments file" by replacing the character number with the location of the outline file.

(e) Repeat the above steps until no memory storage area is available for the line segments file or the outlline file.

(f) Pass control to the Outline Converter Subsystem. Restart on the other Data RAM.

1.3 Outline Converter Subsystem: The integrated circuits and interconnections forming the Outline Converter Subsystem are shown in FIG. 5. Basically, the responsibility of the Outline Converter Subsystem is to convert the outline or contour data stored in the shared Data RAM into horizontal stroke data for the laser recorder. More specifically, the Outline Converter Subsystem executes the following process steps:

(a) Read the identifying data and size data for the first line segment.

(b) Read the distance from the margin to the left side bearing (LSB). Store in an X register.

(c) Read the outline start data for the next character, compute the distance from the LSB of the character to the outline. When necessary, fetch new outline vector data to update the X,Y start data.

(d) Output the sum of this value and the current X value (located in a "X" register") to the appropriate raster line buffer.

(e) Read the next outline(s); repeat step (c) until all outlines have been computed at the level on the page being set.

(f) Read distance to the LSB of the next character; add this to the X register.

(g) Repeat steps (c) through (f) until all characters in the line segment have been computed and output. Then repeat steps (a) through (f) for all other line segments on this level.

(h) When all line segments on this level have been computed, transfer control of the raster line buffer(s) to the laser driver system, and start storing data in an alternate (next) raster buffer for the raster line 1/10th point down the page.

2.1 General: The second digital data defining the characters of each desired font is stored on the font floppy disk. This data is of the "outline" type; that is, it defines the contour of each character with respect to a normalized encoding set of coordinates. In order to compress data, not all the character edge points on the resolution matrix are encoded. The general nature of the encoding scheme is described in the above-referenced, commonly-owned U.S. patent application Ser. No. 905,451, filed May 12, 1978, of Derek Kyte, et al. and entitled "Character Generating Method and Apparatus".

Because the floppy disk is the principal storage medium for fonts, the font data has been structured to correspond to the size of the sectors of the disk. Three separate subfiles comprise a font file as illustrated in FIG. 6:

1. Font Control File;

2. Character Width File; and

3. Outline Data File.

The Font Control File contains all the data which describes the font as an entity, such as a font description, serial number, number of characters, base jump and fixed space data, etc. This file is totally contained in one floppy disk section (125 words). Section 2.2 of this specification describes the Font Control File in detail.

The Character Width File contains the width of each character and other character-related data required by the Input System. This file is contained on one or more full sectors, depending upon the number of characters per font. Section 2.3 of this specification describes the Character Width File in detail.

The Outline Data File contains the starting coordinates of character outlines, the vectors that define the shape of the outlines, character control words and file size controls. This file is contained on one or more full sectors, depending upon the number of characters and the shape complexity of the characters. Section 2.4 of this specification describes this file in detail.

When the Data Management Subsystem (DMS) requires font data, the appropriate font data is read from the floppy disk by the Input System and transferred to Font RAM 82 by the DMS. Section 2.5 of this specification describes the structure and contents of data that has been transferred to the Font RAM.

2.2 Font Control File: The Font Control File contains all the data which describes the entire font or which is applicable to multiple characters in the font. This file is contained on one floppy disk sector, and therefore it is 125 words long. The structure of this file is shown in FIG. 7. The elements on this file, indicated in FIG. 7, are described below:

2.2.1 Number of Characters: The most significant byte (MSB) of word 0 defines in binary the number of characters in the font.

2.2.2 Number of Sectors: The least significant byte (LSB) of word 0 defines in binary the number of floppy disk sectors required by the Outline Data File. This can be utilized by the Input System to allocate disk space if additional fonts are added.

2.2.3 Number of Characters in 16K: The MSB of word 1 defines in binary the number of characters contained in the first 16,384 bytes of the Outline Data File.

2.2.4 Number of Characters in 32K: The LSB of word 1 defines in binary the number of characters contained in the first 32,768 bytes of the Outline Data File.

2.2.5 Number of Characters in 48K: The MSB of word 2 defines in binary the number of characters contained in the first 49,152 bytes of the Outline Data File.

2.2.6 Last RAM Number of Sectors: The LSB word 2 defines in binary the number of Outline Data File floppy disk sectors that must be transferred for the RAM that would contain the last character in the font. The will be a number between 3 and 68, and can be utilized by the Input System to control font data transfer to the ODS.

2.2.7 Y MIN: Word 3 defines in binary the minimum Yn value in data resolution units (DRU's) in the entire Outline Data File. This will be utilized by the Input System to set the Y position (YPOS) for each line segment.

2.2.8 Y MAX: Word 4 defines in binary the maximum Yn value in DRU's that will result from adding each outline dy's to its correlated Yn. This is the lowest point on all characters in the font, and will be utilized by the input system to set the lowest point (YLOW) for each line segment.

2.2.9 BJ for UC Accents: The MSB of word 5 defines the vertical displacement that will move a centered piece accent from its nominal lower case position to the proper location for use with upper case characters. This is an unsigned binary number in 1/432nd's of an em.

2.2.10 Italic Accent Shift #0: The LSB of word 5 defines the horizontal displacement to be made when a lower case piece accent is used on upper case characters. This is nominally only required on italic fonts, and the shift is nominally to the right. This is the default value if an alternate Italic Accent Shift is not specified. The shift value is expressed as a 7 bit unsigned value in DRU's. The eighth, most significant bit is a sign bit, set to 0 for a right shift and set to 1 for a left shift.

2.2.11 BJ/Display Superior: The MSB of word 6 defines the base jump (BJ) to align display superior numerals, in accordance with the bit definition of paragraph 2.2.9. It is nominally equal to half the height of the numerals in the font, and is nominally utilized with 50% point size reduction to achieve top aligning superiors for prices, etc.

2.2.12 BY/Autofraction Numerator: The LSB of word 6 defines the base jump (BJ) to align the numerators of autofraction numerals, in accordance with the bit definition of paragraph 2.2.9 It is nominally equal to 40% of the height of the numerals in the font, and is nominally utilized with 60% point size reduction to achieve top alignment.

2.2.13 BJ/Inferior (Inf.) on Superior (Sup.): The MSB of word 7 defines the base jump to align an inferior onto a first order superior. The base jump (BJ) value is expressed as a 7 bit unsigned value in units of 1/54 of an em (eight DRU's). The eighth, most significant bit is a sign bit, set to 0 if an upward displacement is required and set to 1for downward displacement. It is nominally equal to half the height of the numerals in the font, and is nominally utilized with 4/9th's point size reduction.

2.2.14 BJ/Inf. on Inf.: The LSB of word 7 defines the base jump to align an inferior onto a first order inferior, in accordance with the bit definition of paragraph 2.2.13. It is nominally equal to downward half the height of the numerals.

2.2.15 BJ/1st Order Sup.: The MSB of word 8 defines the base jump to align a first order superior, in accordance with the bit definition of paragraph 2.2.13. It is nominally equal to the 7/12th's of the numeral height, and it is nominally utilized with 2/3rd's point size reduction.

2.2.16 BJ/1st Order Inf.: The LSB of word 8 defines the base jump to align a first order inferior, in accordance with the bit definition of paragraph 2.2.13. It is nominally equal to downward 3/12th's of the numeral height.

2.2.17 BJ/Sup. On Sup.: The MSB of word 9 defines the base jump to align a superior onto a first order superior, in accordance with the bit definition of paragraph 2.2.13. It is nominally equal to the numeral height.

2.2.18 BJ/Sup. On Inf.: The LSB of word 9 defines the base jump to align a superior onto a first order inferior, in accordance with the bit definition of paragraph 2.2.13. It is nominally zero.

2.2.19 Thin Space Width: The MSB of word 10 defines the width assigned to thin spaces for this font. It is expressed in 1/54th's em, and is nominally the same as the period width.

2.2.20 En Space Width: The LSB of word 10 defines the width assigned to "en" spaces for this font. It is expressed in 1/54th's em, and is nominally the same as the width of numerals.

2.2.21 Em space Width: The MSB of word 11 defines the width assigned to em spaces for this font. It is expressed in 1/54th's em, and is nominally 54 units.

2.2.22 BJ/Autofraction Denominator: The LSB of word 11 defines the base jump to align the denominator for autofraction numerals. The jump value is expressed as a 7 bit unsigned value in DRU's. The eighth bit is a sign bit, set to 0 if an upward displacement is required and set to 1 for a downward displacement.

2.2.23 Serial Number: Word 12 defines in binary a number that identifies the number of times that this Master Font Number has been reproduced.

2.2.24 Italic Accent Shift #1: The MSB of word 13 defines the first alternate italic shift, in accordance with the bit definition of 2.2.10. The value is additive to the default values (zero or 2.2.10).

2.2.25 Italic Accent Shift #2: The LSB of word 13 defines the second alternate italic shift, in accordance with the bit definition of 2.2.10. The value is additive.

2.2.26 Italic Accent Shift #3: The MSB of word 14 defines the third alternate italic accent shift, in accordance with the bit definition of 2.2.10. The value is additive.

2.2.27 Italic Accent Shift #4: The LSB of word 14 defines the fourth alternate italic accent shift, in accordance with the bit definition of 2.2.10. The value is additive.

2.2.28 Small Cap %: The MSB of word 15 defines the % reduction in point point size to derive a small cap from a drawn upper case letter. It is expressed as an 8 bit binary fraction.

2.2.29 Spares: Words 15 through 17 inclusive are spare words with zero fill.

2.2.30 Font Layout Number: Words 18 and 19 define in ASCII code a font layout number that correlates the specific characterform in the font with their sequential position in the Character Width File and the Outline Data File Index. Each unique font layout number identifies a unique set of characterforms in a unique sequence.

2.2.31 Master Font Number: Words 20 and 21 define in ASCII code the font master number that uniquely defines the fonts:

2.2.32 Font Description: Words 22 through 29 inclusive are an ASCII coded sequence that forms a 16 character title describing the font face name, style, design size, etc., in an abbreviated fashion that enables ready identification of the font. Spaces are used for fill if required.

2.2.33 Extended Font Description: Words 30 through 77 inclusive are an ASCII coded sequence that forms a 96 character title describing fully the font. Spaces are used for fill if required.

2.2.34 Random Fill: Words 78 through 124 inclusive are filled from a random number generator.

2.2.35 Chain Address: Word 125 contains the chain address (track and section) of the first sector of the Character Width File.

2.3 Character Width File: The Character Width File contains data applicable to individual characters which are needed by a composition system. The Output Data Processing System makes no use of this file. FIG. 8 illustrates the structure of this file. Two bytes are required to define the data for each character; up to 125 characters may be described in each sector. Any unrequired bytes are filled with zeros. All character width groups must be sequenced to correspond with the sequence of characters in the outline index.

2.3.1 Unit Width: The first byte of each character width group defines in binary the unit width in 1/54's of the normalized em.

2.3.2 Flag Bits: Designated bits in the second byte of each character width group define specific characteristics of the character. For example, flag bit 6 is the B bit, which defines that the character is a piece accent that shall not be jumped.

2.3.3 Chain Address: Word 125 (Bytes 250 and 251) contain the chain address (track and sector, respectively) of the next Character Width File sector or the first sector of the Outline Data File.

2.4.1 Character Definitions: All characters are digitally encoded or "digitized" for an outline, relative vector decoding system, where all character outlines are assumed to be closely approximated by straight line elements. Such a system is disclosed in the commonly-owned U.S. patent application Ser. No. 905,451, referred to above.

All characters are defined as a multiple series of "curves". Each curve describes a vertical outline edge with the following components:

(a) An X, Y coordinate defining the highest point of the curve within an em square;

(b) A white-to-black or black-to-white bound;

(c) A series of straight line segments, defined by a series of data bytes which define the slope and length of each segment of the curve; and

(d) Vector direction (downward and left-to-right or right-to-left) of the segments.

Defining the character consists of listing all the curves which outline the character. They are listed in descending order; that is, the curves that start at the top of the character are listed first and the bottom last.

2.4.2 Scale: The principal unit of measurement is the Data Resolution Unit (DRU) which is defined as 1/432 of the traditional em. An extended em square is 576.times.576 DRU's.

Position 0,0 is located at the intersection of the left side bearing (LSB) and the top of the extended em square as illustrated in FIG. 9. Therefore, X (left-right) values can be positive (positive is right) or negative (if a character bound extends to the left of the left side bearing (LSB), but Y (up-down) values will always be positive (positive is down).

2.4.3 Outline Data Words: Each outline will be defined by 3 or more data words: a Y word, an X word, and one or more outline (vector/control) bytes. The format of these data words is shown in FIG. 10. The various parts of the coding shown in FIG. 10 are specified below:

Y Data Word Components:

Yn--This data defines the vertical position of a start point from the upper edge of the extended em.

K--Undefined.

X Data Word Components:

Xn--This data defines the horizontal position of a start point. The left side bearing (LSB) is defined as 0.

X Sign--The sign bit defines the displacement direction of Xn with respect to the LSB.

L Bit--The L Bit defines the direction of the dx of the first vector. A one defines a left pointing vector, a zero defines right pointing.

F Bit--The F Bit or "Flare Bit" defines which vector slope will be used by the decoder in extrapolating the character outline in the region of the grid immediately above the line Yn.

E Bit--The E Bit or "Extrapolation Bit" defines whether extrapolation is or is not used above the start point grid line Yn.

B Bit--The B Bit is the "Boundary On/Off Bit" and defines whether the outline is a left-side (on) boundary or a right-side (off) boundary.

Vectors/Controls Data Byte Components:

Vectors:

dydx--For all values of dy greater than 0, this byte defines the slope of the vector outline of the character from the start point (Xn Yn), or from the last vector end point. All vectors are sequenced serially in the same sequence that they occur on the character outline.

Controls:

For all values of dy=0, this byte defines a control code. The specific control is dependent upon the value of dx (in hexadecimal notation) as indicated below:

0--End of outline.

1--Reverse the dx direction for the next vector.

2--Defines that there are no displacement vectors applicable to the start point defined by the preceding Y and X Data Words. This control is always followed by a zero byte to produce an "End of Outline" control code.

3--Defines the vector with a horizontal displacement of 0 DRU's (a vertical vector) and a vertical displacement greater than 30 DRU's. The next data byte defines the binary value of the vertical displacement. The data byte has a resultant range of vertical displacements of 0 to 255 inclusive, but it is not utilized between 0 and 30 inclusive. (Example: The two bytes 0/3, 2/6 describe a composite vector that goes vertically down 38 DRU's.)

4--Defines a vector with a horizontal displacement of 1 DRU and a vertical displacement of 30 DRU's.

5--Defines a vector with a horizontal displacement of 1 DRU and a vertical displacement of 120 DRU's.

7 through

C--Undefined

D--Defines a rectilinear outline change with a vertical displacement of 1 DRU and a horizontal displacement of up to 255 DRU's. The next data byte defines the binary value of the horizontal displacement. (Example: The two bytes o/D, 2/6 describe an outline made up of 1 DRU vertical and a 38 DRU horizontal displacement.)

E--Defines a rectilinear outline change with a vertical displacement of 1 DRU and a horizontal displacement greater than 255 DRU's. The next data byte defines the binary value of the horizontal displacement in excess of 256. (Example: The two bytes o/E, 2/6 describe an outline made up of a 1 DRU vertical and a 294 DRU horizontal displacement.)

F--Defines a shallow slope vector with a vertical displacement of 1 DRU and a horizontal displacement greater than 15 DRU's. The next data byte defines the binary value of the horizontal displacement. (Example: The two bytes O/F, 2/6 describe a composite vector that goes over 38 horizontal DRU's and down one DRU.)

2.4.4 Outline Data File Structure: The Outline Data File resides on the font floppy disk, and stores a memory image of the data that will be loaded into one or more Font RAMs. The file occupies one or more sectors on the disk, and accordingly it is modulo 125 words long. FIG. 11 illustrates the file structure.

If the total font outline data is less than 16,384 bytes, then the Outline Data File will contain:

1. FSIZE word (No. of bytes in RAM)

2a. CINDEX (Character Index)

2b. Header and Outline Data

3. ENDFNT (Zero word)

4. Sector filler

Items 2a and 2b comprise the RAM memory image, and may not exceed 16,384 bytes.

If the total font outline data exceeds 16K bytes, the File will contain:

1. FSIZE word

2ab. CINDEX, Header and Outline Data (16,384 bytes max.)

3. FSIZE word (No. of Bytes in next RAM)

4. Header and Outline Data (16,384 bytes max.)

5. ENDFNT

6. Sector filler

Items 3 and 4 may be repeated as required if the total font outline data exceeds 32,768 or 49,152 bytes. The data will occupy the Font RAM beginning at address "4000 and may fill through to address "7FFF (where the initial quotation mark (") indicates a hexidecimal number. Addresses in the headers will be absolute; addresses in the CINDEX will be offset absolute ("0000 through "3FFF) with the two MSB's flagging multi-RAM locations.

The specific contents of the Outline Data File are as follows:

2.4.4.1 FSIZE: This word defines in binary the number of bytes to be loaded into a Font RAM. The count does not include the FSIZE word or the ENDFNT word. The count for the first Font RAM includes the entire CINDEX and all header and outline data.

2.4.4.2 CINDEX: The character index is variable length and consists of a character count (CCOUNT) and a relative addressed index.

The CCOUNT is one byte defining in binary the number of characters in the font, and therefore it also defines the word length of the index. It will be a number between 1 and 225 inclusive. The RAM address location of CCOUNT is "4000.

The index contains a one word entry for each character in the font. Each entry is the offset absolute address of the YCOUNT byte for the character (see section 2.4.4.6 below). The two most significant bits of word indicate in binary the RAM that contains the character, where 00 is the RAM that contains the index. The 14 least significant bits contain the offset RAM address (the absolute RAM address less "4000) of the YCOUNT byte of the character.

The first entry in the index is by definition character number 1 and must correspond with the first character width group in the Character Width File. Character numbers proceed sequentially by implication (there are no expressed character numbers or library numbers at any location in the font).

2.4.4.3 ENDFNT: This word defines the end of all font data and consists of 2 bytes of zeros.

2.4.4.4 Sector Filler: Zero data is used to fill through to the end of the floppy disk sector that contains the ENDFNT word.

2.4.4.5 Header and Outline Data: The header and outline data in each RAM contains all of the character digitization data pertaining to each of the characters located within that RAM. The X and Y start locations for characters are listed in the Header File; the vectors and control bytes that define the profiles of characters are listed in the Outline File. The two files are separated by a zero data byte (ENDHDR). FIG. 12 illustrates the file structure of the Header and Outline Data.

A checksum byte follows the Outline File and immediately precedes the ENDFNT word or the FSIZE word that separates RAMs.

2.4.4.6 Header File: The Header File consists of a series of character headers, one for each character in the font. There is no space between headers. Each character header contains (in sequence and without space) a YCOUNT byte, a CSIZE word, and one or more start-pair sets of data words (one set for each pair of starts).

YCOUNT: The YCOUNT byte defines in binary the number of YN entries in the header, which is the same as the number of start pairs. The length in each character header is ten times the YCOUNT plus 3 bytes.

CSIZE: The CSIZE word defines in binary the total amount of data space in bytes that the character fills when loaded once into the Data RAM. Accordingly, it is equal to twelve times the YCOUNT plus the length of all the profile strings addressed within the start-pair data sets.

START-PAIR DATA: YN is the Y Data Word and XN is the X Data Word as defined in 2.4.3. N must be even, since outlines always start in pairs. AN is the absolute address of the initial byte of the profile string of vectors and controls that define each outline shape. Each address will be a number between "4000" and "7FFF. Addresses may be duplicated within the header file in the event that a profile string is shared (the character outline shape is common) for more than one start point. An address may not point to a profile string located in another RAM. The YN, XN, and AN Data Words are sequenced as shown in FIG. 12 and listed without space. Each successive YN value is equal to or larger than the preceding YN value.

2.4.4.7 Outline File: The Outline File consists of a series of profile strings. Each profile string is a sequential series of two or more Vectors/Controls Data Bytes, as defined in 2.4.3. Each string defines a unique vertical character outline and begins at the header start point. A string is terminated by control 0 (end of outline), which is a zero data byte. Filler bytes may not be used within a string; they are permissible before or after any string. The digitization program(s) avoids duplication of identical profile strings, and minimizes the number of RAMS a font used by sharing profile strings for character outlines that closely approximate each other.

2.4.4.8 CHKSUM: A one byte checksum verifies each complete RAM; it is formed with all of the data in the Font-RAM except the CHKSUM byte itself. The checksum shall be formed by initializing to zero; then, for each byte, the checksum is rotated right one bit (LSB becomes MSB) and the data byte is added to form the new checksum. Overflow on the addition is ignored. The final 8 bit checksum is defined as CHKSUM and is entered after the last data byte.

2.4.5 Profile Strings: In general, the profile strings in the Outline Data File are separated from the start points (YN and XN) to permit several start points to reference (address) the same profile string. In this way, different characters within the same font having, as a part thereof, the same basic shape may be defined by the same data, thus achieving data compression.

For example, the following letters may have the identical contour on their left-hand side: "o", "c", and "e". The Outline Data File will thus contain two profile strings defining the inner and outer boundaries on the left-hand sides of these characters. The highest pair of start points in the character header for the "o", "c" and "e", respectively, may therefore address these two profile strings.

Because the dx values in the profile strings may be either positive or negative, depending upon the "L bit" in the X data word (XN), a single profile string can serve for various characters which are symmetrical. For example, portions of the character "b" may be symmetrical with the character "d" and portions of the character "p" may be symmetrical with the character "q". Such characters may be defined with the same profile strings which are directed by the "L bit" to move in opposite directions.

In general, character designers (persons who design character fonts) tend to create a few basic character shapes which are repeated throughout the font, either directly or in mirror image. Consistency dictates that a few shapes be repeated throughout the font; symmetry dictates that mirror images be used. The profile strings utilized in the digital definition of characters in the present system are a useful tool in recreating these basic character shapes. Because the encoding scheme permits the addressing of a single profile string from the start points of various characters, and permits the dx increments in a profile string to have positive or negative values, the quantity of data required to define an entire font is substantially reduced.

2.4.6 Miscellaneous: Within the definition of a single character, there is no restriction on starting two outlines (profile strings) from the same point. There is also no restriction on ending two outlines at the same point. Two outlines may touch, but they may not cross over each other if they change from "on" outlines to "off" outlines.

Broken characters are also permissible in the Outline Data File. There is no restriction on broken (divided, separated) characters.

2.5 Font RAM Format

The DMS utilizes RAM memory to contain the font data for the font(s) to be typeset on the page in order to have high-speed access to this data. The data for the font is supplied to the DMS by the Input System where it is stored on the System Floppy Disk (SFD).

A complete font is stored on one or more Font RAMs, each Font RAM storing no more than one font at a time.

The DMS can contain one to eight Font RAMS. The system will function with only one Font RAM, provided that one Font RAM size fonts are used. Multiple Font RAMs ensure against degradation in throughput speed on pages with font mixing.

Each Font RAM is 16K bytes; each Font RAM card can contain up to 64K bytes of memory: the equivalent of 4 Font RAMS. Units of less than four fonts can be accomplished by depopulating the Font RAM cards in 16K byte increments.

At system reset, the DMS determines which Font RAMs are available for loading by writing a pattern into each RAM location and reading back the results. Any mis-match is recorded as an inactive font position in a font table. After testing each of the eight locations, a message is sent to the Input System defining the number of active Font RAMs; this can be utilized to detect defective RAMs. The font table is used later to record the font numbers stored in each RAM.

In the process of developing a Data RAM, the DMS copies specific character outline data from the Font RAM into the outline file in the Data RAM. If a font change occurs, the DMS will search the table of font numbers loaded. If the font is not already loaded, the DMS will load the new font into the first empty Font RAM(s). If all Font RAMs are in use, the RAM or RAMs least recently used are overwritten with the new font needed.

The data stored in the font RAM is identical in content and structure to the Outline Data File font data on the Input System floppy disk, as defined in Section 2.4.4.

In addition to the font data stored on the Font RAM card(s), the DMS maintains two additional tables per font in the program workspace that are used to regulate data transfer from Font RAM to Data RAM: an In-Seg Table and an In-RAM Table. These are described below.

2.5.1 In-RAM Table: This 512 byte table contains the address within the Data RAM where a character header has been stored. The table is ordered in accordance with the character numbers. Each entry is two bytes. A zero entry indicates that the character data has not been loaded.

Whenever a new character is put into the Data RAM, the corresponding 2 bytes in this table are loaded with the Data RAM address. This table is cleared at the start of building each new Data RAM.

2.5.2 In-Seg Table: This 32 byte table is used to indicate which characters of the font have already been encountered within the line segment currently being developed in the Data RAM. Whenever an address is loaded into the In-RAM Table, a bit is correspondingly set in this table. This table is cleared at the start of each new set level (YSL).

In structure, each bit corresponds to a character number between 0 and 255 inclusive. The address of the bit is computed by: ##EQU1## where Q is the byte in the table and R is the bit within the byte.

3.1 Page Definition: A page position is defined by X, Y coordinates in 1/10 pts. This is called a raster resolution unit (RRU). The top left hand corner is position 0,0. The maximum page size is 11".times.17". That is 7954.times.12,292 RRU. Movement in a page can only be from top to bottom.

The raster position being solved for at any time is called the Y set level (YSL). This value initially starts at 0 and is incremented by one until it reaches the maximum page depth.

3.2. Page Variants: In addition to the normal standard page form defined by section 3.1, five full page variants have been incorporated in the ODS design. All variants are mutually exclusive.

3.2.1 High Resolution: A high resolution laser recorder can have its drum drive gear ratio altered so that each step of the stepper motor 58 drives the drum 56 by 1/20th pt., a high resolution RRU (HRRRU). Horizontal (x) resolution is not increased.

The ODS has a chip switch on the DMS (Z80A) microprocessor which is set for this laser recorder. The DMS halves the .DELTA.Ys, and the OCS increments the set level on every other raster output.

3.2.2 Proof Page: A laser recorder with proof page capability would make 2 stepper motor steps between each raster line, effectively doubling the speed of page setting with proof quality.

The command for proof page will be entered from the Input System. The DMS will set the Ys level depending upon whether the laser recorder is a normal or a high resolution unit, and the OCS will accordingly increment the set level by one or two on each raster output.

3.2.3 Page Width: The laser recorder will have either an 81/2" or a 11" wide drum 56.

The ODS has a chip switch on the DMS (Z80A) microprocessor which is set for 81/2" or 11". The DMS uses an appropriate page width when page complementing the XPOS value for wrong reading output.

3.2.4 Wrong Reading: Any page can be output from any type of laser recorder in right reading or wrong reading (mirror image).

Selection of wrong reading is made by a toggle switch on the DMS (Z80A) microprocessor. The DMS page complements the XPOS location of every character, and complements the X position of every outline on each character and the direction that each outline moves.

3.2.5 Reverse Video: Any page can be output white-on-black or black-on-white (reversed normal).

Selection of reverse video is made by toggle switch on the hardwired processor (HWP). The HWP inverts the polarity of the raster.

3.2.6 Line Rule: Line rule is similar to reverse video, except that an entire line (white-on-black or black-on-white) of defined length becomes a single solid color. This command permits generation of line rules on the page.

4.1 General: This specification sets forth the required data and data format to be transmitted between the Input System and the Output Data Processing System. The transmissions are made through the Input System IOP-80 on a handshake basis of a byte serial transfer. Table 1 summarizes all the interface transmissions to the ODP System.

4.1.1 Page Data: The Output Data Processing System is a page output machine, principally because the laser recorder must expend the time required to expose a full raster even if it only had data for part of a raster. Therefore the throughput of the machine is enhanced significantly by supplying the laser recorder with all of the graphic data needed in each full raster prior to exposing the raster. This requires storing and regrouping random sequence input data into a top down sequence. Due to memory size limitations in the Output Data Processing System, the data must be further packeted into groups defined as "line segments", which is the standard unit of page data to be transmitted by the Input System. Section 4.2 will detail the page data requirements.

4.1.2 Font Data: The Input System stores digital outline fonts on floppy disks in the manner described above. The outline data is required in the solution of the raster on-off points, and this data is transmitted by the Input System on a whole font basis (excluding width data, BLJ dat, etc.). Section 4.3 will detail the font data requirements.

4.1.3 Other Data: In addition to the above job related data, periodic data transfers may be made by the Input System, if desired. These include programs, error messages, restart and program reset. The power-on reset signal may also originate in the Input System.

4.1.4 Notation: Meta-Language notation will be used to describe the syntax of the data requirements. The following notation will be used:

4.1.4.1 " " Terminal--a fixed bit length symbol element (e.g.: all page data elements are 16 bit words).

4.1.4.2 () Non-Terminal--A higher order language element which is composed of one or more terminals and/or one or more non-terminals.

4.1.4.3 { } Optional Repeats--The braces indicate that the enclosed non-terminal(s) may be not used or used as often as desired.

4.1.4.4 / Either-Or--A slash indicates that the non-terminals on either side of the slash are possible alternative elements.

4.1.4.5 * Once only--An asterisk is used to indicate that the non-terminal may not be used more than once within the complex non-terminal being defined.

4.2. Page Data:

4.2.1 Page Data Structure: As outlined in section 4.1.1 above, coded data which describes a page must be packeted into groups defined as "line segments":

Each page can consist of one or more line segments followed by an end page code. A blank page has no line segments. The end page code is a terminating code, and no data relating to the page can be accepted after the code. All functional data received prior to the end page code is not carried over into the next page, and must be repeated as needed.

Each line segment defines a character set, a reverse video set, and/or a line rule set:

The first element in each line segment is the segment number. Separate line segments with unique segment numbers must be defined for each character set with a unique YPOS and point size combination. Separate line segments should preferably be defined for each reverse video or line rule set, and also preferably for a set that is not contained within the Y limits of the extended em square of a character set. All reverse video or line rule sets within a single line segment must have the same YPOS value.

The segment number is followed by YPOS, which nominally is the Y coordinate on the page of the top of the extended em square of the characters in the line segment and/or the upper coordinate of the reverse video or line rule set(s) in the line segment. All the line segments on the page must be sequenced in the order of the YPOS coordinate; there is no sequence requirement between line segments with the same YPOS coordinate.

The line segment can contain one or more character sets, and/or one or more reverse video and/or line rule sets.

The end segment code is a terminating code, and no data relating to the segment can be accepted after the code. All functional data received prior to the end segment code is not carried over to the next segment, and must be repeated as needed.

4.2.1.1 Character Sets: All character sets within a line segment must follow the structure:

with one initial character followed by one or more character pairs. The initial character must follow the structure:

The initial character in a line segment must contain a size, a font number, the x coordinate of the character's left side bearing (XPOS), and the character pair data which is structured:

and where permissible functions are:

The character code is partially a terminating code, that is, although no functions relating to a particular character can be accepted after the code, all functional data received prior to the code is carried over and remains valid until altered by a new function code or a line segment terminating code (END SEG).

All latest function codes are valid in this manner until altered except YLOW, which can only be altered by the issuance of a numerically higher valued YLOW.

Within a character set, an initial character must precede any follow-on character pairs. All other functional codes may be sequenced randomly, subject only to the restrictions described above.

4.2.1.2 Reverse Video Sets: All reverse video sets within a line segment must follow the structure:

The elements of the reverse video set must be sequenced in the above order with no intervening elements.

4.2.1.3 Line Rule Sets: All line rule sets within a line segment must follow the structure:

The elements of the line rule set must be sequenced in the above order with no intervening elements.

Table 2 summarizes the syntax of the page data structure:

4.2.2 Input Codes/Terminal Elements: Table 3 summarizes the input code structure for those terminal elements which are to be sent from the Input System to the Output Data Processing System, with references to the following descriptions of the terminal elements used in the syntax in section 4.2.1.

4.2.2.1 (SEG#)="SEG#"

This is a 13 bit number (the LSB's of the 16 bit field, the 3 MSB's shall be 0's) unique to each line sement in the page. It is the number used to identify each line segment, and will be used by the Output Data Processing System when it needs to call for a specific line segment. It must be the first code of every line segment, and may not be issued more than once in any line segment.

Segment numbers may be any number between 1 and 8191 inclusive (not zero), and it is not required that the segment numbers be sequenced with increasing YPOS values.

4.2.2.2 (YPOS)-"YPOS"

This is the Y coordinate on the page in RRU's of the top of the extended em square of the characters in the line segment and/or the upper coordinate of the reverse video set in the line segment. The top of the page (which is nominally below the top of the sheet of paper) is defined as 0 RRU's. Up to 14 bits are available to describe YPOS values between 0 and 12,292 RRU's (17 inches).

The LSB of the YPOS corresponds with the LSB of the input word. This code nominally follows the segment number, and is only issued once within a line segment.

4.2.2.3 (XPOS)="XPOS"

This is the X coordinate on the page in RRU's of the left side bearing of the character or of the reverse video coordinate that pairs with the YPOS RV coordinate.

The left hand edge of the sheet of paper and the page is defined as 0. Normal margin offsets are controlled by Input System programs. Up to 14 bits are available to describe XPOS values between 0 and 7 and 7,954 RRU's (11 inches). The LSB of XPOS corresponds with the LSB of the input word.

4.2.2.4 (YLOW)="YLOW"

This is the Y coordinate on the page in RRU's of the bottom of the extended em square of the characters in the line segment. It is not necessary that this value be supplied for line segments with average characters; i.e., characters that fall within the extended em square. It must be supplied for logo's that do extend lower than the extended em.

The Input System derives the value from the font data and the point size that the logo is being set at. If this code is issued more than once, the highest value (lowest point on page) is retained by the Output Data Processing System. Scaling, zero placement, and data placement are identical to YPOS.

4.2.2.5 (FONT)="FONT"

This terminal code defines the font number to be used for all characters following until a new font is input. Up to 10 bits are available to input font numbers between 1 to 254 inclusive. The font number LSB corresponds to the word LSB.

4.2.2.6 (CHAR)="CHAR"

This terminal code defines the character number to be output and is a semi-terminating code (see the description in section 4.2.1.1). Up to 10 bits are available to input character numbers between 0 and 255 inclusive. The character number LSB corresponds to the work LSB.

4.2.2.7 (PT SIZE)="PT SIZE"

This terminal code defines the point size to be used for all characters in the line segment. It may only be issued once within a line segment. Up to 10 bits are available to input all half point sizes between 1/2 and 130 inclusive. The word LSB corresponds to 1/2 point, and bits 1 thru 8 define the binary value of the point size directly.

4.2.2.8 (SET WIDTH)="SET WIDTH"

This terminal code defines the set width to be used for all characters following until a new set width is input. If this code is not issued, the set width has a default value equal to the point size of the line segment being set. The set width command(s) must follow the point size commands.

4.2.2.9 (LR CODE)="LR CODE"

This terminal code defines that the immediately following 2 words represent the YEND and XEND respectively in a line rule set in which YPOS and XPOS preceding are the beginning coordinates. The 10 LSB's of the input code are zero.

4.2.2.10 (SLANT)="SLANT"

This terminal code defines the slant amount to be used for all characters following in the line segment until a new slant is input. Up to 10 bits are available to define 5 possible slant conditions; value 0 corresponds to SLANT OFF, value 1 to SLANT+7, value 2 to SLANT+14, value 3 to SLANT-7, and value 4 to SLANT-14.

4.2.2.11 (RV CODE)="RV CODE"

This terminal code defines that the immediately following 2 words represent the YEND and XEND respectively in a reverse video set in which YPOS and XPOS preceding are the beginning coordinates. The 10 LSB's of the input code are zero.

4.2.2.12 (YEND)="YEND" (XEND)="XEND"

These terminal codes define the bottom coordinates on the page of a reverse video set in the same scale as YPOS and XPOS. The two MSB are set to zero. If the YEND input has a value higher than the current YLOW, this value is used to redefine YLOW.

4.2.2.13 "END SEG"

This terminal code defines the end of a line segment, and sets all variable functions contained within that segment to the default value. The 10 LSB's are all zero.

4.2.2.14 "END PAGE"

This terminal code defines the end of a page, and sets all variable functions to the default value. The 10 LSB's are all zero.

This code must be preceded by a line segment which contains a YLOW value equal to the depth of the page. This may be done by either:

(1) Defining a YLOW equal to the page depth within the last line segment of the page; or

(2) Defining an additional line segment with the content: (SEG#) (YPOS) (YLOW) "END SEG",

where YPOS and YLOW equal the page depth, or with the content: (SEG#) (YPOS) "END SEG",

where YPOS equals the page depth.

4.3 FONT DATA

4.3.1 Font Data Structure: Coded data which describes the outlines of characters in a font will be transmitted from the Input System to the Output Data Processing System on a whole font basis:

Each font consists of one set of character outline data for each character contained in the font. Up to 256 characters may be contained in the font, provided that the total contained in one font is less than 15,328 bytes by twice the total number of outlines in the font. The end font code is a terminating code, and no data relating to the font can be accepted after this code.

Each character's outline data is:

where the outline non-terminal is used once for each outline defined by the #OUTLINES terminal.

The character number is identical to the CHAR terminal described in section 4.2.2.6 and is a number between 0 and 255 inclusive. The number of outlines per character is limited to 255.

Each outline consists of start coordinates, vectors and controls as required to describe one edge of the character:

where the permissible controls are:

(CONTROLS)=(END OUTLINE)/(CHANGE DIRECTION) (NO VECTORS)/(LONG VERTICAL)/(SHALLOW HORIZONTAL)

Section 2.4.3 describes in specific detail the outline data structure syntactically covered in the two statements above. Reference should be made to this section; it is this data that should be transferred exactly from the floppy disk storage medium to the Ouptut Data Processing System by the Input System for these two statements.

Table 4 summarizes the syntax of the font data structure:

4.3.2 Input Codes/Terminal Elements: FIG. 13 summarizes the input code structure for those terminal elements which are to be sent from the Input System for a font data transfer.

4.3.2.1 (CHAR)="CHAR"

This terminal code defines the character number assigned by the Input System. Up to 8 bits are available to describe character numbers between 0 and 255 inclusive. Bit 15 corresponds to the MSB and bit 8 is the LSB.

4.3.2.2 (#OUTLINES)="#OUTLINES"

This terminal code defines the number of outlines in this character and originates on the font floppy disk. Up to 8 bits are available to describe between 1 and 255 outlines.

4.3.2.3 "END FONT"

This terminal code is the font transfer terminating code. All 16 bits in the word are zeros.

5.1 General: The data RAM serves as an output buffer for the DMS, and an input buffer for the OCS. As line segment data is input to the DMS, it is reformatted and stored into the data RAM. Two such data RAMs are used within the system, each one is 32 K bytes long. Both data RAMs are accessible by the DMS and the OCS with the following limitations:

(1) A processor may select and operate on only one data RAM at a time.

(2) A processor may not select a data RAM which is selected by the alternate processor.

(3) Once a processor deselects a data RAM (or releases control of it), any data within that RAM is no longer valid to that processor.

5.2 Data Ram Building: The data RAMs are developed by the DMS and passed onto the OCS for processing. Double buffering is used in building up the data and therefore two such RAMS exist. This permits the DMS to develop the next buffer of data while the OCS is processing the other. The basic design is shown in FIG. 14. In developing this data RAM, and DMS attempts to fill it with as much data as possible. By so doing, it should provide the OCS with enough data to word with to avoid the possibility to phototype setting unit (PTU) slow down.

The buffer space is optimized by sharing outline data that has been put into the buffer for other line segments. In order to achieve this, the DMS develops the RAM from two directions. Line segment data, as it is read in and reformatted, is put at the low end of memory, and related character outline data is put at the high end. When these two data sets interfere with each other, the data is backed up to the last complete line segment and the output limit is defined. This process is shown in FIG. 15.

5.3 Data RAM Format: The data RAM layout is illustrated in FIG. 16. The Character Outline section of the data RAM is shown in FIG. 17. In defining the format of the data RAM, the Meta-Language notation, as outlined in section 4.1.4, will be used. The Page, Line Segment File and Outline File structure are indicated in Tables 5, 6 and 7, respectively.

5.3.1 Page Structure: The OCS defines a page as one or more data RAMS:

All pages must begin with a initial data RAM whose format is:

All subsequent data RAMS for the same page have the format:

The difference between the two being the new page element which is itself a terminal element:

All data RAMS must have as its first code, the output limit value:

The data file can be divided into a separate files, the line segment file and the character outline file:

5.3.2 Line Segment File: The Line Segment File consists of all the line segments input and reformatted by the DMS. This has the form of:

As many line segments as there is room for may be put into this file. The last line "segment" must be followed by the end RAM code:

If more data exists for the page, the end data element is used. If this is the last RAM for the page, the end page element is used.

The line segment may be defined as follows:

The structure is very similar to the input format, however, the code structure does vary.

All line segments must start with a start segment element. This is defined as:

YSET can be defined as:

and is always inserted at the start of every line segment by the DMS. This defines the Y coordinate of the page where the line segment is to be processed. The ZERO DATA WORD serves as a two byte pad for use when the line segment becomes active (i.e. processing of outlines begin) or deleted.

Once processing of a line segment starts, YSET is replaced by the YACC PAIR, where:

The YACC PAIR is the next set level in DRU's for the line segment. After the OCS has completely processed a line segment, it replaced it with the link pair.

This link pair is two elements which combined provide an absolute of the next line segment.

5.3.2.1 Character Set: All character sets within a line segment must follow the structure:

with one initial character followed by one or more character pairs. The initial character must follow the structure:

The .DELTA.YS pair defines the change in DRU's per raster resolution unit. It is dependent on the point size.

The scale pair defines the number of raster resolution units for each DRU. It is dependent on the set width.

A character pair is defined as:

and where permissible functions are:

All functions received prior to the outline address are carried over and remain valid until altered by a new function code or a new line segment.

5.3.2.2 Reverse Video Set: The reverse video structure is:

and:

5.3.3 Outline File: The Outline File has the basic structure:

The character outline file is the font RAM data for the particular character reformatted by the DMS into the following format:

The curve update file is the start for every curve in the character

The signs in XN word in the outline file have been complemented by the DMS if wrong reading is in effect.

As the OCS processes outlines, the OUTLINE STARTS data are updated to reflect the current processing point within each outline.

The slope file is:

The controls are defined in sections 4.3.1 and 2.4.3.

5.4 Terminal Elements: Table 8 summarizes all the coded entries in the data RAM entries. The uncoded elements, YLMT, YSACC, and outline elements, within a data RAM are not in the table: they are 16 bit binary values whose position defines the code type. The uncoded data RAM outline elements are as defined in section 2.4.3; the previously defined syntax must be followed.

5.4.1 [OUTLINE ADDRESS]="outline address"

This is a 14 bit number which specifies the address in the outline file of the character outline to process. The address always points to the first curve of the outline in the update file. This is a word address and must therefore be doubled to get the byte address within the 32 K RAM. [Since all data in the data RAM is 16 bit codes, codes will always start on an even byte address).

5.4.2 [YPOS]="YPOS"

This is the same code an entered into the DMS. See section 4.2.2.2 for format and definition. In OCS, it shows an additional function. It not only marks the start of a new line segment, but also terminates the previous line segment.

5.4.2.1 [ZERO DATA WORD]="ZERO DATA WORD"

This uncoded 16 bit field always follows YPOS. It serves as a pad for use when a line segment becomes active.

5.4.3 [YACC HIGH]="YACC HIGH"

This code defines the most significant 8 bits of the set level in DRU's. It contains the 8 most significant integer bits of the value.

5.4.3.1 [YACC LOW]="YACC LOW"

This is a 16 bit field which always follows YACC HIGH. It contains the remaining 2 integer bits and the 11 decimal bits of the set level; the 3 least significant bits are always 0.

5.4.4 [XPOS]="XPOS"

This is the same code as entered into the DMS. (See section 4.2.2.3 for format and definition), except when the wrong reading switch is set, when the page width complement of XPOS (PG WDTH-XPOS) is entered.

5.4.5 [NULL]="NULL"

The null code is used to delete elements within a line segment. The DMS inserts this code to remove font calls within a segment as they are acted on. The OCS inserts this code to remove "outline address" as they are completed.

5.4.6 [.DELTA.Ys HIGH]=".DELTA.Ys HIGH"

This contains the most significant 8 bit of the computation 43.2/POINT SIZE. It consists of a 5 bit integer number and 3 bit decimal.

5.4.7 [.DELTA.Ys LOW]=".DELTA.Ys LOW"

This contains the remaining 8 bit decimal value of the computation 43.2/POINT SIZE. This code will always follow Ys high.

5.4.8 [SCALE HIGHL]="scale high"

This contains the most significant 8 bits of the computation set width/43.2. It consists of a 4 bit integer number and 4 bit decimal.

5.4.9 [SCALE LOW]="scale low"

This contains the remaining 8 bit decimal value of the computation set width/43.2. This code will always follow "scale high".

5.4.10 [BLJ]="BLJ"

The code defines the baseline adjustment in 154ths. See 4.2.2.9 for format.

5.4.11 [LINK HIGH]="link high"

This code supplies the most significant 8 bits of a link address used to skip over line segment(s) that have been completed.

5.4.12 [LINK LOW]="link low"

This code supplies the least significant 8 bits of a link address used to skip over line segment(s). This code will always follow the "link high" code.

5.4.13

[RVY HIGH]="RVY high"

[RVY LOW]="RVY low"

[RVX HIGH]="RVX high"

[RVX LOW]="RVX low"

These codes must be in the above sequence. The lower order 8 bits defines the reverse video limits.

5.4.14 [NEW PAGE]="New Page"

This control defines a data RAM as the first of a new page. It must be issued before the first line segment of a page.

Bit 0 is a 0 for normal, standard operating mode and is a 1 for proof page mode. Bit 1 is a 0 for normal resolution units (1/10 pt. per raster) and is a 1 for high resolution units (1/20 pt. per raster). The other 6 LSB's have no meaning.

5.4.15

[SLANT OFF]="Slant off"

[SLANT +7]="Slant +7"

[SLANT -7]="Slant -7"

[SLANT +14]="Slant +14"

[SLANT -14]="Slant -14"

These codes define which slant constant is to be used for the following outline computation. The LSBs have no meaning.

5.4.16 [END DATA]="End data"

This code follows the last line segment within the data RAM. It informs the OCS that more data for this page will follow in the next data RAM. The LSBs have no meaning.

5.4.17 [END PAGE]="End page"

This code follows the last line segment of the page. It informs the OCS that no more data RAMs for the page follow. The LSBs have no meaning.

5.4.18 [YLMT]="Y limit"

This terminal is the first code in the data RAM. It is a 14 bit number defining the last raster in RRUS that shall be output by the OCS with the data in the data RAM.

5.4.19 "Font Call"

This code is generated by the Z80A to simplify font call processing. It is a temporary code which is replaced with a NULL Code prior to releasing the DATA RAM to the OCS for output processing.

The outline algorithm is the formulae and processes that solve and store the intersections of rasters with the outlines of the characters to be set at any particular vertical level on the page. These intersections are the horizontal locations where the laser beam shall be turned on and off as it strokes out the raster lines.

6.1 Basic Outline Intersections: All characters in the data set are described by a series of initial locations and vectors that trace the outlines of the characters. FIG. 18 illustrates the factors to be used in solving one outline. (Refer to FIG. 9 for correlation).

Vector VN represents one edge of one character. The distance from the left edge of the page to the point on this vector where it is intersected by the current raster level is to be solved. It can be seen that this distance, which is defined here as XREG, is:

The general expression must also provide for the possibility that XN may be left on the left side bearing (XN sign bit is negative) and that the vector may point down and to the left (the data L bit is one) resulting in a negative .DELTA.X:

Since XREG is desired in raster resolution units (RRU's) for ultimate output to the laser, the proper scale factors must be applied. XPOS is input from the Input Systm in RRU's and is maintained in those units throughout the Output Data Processing System. XN and the .DELTA.X factors are all stored in relative character units, to enable setting all size characters from one set of data. The horizontal scaling of characters is based on the set width command to enable character expansion and compression for appearance purposes and/or for copy fit purposes. The expression becomes:

Similar triangles are used to solve geometrically for .DELTA.X: ##EQU2## Substituting:

In the general form of the equation above, XN and YN are the starting locations of the vector VN, whose slope is dx/dy. Ys is the distance in character units, defined as data resolution units (DRU's), from the zero datum on the character extended "em" area to the raster level. Ys is thus a function of the point size to be set and the raster distance in RRU's from the top of the em:

In actuality, the algorithm looks up the incremental .DELTA.Ys in a table, where:

and adds this value to the prior Ys for each new raster line.

6.2. Boundary Bit And On/Off Buffer Storage: Equation (1) is solved for all character outline vectors that intersect the current raster. Within the character data, a boundary bit (B bit) is embedded with each outline series that records whether the outline should turn the laser beam on or turn it off.

The final solutions of each intersection are stored in either an ON RAM or an OFF RAM dependent upon the state of the B bit. These RAMs, which are the so-called "raster line buffers", are 8 K long each and each contain one bit for each horizontal resolution element (Ref: Section 7.5). Each XREG computation results in the setting of the particular RAM bit that corresponds to the value of the XREG solution. Incrementing techniques cause an adjacent bit to be set if a solution is identical to a prior solution on the same raster line which can result when characters are overlayed or kerned.

6.3 Long Verticals: To enhance data compression, a pair of data control bytes are available for use in the vector data stream to define vertical vectors longer than the nominal longest allowable sloping vector, VN (whose maximum length is limited by the 15 DRU limit on dy and dx).

A control byte is used to define this condition, where dx (and therefore .DELTA.X) are zero, and the following byte defines the dy length of the vector in binary code (Ref. Section 2.4.3 "Controls", where dx=3). This data byte pair should define any vertical vector with a length between 1 and 255 DRU's. However, its use has been restricted to vectors with lengths between 31 and 255 DRU's because (a) lengths between 1 and 15 DRU's can be expressed with less data using the non-control dydx definite vector coding available and (b), lengths between 16 and 30 DRU's can be processed faster in the 8.times.300 microprocessor if they are expressed as two definite vertical vectors.

6.4 Steep Slopes: There exist character outlines that have such a steep, almost vertical, slope that they cannot be adequately defined by using the steepest non-vertical dxdy pair due to the 15 DRU limitation. If these outlines were defined by the combination of a long vertical and a 1/15 vector, (which pair would accurately define the end points of the steep vector desired), the output that is being used would accurately output each component part of the pair; however, this would give a bumpy appearance to a character edge at large point sizes, particularly if the vertical component of the pair is longer than the sloping component.

A series of control bytes have been defined that enable all steep vectors to be defined by the combination of a vertical component and a longer sloping component. The control bytes (Ref. section 2.4.3 "Controls", where dx=4, 5 and 6) define vectors with slopes of 1/30, 1/60, and 1/120; they are used with vertical vectors or long verticals as required to complete the exact steep vector.

Since the hardwired processor (HWP) (Ref. section 7) does not accept 1/30 etc. slopes, the solution of these control vectors is accomplished by having the SMS300 microprocessor input compensated values to the HWP. The SMS300 inputs a 1/15 vector value to the HWP, but reduces the Ys and YN values so that a halved (or quartered or eighthed) .DELTA.X solution results.

6.5 Shallow Slope: There exist character outlines that have such a shallow slope that they cannot be defined by one dxdy pair due to the 15 DRU limitation. In this case dy is one DRU and dx needs to be greater than 15 DRU's. A control byte is used to define this condition, and the following byte defines a multiple (1 thru 15) of 16 DRU's displacements, plus a 0 thru 15 additional DRU displacement. (Ref. Section 2.4.3: "Controls", where dx=F). Any vector with a dx between 16 DRU's and 255 DRU's in combination with a dy of 1 can be defined by this byte.

The multiple above is defined as K, and this alters the basic equation to:

6.6 Squared Outline Shapes: Character outlines such as the right side of an upper case Helvetica "E" have a squared shape. These shapes could be treated as a series of individual vertical outlines, but this would require a significant data load to record the start locations of each outline. Rather, two control bytes have been provided to define the condition of an outline that displaces horizontally in a rectilinear fashion during one vertical DRU (as contrasted with the shallow slope, which displaces horizontally during one vertical DRU in a sloping fashion). One control byte (Ref. section 2.4.3: "Controls", where dx=D) defines horizontal displacement up to 255 DRU's; the second control byte (Ref. Section 2.4.3: "Controls", where dx=E) defines displacements between 256 and 511 DRU's inclusive.

The 8.times.300 microprocessor outputs dx and K as zero within this vertical DRU zone in order to accomplish a squared off outline shape.

6.7 Extrapolation: When the current raster level approaches within 1 DRU of the initial YN (start) of an outline, a solution is made. This is an extrapolated solution, because the raster level has not yet reached the vector. In this case, Ys is less than YN which makes the term (Ys-YN) negative, which in turn reverses the polarity of the sign that precedes the term in equation (2) above (.DELTA.x is reversed in sign).

The slope of the next following vector is used while Ys is greater than YN-1 and is equal to or less than YN, provided that the flare bit (Ref. section 6.8) is not on. True extrapolation results in this zone.

6.8 Flare Bit: In two cases, extrapolation with the next following slope is insufficiently shallow for quality character reproduction. In case 1, a character editor, based on a visual review of real or simulated character output, desires a shallower entry slope in the extrapolation zone. A flare bit (F bit) is provided in the XN data word, which he may set on. In case 2, a very fine vertically reentrant feature of the character produces start points without any following vectors; in this case, the character digitization program automatically sets the flare bit to a one. The digitization program also sets the flare bit to a one when the initial vector has a slope that is less than 30.degree..

When the flare bit is on, the K value in equation (2) is set to a value of one for the extrapolation zone only, thereby decreasing the slope in this zone. This is controlled by the OCS.

6.9 Extrapolation Cancel Bit: The E bit in the XN data word will cancel extrapolation. This bit can be set to a one by the character editor.

6.10 Slant: Electronic slanting of a character consists of taking character features within the em square and parallelogramming them a specified number of degrees in either a right or left direction depending upon the slant request. This is done about a central neutral axis which remains in place; Ys of 344 DRU's is the level of the neutral axis.

The change in the x position oulines caused by this slanting is therefore:

The allowable slant angles have been selected as 7.1 degrees and 14 degrees because their tangents are 1/8 and 1/4 respectively, and therefore allow scaling the SLC by bit shifting.

Incorporating this final factor into equation (2) results in:

which is the equation solved by the OCS and the HWP

6.11 Reverse Video: Reverse video does not follow the outline algorithm equation; rather it simply reverses the resultant output from the ON/OFF buffers (Ref. section 6.2) between two horizontal resolution locations. These locations are specified from the Input System, which also specifies the vertical limits over which the reversing shall be effective. The points are stored in a RAM similar in data structure to the ON/OFF RAMS.

This capability can easily accomplish horizontal rules, vertical rules, or reverse video blocks. Overlapping reverse video areas will re-reverse where they overlap.

Borders can be made by defining one area just within a second one. Underlining can similarly be defined by two RV areas. However, care must be used in specifying overlapping RV's. For example, horizontal rules that cross vertical rules will have white intersections, and RV underlining that intersects characters will be white in that intersection.

6.12 Line Rule: Line ruling commands, defined in a similar manner as reverse video commands, have been provided so that rules can be accomplished that do not reverse in video when they cross characters or other rules.

Bits are set in the appropriate ON/OFF buffers, but no special hardware has been incorporated in the HWP for this facility. The 8+300 outputs on XPOS value to the HWP that corresponds to the appropriate end of the rule (LRX HIGH, LRX LOW), and output a zero dx and a zero XN with proper boundary bits set to accomplish the function.

6.13 Wrong Reading: To accomplish wrong reading, XPOS is obtained by subtracting the input XPOS from the page width, in order to obtain the distance from the right edge of the page. The sign of each XN and its L and B bits are also complemented to effectively wrong read each character. These tasks are all done during Data RAM building by the DMS (Z80A) and are transparent to the OCS, which will output the data in the usual way for the normal left-to-right laser beam stroking.

7.1 General: The hardwired processor (HWP) receives data and controls from the Outline Converter Subsystem (OCS) and from the Laser Recorder or "Phototypesetting Unit" (PTU) and sends a "video signal" as well as other control signals to the PTU. These functions are carried out by executing the following general operations:

(1) Using input data and control information from the OCS, the on/off points and reverse video points are computed, and "one" bits are placed into an "ON" register, an "OFF" register or a "Reverse Video" register, all of which are referred to herein as "output raster line buffers".

(2) Using an input clock and controls from the PTU, a video NRZ control signal is generated which has transitions dependent on the "ones" stored in the output raster line buffers.

(3) Using internally generated flags, the input and output average data rates are adjusted to compensate for the wide variability in duration (or quantity) of input data.

(4) Using input controls from the OCS and PTU, the raster line buffers are cleared and required control (or reply) signals are sent to the PTU, for example to trigger the PTU advance roller drive motor (element 58 in FIG. 2).

The video control signal is the beam modulation signal for the Laser Recorder. On-off transitions occur at character boundaries. Input data is based upon the characters in each "line segment" that is to be printed. Since line segments are allowed to overlap, characters may overlap. To accommodate this, a technique using an "ON" buffer and an "OFF" buffer is used. Each buffer is a memory of sufficient size to store a bit representing each element of a single scan line. The "ON" buffer stores a one bit in those elements that represent a black to white transition of a character and the "OFF" buffer does the same for the white to black transition. For cases where two characters have a transition at the same element, the next higher address element is set to one for the "ON" transition and the next lower address element is set to one for the "OFF" transition. To generate the output video signal the successive ones in the buffers are detected and those from the "ON" buffer decrement an up-down counter and those from the "OFF" buffer increment the counter. If the counter is decremented to a negative number, the video output level (low to the PTU) is such as to turn the laser beam on.

To provide for reversal of the usual black on white output a "Reverse Video" (RV) buffer of the same size as the ON/OFF buffers is used. A one bit in the RV buffer inverts the outputs video signal polarity for that element and all subsequent elements until the next one bit in the RV buffer. A switch called "Global RV" provides for reversing the video polarity for a whole page.

To facilitate rapid throughput, the output raster line buffers are duplicated so that read-in (loading) and read-out (unloading) can be performed simultaneously. The buffers are either in a mode receiving the patterns of ones representing a line or outputting to generate the video signal. Both modes are read-write operations on the memory since, for the output mode, zeros are written in to replace the ones. The buffers then can be initiated to an all zero condition by parallel ports from the OCS that perform two unload operations.

7.2 Input: The hardwired processor receives its input from three "IV" bytes. Each IV byte is 8 bits. Two of these bytes are "data bytes" (DATA HIGH, DATA LOW), and the third is a "control byte" which directs the HWP on what function to execute next. The control byte format is shown in FIG. 19A.

When the handshake bit is zero, it indicates that nothing is to be done. When the OCS wants the HWP to perform a function, it loads the data byte(s) with any required data, and then sets the control byte with the proper control, which includes the handshake bit. After the HWP accepts the requested function, it resets the handshake bit back to zero.

The control information is one of three basic types:

(a) Load Variables;

(b) Solve; and

(c) Output.

The Data Load controls are listed in Table 9. The Output controls listed in Table 10 all communicate with the PTU, except for the RVX (Control 88), which commands a reverse video limit store.

The Solve Control (87) is duplexed with the dydx data loading, and is the one which requires the HWP to solve the equation. Prior to solving, the data IV Byte must be input.

7.3 On/off Solving: The equation to be solved is:

where:

The sign of the .DELTA.x result is dependent on the result of the (Ys-Yn) subtraction and on the direction of the curve (this is loaded under control 84 in Bit 0 Data Low). The parameters used in this equation are handled by the HWP as follows:

XREG--This result is the address of the bit on the on or off buffer to be set. The buffer (on or off) is selected depending on Bit 1 Data Low under control 84, (if this B bit is on, the buffer shall be selected).

XPOS--This value is input under control 82 and is 14 bits.

Xn--This is a 10 bit integer input under control 84. Bit 6 Data High defines whether this is a + or -. The value will be expressed as a two's complement if the sign is negative.

Ys--This variable is input under control 86. In dealing with the decimal, only bits 5 to 2 in Data Low will be needed for this computation.

Yn--This is a 10 bit negative integer input under control 83 in 2's complement form. NOTE: (Ys-Yn) will always yield a number >-1 and <15 (see dydx & K descriptions).

dydx--This variable is input under control 87. This value is always used as supplied when (Ys-Yn) is positive. If the results of the subtraction are negative, dydx is used or effectually replaced by dx/dy of 16 in accordance with the flare bit status. See Table 11.

K--This value is input under control 83 and defines in Bits 5 to 2 of Data High. When (Ys-Yn) is less than zero, K is employed according to Table 11.

SLC--The slant constant is input under control 85. Only the 4 MSB of Data Low are needed. The sign in Bit 7 of Data High defines .+-.. The value will be expressed as a two's complement if the sign is negative.

SCALE--This value is loaded under control 81. The 2 LSB of Data Low are not required for accuracy. The scaling constant is equal to the set width divided by 43.2.

E--Bit 7 of Data High under control 84 will cancel the solve requirement if the result of the (YS-Yn) subtraction is negative. (E bit on cancels).

FIG. 19B summarizes the controls and the data formats. FIG. 20A is a Functional Block Diagram of the On/Off Solving section of the HWP.

7.4 Reverse Video Input: Reverse video limits are input via the XPOS input. The HWP will accept an XPOS input under control 82. If this value is an RVX point, control 88 will be input to cause the HWP to store the XPOS value in the RV buffer (see section 7.5.2).

7.5 Raster Line Data Storage Buffers: The HWP contains six data storage registers to store the data that controls the laser recorder. While the data in three of these registers is being output to the laser recorder to form one raster line, the other three are storing the result of the character outline computation.

The three registers that comprise a control set are dedicated to storing the locations where the laser recorder should turn on (the ON BUFFER), storing the locations where it should turn off (the OFF BUFFER), and storing the locations where the normal on-off conditions should be reversed (the RV BUFFER). Each buffer has a bit dedicated to each horizontal resolution element in the raster, and therefore has a minimum length of 7,954 bits.

FIG. 20B is a Functional Block Diagram of the raster line data storage buffers section of the HWP.

7.5.1 ON/OF Buffer Input: The output of the XREG adder contains the address of the bit in the On or Off Buffer that should be turned on to store the on or off function. The Boundary (B) Bit state determines whether the On or the Off Buffer is to be loaded.

Loading of the buffer must recognize whether the bit being addressed has already been loaded from a prior solution. If the bit addressed is a zero, a Read-Modify-Write cycle on the RAM may be performed. If the bit addressed is already a one, the contents of XREG must be incremented (or decremented, if the Off Buffer is being loaded), and another load must be attempted. This is repeated until a valid Read-Modify-Write is performed. This process ensures that some bit is turned on for every input requirement.

7.5.2 Reverse Video Buffer Input: The output of the XREG adder contains the address of the bit in the RV Buffer that should be inverted to store the reverse video function. A Read-Modify-Write cycle on the RAM is performed for every RV input. This will invert the bit, whether its current state is a zero or a one.

7.5.3 Buffer Output: The contents of the On and the Off Buffer is clocked out through an up-down counter with a minimum range of .+-.15. Bits in the On Buffer that are one will decrement the counter, whereas one bits in the Off Buffer will increment the counter. A negative count detection circuit controls the output to the laser recorder, the laser beam being turned on whenever the count is less than zero. (If both the On and the Off Buffer bits at any address are identical, the count does not change). The RV Buffer is clocked out at the same rate, and will reverse the normal output from the zero detector. All buffers are reset to zero after the outputting cycle.

The clocking is controlled by the "read clock" signal from the PTU. A toggle switch is provided on the HWP card (board) to enable a set-up operator to command Global Reverse Video; i.e., a black/white reversal for the entire page.

7.6 The HWP Structure: The HWP will first be described using a simplified block diagram. Then the overall timing will be given followed by a detailed explanation of the logic design.

7.6.1 Basic Sections: FIG. 21 shows the basic sections of the HWP. Data and control words are loaded into a buffer and storage section 100 under control of a "handshake" signal from the 8.times.300 in the OSC. One of the control words initiates an address computation performed in the next section 102 and this same control initiates a load into that address location performed by the last section 104. Each section has its own control and sequencing logic.

The control and PTU interaction logic 106 generates the output video signal in synchronism with a read clock from the PTU. This section also utilizes synchronizing signals from the PTU to generate flags that indicate when the data buffers are full. The conditions of three flags are used to control the input of data (through the handshake signal) and to interrupt the output of data if the quantity and duration of the input data is such that the output cannot keep up. This section also receives condition signals such as "ready for load" and "read for scan" to generate outputs such as "load command", "scan enable" and "step enable".

7.6.2 Data and Control Buffers and Storage: FIG. 22 shows the first section 100 in block form. Data and control bytes come over three 8 bit busses from the 8.times.300. Data High and Data Low are buffered to drive a set of latches. The abbreviations used for data will be explained in next section. The notation used in FIG. 22 indicates the bit range and bit weight for each piece of data transmitted. The control byte, except for the single bit called the Handshake (HS) bit, is stored directly in an 8 Bit latch 107. When the HS bit is set to one a load clock pulse is generated and supplied to the control byte latch. The output of this latch provides a three bit word and a set of four single line control signals. The 3-bit word is decoded and used to generate 4 clock pulses for selected data latches.

The control line 87 initiates a computation and an ON/OFF buffer load. The control line 88 initiates an RV buffer load. The control line 90 enables the generation of a load command to the PTU if the "Ready for Load" condition exists. The AO control signal resets the "Ready for Scan" and "Scan and Sync Enable" output conditions. The CO control signal indicates that a data buffer has been loaded and is available for unload.

7.6.3 Output Data Buffer Address Computation: The second section 102 computes the output data (raster line) buffer address that is basically given by:

where

The notation used here is slightly different than that used previously:

X.sub.R =XREG

X.sub.P =XPOS

SC=SCALE

The signs of these quantities and the result can be handled by understanding that X.sub.n, .DELTA.X, SCL and SC are two's complement numbers. All but .DELTA.X are provided from the OSC as two's complement numbers. The quantity .DELTA.X is computed in this section where dy/dx is generated by pair of ROMs 108 and 110, the second one providing the two's complement, as shown in FIG. 23. The K value is provided as a positive number and, when the quantity (dx/dx+16K) is a negative number, the logic formulates a full 16 bit two's complement number. The sign of this quantity is under control of the L bit (see FIG. 22).

There are other control modifications to the X.sub.R calculation. If the sign of Y.sub.s -Y.sub.n is negative and the Flare Bit F (FIG. 22) is one, K is forced to equal one. If Y.sub.s -Y.sub.n is negative and the E Bit is one, the X.sub.R computation is abandoned. This latter event is described as part of the control logic for this section.

Note that both the quantity .DELTA.X and the second term of X.sub.R require multiplication. This multiplication is effected by a time-shared 16 bit multiplier 112.

The operation of the section 102 will now be explained with reference to FIG. 23. The quantity (Y.sub.s -Y.sub.n) is provided by an adder 114 on the bits (0,3) (this notation meaning bits 0, 1, 2 and 3) because Y.sub.n is always supplied as a two's complement negative number. Since Y.sub.n has no fractional digits, the adder output and the bits (-4, -1) of Y.sub.s go to tri-state buffers 116 and 118 and form the multiplicand for the first multiplication. The sign for the (Y.sub.s -Y.sub.n) computation is determined by the adder and is zero for a negative result. Note that the input data provides only bits (-4, 3), the remaining bits must be provided by the sign bit. The existence of bits (-6, -5) will be clear when the multiplier is described.

The inputs dy, dx provide 8 bit addresses to the 256 word by 8 bit ROMs 108 and 110 that generate the ratios dx/dy and dx/dy (the double overline meaning two's complement). The ROMs have tri-state outputs which are wired together and they are selected by the L bit.

The circuits after the two ROMs 108 and 110 generate the quantity (dx/dy+16K) in two's complement form. The K variable goes to a muliplexer 120 which is under control of the Flare Bit F and the sign of (Y.sub.s -Y.sub.n), S.DELTA.Y. The other input of the multiplexer is the constant one, which is selected when (Y.sub.s -Y.sub.n) is negative and F is one. For a non-negative condition (L=0), the dx/dy ROM 108 is selected and the K value is added to zero and forms the weight positions (4,7) of the quantity. This quantity goes to a tristate buffer 122 and is the multiplier for the first multiplication.

If the L bit is negative, the dy/dx ROM 110 is selected. The output of the multiplexer 120 goes to an exclusive or circuit 124. For L=0, its input to the exclusive-or inverts the K value. If dy/dx is not zero, the EX-OR output is added to zero by a full adder 126 forming the two's complement number. If dy/dx is zero, the K value must be a two's complement (not just inverted). The zero condition is detected by an inverter 128 and an 8-input NOR 130 and generates a carry to the adder 126 to produce the two's complement of K.

The multiplier 112 is made up of two 8 bit multiplier chips and three, 8 bit serial shift register chips forming a shift register 132. For each shift clock a 1 bit by 16 bit product is formed. Thirty two shifts generate a 32 bit product (which is located in the shift register), but since the full product is not required, only 24 bits are retained. Since operations are performed least significant bit first, the lowest 8 bit positions are lost. The multiplicand (Y.sub.S -Y.sub.n) and the multiplier (dx/dx+16 K) are loaded to have the binary point between weight positions 5 and 6 of the multiplier and shift register. The first product then retains weights (-4, 11) and is located in the two lower order shift register chips. The positions (12,19) are not used. The first product is ANDed to by an adder 134 and the result is added to SLC by an adder 136. This result goes to a tri-state buffer 138 to provide the multiplicand for the second product. Note that the tri-state buffers and the tri-state ROMs have output controls provided by the appropriate polarity of an M signal.

The second product provides:

The scale constant SC comes from a tri-state latch in the storage section 100 (FIG. 22) which has an output controlled by the Reverse Video control signal 88 and a pulse SP that occurs at multiplier load time. If the Reverse Video control signal 88 is present, the SC latch is held in an open circuit state and the shift register product P is forced to zero.

For the second product the binary point for the multiplicand is between weight positions 3 and 4 and the multiplier binary point is between weight positions 11 and 12. The product weights are (-8, 15) but only the integer values are required which reside in the upper two (MSB) shift register positions.

The product P1 is added to X.sub.P by a full adder 140 to form X.sub.R, the data register address. Note that for RV control 88 present, P1 is zero and X.sub.P becomes the data register address.

7.6.4 Output Data Buffers and Video Logic:

FIG. 24 is a block diagram of the third section 104. The register address goes to a multiplexer 142 that is controlled by the most significant bit of X, the sign bit. The register address as calculated is usually a positive number whose MSB is of weight 12, which provides 8192 positions, which are more than the number of elements on a scan line. If the address calculation results in a negative number, which can happen at the beginning of the scan, the sign bit of X.sub.R (S.sub.XR) is used to switch the multiplexer and load a one into address 32. If the sign bit is not one but bits 13 or 14 are one, the load operation is abandoned.

The output of the multiplexer 142 is parallel loaded into an up-down counter 144. The counter output goes to multiplexers 146 and 148 that select which of the two output data buffers are to be used for loading. The other buffer address is selected for unloading and this comes from a counter 150 that is operated from an output clock in the PTU.

The address multiplexer 146 or 148 is selected by the Buffer Select (BUSE) signal. When this signal is low, Buffer One is selected for load and Buffer Two for unload. Considering first the load operation, the controls to the memory chips are the Chip Enable (CE) and the Write Enable WE. The CE signal is gated by control logic (not shown) that selects a particular chip in the ON, OFF, or RV buffer under control of the Boundry Bit (B) for selecting either ON or OFF and the Reverse Video Control signal 88, for selecting the RV buffer. If there is a zero in the selected ON or OFF buffer location, the WE signal occurs and writes one into this location. If one is found in the ON buffer location, the WE signal is delayed until the address is incremented by a clock to the up-down input address counter 152. Note that the direction of counting is controlled by the B bit. If there is a One in the OFF buffer, the address is decremented. If the RV buffer is selected and a one is found, a zero is written, and vice versa. The input for the RV buffer is via a latch 154 and a multiplexer 156.

For the unload sequence all three buffers are enabled by the CE and WE signals. For the ON and OFF buffers zero is loaded via the BUSE signal level and for the RV buffer input, a ground level is selected by the multiplexer 156. The output level of each buffer is stored in the latch 154. Exclusive-or circuits 158 detect if the bits in the ON and OFF buffers are different and, if so, the output up-down counter 152 is enabled via a multiplexer 160 that selects the condition from the appropriate buffer. If there is a one in the OFF buffer this goes to the up-down control of the counter 152 and causes an up count. If there is a one in the RV buffer this causes a JK flip flop 162 to change state and the flip flop output is exclusive-road with the MSB of the up-down counter. The output of this exclusive-or circuit 164 goes to a D-flip flop 166 which is cloked by a Read Clock pulse from the PTU. The output of the D-flip flop 166 is exclusive-ored with a Global RV signal and this output is buffered to provide the video output signal to switch the laser beam on and off.

8.1 General: The general architecture, detailed structure as well as the operation of the Output Data Processing System (OCS) have now been described in such a manner as to enable a person skilled in the digital typesetter art to construct and use a typesetter according to the present invention. While it is believed that a computer programmer of average skill could write the necessary program for the Z80A and the 8.times.300 microprocessors, the preferred embodiments of such programs are set forth in the attached Appendix for the convenience of the reader.

While there has been described what is believed to be the preferred embodiment of the invention, those skilled in the art will recognize that various changes and modifications may be made thereto without departing from the spirit of the invention, and it is intended to claim all such embodiments as fall within the true scope of the invention. ##SPC1## ##SPC2## ##SPC3## ##SPC4## ##SPC5## ##SPC6## ##SPC7## ##SPC8## ##SPC9## ##SPC10## ##SPC11## ##SPC12## ##SPC13## ##SPC14## ##SPC15## ##SPC16## ##SPC17## ##SPC18## ##SPC19## ##SPC20## ##SPC21## ##SPC22## ##SPC23## ##SPC24##