JPS5914758B2 - character generator - Google Patents

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention This invention relates to character generation systems, and more particularly to character generation systems in which data storage requirements are minimal, and in which character display control is performed using a small number of storage units in which each character is encoded. This invention relates to a digital character generation system derived by calculations based on parameters.

Character generators have many applications, a common commercial application being phototypesetting.

Early such devices were essentially optical and used masks formed into character structures. CRT displays have been developed in which patterns can be defined by optical or digital signals. As an example, in U.S. Pat. No. 3,324,346, a flying spot scanner is optically coupled to a character representation matrix;
Derive a digital signal for the pattern. Other CRs
The T system used scanned masks for similar purposes, as in US Pat. Nos. 2,275,017 and 2,379,880.

One brute force method of character encoding is to identify each segment or dot of the dot matrix that corresponds to a character segment when the characters are superimposed on the matrix.

The dot type generation system is disclosed in U.S. Patent No. 31651.
It is described in No. 45. The fatal flaw in this attempt is that
Even medium or low quality playback requires an extremely large amount of memory. Other encoding methods developed by the prior art generally involve separating the character area into narrow bands and storing the starting coordinates and length of each narrow band.

This technique is disclosed in US Pat. No. 3,305,841. An improvement to this technique is described in U.S. Pat. No. 3,471,848, which uses an improved method of encoding successive narrow bands of terminal points. This generally helps reduce the memory required for encoded character data. An alternative, more recent prior art attempt is described in U.S. Pat. No. 3,422,419, in which controlled to specify multiple patterns consisting of multiple line segments of length,
A set of characters is analyzed. Each character Gζ is encoded as consisting of a combination of selected ones of these common patterns. Although such systems reduce memory requirements, they impose severe constraints on the glyph style and introduce some distortion in the generated characters. SUMMARY OF THE INVENTION According to the invention, all characters of all fonts to be stored in memory are encoded in relation to standardized quadrilaterals (Quads).

Generally, the quadrilateral corresponds to the largest point size character to be displayed. The valley characters are analyzed in relation to the coordinates of the quadrilaterals, especially the paired limbus.

The paired limbus contain one segment of the character between them, thus defining the boundaries of the segments. Each character is thus defined by one or more limbal pairs. The information encoded regarding the parameters of each character is
It includes the Y-coordinate of the starting point or point of each pair of annulus, as well as the slope and curvature of such annulus (variable directional parameters). In a quadrilateral, the X coordinate space or bit position along the X coordinate is defined as a unit value. Therefore, all slopes are defined by incremental changes in the Y-coordinate of the limbus for successive X-coordinate positions. The curvature is encoded as a predetermined radius of curvature that corresponds to the character limbus. Such curvature determines a slope that changes continuously and incrementally. Additionally, the rate of increase in slope varies. Thus, the predetermined curvature defines a successive incremental change in slope in each change period, and the increase in slope determines a successive incremental change in Y coordinate position. From this encoded and stored parameter information, a character is generated according to successive calculation cycles corresponding to successive X coordinate positions.

The sequence of consecutive calculation cycles is a function of time depending on the number of calculations to be performed, which in turn depends on the number of limbus to be processed. The final character display is achieved as a function of blanking or unblanking of the scanning beam with successive horizontally spaced vertical amplitudes. Each limbus pair allows the scanning beam to be scanned unblanked through the vertical excursion of the limbus pair. The horizontal scale factor is for correlating the stroke function at the desired stroke density with the calculated Y coordinate change value. Y coordinate change values are generated in successive calculation cycles as a function of stroke density and required display point size. DETAILED DESCRIPTION OF THE INVENTION SUMMARY DESCRIPTION Each character that may typically be displayed is generally represented by a font (FOn).
t) or one of a set of characters called a font style.

Each such font style must be able to be used for display in any size range. Typical terminology in typography relates character size to approximately 1/72 inch points as the basic type unit. Therefore, a 9 point character is 9 x 1/
Defined within a 72 inch = 1/8 inch quadrilateral. Similarly, a 72 point character has a 1 inch square. In this invention, each character is encoded for storage according to a standardized quadrilateral that is common to all characters of all fonts.

This quadrilateral is assigned, for example, a coordinate system with 1024 coordinates or bit positions in the X direction and 1024 coordinates or bit positions in the Y direction. As explained in detail below, the present invention requires minimal storage or memory for encoded character data.

In particular, each character is encoded with respect to certain parameters regarding the limbus of the character valleys that are correlated to the normalized quadrilateral. Limbs are therefore associated pairwise and are essentially the boundaries of each character segment. Although a standardized quadrilateral is assumed as the basis for character encoding, the system is not inherently constrained to a predetermined quadrilateral structure in the sense of the typical square of typography. . Conversely, the horizontal dimension or width of the quadrilateral may effectively vary according to the width of the character. The encoded and stored parameters for each character include the Y coordinate for the starting point of each loop, the slope and curvature of those loops.

With respect to quadrilaterals, each bit position of the X coordinate of the quadrilateral surrounding the character is considered as a unit space. Also, the calculation is
It is executed with respect to the stored parameters for successive X bit positions of the quadrilateral surrounding the character, ie, a calculation cycle, to calculate the Y coordinate of each limb at that bit position. The display of each character takes place on a cathode ray tube (CRT) display screen, which is of high resolution both in terms of the quality of the luminescent screen and the control sensitivity of the scanning electron beam.

Since the scanning beam is displaced by a vertical stroke,
Depending on the calculated Y coordinate of each successive pair of limbus, the beam is alternately blanked or unblanked to fill in that vertical portion of the character segment between the limbal pair. The character generator of the present invention can also be applied to display CRTs of any desired scanning density. For example, a display CRT may have a display line that is 11 inches across. Additionally, a constant increment for the successive excursions of the vertical scan over this maximum width is set, as an example, 214 positions or A bit position may also be provided. Typically, the scan density or resolution is adjustable, up to 1/1488 inch or 1/744 inch (in other words, one scan or stroke for each bit position or all other bit positions). can be selected. In a high-resolution display CRT of the type to be used with this character generator, the scanning CR
The T-beam spot size is very precisely controlled.
In this example, a spot size of 0.0015 inches is used. The overlap between adjacent strokes is 1/
A scan density of 744 inches or stroke deviation is achieved. Important points to recognize lζ In this invention,
Although the bit position or memory of the quadrilateral is independent of the scanning raster, and therefore also in character encoding, clearly the two must be correlated to achieve the display function. . In particular, it is encoded for the maximum point size of the display within the character & circle normalized quadrilateral. Horizontal and vertical scale factors are scan CR
It is introduced to transform the coordinate data calculated for the control of the T-beam according to the desired point size of the display. Therefore, just one set of character data encoded for any font is sufficient to display all characters of that font at any point size within the full range of available point sizes. . The minicomputer receives human data specifying the font style and display dimensions as well as the specific data to be displayed and positions the scanning beam at the appropriate line and desired character space location. The number of character arrangement lines displayed on the CRT is selected by paying attention to the font size to be displayed under the control of a computer.

In summary, every character of each font is encoded in relation to a standardized quadrilateral, and the data needed to reproduce the character is the initial coordinates of the character within the quadrilateral, i.e. the starting position of the character limbus, It includes changing parameters such as the direction and curvature of the straight lines and curves that make up the annulus.

The generation of the character limbus occurs simultaneously with the display of the character according to calculations performed sequentially in time during the stroke of the CRT display beam. However, a 1:1 correspondence between calculation period and stroke period is not required. Furthermore, regardless of the point size of the character to be generated, the same number of calculations to define the character limbus are performed, but the Y coordinate, which is the output for controlling the unblanking of the scanning beam, is Generated in relation to the scale factor. The horizontal scale factor is the number of calculation cycles, CR
Relate to the desired point size and stroke density of the T-beam. Scale: Calculations and Actual Display As an example, in a system with a maximum 72-point dimension display, assuming that all characters are encoded (for that dimension) and that the standardized quadrilateral is 210 bits (1024 bits): The scale factor is calculated as follows.

It will be appreciated that the above calculation yields a fractional value.

These can be expressed as binary equivalent values, and in fact were developed in this way for processing by this system. The actual number of strokes per character is related to the calculation cycle in the expression below.

However, the integer part of the stroke can be obtained by rounding down the fractional part of the answer. To facilitate understanding of the above,
Please refer to FIG.

FIG. 1 shows, on an enlarged scale (see FIG. 3A), a 72 point serif display produced by a scanning CRT with a stroke density of 1024 strokes per inch. FIG. 1 takes the stroke density, for illustrative purposes, to be the number of strokes per inch equal to the number of bit positions in the standardized quadrilateral. Furthermore, Figure 1 shows that
For maximum 72 point size character displays, the calculation of the limbus is related to the initial encoding of the character by 1.1. From the above equations (1) and (2), HSF-VS
It becomes F-1. formula? ? ? Therefore, the number of strokes per character is equal to the number of calculation cycles. In Figure 1, the initial coordinates are X-200, Y-75 for the lower annulus.
0, X-200, Y-800 for the upper limbus.
Given this initial information, the CRT beam immediately
- the first line is scanned at the 200 position, the beam is initially blanked and then the Y-75
It is unblanked with 0 and blanked again with Y-800. The vertical position of the beam during the stroke is determined by counting the pulses of an 8 MHz clock, which identifies the actual physical position of the beam by equation (2) for any slope rate. do. For a given stroke period, the system reserves the character limbus position for the next stroke.

Since HSF-1 in FIG. 1, a calculation is performed for each successive horizontal bit position and a stroke is performed for each bit position. As will be explained in more detail below, character encoding such as that of FIG. 1 identifies serifs as unchanged in the superior and inferior limbus from calculation cycles 200 to 250. Similar blanking and unblanking of the beam is therefore performed for a period of 50 calculation cycles. However, in cycle 250, a change occurs in the lower annulus consisting of a downward curvature that continues in a more or less regular manner until calculation cycle 300. As shown in FIG. 1, the curve is approximated by successive increasing steps, and the Y-coordinate of the lower annulus decreases in successive steps along the X-axis for a predetermined number of calculation cycles. For example, the first change in the low Y coordinate is calculated from calculation cycles 256 to 260 (
5 cycles), another change occurs from cycles 261 to 264 (4 cycles), and various other Q changes occur. FIG. 2 shows a more typical situation in which the stroke density does not correspond 1:1 to the bit positions of the standardized quadrilateral, but instead a stroke density of 744 points per inch is illustrated.

Additionally, a 4 point serif is shown, which is 1/18th the size of the 72 point serif of FIG. From equation (2), as shown in Figure 2, VSF=1
It is 8. Since the serifs are encoded on a standardized quadrilateral basis, the serifs are shown in FIG. 2 as having the same dimensions as in FIG. However, Figure 1 shows the serifs at 62.5 times the actual display size of 72 point characters, while the display size of the 4 point serifs in Figure 2 is suggested by the illustration above Figure 2. . The difference in scale can be appreciated by comparing the 0.0015 inch diameter CRT spot shown in FIG. 1 with a similar spot shown in FIG. 2 for a 4 point serif.
The beam position during each vertical stroke is identified by an 8 MHz clock, but instead of each clock pulse counting as 1 as in FIG. 1, each clock pulse causes the counter to count up by 18.

Therefore, the tilt rate of the scanning beam remains constant.
From equation (1), as shown in Figure 2, HSF=
It becomes 24.774.

This means that one vertical stroke is performed during each calculation cycle of 24.77 times. Since all the numbers or integers in the computation cycle must be related to a single stroke, therefore, in order to embody the above, we can vary the integer number in the computation cycle for successive strokes and calculate the average value. HSF-24.77
A specific circuit to achieve 4 is provided and disclosed below. With reference to the illustration of FIG. 2, it will be appreciated that five strokes 10-14 are performed by the CRT to display the serif portion of the character shown.

FIG. 2 also shows the trajectory of the spot in dotted lines, and the bold lines in FIG. 2 represent the calculation cycles in which the actual strokes are performed. Of course, the resolution of the text will drop considerably, but
Given its reduced dimensions, the displayed characters are of high quality,
In other words, it is descriptive. Also, once a stroke begins, the system calculates the character loops, or scanning beam blanking and unblanking positions, for the next stroke, requiring multiple calculation cycles.
Encoding Character Data In this section, the basic techniques of character encoding are considered in more detail.

In Figure 3A, a "J" is shown in large block letters, and the third
Figure B shows the associated table of instructions for generating this character. The same letters are shown in Figure 3C to illustrate the letter loops. FIG. 3A includes boxed areas corresponding to the lines of FIGS. 1 and 2 for correlation. The characters have an initial X. of 0.400. It is shown as having a Y coordinate and occupying 500 calculation cycles. As shown in Figure 3A, the letter is completely C
The paired rings to be written by RT strokes are marked YIJ in FIG. 3C to indicate the boundaries of these writing areas. Start of character (
In X-0), the first pair of rings 1, 2 is defined and in cycle 200 a new pair 3, 4 is created.
As shown in FIG. 3C, angles θ, which are within ±87.2 degrees, are measured with respect to the horizontal. The line defining the angle is a tangent to the curved boundary or ring of the character, as will be explained in more detail below. Every letter always contains at least one pair of loops at their initial starting point, in the case of the illustrated letter "J", loops 1, 2 having a common initial Y coordinate 400.

Since recognition of the slope of limbus 1 is essential in the first calculation cycle, the starting limbus pair (BOL
P) The command must also express the need for this additional information. This means that on the final side of Figure 3B,
0 "Computation cycle to next instruction". At the starting point, the limbus has a vertical downward direction. Ring 1 is also encoded as having a curvature change command (CK) of curvature value +1/200 at the starting point. This CK instruction is also encoded into the next instruction for 100 calculation cycles. For a suitable horizontal scale factor, the strokes and calculations are performed in a calculation cycle (C.
C. ) Progress to 100. The CK instructions for changing the slope and the CK instructions for changing the curvature for the annulus 2 are then given and these are propagated for the duration of the following 100 computational cycles.

Changes in command for limbus 1 are not coded, so limbs 1 and 2 are each defined in a curved structure as shown until calculation cycle 200. C. C. At 2OO, paired hoops 3, 4 are created according to another BOLP instruction encoded for each Y coordinate 750, 800. C. C. At 25O, there is a curvature change command CK for limbus 3 encoded for curve K-1/50, which lasts for a period of 50 calculation cycles. limbus 2, 3
It should be noted that the initial portion of and the entire annulus 4 are horizontal, and that no CM instructions are needed to specify the initial slope for these annuli. It should also be noted that there is no need for a CM command prior to the CK command for limbus 3 since there is no change in tangent at the point where the curve begins. Rings 2, 3 cease to exist in cycle 300. This is the end of the limbus (EOL)
P) established by command. 100 calculation cycles are encoded in the EOLP instruction.

As will be detailed, the paired limbus that continue to define the rest of the character are now limbus 1 and 4. Therefore, unblanking continues between loops 1 and 4, with loop 4 remaining straight and loop 1 continuing to follow the curvature of the previous CK command. In cycle 400, limbus 1 is further defined by a special type of BOLP instruction, and in cycle 400,
The value of approximately Y=400 is directly reproduced to the value of Y-700.

The BOLP command, described below, is generally used at some intermediate X position to alleviate character limbus discontinuities. Ring part 1 is also θ1+87.2th, 0C. C. CM command of and CK of curvature K-1/50
The CK instruction is encoded in the 50C. C. The period is encoded. In cycle 450, the CM instruction sets the slope change to 50C. C. encoded into θ-0M for
Limb 1 progresses as a horizontal line during the next 50 calculation cycles. Finally, the end character (EC) instruction is
It is used to confirm the completion of calculations for character display control upon completion of character encoding. FIG. 4 represents the instruction format used in an actual operating system.

Instructions are based on 16-bit words and are compatible with BOLP and C
Each DY consists of two words. The seven commands shown are the total commands for generating any character. With the exception of NOP and EC, each instruction identifies the operation to be performed, the parameters to be controlled, the particular limbus involved, and the number of computational cycles until the next instruction. The operation code in bit positions 1-5 identifies the specific instruction and
Each instruction except instructions contains a number of bits to encode a computational cycle.

The 4 bits for this purpose are CDY, . CM,. The CK and EOLP instructions, while the NOP instruction contains an 8-bit position for this purpose. Thus, with 4 bits it is possible to encode O-15 computation cycles into the next instruction, while a NOP instruction
This allows encoding of 255 deliberation cycles. Therefore, NOPs are useful when character extensions are to be generated without the need for any limbal information. B
The two words of the OLP instruction are the subscript S. for the lower and upper Y coordinate values. For a given one, as specified by L
Sequentially associated with paired superior and inferior limbus.

Even if Y is a 16-bit number in calculation, the Y value is encoded with 9 bits (bit 8 - bit 16).
The most significant bits 10 identify the integer value of Y as part of 1024, and the lower 6 bits are the fractional bits needed to approximate the fractional or non-integer part of the slope value. However, B
Control of beam position based on OLP commands is limited to convenient 9-bit values, allowing the initial Y coordinate to be set to any of the 512 even values out of 1024 bit positions on the Y coordinate axis.
The BOLP instruction also uses 4 bits J1-J4 in each word.
, which adorns the particular limbus to which the command relates. This allows for a total of 16 loops and thus 8 pairs of loops along the vertical portion of the character. The remaining instructions are each encoded with bits J1-J4 identifying the limbus to be modified based on them. The CDY command is used to define a rim, which is a straight slope line. As mentioned above, curves and slopes are achieved by incremental changes in the Y coordinate value. These increment values are encoded in the 14 bits ΔY1 - ΔYl4 plus the sign bit of the CDY instruction, with negative ΔY values defined as one's complement numbers. As a result, +255.63/64 to -255.
The slope range of the straight line of the character ring up to 63/64 and 1/
64 resolutions are provided. The provision of the CDY instruction is a compromise in terms of efficient use of memory capacity for slope changes and curvatures, which will be discussed later. In general, if the generation of curved gradient lines is performed by calculations based on gradient change commands (CM), it is possible to generate accurate straight lines as required in high quality displays such as printed art. This requires extra storage of gradient information. On the other hand, if only a reasonable number of slope changes are encoded, then the occurrence of a long straight slope line CM command (as "M" or "Kan") will result in an uneven, stepped line structure; This results in unacceptable quality. Therefore, it includes a gradient change command (CM) & ζ6 bits of gradient information (26=64 gradients are possible, which is considered a reasonable number). The curve change command (CK) includes 7 bits of slope information, K7
is a sign bit and gives ±26 to ±64 curvatures. CM and CK are encoded with limb numbers J1-J4, respectively. The limbal termination order (EOLP) is
Requires encoding to identify the limbus number of limbus simply by the lower Y coordinate. The limbus always starts in pairs;
The term "pair" appears in the EOLP. An end character command (EC) is encoded and recognized by the system as the end of the character generated. Illustrative Examples of Calculations To facilitate the provision of the present invention, illustrative examples are described utilizing a significantly reduced level of data and parameters to simplify the description of the calculations performed.

Accordingly, a quadrilateral of 64 units is shown in FIG. 5 in each of the X and Y coordinate directions. Consistent with the more complex illustration described above, a basic line is established within the quadrilateral at Y=21. This is a letter such as a "Q" or a P. with a portion extending below the base line to be included in the quadrilateral. q. Provided for many lowercase letters such as y. The X coordinate is defined as a 6 bit binary number and the Y coordinate is defined as a 12 bit binary number. Therefore, the points in the quadrilateral are X-6, Y=7.3
3/64 (note that 26 bits define the integer value of Y and 26 therefore identifies a fraction expressed as 1/64). In fact, Y is resolved to this precision, and the coordinate that ultimately controls the unblanking and blanking of the scanning beam is only a 6-bit number, defining only the integer or total value of the Y coordinate. The need for fractional Y values arises in the encoding and calculation of slopes and curves, as will become clear from below.
Gradient In FIG. 6, a diagram of an "encodable" gradient is shown.

This diagram shows rays with gradients M=0 to M=7, and the adjacent coordinate diagram relates the angle θ to the X coordinate. As a result, since the X increment ΔX corresponds to successive calculation cycles or bit positions and can be taken as a unit value by definition, ΔX−
It can be said that it is 1.

Therefore, from the description of the stop notes, it can be seen that slope and curvature are achieved in incrementally changing the Y coordinate of successive strokes. These changes are therefore of the value ΔY. It can be seen how these changes are related to the slope function M related to the angle θ. It will also be appreciated that to achieve the desired slope of the character ring, ΔY must be adjusted according to equation (5). The slopes M-0 to M-7 are converted into 3-bit binary numbers as shown in the table of FIG.
The angular value of , the value of Tanθ and the calculated value of ΔY(M) are also shown.

FIG. 8 is a schematic geometry to illustrate the "update" operation of Y as a function of the gradient value M.

FIG. 9 is a summary list of encoding instructions corresponding to the occurrence of the structure of FIG. The BOLP command executes Y for the pair of limbus 0 and 1, as shown in FIG. =2, Y1 =
24 initial coordinate information. The X coordinate is arbitrarily set to X=4
Selected as θ with respect to limbus 0. =22.5, and θ1=-45 for limbus 1, and from FIGS. 6 and 7, these are encoded as parameter information in the corresponding CM command for convenience.M1=1, MO-5 corresponds to The system performs the ΔY value calculation or Y update function and generates the indicated gradient rings 0, 1 until the EC command indicates the end of the character. In FIG. 9, the ΔX column corresponds to the number of calculation cycles in FIG. 4, and is 14 in the case of a CM instruction for limbus 1.

It should be noted that the letters in FIG. 8 extend from X-4 to x=18, so ΔX is 14.
Therefore, 14 vertical scans (assuming a 1:1 ratio between scans and calculation cycles) are required to display the characters in Figure 8, plus 13 vertical scans in limbus 0 and 1. The ``
"Y update" is required.

The equation below shows the update process and is defined as ΔX-1 and j number of flowers for each update. Furthermore, since θo-22.5 and θ1-45, from the general description of hole 6), a specific update function can be described for limbus 0 and 1 as below. .

The table in FIG. 10 shows X for continuous values of X. (X), Y1
The calculated coordinate values of (X) are shown. By further considering from the corresponding CRT display diagram of FIG. 11 that only integer values of Y control the unblanking of the scanning beam, limbus 0 is changed in successive groups of Y coordinate values. It will be understood that steps are taken. However, from FIG. 10, fractional Y values are accumulated and finally Y. It will be appreciated that this affects the integer value of (X). The reason for ignoring the fractional value of Y1(X) is, for example, when a 6-bit D/A converter
The point is that it is used to generate an unblanking function associated with the sweeping of the T-scan beam. For convenience, the formula (5P calculation is performed by read-only memory (ROM),
In the illustrated system, the memory (ROM) has a 32-bit capacity with a bit pattern "programmed" to provide four 8-bit words, as shown in FIG. Figure 12 corresponds to the binary code value now, N1 is the integer part of △Y, and N2 is the numerator value of the fractional part of 1/64 of △Y, which can be expressed as follows ~ To the ROM in Figure 12 It should be noted that the inputs of are M1+, M2+, which correspond to the values internally decoded for the ROM-J)M inputs according to the logical [exclusive sum] formula.

The ROM bit patterns for different values of M+ in binary digits are shown in the table of FIG. 13A, and the corresponding values for base 10 are shown in FIG. 13B.

It can be seen from the above that the ROM is addressed with only 2 bits and this can be done in view of equation (A5) by responding to the third bit M3 to control the sign of ΔY. be done. As understood from the above, the value of Y is updated by addition or subtraction of the ΔY value (the subtraction is performed by one's complement techniques).

A complete representation of the M address function is shown in the logic table of FIG. 3C. A schematic diagram of the Y update mechanism is shown in FIG.

The 12-bit value, initially derived from a data source (such as a BOLP instruction), or consisting of the current computed Y value as described below, becomes the input to the 12-bit adder 10. The 3-bit word defining the M slope value as derived from the CM instruction is applied to ΔYROMl3 via appropriate gates 12a, 12b and is shown in FIGS. 12 and 13.
In the manner shown in Figure A, ΔYROMl3 is addressed and the stored ΔY value from ROMl3 is provided to one complement gate 14 (also responsive to the M3 value). gate 1
4 is an addition or subtraction to the currently supplied Y value (
ΔY is supplied to a 12-bit adder 10 for one-complement function). The resulting sum Y+ΔY is supplied to a 12-bit latch 15, which stores the obtained Y+ΔY in the Y coordinate RAM as the new Y coordinate to be used in the next 0 display scan. This new Y value is a different Y coordinate R
It can be supplied to the OM to access and control the scanning beam. The current Y coordinate value will be used in the next Y update operation, so 1
Returned to 2-bit adder 10. Curvature Next, we consider the encoding of curvature.

In FIG. 15, a system of four radii of curvature in two polarities of curvature, labeled K-0 to K-7, is shown. These curvatures in 3-bit binary numbers are 4
encoded as one of two radii (2 bits) and one of two polarities (1 bit).

Therefore, Kl and K2 are R. define the desired radius of
The third bit K. defines the sign of curvature. RO=32
The table of FIG. 16 relates the curvature K to its binary representation and to the desired value of RO. 1st
Figure 7 relates the above to the occurrence of curvature.

The rays M-0-M7 correspond to those of FIG. 6 and indicate the direction of inclination that can be approximated by successive Y updates when M=constant. However, although a straight line slope is generated by a constant M, if the limbus slope changes as the value of M changes, it curves the limbus. In FIG. 17, the generated shape is at each position X. - A change in slope occurs at X7 16
It is a polygon with angles. Therefore, in this system, the M value is updated to approximate the curve, ie, the circle in FIG. Also, in Fig. 17, the increase M, that is, (M→M
+1) creates a positive curvature, while the decrease M, i.e. (M-+
M-1) has been shown to produce negative curvature. If the curvature parameter K should represent the number of Y updates for every M update, a curved line would be obtained and would not approximate a circular arc.

For example, the fourth update of K-4 and Y (i.e., Y
-+Y+△Y) and (M-+M+1), then the 18th
A continuous straight line segment as shown in the figure is obtained, which does not coincide with the circular arc indicated by the dotted line. To approximate the polygon by the desired circular arc, the system introduces a new parameter S that is a function of M for a constant curvature K and that allows updates of M after a variable number of Y updates.

This is shown in the table of FIG. However, the number of Y updates to be performed at a predetermined M before SN-M is updated S = the number of Y updates after the last M update K-6 (corresponds to Rc-32) Table in Figure 19 , the straight line segment is R in Figure 20.
defined by matching the arc of c-32.

By the coordinates shown and the schematic example considered here initially Y-0, . For M-4, the sequence is as follows from the table in FIG. Y-+Y+0(6
times, when M-4, M → 5) Y-+Y+27/64
12 times, when M=5, M→6 becomes Y-+Y+1 (9 times, when M-6, M→7 becomes) Y-+Y+2・27/6
4 (5 times) These update functions are fully described in the table of Figure 21, and the calculated values listed here create a continuous approximation to the desired curve of radius Rc-32 as in Figure 22. A Y coordinate is generated that changes incrementally.

In the M update operation, the configuration for performing calculations when increasing or decreasing M is programmed into the 64-bit ROM (6 binary inputs correspond to 26-64) shown in FIG. One output δ is derived from this.

This ROM is parallel to the Y update ROM of FIG. Referring to the corresponding ROM bit pattern table in FIG. 24, if it is S-SN and S is set to 0001 at this time, then δ=1 and then as long as δ-0, S will only be 0001. will be increased. From the ROM bit pattern table in Fig. 24, in each successive operation, the accumulated value of S is
Values of N-6, 9, 1, 2, 5 (M+-011, respectively
2 and 3), it becomes δ-1.

It should be noted that ΣSN-6+9+12+5-32, that is, the desired Rc. Yumasa's curvature R in the above explanation.

We have only explained the occurrence of -32. Generation of different curvatures is accomplished by programming the ROM to the desired curvature. This requires a large amount of memory and is expensive and undesirable. Another preferred approach is ΔM, ΔS and S.

(S offset value when S=SN) is changed as a function of curvature K. The table in Figure 25 shows the values △M,
ΔS and S. represent various values for such changes in ΔM, ΔS and S. As can be understood by the unit value of R. =32 is the basic radius. In FIG. 26, a block diagram of a system embodying the curve generation function described above is shown.

The curvature value K from the CK instruction is supplied to the decoding logic 16, which also inputs the ROM17 (23rd RO
δ output corresponding to M) is received. The logic device 16 outputs an output △ defined as a function of the value K as shown in the table of FIG.
M, ΔS, and the outputs ΔM, ΔS are supplied to adders 18, 19 with corresponding M and S values, respectively, and the adders 18, 19 provide updated values M+ΔM and s+Δ S
Output. Of course, the update operation of the circuit in FIG. 26 is δ
If it is -0, it is assumed that ΔM-0. Updated value M+△M
and S+ΔS are inputs to the ROM in FIG. SUMMARY In summary, the above discussion first allows a character to be encoded as a function of a finite number of initial parameters defining one or more pairs of limbus, as well as the slope and curvature of the limbus of an arbitrary structure. The basics have been disclosed.

Also disclosed are instruction words that are stored in memory and based on which various calculations are performed to generate desired characters of any font in memory. Additionally, a schematic block diagram depicts the hardware required to perform the calculations to generate the Y coordinate values that provide control signals to the scanning beam in reproducing each character on the CRT.
It will be appreciated that the storage requirements of this system are minimal compared to the prior art, further providing an extremely flexible and efficient system. A desired font encoded as described above can be stored in memory and the character reproduced in the desired font size. This system can be easily adapted to desired CRT scanning densities according to scale factors. The calculations are performed concurrently with the CRT scan and are typically completed well in advance of each scan stroke due to processing speed. Therefore, speed constraints are essentially only imposed on the scanning deflection circuit itself, so high-speed operation can be easily achieved. Of course, it will be understood that displays other than CRT can also be used. Detailed Program Diagram Before beginning the description in this section, it should be noted that although the character generator of the present invention is merely a part of the entire character generation display system and therefore has a time function etc. that is on par with the generator, it should be noted that the CRT The scanning electronics are not an integral part of the generator itself.

Further, the entire sequence of operations and peer relationships are controlled by a minicomputer, and a commercially available minicomputer is sufficient for implementing the present invention. Therefore, these and other elements of the overall system are not shown. However, their communication is shown in the following block diagram. In these block diagrams appears the name PACT, which is an acronym for Limbal Algorithm Computing Technology and is a term that adequately characterizes the character generator of this invention. Mini computers use magnetic tape, punch cards,
Receives input to be generated from tape or the like in a suitable encoded form compatible with the computer.

Data input to the computer specifies the font to be displayed and the point size of the display. By means of a suitable memory or direct data input, the computer obtains the necessary information regarding the legality of the characters in the line representation, character spacing, line spacing and other information.

Figure 27 shows all major subsystems and character generator inputs and outputs.

FIG. 2 is a schematic system block diagram. The computer interface issues IORESET to the CPU 26, and this signal starts the entire system via the CPU 26. Computer interface 20 provides strobes DS64, DS65 and introduces vertical and horizontal scale factors (16 bit words shown as data 0-data 15) to scale, stroke and video control unit (SSVCU) 22. The computer also receives data from disk or other mass memory consisting of instruction words of FIG. 4 relating to a predetermined font to be displayed, which is stored in PACT buffer 24.
Buffer 24 provides high speed access for mathematical and computational functions. A PACT start signal supplied to CPU 26 initiates calculations for the display of each character and is issued by PACT buffer 24 under computer control. 1 expressed as PACTO-PACTl5
The 6-bit instruction word is sent to the PA in response to a PACTREQ, issued by the CPU 26.
The data is supplied from the CT buffer 24 to a central processing unit (CPU) 26 and a computational storage unit (VCSU) 28, respectively. The basic system clock at 2MHz is scanned by the scanning electronics 3.
It is derived from the original oscillator in 0 and generally supplied to (CPU) 26, from which clocks are issued to other operating systems.

(SSVCU) device 22 provides the SVS signal to scanning electronics 30, which operates to initiate and terminate each vertical stroke of the CRT scanning beam.
The SVS is also supplied to the CPU and indicates the stroke currently in progress, allowing the system to equate, for example, fast calculations with slow stroke periods in successive calculation cycles. When a stroke is not in progress and SSVCU 22 performs certain storage and transfer functions as described below to initiate a stroke, CPU 26 directs SSVCU 22 to P
Issue FTF. Additionally, PFTF requires that the calculations for the next stroke are finished. Therefore, the configuration requires a large amount of calculations to be performed and the previous stroke ends before the calculations for the next stroke are finished, in other words the CRT scan has to be reversed. It is also given in the situation of In reality, this situation almost never occurs. One of the transfer functions is to transfer the calculated Y change coordinate values RY7-RYl6 from CSU 28 to SSVCU 22.

A 10-bit word is thus transferred for each limb of this scan.

In this regard, the CPU 26 sends J1-J4 to the CSU 28.
and calculate the parameters of the next limbus (J1-J
(for each of the two or more limbus identified in 4)
Each ring being scanned is identified along with PFF, which is a command CSU 28 for Calculations by the CSU 28 are blocked and the BOLP. D.Y. K. C.K.
Certain limbal parameters such as those shown in CM are
is supplied to CSU 28 by. Clear and clock control is also provided by the CPU. As explained above, the output RY>B from the CSU 28 to the CPU 26
Y is the period limbus number J1-J4 of the BOLP instruction.
Ordering by increasing value of coordinates, from CSU28 to SSV
Operates to order transfers RY7-RYl6 to CU22. When ordered in this way in the temporary memory of SSVCU 22, a very simple blanking/unblanking operation is performed. As mentioned above, 8MHz
A gated clock signal is generated by the scanning electronics 30, and the gating function generates this signal when a vertical stroke is initiated. The 8 MHz clock drives a counter in the (SSVCU) which operates to identify the physical position of the beam in its vertical stroke relative to the known slope function of the scanned beam.

This count is scaled by VSF, which
An unblank signal is generated when corresponding to the Y coordinate in the SVCU's temporary memory. The unblank signal is applied to a high voltage video coupler 29 which controls the unblanking of the VRT scanning beam and the blanking and thus coloring of the characters on each vertical stroke. The beam is normally blanked and unblanked when the beam position reaches the first Y coordinate value and thus corresponds to the lowest limbus of the character. A very simple unblanking and blanking operation is achieved through the use of paired annuluses, with each successive Y coordinate value changing the blanking/unblanking state from its current state to its opposite state. Simple symbols consist of other signals.

The "S" counter in SSVCu is set to the horizontal scale factor and is counted down by one in each calculation cycle by the PFFT from CPU 26. When the count equals O, a SXZ signal is issued to CPU 26. CPU 26 stops calculating after the current CRT scan is completed, transfers the Y coordinate RY7RYl6 to SSVCU 22, and requires SXZ (relative to S-0) to begin the stroke. The S counter function thus serves to relate strokes to calculation cycles according to the HSF. CPU 26 includes an I counter that receives a 2 MHz clock from scanning electronics 30.

The number of calculation cycles of the instruction to be processed is introduced into the I counter. This is similar to the case of S counter, PFFT
Under the control of , the 2 MHz clock decrements the count by one. Therefore, (CPU), (CSU
) and (SSVCU) consist of the main functional blocks of the PACT processor.

Furthermore, (CPU) and (CSU) are (SSVCU)
The Y coordinate and limbal parameters M, K,
S and processes the PACT commands needed for subsequent strokes while the CRT is being scanned on the current stroke. The above discussion of FIG. 27 and its functionality will be more easily understood by reference to the logic flow diagram of FIG. 27A.

Here, the function PACT start sets both the I and S counters to O, places the system in the P state (basic or restored state), and indicates that one of the following four operations will be started from the P state: Must pay attention. (1
) Instruction processing (PT) (2) Limb calculation (PFF) (3) Y coordinate transfer (PFTF) (4) Waiting for current CRT stroke until completion (PF
TT) The block diagram of FIG. 28 shows (CSU) 28 in more detail.

(CSU) includes an S unit 32, a K unit 34, an M unit 36, and a Y unit 38, the latter three units being connected to the PACT from buffer 24 as shown.
1-15" data. Each unit receives from the CPU the system clock and bits J1-J4 which identify which of the 16 possible limbus will be processed and calculated, as will be recalled. Logic unit 39
is the PFF (calculation command) and instruction BOL from the CPU 26.
Receives P, CM, CK as well as status signals (set by the CK command word). Unit 39 issues commands to write to these units, SWE, KWE, MWE and YWE, to the corresponding units. J1-J4
is supplied to each unit to identify the limbus for which parameters are to be calculated. Various patterns of data flow and instructions are discussed below. As known, (CSU)2
The output of 8 is the Y coordinate R supplied to (SSVCU) 22.
It is Y7-RYl6.

These values are calculated and updated based on the M parameters supplied from the M unit 36 to the Y unit 38;
Shown as M3-RM8. The M parameter is controlled by the K and S units. These basic programs are considered separately. In FIG. 29, the Y unit includes YRAM 40 and CDYRAM 42, which receive Y coordinate data from instructions provided by PACT buffer 24.

CDYRAM 42 specifically receives Y increment data from the CDY instruction, while YRAM 4O receives Y increment data from buffer 24 or 2.
-1 receives Y coordinate data from latch 46 via data selector 44; Latch 46 stores the Y+ΔY output (updated Y coordinate value) of 16-bit adder 48, which will be described later. The selector 44 is controlled by the PFF (see FIG. 27) and passes the Y value of the latch 46 while computing limbal parameters in a given calculation cycle, and passes the Y value from the buffer 24 when a new command is received. Pass the Y value. C
DYRAM 42 is installed for the CDY instruction to generate such a line instead of attempting to calculate a long linear slope line from the M value given by the CM instruction.

RAMs 40 and 42 each have one memory address as identified and addressed by inputs J1-J4 from (CPU) 26.
It has the ability to store 16 16-bit words corresponding to 6 limbus.

The Y unit also includes a programmed ROM 52 (PROM) that stores the ΔY(M) value. In an actual system, 64 values corresponding to 64 slopes are stored, thus providing 64 corresponding ΔY values to PROM 52. Adjacent ΔY decoding logic 54 provides the most significant bit ΔY8ΔYl6, the most significant bit being PROM
52, saving circuitry. Another 2-1 data selector 56 normally selects ΔY from ΔY PROM 52 and composite logic 54 and supplies it to adder 48 . but,
D from the CPU 26 and RAM 42 by the CDY instruction.
Y is derived, and the output DYF causes the selector 56 to select CDY
The output of RAM 42 is passed to adder 48. Finally, the updated Y value is supplied to YROM4O as described above, and the CRT spot unblanking/blanking control bits RY7-RYl6 are transferred from YRAM4O (SSV
CU) 22. A period comparator 50 is used to reorder the limbus numbers when a new pair of limbal changes begins during a BOLP command.

This compares the existing value of the Y coordinate (RY) from RAM 4O with the Y coordinate value (BY) of the new limbus to be started based on the BOLP command from the buffer 24, and creates the ring based on these Y coordinate values. It operates to determine the position in the sequence of parts where the new limbus number should be. 29th A
The flowchart in the figure helps illustrate the above.

From the PFF instruction for calculation, if DYF is a false value (0), the MSB of M3M8 bits from M unit 36 (
That is, bit 8) determines whether Y should be increased or decreased by ΔY. The M8 bits can be thought of as sign bits. M is either O or 1
The effect of 8 bits is easily understood from Figure 29B, which shows its effect in generating positive slopes and curvatures.
It should be noted that the curvature polarity is defined by the seventh curvature bit K7. If K7=O, the curvature is negative and M is decreased by ΔM; if K7=1, the curvature is positive and M is increased by ΔM. In FIG. 30, the M unit 36 of FIG. 28 is shown.

During the calculation, the M parameter addresses the △YPROM or lockup table (see Figure 29), and the value of M itself is
Updated in unit 36.

In particular, an MR that stores the M parameters for each existing limbus.
AM6O is illustrated. (Address input J1-J4
reference). The encoded gradients for up to 16 limbus each are shown. The M value that defines the 64 possible slopes includes 6 bits that define the integer number of slope values from 0 to 63 and 2 bits that define the fractional parts 0, 1/2, 1/4, and 3/4. It consists of 8 bits.

Bits M3-M8 defining the integer value are provided to Y unit 38 as described above. Similar to the function of Y unit,
The M parameter is an updated value created by the PACT instruction and the value is incremented for the duration of the calculation cycle. Therefore, the 2-1 data selection circuit 62 selects between these inputs, in other words, the PACT buffer 24 (
Consists of bit positions 2-7 of the CM instruction word in Figure 4)
or is arranged to select the input from the update circuit. The update circuit includes an 8-bit adder 64 and an 8-bit latch 66.

As will be recalled, the M value is updated under the control of the K and S units. Furthermore, △M to be changed
The absolute value of 1ΔMI is supplied by the K unit 34. Bit RK7, which is the 7th bit of the K parameter, has a positive curvature (M increases) or a negative curvature (M decreases), as shown in FIG. 29B, depending on its bit value 1 or 0. ). The MZ signal from logic unit 39 is a true value for the BOLP instruction;
Set M=31·3/4 (for which ΔY=O) at the starting point of the paired annulus. The effect of MZ is essentially to drive selector 62 non-1 and no input is provided to the MRAM. The conditions under which M changes are better understood from the flowchart of FIG. 30A.

As shown in FIGS. 30 and 30A, the CPU 2O issues a control PFF to start the M calculation. First judgment PK
In F, the flag bit in K unit is MN-MN-
The o judgment in 1 indicates that there is no curvature for a given ring (J1-J4), and M should not be changed, in other words, no curvature value is recognized for that ring in KRAM. It means that. The second determination determines whether the increase to M caused an overflow, and if so, M does not change again.

The third judgment is the main judgment, and is the turning point of the limbus that draws the curve, and it is judged whether the SN value stored in the SN table exceeds the S parameter.

If it is a false value, the logic requests an update of M, and in the final test, RK7 (sign bit of curvature) determines whether it results in an increase or a decrease. If it is a true value, M remains constant. Returning to FIG. 30, when an M update occurs, a new value to be written from adder 64 to MRAM 60 via data selector 62 is stored in latch 66.

FIG. 31 is a detailed block diagram of the K unit 34.

KRAM 7O receives 7-bit curvature information from buffer 24 in accordance with the CK command. It is understood that unlike the Y and M units, KRAM is started by a PACT instruction, so the K value is not updated during calculations. Bits Kl, K2, K5, K6 are K decoding logic 72
and K decoding logic 72 provides IΔMI.
RK7 controls the one complement circuit 73 and supplies lΔMI to the M unit 36. In FIG. 31A, the truth table for the K decoding logic is shown. For basic radius K6-1! △
MI-△S1-1, △SF-0, K5-K2-K1-
It should be noted that it is 0. The K3 and K4 bits identify and select the base radius.

Four SN tables are used, corresponding to the four basic radii, and the desired other curvatures are made easier by scaling the ΔM and ΔS values in the SN tables that are most closely related to the encoded curve. is approximated by

The table in FIG. 31B shows the selection of SN for K3, K4 and the values of IΔY(M) for M+ from 0 to 31. (One SN table is used and the desired number of radii as one extremum and opposite extrema is calculated by scaling ΔS and ΔM, resulting in 64 different SNs corresponding to 64 radii or curvatures.) A data selector 76 performs a selection function in response to K4 for this purpose. M.
The K decoding logic 78 resets the S parameters using K1, K2, K5, and K8 as functions. value S.

The truth table for this input is shown in Figure 31C. In FIG. 32, a detailed block diagram of S unit 32 is shown.

The SRAM includes a 4-bit section 80 that stores the integer value of S and a 4-bit section 82 that stores the fractional value of S for each of the 16 rings as identified by the J1-J4 addresses. The K function of the S unit is provided by a comparator 84 which converts the SN value from the K unit into an increasing integer value of S supplied from SRAM 8O, i.e.
Contrast with (S+ΔSF)1. If the comparator 84 outputs A<B, the carry value from the adder 86 and ΔS1 from the K unit 34 are sent to the adder 8 by the data selector 89.
8 is added to the S value supplied to 8. This S value is S when the S count exceeds SN. Therefore, S. Or it is set to the current value of (S+ΔSF)1. The increased value is supplied to RAM8O as an updated value. Adder 86
is the fractional increment △SF and the memory increment S from RAM82.
F and supplies the sum SF+ΔSF to the RAM 82 for updating.

The complement logic output A>B from comparator 84 is obtained from latch 86.

Therefore, when (S+ΔSF)1>SN, the output is CS
It is applied to the calculation logic unit 39 of U28 (see FIG. 28). As will be recalled, when S>SN, M is increased or decreased by 1ΔMI according to K7, and the M-K decoding logic is S.

, and Y is increased in the next calculation as a function of the new value of M. Therefore, subsequent segments of the curve fit are generated according to the new Y coordinate change values. (SSVCU) 22 is shown in detail in the block diagram of FIG.

The principal elements are a horizontal scaling unit 90 and a vertical scaling unit 92, which receive scale information from the O-15 input of data from the computer interface 20 in response to corresponding strobes DS65 and DS64, respectively, as described. do. Temporary Y(T
Y) Coordinate memory 94 receives bits RY7-RY16 representing the Y change coordinate of each limbus. Based on the command S1 from the CPU 26, the Y coordinate is supplied by the CSU 28 in a numerically decreasing direction. TY memory 94 stores sixteen 11-bit words, with the last bit typically being an O. When the final Y change coordinate is read, the cycle output from CPU 26 sets the 11th position pit to "1". This bit serves to identify the last wheel to be processed in a given calculation cycle, as described below. TY memory 9
4 is addressed by a TV address counter 96 which is clocked by the TK clock from the scanning video unit 98.

Each Y coordinate value read from the TV memory 94 by the address counter 96 is compared with VY'! Comparator 1 for 2TY
00. As will be recalled, a character of the desired point size is generated from the encoded character data associated with the 72 point size or maximum point size for the coordinates of the quadrilateral set by the vertical scaling function.

An 8 MHz clock supplied from scanning electronics 30 to vertical scaling unit 92 causes a counter installed in unit 92 to step up by a value corresponding to the scaling of the character to be displayed. For example, when a 72 point character is to be displayed, the counter increments by one unit for each clock pulse input. Conversely, when a four point character is to be displayed, the counter increases by the ratio of the display point size to the encoding standard size, ie, 18, for each clock pulse. Therefore, the scaled CRT spot coordinate position output from scaling unit 92 is compared with the Y change coordinate read from TY memory 94 in comparator 100.

When the result of the comparison is VY>TY, the output is provided to the scanning video unit 98 and the scanning beam is unblanked by control of the video coupler 29. As is known,
The beam is blanked initially and unblanked by the first comparison. Based on the concept of paired limbus, each subsequent comparison switches the beam from its current state to the opposite state, so that the next comparison again blanks the beam. Scanning video unit 98 is connected to comparator 10
Immediately upon receiving the comparison output from 0, TYCLK is supplied to the TY address counter, the TY memory 94 is addressed, and the next changed coordinate is read out. When the final Y coordinate is provided to comparator 100, the 11th bit stored in TY memory 94 produces the TYF output.

By definition, when Y>TY occurs, the beam is blanked for the duration of that stroke. TYCLK is undriven and SVU switches the SVS signal to the opposite logic state, causing CP
Indicates to U26 that the stroke signal is now finished. The scanning electronics 30 therefore completes the subsequent stroke and prepares itself for the next stroke function. Therefore, the scanning circuit does not scan the beam according to a fixed raster. Therefore, this function enables high-speed operation. Details of horizontal scanning unit 90 are shown in FIG.

It will be recalled that in this system non-integer scale factors are possible. An integer latch or register 110 stores the integer part of the horizontal scale factor;
Latch 112 stores the fractional portion supplied by the data input from the computer interface and introduced by DS65. Under the control of (CPU) 26, the integer part from latch 110 is introduced into S counter 114. After the calculation cycle, the CPU 26l issues an FFT and decreases the S counter 114 by the next CLK input. When the counter 114 reaches the value of O ([MIN] output), the SX
Z is supplied to the CPU 26 and further calculations are inhibited until a new stroke begins. The value in fraction latch or register 112 is provided as input A to a 10-bit adder and is combined with the fraction currently stored in fraction sum register 118.

Register 118 is initially set to CLE of (CPU) 26.
It is cleared to 0 value by the AR command, and this command also sets the S counter 114. When a carry output occurs as a result of addition in the adder 116 (for example, 4 for a scale factor of 5 1/4)
), the carry output is carried out by the CPU.
S2 from 26 gates the S counter 114 and increments that counter by one.

Of course, this operation is performed when the S counter 114 receives the introduction command S1.
This is done after being preset to the integer part of the horizontal scale by. The above operation of the S unit will be understood to accomplish the purpose of handling non-integer horizontal scale factors.

This is very important because it not only scales accurately to the desired point size, but also allows a variety of CRT scanning densities to be used. The central processing unit (CPU) 26 is a 35th CPU.
It is shown in more detail in the figure and includes as its basic elements a processing state unit 20, an OP decoding unit and a limbal sequence unit (0SU) 126.

OP code decode unit 122 receives the first five bits of each PACT command and provides the corresponding command to the appropriate unit as shown.

Bits 8-11 of the instruction, including limbal bits J1-J4, are provided to limbal sequence unit (0SU) 124;
The limbal sequence unit (0SU) stores values J1-J4 for output to (CSU) 28. Bits 12-15 of the instruction that identify the number of calculation cycles until the next instruction
is supplied to (PSU) 120. Central processing units in data processing systems are well known, and designs embodying the necessary basic sequence and control functions for such systems will be obvious to those skilled in the art.

Therefore, the description of this CPU will be limited to only the important points directly related to the processing control required in this system. FIG. 36 shows more details of the processor state unit and reveals the various outputs as shown in FIG.

The instructions BOLP, EOLP, CK, and CDY are supplied to the corresponding flip-flops B, E, K, and DY, respectively.
When performing FTF, we force the system to leave the P state. The sequence of states is more easily recognized from the flowchart of FIG. 37, where the EOLP and BOLP instructions are further shown establishing a sequence of auxiliary states. Of particular interest is counter 13.
0, and counter 130 registers PACT bit 12-1.
5 and also included in bit 811 for NOP instructions.

The NOP therefore causes gate 132 to pass these additional bits into the counter. In this case, the counter is 1 in the final calculation of each consecutive calculation cycle.
The count is decreased by 0, and the output IXZ as described above is generated. In FIG. 38, limbal sequence unit 124 is shown.

Two 16-bit serial input parallel output shift registers are installed.

Register 140 stores a 4-bit limbus number, and register 142 stores a 1-bit identifying the valid limbus number in register 140 in the corresponding location. The primary functions to be performed include storing the limbus number provided by the BOLP command and, in response to receiving the EOLP command, removing the previously stored limbus. It will be done.

Unit mouth 24 also establishes a limbus by increasing the Y coordinate value. Of course, the allocation of the number of limbus is arbitrary within the range of 0 to 15 for J1-J4. but,
Once assigned, the parameters for that limbus are
various memories (i.e. Y, .M, .K and S) at addresses defined by the individual limbus numbers J1-J4.
is memorized. In the above example of character encoding, previously unrelated limbus may form a pair when a new paired limbus is created or when an old paired limbus ends. It was shown that

Alternatively, the new limbus may have coordinate values intermediate to the existing limbus, forming a new pair. These changes occur in response to the BOLP and EOLP commands.
processed during the state. Therefore, although there is no ordered sequence for limbus number assignment, it is necessary that the limbus numbers be stored according to an ordered sequence of their respective Y coordinate values.

This constraint is imposed to enable direct comparison operations between the TY memory read and vertical scaling discussed in connection with FIG. 33, which provide unblanking control over the stroke.

Therefore, the limbal sequence unit 124 is provided to provide an accurate ordering of the limbal numbers taking into account the Y coordinate values of the individual limbals. Therefore, from CSU to the 33rd
In the Y coordinate transfer operation to the TY memory in the figure, the YRAM of the Y unit 38 (part of the CSU 28, see Figure 28)
4O (FIG. 29) is addressed by limb numbers J1-J4. The ring numbers J1-J4 are read out from the CPU 26 in the exact order of the ring numbers corresponding to the reading of the Y-coordinate change value in the required earth elevation direction. From Figure 29, comparison RY
It will be recalled that >BY was output to CPU 26.

The outputs 1i to CPU 26 are shown in the more detailed block diagram of FIG. 35, particularly as provided to logic unit 126 for further processing. These operations are described somewhat schematically in the 38th section to facilitate understanding of the operations.
It is shown in the figure. The B and E states corresponding to the BOLP and EOLP instructions are in logic unit 1 with RY>BY.
26 (FIG. 35), unit 126 then provides an output to control unit 150 indicating whether the Y coordinate in the command is less than the Y coordinate value currently read from memory. Recall again that the Y coordinate values read out are identified by the limbus numbers J1-J4. The limbus number at any given time is J1-J from the data selector 152.
4 outputs, which are used by the CSU for addressing.
Given to 28. In operation, registers 140, 142 continuously recycle under control of the clock.

As explained below, register 142, along with decoding logic 144, identifies the location of the limbus with the lowest Y coordinate value at any given time. 8-1 data selector 143 selects the maximum Y
a shift register stage is identified when a coordinate is currently being addressed and indicates that another shift register stage must be selected;
and is loaded into storage flip-flop 146.

Therefore, the 8-1 data selector 141 selects the ring number of the minimum Y coordinate from the data selector 15 during the next clock period.
The data is controlled to be read at 2. If the Y-coordinate value of the BOLP instruction is less than the minimum Y-coordinate value of an existing limbus thus identified (i.e., RY>BY), the limbus number from the shift register is taken out of recirculation and gated. The limbus number for the new limbus held in storage unit 152 and identified by the BOLP instruction is introduced through unit 152 into the input stage of register 140.

More specifically, the BOLP includes two words and two Y coordinate values corresponding to a pair of two limbus, and the lower coordinate values are the first BOLP word and the second BOLP.
Recalling from Figure 4 that within a word, it will be appreciated that two consecutive limbal numbers corresponding to the two Y coordinate values of the two BOLP words are introduced in succession. Conversely, if the Y coordinate value of the BOLP instruction is greater than that of an existing limbus, gate and storage unit 152 recirculates the existing limbus number until the point where the comparison RY>BY is obtained.

The opposite situation is obtained with the EOLP instruction, in the sense that the existing ring number should be removed from the shift register.

This operation is more easily constructed and therefore easier to understand. Note from FIG. 4 that the EOLP command is encoded with the limbus number being the lower Y coordinate of the paired limbus to be terminated. Therefore, when this limb number is provided to units 150, 152 from the PACT buffer, the comparison is requested to control unit 150 when the limb number is recycled by register 140 and a comparison is obtained. The control unit also removes the loop by controlling the gate unit 152 to inhibit recirculation of the loop. Additionally, unit 152 switches from last stage output 140A to last last stage output 140B, thereby causing register 14
Advance all consecutive limbus numbers by one stage at 0. This helps preserve all existing loops in successive stages of register 140, allowing more efficient processing. A flag bit for identifying the valid limbus number is introduced into or removed from shift register 142 by substantially identical control of gate unit 154, thereby causing gate unit 154 to It will be readily understood that the sections are substantially parallel to each other.

As a final point, only 8 outputs are in register 1
40,142, respectively.

This is a result of the unique relationship of the paired limbus. With proper timing, these 8
The parallel outputs can always correspond to a lower Y-coordinate limbus, in which case the system essentially knows that the next limb in memory is the associated higher-Y coordinate limbus in the pair. Know. Reliability in this relationship was explained in connection with the introduction of a new pair of high Y-coordinate wheels from the BOLP instruction. As mentioned above, for the EOLP instruction, only the low Y coordinate ring is encoded. The cancellation function performed by control unit 150 therefore causes cancellation of both the limbus in which the comparison is accomplished and the next succeeding limbus. Therefore, the EOLP instruction does not require a second word to identify the high Y-coordinate wheel since this is redundant. Briefly, the limbus sequence unit 124 operates to maintain the limbus in a precise sequence of ascending Y coordinate values.

Furthermore, the limbus number is maintained in successive stages of register 140. A flag bit correspondingly indicating a valid ring is maintained in register 142. The significance of the flag bit and decode logic 144 is that the processing function begins simultaneously with the lowest Y coordinate wheel and progresses through all valid stored wheels. Therefore, the process is not time bound to complete the recirculation of each register. This saves valuable computation time. For example, if only one pair of limbus is stored, the system can immediately identify the location of the limbus;
Two limbus are processed and then activated for subsequent processing functions. Therefore, the remaining 14 stages of the shift register do not need to be considered. In this example, compared to a situation where all 16 stages of the shift register have to be inspected and processed,
It is sufficient if only 1/8 of the processing time is consumed. Conclusion In the above description, we have described the encoding technique of the present invention, the very basic form of the processing circuit for calculating the coordinates of the characters to be generated, as well as the more complete detailed configuration of the processing circuit corresponding to an actual operating system. Ta.

Those skilled in the art will appreciate that numerous variations and applications of the techniques and specifically constructed systems of the present invention may be readily accomplished.
As an example of a variation, although the invention has been disclosed in relation to a quadrilateral in rectangular coordinates, it will be clear that other coordinate systems may also be used. ``Furthermore, it is possible to use other than rectangular coordinate type display systems.Furthermore, encoding q
The rigid reference system need not be directly related to the scanning pattern. For purposes of clarity, the system specifically disclosed uses a rectangular coordinate encoding quadrilateral and raster scan pattern. However, the invention is not limited to such a linear relationship, and the apparatus may instead be used to encode rectangular quadrilaterals with a circular scan pattern or a polar coordinate encoding system and a raster or circular scan for display. It can also be a shape. The techniques necessary to correlate the results of the calculations obtained to define the coordinates of the limbus with respect to the encoding system and display control of the desired scan pattern used for display are within the skill of those skilled in the art. It's self-evident. Furthermore, important advantages in certain applications are provided by the utilization of the paired limbus concept as described in the detailed description of the preferred embodiment of the invention, so that the character structure It must be understood that there is no need to be defined by. Rather, a character can be defined by only one limbus. This is easily understood in relation to letters such as [E], "F", "", etc. If a single limbal approach is adopted, the basic encoding description as described here is still valid.

In this case, the encoding includes both increasing and decreasing values of calculation cycles identifying the degree of application of a given encoded coordinate value or parameter values such that the locus of points defining the limbal position is calculated. Furthermore, limbus with two or more letters,
It must be recognized that the concept of relating the limbus as a pair, even as defined by , is useful primarily for understanding the encoding function. In fact, it is clear that the valley ring can be defined separately. In all variations as described above, the basic consideration is that the integrity of the character is maintained by using one or more limbus in the encoding and calculation functions taught by this invention. . Accordingly, the appended claims are intended to cover all modifications and applications falling within the true spirit and scope of the invention.

[Brief explanation of the drawing]

FIG. 1 shows horizontal and vertical unit scale factors (HSF=VSF=1
), Figure 2 is generated from the same encoded instruction, SF=18, HSF-25.7
74 and a magnified view of the serifs in Figure 1 shown in 4 points, the illustrations represent the actual display at the same enlarged scale as in Figure 1, and Figure 3A shows the
Represents the characters generated by the commands in the table in Figure B, and
Figure C shows the same letter rings as Figure 3A numbered in the instructions of Figure 3B, Figure 4 is a table of instructions showing these formats, and Figure 5 is a schematic quadrilateral showing the character encoding. , 6th
7 is a table of values for the gradients of FIG. 6; FIG. 8 is a schematic quadrilateral with a simple character structure to explain gradient encoding; FIG. 9 is a table of commands for the characters and quadrilaterals of Figure 8, Figure 10 is a table of calculated Y coordinate values relative to successive X coordinate positions calculated for the command of Figure 9, and Figure 11 is a table of the calculated Y coordinate values relative to the successive X coordinate positions calculated for the command of Figure 9. FIG. 9 is a diagram of the characters generated from the encoding command; FIG. 12 is a block diagram of the ROM that stores the incremental change values of the Y coordinate value for the corresponding gradient value; FIGS. 13A and 13B are the bases, respectively. FIG. 13C shows the bit pattern of the ROM of FIG. 12 for 2, 10, and FIG. 13C shows the displayed slope value of FIG. The table and Figure 14 are a schematic block diagram of the mechanism of the Y update function in gradient calculation.
Figure 5 is a schematic diagram of the curvature, Figure 16 is a table relating the curvature of Figure 15 to the binary system and the corresponding radius, Figure 17 is a diagram of a continuous gradient approximating an arbitrary radius of curvature, and Figure 18 is X
A comparison diagram of gradients generated continuously for equal intervals of coordinates and curves to be approximated and encoded. FIG. A predetermined number SN of Y coordinate change calculations to be made to each gradient value as utilized to closely approximate a circular arc.
20 is a comparison diagram of the ring generated according to the curvature table of FIG. 19 and the circular arc to be approximated, and FIG.
A table showing the Y-coordinate update values calculated according to the table in Figure 22, a comparison diagram of the ring and the approximated arc generated from the values in the table of Figure 22 & Figure 21, Figure 23 is a comparison diagram of the approximated arc. A schematic design diagram of the ROM used to embody the M update function according to the table in Figure 19, Figure 24 is the ROM bit pattern table in Figure 23, and Figure 25 is the basic radius when K = 6. A table showing the occurrence of each individual curvature from one ROM of FIGS. 23 and 24 programmed for Rc=32, FIG. 26 shows the M update function according to FIGS. 19-25. 27 is a basic block diagram of an actual embodiment of the character generator according to the present invention; FIG. 28 is a detailed block diagram of the calculation and storage unit of the block diagram of FIG. 27;
Fig. 29 is a detailed block diagram of the Y unit shown in Fig. 28, Fig. 29A is a flowchart of the Y calculation function of the Y unit of Fig. 29, and Fig. 29B is an incrementally updated version of M and Y. 30 is a detailed block diagram of the Y unit of FIG. 28; FIG. 30A is a flowchart of the function of the Y unit of FIG. 30; FIG. 31
The figure is a detailed block diagram of the K unit in Figure 28, and Figure 31A.
Figure 31B is a table of values for the K decoding logic of Figure 31; Figure 31B is a table of values stored in the SNPROM of Figure 31;
Figure C is a logical truth table of the MK decoding logic block in Figure 31, Figure 32 is a detailed block diagram of the S unit in Figure 28,
33 is a detailed block diagram of the scaling, stroke and video control unit of FIG. 29, and FIG.
35 is a detailed block diagram of the central processing unit of FIG. 27, FIG. 36 is a detailed block diagram of the processing state unit of FIG. 35, and FIG. 37 is a system state sequence. 38 is a detailed block diagram of the limbal sequence unit of FIG. 35. In the figure, 20 is a computer interface;
2 is a scaling, stroke and video control unit (SSVCU), 24 is a PACT buffer, 26 is a central processing unit (CPU), 28 is a calculation and storage unit (CSU), 29 is a high voltage video coupler, and 30 is a scanning electronics circuit. show.

Claims (1)

[Claims] 1 A character generator that generates characters encoded according to a standardized quadrilateral of XY coordinates, the character generator
The coordinate values correspond to a calculation cycle in which each character is encoded into successive data instructions associated with the contours of the character segment, and a starting line identifying the initial Y coordinate of each of the associated pair of contours. instructions, contour modification instructions for specifying various fixed and variable orientations of contours and the ending points of a pair of contours, and instructions for specifying the end of a character, each modification instruction including the identification of an associated contour. , and a number specifying the computational cycle until the next instruction, and in response to a start row instruction, identifies an initial pair of contours and stores the respective Y coordinates in association with the contours. means for calculating a new Y-coordinate for each identified contour in accordance with an associated modification instruction in each successive calculation cycle, including maintaining an initial Y-coordinate for the unchanged contour; means for updating the Y-coordinate value in said storage means in accordance with a new value of the Y-coordinate value computed in a computation cycle; and means for requesting and receiving subsequent instructions for a character based on the completion of an encoded number of computational cycles of instructions.