H BBS Bn» NHS? HSU m mm it 111 web Ira ESSa* BaSst HHMMm 1 HntSn 1 m I 111 HSR HB^^fi»m HH fflm. m& gw smffi Bail mmsmm ml KR9 SHE B LIBRARY OF THE UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAICN 510.84 no.5U-5ie> cop- 2 2 UIUCDCS-R-72-515 Jy^tCt/^- coo 2J469-0209 RASER: RANDOM TO SERIAL CONVERTER by TAKEHIKO KATOH June, 1972 Digitized by the Internet Archive in 2013 http://archive.org/details/raserrandomtoser515kato UIUCDCS-R-72-515 RASER: RANDOM TO SERIAL CONVERTER by TAKEHIKO KATOH June, 1972 Department of Computer Science University of Illinois Urbana, Illinois 6l801 This work was supported in part by Contract No. AEC AT(ll-l)lU69 and was submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Electrical Engineering, June, 1972. Ill ACKNOWLEDGEMENT The author wishes to express his gratitude to his thesis advisor, Professor W. J. Poppelbaum, for suggesting this problem and for his interest and help during its solution. Deep thanks also go to Professor W. J. Kubitz for his counsel and suggestions. The author wishes to thank Mr. Frank Serio and his technical staff in the fabrication group and machine shop for the work they contributed toward the building of RASER. Thanks also go to Mrs. Barbara Bunting for typing the thesis and the staff in the drafting group for drafting the figures and staff in the printing shop for duplicating the thesis. Finally, the author thanks his wife, Linda, for her support and encouragement through this project. iv TABLE OF CONTENTS Page 1. INTRODUCTION 1 2. OPERATIONS PERFORMED BY RASER 6 2.1 Random to Serial Conversion 6 2.2 Rotation and Reflection Around the Axes 7 2.3 Shrinking a Picture 11 2.1+ Picture Expansion 16 3. LOGICAL DESIGN OF RASER 23 3.1 Random to Serial Conversion 23 3.2 Rotation and Reflection Operations 28 3.3 Shrinking Operation 30 3.1+ Expansion Operation 33 3.^.1 First Step of Expansion Operation 33 3.^.2 Second Step of Expansion Operation (Format Conversion) 35 k. CONCLUSION 1+0 LIST OF REFERENCES 1+1 APPENDIX 1+2 VITA 51 V LIST OF FIGURES Figure Page 1. Fundamental Diagram of RASER 2 2. Word and Bit Location in the Binary Video Memory for (x, y) Pair k i i 3. Relation Between (x, y) and (x , y ) for Rotation 8 k. Relation Between (x, y) and (x , y ) for Reflection Around the Axes 10 5. Examples of 1/2 x 1/2 Shrinking in 2 Typical Cases .... 13 6. Location of M(x , y ) in the Binary Video Memory 17 7. Address, A (x , y ) , in the Binary Video Memory 21 8. Block Diagram of RASER 2k 9. Timing Chart of Display Clock and Master Counter 25 10. Timing Control Circuit 27 11. Logic to Video Converter Circuit 29 12. Circuits for Picture Rotation, Reflection and Shrinking Operation 31 13. Sample Rate Control Circuit 32 Ik. Difference Computation (for Expansion) Circuit 3^- 15. Matrix Pattern Generator Circuit .- 36 16. Format Conversion Circuit 38 17. Timing Chart of Format Conversion 39 1. INTRODUCTION In the computer graphic display field the raster format display (of which standard television is a particular type) has certain advantages over a random format type display in some applications. These advantages are 1) The efficiency of this format is greater than for a random for- mat display when the amount of information to be displayed is large but the scanning mechanism does not have a high frequency response, such as for large screen displays using lasers or large screen CRT's. 2) There is no overflow problem when the amount of information to be displayed is large and the duration of the display time is limited (typically 1/30 of a second). 3) Costs are generally lower for the display, because of the high volume of production of raster T.V. equipment. The system described herein is named RASER (RAndom to SERial Converter) since the raster format is SERIAL while the data from the computer is RANDOM. This kind of conversion may also be called D/V conversion (Digital- to- Video Conversion) or simply scan conversion. The purpose of RASER is to bridge the gap between the random output of a computer and the serial requirements of a raster display. The fundamental idea of RASER is shown in Figure 1. Coordinate pairs (x, y) corresponding to random locations in an xy-plane come into the converter at a fixed rate (e.g. a pair every 3 M>s). In RASER a kK by lo bit core memory with a full cycle time of 1 u-s, and a split cycle time of 1.1 \xs plus time, is used as a binary video memory. Because of the size of the binary video memory, the (x, y) pairs are 8 bits accuracy each. o Q ■p 64 •H &4 Let (x, y), the input to RASER, be defined as 7 i 7 i x = Z x. -2 and y = Z y. -2 i = X 1= X where x and y are sign digits and negative numbers are expressed as 2's complements. Let (x 1 , y') be input to the binary video memory. The x ' = x * 2^ + Z x . • 2 1 ' i = X 7 6 - i and y' = y -2 ( + Z y -2 . ' i = Then "1" will be stored in the binary video memory at the bit b of the worl at address A, where i 11 6 - ±+h - 3 6 i-k b= L x.-2 and A = y "2 + Z y. '2 + x^'2 + Z x.*2 i = x 7 i = o - 7 i „ U i See Figure 2. "1" corresponds ultimately to an unblanked electron beam and "0" to a blanked electron beam on the CRT screen. For the display, the binary video memory is read out serially from A = to A = ^095. The first point in the ith raster line is b = and A = 16' i. The second point in the ith raster line is b = 1 and A = l6'i. The (16 'j + i)th point in the ith raster line is b = I and A = l6"i + j. With simple operations on the incoming (x, y) pairs, RASER car. dis- play the picture rotated through 90 , 1^0 or 27O (clockwise) or reflected about the x = y, x = -y, x = or y = axis. By shifting x and y by one or two bits RASER can display a shrunken picture. Let (x', y') be the input to the binary video memory, 7 6 ^ i 1 then x* = x''-2' + x -2 + Z x.'2 ' ' ± = 1 X -> x 1 Y' V A=0 t t "5 5 C A = 16 A=l t V A= 17 rz A=15 RASTER LINE $ RASTER LINE i + 1 b=0 b=Zxi-2' i=0 a b = i b = 15 BINARY VIDEO MEMORY 4K by 16 bits I A=4095 Figure 2. Word and Bit Location in the Binary Video Memory for (x, y) Pair 7 6 ^ — i 1 and y' = y"-2' + y "2 + Z y. *2 for a 1/2 x 1/2 picture, 7 ' i = 1 X z- j- 6 . or x' = x"-2 7 + x"-2 + x^-2- 5 + Z x*2 7 6 7 i = 2 i 7 6 5 IP and y* = y"-2' + y>'-2 + y «2 3 + Z y. «2 " for a 1/U x 1/1+ picture, 7 o r ± = 2 i where x", x^. y" and y!-' are specified to RASER by the operator, and the 7 o 7 o incoming coordinates are: 7 7 i i x = Z x. *2 and y = Z y. '2 . " " i i = i - RASER can also display an expanded picture of 2 x 2 or 1+ x h with straight-line-interpolation requiring no additional computation time. How- ever, for this operation the (x, y) pairs are required to be random with respect to the raster scan but the distance, d, of successive incoming (x, y) pairs, (x . , y.) and (x . , . y. n ), must be less than 2, where d is defined as J J j+1' J+l d=maxl|x. -x. +1 |, |y. -y J+1 |). 2. OPERATIONS PERFORMED BY RASER There are four main operations performed by RASER. The first and main operation is random to serial conversion. The second is rotation and reflection of a display, the third is the shrinking and placing of the display in the proper portion on the CRT screen. The last and most interesting operation is expanding a part of the picture with straight line interpolation. 2.1 Random to Serial Conversion In RASER the random to serial conversion is accomplished simply by storing l's at the correct bit locations of the binary video memory and read- ing out these bits at the proper moment. Since the raster timing must provide overall control of the conver- sion as well as display process, all digital and display functions must be timed to the raster synchronization pulses. Every raster line contains 256 points (16 words) and has about 6k usee duration. In order to make the display square, both the left and the right ends of the raster must be blanked. Therefore, each point displayed on the CRT screen is about 200 nsec in duration and is generated by an unblanked or blanked electron beam on the CRT screen. This means a word (16 bits) must be read out every 3.2 usee from the binary video memory. A R/R cycle (Read Restore cycle) requires 1 usee, and there are, thus, only 2.2 usee remaining between cycles. In order to store l's for points in the binary video memory without erasing the l's already there, the R/M/W (Read Modify Write) cycle must be used. It takes 1.1 usee plus wait time. There- fore one R/R cycle and one R/M/W cycle can be completed in 3.2 usee. 7 The binary video memory contains 256 raster lines but there are only 2^1 visible raster lines in one television field (1/60 of a second or 1/2 of a frame). RASER displays the whole content of the binary video memory for each field. Thus, 8 lines of the top and 8 lines of the bottom of the binary video memory will not be displayed. 2.2 Rotation and Reflection Around the Axes Let the (x, y) pair be the input to RASER and (x*, y') be input to the binary video memory as defined previously. Then x' and y' for the various rotations are given as follows : Case i) No rotation x* = x -2 7 + Z x. '2 1 . 7 i =0 x 7 6 - i y' = y 7 -2 ( + Z y -2 . ' i = See Figure 3(a). Case ii) 90° clockwise rotation - 7 6 i x' = y -2' + Z y. -2 . y i = 1 - 7 6 i y* = x-2' + Z x. '2 . 7 i = X See Figure 3(b). Case iii) 180 clockwise rotation 7 6 - i x'=x'2'+ Z x.-2. 7 i = X - 7 6 i y» = y -2' + Z y -2 . ' i = See Figure 3(c). 8 -^x >> x ) ^ ( x ,y) «^ A y (a) No rotation (c) 180° rotation FT ->X" \ \ \ \ \ \ \ \ ^90° \m \ f(x,y) ST - ■■■■ (x,yK ^ pi y ~270° (b) 90° clockwise rotation (d) 270° clockwise rotation Figure 3. Relation Between (x, y) and (x , y ) for Rotati on Case iv) 270 clockwise rotation x ' = y • 2 7 + Z y . • 2 1 . ' i = 7 6 - i y' = x # 2' + S x.-2 . ' i = X See Figure 3(d). Case v) Reflection around the axis, x = 0. 7 6 - i x* = x *2' + Z x.-2 . 7 i ^O 1 7 6 - i y* = y 7 '2 f + Z y -2 . ' i = See Figure 4(a). Case vi) Reflection around the axis, y = 0. x' = 3L*2' + Z x. -2 1 . 7 i ^o 1 - 7 6 i y* - y -2 ( + z y "2 . ' 1 = See Figure 4(b) . Case vii) Reflection around the axis, x = y. x' = y -2 7 + Z y -2 1 ' i = y' = x -2 7 + Z x -2 1 . ' i = See Figure 4(c). 10 \ i \ 1 \ \(x',y) $ » / s Y ^* ^ s i y (a) Reflection around the axis, x = 0. (c) Reflection around the axis, x = y ^ x x = -y (b) Reflection around the axis , y = (d) Reflection around the axis, x = -y Figure h. Relation Between (x, y) and (x , y ) for Reflection Around the Axe* 11 Case viii) Reflection around the axis, x = -y. 7 6 - i x' = y -2' + Z y. '2 . ' i = - 7 6 i y* = x_-2' + Z x *2 . ' i = See Figure ^(d). 2.3 Shrinking a Picture RASER can shrink a picture to 1 x 1/2, 1 x l/k, 1/2 x 1, 1/2 x 1/2, 1/2 x l/k, l/k x 1, 1/k x 1/2, 1 or 1/1+ x 1/k x 1/k. Again let the pair (x, y) be the input to RASER and (x 1 , y') be the input to the binary video memory, where 7 7 i i x = Z x. "2 and y = Z y. "2 , i « i = and x and y are sign digits. 7-6 ^ i-1 If x' = x""2 + x '2 + Z x.«2 , the picture is shrunk by one half in the x direction. If x" = 0, the picture is placed in the left half of the screen. If x" = 1, picture is placed in the right half of the screen. Similarly, if y' = y"-2' + y '2 + Z y •2 1 ~ , the picture is ' ' i = l 1 shrunk by one half in the y direction. If y" = 0, it is placed in the upper half of the screen, and if y' T = 1, it is placed in the lower half of the screen. 6 If x' = x"-2 7 + x"-2 + x -2 5 + Z x.-2 1-2 , the picture is 7 6 7 is2 i shrunk by 1/k in the x direction and if y . B y ». 2 7 + y"-2 6 + y'2 5 + I y *2 I • i „ o x 12 1-2 1.2 the picture is shrunk by 1/U in the y direction. For x", x'J, y" and y'J there are 16 possible combinations and thus 16 different possible locations for the shrunken picture for l/k x 1/4 case. Therefore the (x , y ) pair can be any pair of f> 6 x' = x -2 7 + Z X.-2 1 , x"-2 7 + x -2 + Z X.-2 1 " 1 , or 7 i = X 7 7 i = 1 x 7 6 R^iP 7 ^ i x ... 2 r + x .,. 2 o + - 2 P + 2 x . . 2 ±_ ^ and y' = y. -2' + Z y.-2, 7 6 7 i =2 1 x i = a 6 6 y"'2 7 + y-2 6 + Z y'2 1 " 1 or y'"2 7 + y;-2 6 + y„-2 5 + Z y '2 1 ' 2 . ' ' i = 1 ' ' i = 2 However, difficulties arise for certain cases. Let the four succes- sive (x, y) pairs be (x., y.) for j = 1, 2, 3 and h. Let J J 7 7 x. = Z x. .-2 1 and y. - Z y . . '2 1 for j = 1, 2, 3 and 4. Let (x'., y 1 .) j i = j,i j i = j.i u J' "y for j = 1, 2, 3 and 4 be the input to the binary video memory. Also let _ >» 6 . , ^ 6 . , x: = x"-2 ( + x~-2 b + Z x -2 1 " 1 and y« = y"-2 7 + y *2 b + Z yTT' 21 " J i J j I j_-iJ5- 1 - J i Jsl i — 1 ' for j = 1, 2, 3 and k for a case of 1/2 x 1/2. Suppose |x.-x. ,-| =1 and ly.-j. , I = 1 for j = 1, 2 and 3« Then 1 J J+H |0 J J j+l! (x . , y.)' s are located diagonally in the original picture. (See an example J J in Figure 5(a).) Suppose d'(j, j + 1) = 1 for j = 1, 2 and 3 where d'(j, j + 1)= max (|x' - x' . I, ly! + y • . I) for j = 1, 2 and 3- Then after being shrunk to 1/2 x 1/2, four distinct points are still on the screen. (See Figure 5(b).) For this case the shrunken line is two points thick. 13 - 1 m - * x I X ^ rO - ro ro GO CJ I— I II • — "I o " ro - ro X o II ro CVJ 2 or i /Tk ro CVJ CVJ X ro * X X \ 7 \- y ■-I 1-1 l-H >— - >s p4 X T3 C o ro ce CVJ 2 or x CO s» ro X T X I r-t CM " X •—> ,_^ l—t to (0 C\J H II I 05 to 1 then j + 1 will start a new set. If I contains only one number then this point is called an isolated point. Let I 1 = {k + 2n(k + 2nel for n = 0, 1, 2, ..., and k is the smallest number in 1} . Let a 1 be written in the binary video memory at (x ! . , y'. ) for jel' J J instead of jel. Then the line is not two points thick for jel and there is no discontinuity along the line for jel. Consider the following possibilities: Case i) I contains only one number. Then I = I' by definition of I'. Case ii) j and j + lei and d'(j, j + 1) = then (x!, y') = ( x j +1 > y^ +1 ) and also d€j' or j + lei' whichever it may be it does not break the line. See an example in Figure 5(c) and (d), j -2. Case iii) j and j + lei and d'(j, j +1) =1 a) j £ I' Then there is j - lei' by definition of I'. 15 1) d*(j - 1, J) - 1 Since d is a distance function d(j - 1, J + 1)£ d(j - 1, j) + d(j, j + l) = 2. d'(j - 1, j + 1) = maxt l x 'j-i - *W' ly'j-i- y'l+iH ^maxtd^-ix.^-x.^i, (V«-|y d .i-y J+ il) - (l/2).max{|x._ 1 - x . + J , [y.^ - y. +1 |} * (l/2)'d(j - 1, j + 1) Since the picture is shrunk to l/2 x l/2 the distance between two points is shrunk by 1/2, too. Therefore d'(j - 1, j +1) = (l/2)'d(j - 1, j + 1).<: (1/2) '2 = 1 and the line is still continuous . 2) d'(j - 1, j) = This is case ii. b) j + 1 { I 1 . If j + 2 i I then the end point-point is lost but it does not effect continuity. If j + 2el then j + 2€l' and this is case iiia. In any case the continuity of a line is not effected by changing I to I'. Clearly, the thickness of a line is reduced since I' contains half as many elements as I and I ' is large enough not to break the continuity of a line. Shrinking a picture to l/U x l/k is the same as shrinking to 1/2 x 1/2 twice. 16 2.U Picture Expansion Let the pairs (x, y) be the input to RASER and let them be in the part of the original picture which is to be expanded. Let (x', y 1 ) be the input to the binary video memory and d, d' and I be defined as in the previous section. Then for h x k expansion d*(j, j + 1) ^maxtlxj - xj +1 |, \y' - jr' |) =max(V|x. - x. +1 |, *|y. +y J+1 |) = U-max{|x. -x. + J, |y. + y. +1 |} = U-d(j, j + 1) for j, j + l€l. Since d(j , j + 1) = 1 for j , j + lei, d'(j, 3 + 1) - Vl - ^ for j, j + lei. Therefore the points between (xl , y'. ) and (x . , , , y . , , ) must be J J J+l 0+1 interpolated. Because of the hardware realization implementation a first order polynominal interpolation is employed here. However, for an isolated point there is no way to interpolate and, unfortunately, an isolated point will vanish, for the expansion operation. Since R/R cycle takes at least 1 [isec and one point in the raster scan takes 200 nsec, expansion must be a two-step operation. However, for simplicity suppose expansion is a one-step operation. Since a word in the binary video memory is 16 bits long, let the word be a h x h matrix, rather than a 16 bit string, or a 1 x 16 matrix along the raster scan. Then (x', y') pairs and matrices are in a one-to-one correspondence. Let a matrix corresponding to (x', y') be M(x', y') and 17 ■» X' y M(0,0) M(1,0) M(2,0) • • • M(63,0) M(0,1) M(l,l) M(2,l) • • • M(63,I) M(0,2) M(l,2) M(2,2) • • • M(63,2) ^ M(k,l)^*3 , uuu j( • • • RASTER SCAN >». X X X X xxxx T v. • • • • • I 1 • | • 1 »j • • • 1 * • • • M(0j63) M(l,63) M(2,63) M(63,63) I I Figure 6. Location of M(x , y ) in the Binary Video Memory 18 b b l b 2 b 3 \ b 5 b 6 b 7 b 8 b 9 b lO b ll b 12 b 13 b lU b l 5 where b, is bit i of the word in the binary video memory, and the address for M(x', y 1 ), A(x', y 1 ), is defined A(x', y') = E (v-2 6 + x ) 2 1 . i = M(x', y') is defined as follows: Case i) x*. - x' = -1 for j, j + l€l If yj - n+i • - 1 ' M(x!, y<) = 10 10 10 1 if r A - y- +1 . o. M(x!, yj) = J J 1111 if yj - y^ +1 - i, r 1 1 1 19 Case ii) x\ - x' = for j, j + lei if yj - yj +1 - -X, M(x', yl) - J J 10 10 10 10 If y j " y j + i * l > M(x j> y j + i : 10 10 10 1. ° ° 2_ Since d(j, j + 1) = 1 for j, j + lei, there is no such case that x j - Si ■ ° «* yj - Si = °- Case iii) x! - x' = 1 for j, j + lei J J + i if y: - y: +1 • -i M(xj +1 , yp = 10 10 2 1111 j j+i M(x j+1 , yj) . 20 If y j " y j+l " l ' M ( x j + 1' y j + l } 10 10 10 1L ° ° _L The relation between M(x ' , y') and the binary video memory is shown in Figure 6. The raster line consists of ith row of 6h matrices and every h points a new word must be read out and the R/R cycle must be 800 nsec - this is too fast for RASER. To solve this problem let the 1 x 16 matrix, M'(k, Z) , be defined as • M'(k, Z) M*(k + 1, Z) M'(k + 2, Z) M*(k + 3, &) = [M(k, Z) ' M(k + 1, Z) ! M(k + 2, &) \ M(k + 3, A)] k x 16 k x 16 where k and I are non-negative integers less than 6k and k is divisible by k, then M"(k, Z) , M'(k + 1, Z) , M' (k + 2, Z) and M'(k + 3, &) are the matrices stored in the binary video memory for display. Conversion from M's to M"s is called format conversion. Let an address of M(x', y') A'(x', y'), be redefined as A'(x', y') = S y"2 + x -2 J + x -2 + Z x.2 . X = U i = 2 X (See Figure 7). Read out the k words, at A*(k, Z) , A'(k + 1, ^), A'(k + 2, Z) and A'(k + 3, Z) of the binary video memory. Then form a h x 16 matrix 21 ■> X 4 8 , 1 5 9 , 2 6 10 3 7 11 64 68 72 65 69 73 | 66 70 74 67 71 75 , ■ 128 132 136 4035 4039 4043 60 61 62 63 124 125 126 127 187 4095 I I 7 Figure 7. Address, A (x , y ), in the Binary Video Memory 22 [M(k, I) \ M(k + 1, i) ! M(k + 2, I) M(k + 3, &)]. Then put these matrices back into the binary video memory such that the first row of the matrix is at A'(k, I) and the second row is at A'(k + 1, I) and the third row is at A*(k + 2, 4)and the fourth row is at A'(k + 3, I). This is the second operation of the two-step operation. For a 2 x 2 expansion a matrix M(k, I) is divided into four 2x2 submatrices such that M(k, £) M i:L (k, Z) M 12 (k, i) M 21 (k, |) M 22 (k, i) Two l's are placed in each submatrix for each (x', y') pair and submatrices and (x ' , y 1 ) pairs are in a one-to-one correspondence. 23 3. LOGICAL DESIGN OF RASER In this section the logic of the various operating modes of RASER is described. A block diagram of RASER is shown in Figure 8. 3.1 Random to Serial Conversion In this operation the random input, (x, y) pair goes through the samp- ing rate control and the shrinker unchanged and is decoded at the bit decoder before going into the data selector. Since the TV monitor sync signals cannot be changed, all timing in RASER is synchronized to these sync signals. The 5MHz display clock is gen- erated by two cross-coupled monostable multivibrators. The shaped negative edge of the delayed TV horizontal sync pulse (an approximately 500 nsec nega- tive pulse) stops the display clock, and the shaped positive edge starts the display clock again. The delay for the horizontal TV sync pulse is adjustable and this determines the distance between the left edge of the picture and the left edge of the CRT. The display clock output is fed to a U-bit master counter, which is cleared by the delayed TV horizontal sync pulse. When the master counter is in the state 0011 a R/R (Read Restore) clock signal is produced (Refer to the timing chart shown in Figure 9)« At count 1000 of the master counter the out- put of the memory is transferred into a 16 bit (two 8-bit) parallel- in/serial- out shift register and shifted by the display clock. The output of this shift register is an unblanked binary video signal. At count 1011 of the master counter a R/M/W (Read Modify Write) mode signal and a R/R clock signal are pro- duced. The R/R address counter is also incremented by one at this time. At count 1111 of the master counter the C/W (Clear Write) clock signal of the R/M/W mode is produced. 2k o •H a M o o H oo 0) •rl L b 25 i ? I I I I I I I I I t, 3. °2 8 J r" ? *• o c .(TOM »> « O C (I o 2 J« u o 2 • o o o ^_ o -« m 2 -M I o c C o o >. o a m u o o '5> ■ ac * 2 u O u * Q u 5 (T V) IS £! urn kw. »»'» I » ,. -JK-1-.. "• nr *m#t ttcrrrrT „;r it ■ it « i i -~^t%D S£r® a tor .^-^ .. >siaj ' " 0^ »w ' -13UU.. . t© 1 • tiT HW1 •Miit/** 1 ,, • IIMIH «■ "^O O-F© ""»' . .. hur> e*o *oo cowTiCL • fit * cowTiot (wqiim*l. « k. . ., *l»iTt ado • o*t* ^M£> WEIIOHr DATA OUT COOiTKOL ^E>i :^o S»> :^I>^ ^3©^ ^>^. ^& Qilt*L*T CLOCH/ m OlVLAT CLOCK ::r ; ■» , ©4^Z>HI] fit rr^ I J Figure 10. Timing Control Circuit 28 I the video signals are blanked the display clock is kept running to generate R/M^ mode signals. The display clock is stopped for about 500 nsec by the negative edge of the delayed TV horizontal sync pulse and is started again by the positive edge of the delayed TV horizontal sync pulse in order to maintt! . Clearly, it is very important to adjust the clod fre<; :h that the delayed TV horizontal sync pulse occur after the gen- eration of the last C/W clock signal of the R/M/W mode but prior to generating the next R/R clock signal. See Figure ^. The logic to video converter circuit is shown in Figure 11. T, and T represent the drivers of a dual peripheral positive-AND driver, the SN75^1P, and T, is a Si low power switching transistor. D is a Ge switching diode and D„ is a Si switching diode. When T, and T are off (i.e. the binary video signal is 1), the base voltage of T , V , is equal to the forward voltage drop of D and D at about 36mA, and V, = 1.4V. The base-emitter voltage drop of T^ is .6V and V = .&V. ' b 3 out When T is on V, = .3V and V = --3V. When T is off but T n is on, 2 b out 2 1 V = .6V and V = OV. For binary video, the blanking and the video signal zero level can be the same. 3.2 Rotation and Reflection Operations The rotation and the reflection operations are done by using 1, 2 or 3 of the following 3 operations; 1. interchanging the 8 x bits with 8 y bits 2. complementing the 8 x bits and 3- complementing the 8 y bits. 29 o u •H o Q) -P ^ > a o o o 1 signal of the counter is called the controlled sample clock signal and > 1 (x., y.) is transformed to the next stage with the controlled sample clock J J I T (i.e. (x , y ) is the output of the sampler), where J J ' - 7 6 i 7 6 - i x = x -2 + Z x '2 and y = y -2 1 + Z y '2 1 . J J ' ' i=0 J ' J J '' i=0 J ' 31 fc fe^ S-53^ ~* / 121 —\ y -J, B^ Q»y — - O^ - 3^ Q*y B^ 0^ B^ B^- IB^ Figure 12. Circuits for Picture Rotation, Reflection and Shrinking Operation 32 Figure 13. Sample Rate Control Circuit 33 This counter then counts down by 1 if d(j, j+l) = 1 where d(j, j+l) = maxllx^ - x J+1 |, |y - y J+1 |). If d(j, j+l) = d(j+l, j+2) «= d(j+2, j+3) = 1 then the counter reaches the 000 state. With the next clock signal 0011 is reloaded into the counter if d(j+3, j+k) =1, (x • m y-jj,) i- s output from the sampler and the counter starts counting down again. d(j+k, j+k+1) / always (k a positive integer) since if two succes- sive input pairs are the same then the clock signals of the counter are inhibited and the second pair is ignored. If d(j+k, j+k+1 ) > 1 then 0011 is reloaded into the counter, (x . , ., , y . , , ) is output from the sampler and the counter starts counting x j+k+1 j+k+1 down again. For the 1/2 x 1/2, l/2 x 1/1+ or l/k x 1/2 mode 0001 is loaded into the counter instead of 0011. Otherwise, 000 is loaded into the counter and every (x . , y.) is output from the sampler. J J 3.4 Expansion Operation Suppose that (x . , y.) and (x . n , y. .. ) are the successive input J J J+l J+l I I I I pairs to RASER with d(j, j+l) = 1 and that (x . , y.) and (x . , y. ) are the ii it successive outputs from the sampler. Suppose also (x., y.) and (x . n . y. ,) J J J+l J+l are the points in the portion of the picture for which expansion is desired. 3.4.1 First Step of Expansion Operation The sequence of operations is as follows : it it First, the difference betwen (x . , y.) and (x . .. y. , ) is computed. J J J+l J+l The logical circuits for this are shown in Figure Ik. 3U pG> ! X" m ; A A A 1 -f ft 6 ft -9 o •H o 3 m O a o •H o ,- - : s MS 2 ■» « 2 -!&**->-. Figure 16. Format Conversion Circuit 39 U. CONCLUSION RASER has shown the feasibility of the binary digital hardware scan converter. RASER has one outstanding advantage over digital software scan conversion, namely speed. RASER performs on-line conversion from a random scan to a raster scan except in the expansion operation. Even in the expan- »n operation the conversion is almost in real time. RASER illustrates that specialized operations may frequently be done faster and/or cheaper by using job oriented, special purpose computer or ter- minal rather than a general purpose computer. In RASER the expansion mode requires a two-step operation, and it is almost real-time operation. If 6h bits (instead of 16 bits) can be read out from the binary video memory in one cycle then the second step of the expan- sion operation (format operation) can be eliminated. However, it is not eco- nomical to read out 6k bits from a core memory in one cycle but with MOSRAM's or a plated wire memory it is relatively easy and in the near future they may be cheaper than a core memory of the same capacity. The limitation on the maximum word length of a core memory prohibits the construction of a digital hardware scan converter having the speed of RASER and with gray levels. This is so even if the bits capacity of the core memory is large enough. Again, with MOSRAM's or a plated wire memory it can be accomplished. RASER has distinct advantage over analog storage tube scan converters in having less wear and tear, better reliability, better noise immunity, better registration for writing or erasing and speed of erasing. Disadvantages are lower resolution and lack of gray levels. 1+1 LIST OF REFERENCES 1. Chu, Y. Digital Computer Design Fundamentals . McGraw-Hill Book Co., New York, 1962. 2. Luxenberg, H. R., and Kuehn, R. L. Eds. Display System Engineering . McGraw-Hill Book Co., New York, 1968. 3. Metzger, R. A. "Computer Generated Graphic Segments in a Raster Display", AFIPS Conference Proceedings , Vol. $k, 1969, pp. 161-172. k. Noll, A. M. "Scanned-Display Computer Graphics", Communications of ACM , Vol. lU, No. 3- March I97I, pp. 143-150. 5. Rosefeld, A., and Pfaltz, J. L. "Distance Functions on Digital Pictures", Pattern Recognition , Pergamon Press, Vol. 1, 1968, pp. 33-61. 6. Sherr, S. "Applications of Digital Television Displays to Command Control, Proceedings of the SID , Vol. 11, No. 2, Second Quarter 1970, pp. 6I-7O. 7. Thoman, R. E. "A Computer-Generated Television Display System", Proceedings, 3rd. National Symposium. Society for Information Display , San Diego, California, February 1964, pp. I46-I67. k2 APPENDIX , , „ v v v ^3 o U ■H O u o -p o 0) H 0> K3 ra w 0) ?H Uk III; ■- - iltf Hull' V^J ;:|l *i li Si A si is Si «iUj -i &::# J o •H ■P cd Sh H a; o •H +3 CO +3 O -p o •H !m O ra -P o •H O O u -p o o ■ • m 3 -s - ■0101 00Q BED mm >v ^cgo v ^m 7>CTi£(/i(/>irti/> — M- ►Jif^^OaJ »x 4. II!= , E>-| 3>- ^O^ i£^£^P -m—> v. ii ^^E>- J (]> ^J^ rT^GD- s ©" 1. jr^iJ l ^-E>-u. SeS IO0H HA w .■ 1143 £_i rU W iv® - ± ■ ' L.W.J WT^ ± • ■ _!H! ®ic 3>a 3. JW « Control Circuits for Overwrite Protection and Memory Reflesh Rate V7 o •H -P 0) O o 1-3 in O o •H O H O ?h -P d o u zzzzzzzzz • BEBE EBEE Q000 1 HE HE U O < 4 SN7400 5 SN7404 SN74O0 SN7410 SN74M11 5N7420 SN74HZ) 2 SM7474 1 SM74|« 1 SW74107 1 SN74121 1 SN74I54 L SW75451P -> '• ^9 i+5V 2KH 1W l.SKfl 1/4W + 5V .0039 M f A > N ONE SHOT SN74121 2 T 300pf 13 jg 20Kft 1/4 W 15 ONE + 5V SN74123 0-^-1 c L l 5 4 DELAYED HORIZONTAL SYNC > B BINARY VIDEO [7 L > K > M.S. EK R > 12 > 13 > 14 >= 15 > 16 > 17 > 18 > 19 >- | SN7545IP i> 75&1/4W COMPOSIT VIDEO wv > -15V E/FC 2 -> U FC on 11 k. 1U 13fs. Id 20 > K>o-i-j6>o-l E/FC 3 -^ V E/FC 10 ■> W E/FC 13 -> X Logic to Video Converter and Delayed Horizontal Sync Pulse Circuits 50 ALL TRIM POTS ARE 6W ALL RESISTORS ARE V 4 W ALL DIODES ARE IN41S1 FUSE IS lA 1 CD4001E COSMOS ] SN7400 1 SN7404 1 SN7408 2 SN7490 DECADE COUNTER 1 SN7492 DIVIDE- BY- TWELVE COUNTER 1 SN7493 DIVIDE- BY- SIXTEEN COUNTER 4 SN74123 TOTAL 12 IC'S 51 VITA Takehiko Katoh was born in Osaka, Japan on May 11, 19^1. In March 1964 he received his B.S. in Instrumentation and Measurement Engineering from Keio University, Tokyo, Japan. Mr. Katoh continued his education in Electrical Engineering at the University of Illinois in September 1964. In February 1966 he joined the Circuit and Hardware Systems Research Group of the Department of Computer Science under Professor W. J. Poppelbaum. He has been employed by the Department of Computer Science as a Research Assistant in that group since February 1966. He received his M.S. in Electrical Engineer- ing in June 1967 and since then has continued to work under Professor W. J. Poppelbaum toward a Ph.D. degree. FormAEC-427 U.S. ATOMIC ENERGY COMMISSION (6/68) AECM 3201 UNIVERSITY-TYPE CONTRACTOR'S RECOMMENDATION FOR DISPOSITION OF SCIENTIFIC AND TECHNICAL DOCUMENT C S»» Instruction! on Rrvrm SM» I 1. AEC REPORT NO. coo 1U69-0209 2. TITLE RASER: RANDOM TO SERIAL CONVERTER 3. TYPE OF DOCUMENT (Check one): £3 a. Scientific and technical report O b. Conference paper not to be published in a journal: Title of conference Date of conference Exact location of conference. Sponsoring organization □ c. Other (Specify) 4. RECOMMENDED ANNOUNCEMENT AND DISTRIBUTION (Check one): F1 a. AEC's normal announcement and distribution procedure* may be followed. ~2 b. Make available only within AEC and to AEC contractors and other U.S. Government agencies and their contractors. ~2 c. Make no announcement or distribution. 5. REASON FOR RECOMMENDED RESTRICTIONS: 6. SUBMITTED BY: NAME AND POSITION (Please print or type) Takehiko Katoh Research Assistant Organization Department of Computer Science University of Illinois Urbana, I llinoi s 6lfi01 Signature f Date June, 1972 FOR AEC USE ONLY 7. AEC CONTRACT ADMINISTRATOR'S COMMENTS, IF ANY, ON ABOVE ANNOUNCEMENT AND DISTRIBUTION RECOMMENDATION: 8. PATENT CLEARANCE: I I a. AEC patent clearance has been granted by responsible AEC patent group. I I b. Report has been sent to responsible AEC patent group for clearance. I I c. Patent clearance not required. i. Title and Subtitle BIBLIOGRAPHIC DATA SHEET 1. Report No. UIUCDCS-R-72-515 RASER: RANDOM TO SERIAL CONVERTER 3. Recipient's Accession No. 5- Report Date June, 1972 6. J. Author(s) Takehiko Katoh 8- Performing Organization Rept. No. |. Performing Organization Name and Address Department of Computer Science University of Illinois .Urbana, Illinois 61801 10. Project/Task/Work Unit No. US AEC AT (11-1) 1U69 11. Contract /Grant No. US AEC AT (11-1) 1U69 12. Sponsoring Organization Name and Address US AEC Chicago Operations Office 98OO South Cass Avenue Argonne, Illinois 60^39 13. Type of Report & Period Covered 14. 15. Supplementary Notes It. Abstracts RASER (RAndom to SERial Converter) is an on-line binary scan converter, designed to convert pictorial information from the computer output form (RAndom) to a raster format (SERial) suitable for a TV display. RASER also performs picture rotation, shrinking or expansion along with the scan conversion. 17. Key Words and Document Analysis. 17o. Descriptors Scan converter Raster format Random to serial converter 17b. Identifiers/Open-Ended Terms 17c. COSATI Field/Group II. Availability Statement unlimited distribution 19. Security Class (This Report) UNCLASSIFIED 20. Security Class (This Page UNCLASSIFIED 21. No. of Pages 58 22. Price c ORM NTIS-3S (10-70) USCOMM-DC 40329-P71 SEP 21 1972 I luiHiiiMMiil