Computer Based Display System Allowing Mixing and Windowing of Graphics and Video

The present invention relates to the field of computer based video/graphics display systems. In particular the invention relates to a system for combining video signals and graphics signals on a common display. The graphics signals are generated by a computer, and the the video signals are from a video source, such as a video disk player.

There are many problems associated with combining video signals with computer graphics signals, but probably the most significant problem is that of synchronising the video and graphics signals so that correct placement of the signals occurs on the combined display.

A method of synchronising the computer graphics signal and the video signal is, for example, known from U.S. Pat. No. 4,599,611. In this method, the computer graphics system takes all its timing signals from the sync signals of the video source. Thus the graphics and video signals are accurately aligned relative to one another. The combined video output signal is also therefore based on the video source timing.

However, such a system limits the graphics display modes in which the computer can operate to only those modes that are compatible with the video source, ie. that have the same number of horizontal display lines per full screen of display, and that have a line rate substantially the same as that of the video source. This generally restricts the available graphics modes to low resolution modes, thereby not making full use of the graphics capability of the computer.

A further problem is that the aspect ratio for the computer graphics is usually not the same as that for usual standards of video image. Correction of the graphics aspect ratio relative to that of the video cannot easily be achieved when the computer graphics are synchronised to the video source, and hence in the combined display the computer graphics will be distorted. A circle drawn in computer graphics will appear as an ellipse in the combined display. This may limit the usefulness of the computer graphics to that of adding alphanumeric text to the video picture.

A further constraint imposed by synchronising by a generator sync for the computer graphics to the video source is that both the graphics and the video displays necessarily fill the whole of the display area on the screen.

In the present invention, the computer graphics signal and the video signal are combined using synchronisation from the computer graphics generator. The video signal from the video source is digitally encoded and asynchronously converted to a video signal synchronised to the computer graphics generator. This allows a variety of different video source signal formats to be combined with computer graphics. The combined display can be displayed using a high resolution computer monitor capable of displaying the full resolution of all the computer's graphics modes. Since the video signal is converted asynchronously, effects such as video picture magnification and reduction, as well as aspect ratio correction can be achieved.

A preferred feature of the invention is that the asynchronous video signal converter uses a dual port picture fieldstore. The input port can be clocked at a rate compatible with, and synchronised to, the video signal, and the output port can be clocked at a rate compatible with, and synchronised to, the computer graphics generator. Preferably the asynchronous converter contains two fieldstores. Complete fields of video signal can then be written into each fieldstore alternately, and the converter output can be taken from the fieldstore that contains the most recent full field, ie. the fieldstore that is not currently being written into. This improves the picture quality of moving video images in the combined display.

Another preferred feature of the invention is a timing synchroniser that is itself synchronised to the computer graphics generator. The synchroniser supplies all timing signals for the video/graphics combining system, as well as output sync signals for the display monitor.

By altering the relative timing and phase of these feature ASIC timing signals, the synchroniser can control the aspect ratio of the video image in the combined display. The synchroniser can also synthesize timing signals and sync signals for the display monitor in the event of an interruption in the sync signals from the computer graphics generator. Such an interruption will occur whenever the graphics mode of the computer is changed. The synthesized output from the synchroniser allows the video display to be maintained on the display monitor. During graphics mode changes, the graphics signals are blanked. When a new graphics mode is started, the sync signals from the graphics generator will reappear, however these will be in random phase to the synthesized sync signals being supplied to the display monitor. To maintain a continuous picture on the display monitor, the synchroniser pulls its synthesized signals back into synchronised lock with the graphics' sync signals, in a controlled manner. Once the timing signals have been resynchronised, the graphics signals can be unblanked. Thus the synchroniser will attempt to cover over the glitch that normally occurs on the display monitor whenever the graphics mode of the computer is changed.

Another preferred feature of the invention is a fading/mixing matrix for combining the video and graphics signals. The matrix is controlled on a pixel by pixel basis, enabling full video, full graphics, or a mixture of video and graphics, to be displayed at each individual pixel position in the combined display. The matrix allows windowing and overlaying of the video and graphics images in the combined display. The matrix can be controlled by the logical colors used in the graphics display. Each logical color is assigned keying/fading attributes, as well as a physical color in the normal manner.

When more than one fieldstore is employed in the asynchronous video converter, another preferred feature of the invention is a dual input switch allowing the video input to be rapidly switched between two video inputs. The switch is controlled in real time to input video fields alternately from the two sources, the inputted fields being stored alternately in the fieldstores. Thus the first fieldstore will contain the video image from one input, and the second fieldstore will contain the video image from the other input. Two independent video windows may then be displayed together with computer generated graphics in the combined display.

The invention therefore provides an apparatus for combining in a visual display on a computer display monitor, a computer graphics image and a video image. The graphics image corresponds to a graphics signal generated by, or input from, a computer graphics generator, and the monitor is synchronised to the graphics generator. The video image is derived from an input video signal asynchronous with said graphics signal, the input video signal being generated by a video source that does not need to be synchronised with the graphics generator.

A further aspect of the invention is a visual display on a computer display monitor, the visual display containing a computer graphics image and a video image. The computer display monitor is synchronised to the computer graphics generator generating the graphics signal, and the video image is derived from a video input signal asynchronous with the computer graphics signal.

A preferred feature of the invention is the display of the video image in a predefined display window area of the visual display. The displayed video image can be a cropped version of the input video image, the displayed image corresponding to a selectable input region of the input video image. The displayed video image can also be magnified or reduced in size compared to the selected region, and the display window positioned anywhere in the visual display.

The term "video image" refers to an image corresponding to a continous video signal. The image may be a moving or animated image, or a still image, or a frozen frame of a moving image.

The term "graphics image" refers to an image corresponding to a computer graphics signal from a computer graphics generator. The graphics image may include alphanumeric characters.

A preferred feature of the invention is a scaling means for selectively magnifying or reducing the scales of the image, independently in the horizontal and vertical directions. Thus the full image can be displayed in a reduced size window on the screen. Alternatively, a selected region of the input video image can be magnified, and displayed in an enlargened window on the screen. The displayed image can also be distorted in its aspect ratio by using different scaling factors in the horizontal and vertical directions.

Another preferred feature of the invention is a means for interrupting the video signal, to freeze the video image that is displayed. By freezing the video image in this way, the video input signal can be blanked without affecting the display on the computer monitor. This is particularly useful when the video source is, for example, a video disk player, allowing the disk player to search for a next image to read from the disk without disrupting the video image that is displayed.

Another preferred feature is a second display window area in the visual display in which a second video image can be displayed. The second video image can be derived from the same input video image as the first video image, or alternatively the second video image may be obtained from a different video source. The invention provides first and second fieldstore means for storing first and second video signals, respectively, corresponding to the first and second displayed video images.

When there are two video images present in the display, a further preferred feature is a display priority that can be assigned to one of the display windows, such that where the two video images overlap one another, the video image with the display priority is displayed in preference to the other video image, such that the video images appear one in front of the other, i.e. one image in the foreground of the display, with the other image partially obscured by it. The invention provides means for assigning a display priority and for displaying the images one in front of the other in a region of overlap.

In another aspect, the invention provides first and second fieldstore means for storing first and second video images, respectively, for display on the computer monitor. The image in one fieldstore can be merged or "pasted" into the image in the other fieldstore, to produce a compound image. This allows a picture to be built up in, for example, the first fieldstore by successively inputting a new image into the second fieldstore, and pasting each new image into the image currently stored in the first fieldstore. The system therefore provides means for creating a new video image picture by successively combining several other video images. The built up image can be combined with computer graphics from the computer graphics generator, the combined image being displayed on the computer display monitor. The built-up picture can also be saved, for example, to a magnetic disc, for later retrieval.

For a better understanding of the details and scope of the invention, an embodiment of the invention will now be described by way of example with reference to the accompanying drawings, in which:

FIG. 1 is a perspective view showing a PC carrying a video adaptor.

FIG. 2 is a block diagram showing the computer based video/graphic display system.

FIG. 3 is a block diagram showing a video input stage of FIG. 2.

FIG. 4 is a block diagram showing the asynchronous converter of FIG. 2.

FIG. 5 is a block diagram showing the fieldstore in the asynchronous converter of FIG. 4.

FIGS. 6 and 7 are diagrams illustrating line doubling in the asynchronous converter of FIG. 4.

FIGS. 8, 9 and 10 are tables showing the filters used in the asynchronous converter of FIG. 4.

FIG. 11 is a block diagram showing a framestore in the asynchronous converter of FIG. 4.

FIG. 12 is a block diagram showing the YUV input to the fading/mixing matrix of FIG. 2.

FIG. 13 is a block diagram showing a part of the fading/mixing matrix of FIG. 2.

FIG. 14 is a block diagram showing the output stage of the fading/mix/rig matrix of FIG. 2.

FIG. 15 is a circuit diagram showing a detail of FIG. 14.

FIG. 16 is a block diagram showing a detail of FIG. 12.

FIG. 17 is a block diagram showing a palette.

FIG. 18 is a diagram illustrating the inputs to the overlay palettes of the palette of FIG. 17.

FIG. 19 is a diagrammatic view illustrating an example of the palette.

FIG. 20 is a graph illustrating fading using the palette.

FIG. 21 is a block diagram showing the synchroniser timer of FIG. 2.

FIGS. 22 and 23 are circuit diagrams showing the sync polarity inverters of FIG. 21.

FIG. 24 is a block diagram showing the horizontal pulse generator of the synchroniser timer of FIG. 21.

FIGS. 25 and 26 are diagrams illustrating operation of the horizontal pulse generator of FIG. 24.

FIG. 27 is a block diagram showing the audio channel of the computer system.

FIG. 28 is a block diagram illustrating real-time input switched mode of operation.

FIGS. 29 and 30 are graphs illustrating operation of the fading palette as in FIGS. 19 and 20.

FIG. 31 is a block diagram showing further detail of the asynchronous converter of FIGS. 2 and 4.

FIG. 32 is a diagram illustrating cropping of an input video image.

FIG. 33 is a diagram illustrating the output video and display windows in the output display.

FIGS. 34 to 41 are diagrams illustrating how the video image can be manipulated in the visual display.

Referring to FIG. 1, a conventional computer 10 includes a computer system unit 12, a high resolution computer display monitor 14 and a keyboard 16. The system unit 12 includes expansion slots 17 into which add-on beards for the computer 10 my be plugged.

A video/graphics adapter 20 is plugged into one of the expansion slots 17 to allow the computer 10 to operate as a computer based video/graphics display system. The adapter 20 includes a base beard 18, and one or more daughter boards 19 carried by the base board 18. The base board 18 is formed with a plug portion (not shown) to plug in to an expansion slot 17.

Referring to FIG. 2, the computer includes a system motherboard 22 carrying memory 23, a central processor 21, and a computer graphics generator 24. The central processor 21 communicates with the memory 23 and the graphics generator 24 by means of a system bus 27. The graphics generator 24 generates graphics signals to be used in the adapter 20. The graphics signals are supplied to the base board 18 by means of a graphics extension bus 29, which also supplies associated graphics timing, synchronisation and vertical blanking data of the graphics signal. The system beard 22 also includes a computer audio system 8, including a sound generator and a loudspeaker.

The base board 18 includes a video input stage 26 connected to a video source 28 such as a video disk player. The video input stage contains circuitry for converting the video signal from the video source 28 into a digitally formatted video signal. The output from the input stage 26 is passed by means of a digital video bus 32 to the daughter board(s) 19. Each daughter board contains an asynchronous video converter 30 for storing and converting the video signals into signals synchronised to the graphics generator 24. The base board carries a timing synchroniser 25 connected to the graphics timing bus 29. The output from the timing synchroniser is connected to the asynchronous converter 30 to supply the correct timing signals to the asynchronous converter 30 to synchronise the output to the graphics generator 24. The output from each asynchronous converter 30 is fed to a display bus 33 which returns to the base board 18, and is coupled to a video input of a fading/mixing circuit 34. The graphics extension bus 29 is also connected to another input of the fading/mixing circuit 34. The fading/mixing circuit 34 combines the video and graphics signals on a pixel by pixel basis. The circuit 34 can select for each pixel position, full graphics, full video, or multiplicative ratio mixing of graphics and video.

The output from the fading/mixing circuit 34 is the combined display signal and this is fed to the computer's high resolution display monitor 14. The base board 18 also includes a control microprocessor 38 that directly controls each of the different parts of the video/graphics adapter 20 by means of a microprocessor control bus 36. The control microprocessor 38 is itself programmed by commands from the computer's central processor 21, sent on the system bus 27.

The different parts of the video/graphics will now be described in more detail.

The video input stage 26 is based on a conventional chip set for decoding color composite video signals to YUV signals. Such a chip is, for example, the Philips SAA9050 digital color decoder.

The video input stage has inputs for accepting video signals in RGB format, composite (PAL or NTSC) format, or SVHS format. SVHS signals are similar to composite signals, but have separate chroma and luma/sync signals. Referring to FIG. 3, the video input stage 26 has six analogue video input channels, arranged in pairs as two RGB inputs 50a, 50b, two SVHS inputs 52a, 52b, and two composite (ie. composite video, blanking and sync, CVBS) inputs 54a, 54b. The SVHS inputs 52 and the CVBS inputs 54 can support both PAL and NTSC video standards. The input channels are fed to an input selector matrix 56 that can select any one input channel and route it to a respective RGB input line 58, SVHS input line 60, or CVBS input line 62.

The RGB input line 58 is coupled to an input of an RGB to CVBS converter 64. The converter 64 may be a conventional RGB to NTSC composite video encoder, based on, for example, a Sony CXA1145H video encoder chip. The SVHS input line 60 is coupled to the input of an SVHS to CVBS converter 66 that combines the separate phase encoded chroma and sync/luma signals into a composite CVBS signal.

The outputs from the RGB to CVBS converter 64, and from the SVHS to CVBS converter 66 and the CVBS input line 62 are fed to respective inputs of a three way selector 68. The output from the selector 68 is fed to a conventional high speed video A/D converter 70 which converts analogue CVBS signals to digital form.

The output from the A/D converter 70 is fed to the digital video input of a digital multistandard color decoder 72. The color decoder 72 includes a chroma extractor 74 and a luma extractor 76 for separating the luma and chroma information in the CVBS signal. The chroma extractor 74 comprises a bandpass filter set to a centre frequency that depends on the format of the CVBS signal. For PAL the centre frequency is 4.43 MHz and for NTSC the centre frequency is 3.58 MHz. These frequencies correspond to the chroma sub-carrier frequency of the CVBS signal. The chroma data then passes through an automatic color control stage 80 and then to a quadrature demodulator 82 which can be selected to decode either PAL or NTSC encoded signals. The hue is controlled for NTSC by changing the phase of the quadrature demodulator 82. The luma extractor 76 comprises a chroma trap filter that rejects the centre frequency of bandpass filter chroma extractor 74. The luma signal is fed through a sync extractor 84 that removes the sync signal from the luma. The luminance then passes through a delay circuit 86, to match the delay in the chroma quadrature demodulator. The decoded separate digital luma and chroma signals appear at the output 88. The luma and chroma signals are multiplexed in a ratio of 4:1:1, the data rate for the luma being 13.5 MHz, and the data rate for the chroma being 3.375 MHz. Thus each sample of chroma information is associated with four samples of luma data. The chroma U and V information is multiplexed on a single 4-bit digital channel, and the luma Y information is a 7-bit signal.

Synchronisation between the different parts of the video input stage 26 is achieved by using a clock generator 90 that supplies clock and timing signals via a clock/timing bus 92. The input selector matrix 56, the three way selector 68 and the digital color decoder 72 are controlled by signals from a video input controller 95, under the control of the microprocessor bus 36.

The digital output from the video input stage 26 is fed via the video bus 32 to the input of the asynchronous video signal converter 30. Referring to FIG. 4, the asynchronous video converter 30 includes horizontal and vertical, filters and interpolators for processing the input chroma and luma data streams.

The digital input 88 is fed to a horizontal luma filter 100 comprising a set of finite impulse response filters. The particular set of filters used at any instant determines the overall bandwidth of the video input. The five filter combinations configurable in the horizontal luma filter 100 are shown in FIG. 8. The first combination A does not impose any filtering in the video signal, allowing the full video bandwidth to pass. The other filter combinations effect narrower bandwidth filtering, the narrowest band pass filter being with combination E.

The output from the filter 100 is fed to a direct input of a horizontal luma interpolator 104, and through a delay circuit 102 to a delayed input of the interpolator 104. The delay 102 introduces a predetermined delay, such that the luma sample signal sent through the delay 102 arrives at the interpolator 104 together with a subsequent luma sample sent directly to the interpolator 104. The interpolator 104 can perform horizontal scaling of the luma signal by sub-sampling at a predetermined clock rate. Horizontal scaling allows the video to be displayed in a reduced size window of the graphics screen. When sub-sampling, the interpolator selects either the current luma sample value, or an average of the current sample and the previous sample (sent via the delay 102), depending on the time dependent phase of the sub-sample clock pulse relative to the video input sample at that instant of time. If the clock pulse is nearly in phase with the luma input sample, the current luma sample value is selected, otherwise the average value is selected. When sub-sampling, the horizontal filter 100 reduces the bandwidth of the input luma data stream from the 13.5 MHz input rate, so that aliasing caused by the sub-sampling is reduced. Referring to FIGS. 8 and 9, the particular configuration of the filter 100 is dependent on the degree of horizontal scaling, the narrowest band pass filter combination E being selected for the smallest video picture size.

Referring again to FIG. 4, the converter 30 also includes a horizontal chroma interpolator 106. Processing of the horizontal chroma data stream to effect horizontal scaling is complicated by the 4:1:1 rate multiplexing of the luma and chroma signals, which means that each chroma sample has four luma samples associated with it. The chroma interpolator 106 uses a simple "nearest neighbor" selection algorithm to effect horizontal chroma scaling, because the multiplexed chroma signal is not in a suitable form for computational processing. Significant logic would be required to demultiplex, process and remultiplex the chroma information. Since the input sample rate of the chroma data stream is only 3.375 MHz (a quarter of that of the luma input rate), no horizontal filtering of chroma information is necessary, aliasing not being a problem.

Once the video data streams have been processed horizontally, they are then processed vertically and stored in a dual port framestore 110 for asynchronous conversion. The vertical processing is performed either before or after the data stream is stored in the framestore, depending on the degree of vertical scaling required. Vertical scaling is required because the video source has either 240 lines (NTSC) or 288 lines (PAL) of active video per field. Using a full computer display screen, the video image has to be displayed in 350, 400 or 480 scan lines of the 31.5 MHz output display, depending on the computer graphics mode in use. If a windowed screen is used, ie. the video is displayed in a reduced size window of the computer graphics screen, more vertical scaling is required.

The output from the horizontal chroma interpolator 106 is fed to a first selector matrix 108. The selector 108 has an output connected to the dual port framestore 110. The selector 108 also has input/output lines connected to a vertical scaler 116. The output from the dual port framestore 110 is fed to a second selector matrix 114, similar to the first selector 108, and having an output line 122, and input/output lines connected to the vertical scaler 116. In use, if the number of computer monitor display lines for the video image is less than the number of video input lines, the video signal is routed through the first selector 108 to the vertical scaler, and then returned through the first selector 108 to the dual port framestore 110. The output from the framestore 110 is fed directly through the second selector 114 to the output 122. If the number of display lines is equal to or is greater than the number of video input lines, the video signal is fed directly through the first selector 108 to the dual port framestore 110. The output from the framestore 110 is fed thorough the second selector 114 to the vertical scaler 116, and returned through the second selector 114 to the output 122.

The vertical scaler 116 includes a line store 118, a vertical luma filter and interpolator 120 and a vertical chroma interpolator 121. The linestore 118 is of sufficient capacity to hold a full width line of video data. If the video data has been horizontally scaled, more than one video line may be stored in the linestore. For example, if the video is horizontally scaled to half its original width, two video lines may be stored in the line store 118. The vertical scaler has two modes of operation, depending on the degree of vertical scaling required.

When the number of video input lines is greater than the number of display lines, the vertical scale 116 sub-samples the input lines. Each video input line is stored in the linestore 118. Vertical filtering and interpolation is performed using a subsequent video line. If the video image is being reduced in size in proportion, ie. maintaining its aspect ratio, two or more video lines may be stored in the linestore 118. This enables more extensive interpolation to be effected by using more vertical samples. The vertical interpolator sub-samples video input lines in a similar manner to the horizontal sub-sampling of the horizontal luma interpolator 104. The output from the interpolator can be either the current vertical sample, or an average of the current and previous (ie. from the linestore 118) vertical samples. The vertical filter is similar to the horizontal luma filter 100, comprising a set of finite impulse response filters to reduce the bandwidth of the vertical samples, and hence reduce the effects of aliasing. The two filter combinations that can be configured for vertical filtering are equivalent to the combinations A and B in the horizontal filter 100. FIG. 10 shows the filter combinations used depending on the degree of vertical scaling. The vertical filtering does not have to be as extensive as the horizontal filtering. The vertical chroma interpolator is similar to the horizontal chroma interpolator 104, and uses a simple "nearest neighbor" algorithm for interpolation of the vertical chroma data. As in horizontal chroma processing, no vertical filtering of the vertical chroma samples is necessary.

When the number of display lines is greater than the number of input video lines, the vertical scaler 116 interpolates (over-samples) between the video lines to generate new lines. This is equivalent in operation to "line doubling" in which new lines are generated between all of the video lines, the original and new lines being sub-sampled to obtain an appropriately vertically magnified video signal. The new lines can either duplicate the original lines, or they may be generated by interpolating between two adjacent lines.

An example of line doubling is illustrated in FIGS. 6 and 7. Say that the input video signal is in NTSC format having 240 lines of active video per field. Say that the computer graphics display is in a 480 line mode, and that it is desired to display the video image in a window set to 71% of the full screen size. Therefore, the number of display lines in the video window is 71%.times.480=340 full lines. This is more than the number of video input lines, and so line doubling is effected. The scaling factor applied by the vertical scaler is 340/240 lines=1.4. Thus the input image needs to be magnified by a factor of 1.4 to fit the display window.

Referring to FIGS. 6 and 7, the video input lines are shown diagrammatically as horizontal lines 130, and these are labelled 0, 1, 2, etc. These input lines are stored in the framestore 110, and are fed to the vertical scaler 116. The vertical scaler 116 generates new lines shown diagrammatically as dotted lines 132, between the original lines 130. The new lines are labelled 0.5, 1.5, 2.5, etc. The new lines may be duplicates of the original lines 130, eg. the line 0.5 duplicating the line 0, the line 1.5 duplicating the line 1, etc. The new lines may alternatively be averages of the adjacent original lines, eg. the new line 0.5 being the average of lines 0 and 1, the line 1.5 being the average of the lines 1 and 2, etc. The number of lines now present is double the number of original input video lines, ie. 2.times.240=480. As explained above, a scaling factor of 1.4 is required, and so the vertical scaler 116 sub-samples the doubled lines 130 and 132 to obtain the 340 required lines from the 480 doubled lines. The sub-samples are taken at interval numbers of 1/1.4=0.7, i.e. 0, 0.7, 1.4, 2.1, 2.8, 3.5, 4.2, 4.9, 5.6, 6.3, 7.0 etc. When sub-sampling, the nearest new line 132 or original line 130 is selected, the selected lines therefore being 0, 0.5, 1.5, 2.0, 3.0, 3.5, 4.0, 5.0, 5.5, 6.5, 7.0, etc. The remaining lines are discarded. This gives the correct number of lines for the display. The method of line-doubling and discarding unwanted lines does not waste video information in the display, as the new lines are merely copies of the original lines. For the purposes of illustration of the effect of the vertical scaling, this example has been described in terms of steps of line doubling, and sub-sampling. It will be appreciated that the vertical scaling is performed sequentially on the video lines by interpolation (over-sampling) as the lines are read out from the framestore 110, thereby providing a continuous data stream of video lines.

Referring to FIG. 11, the dual port framestore 110 includes two dual port fieldstores 32a and 32b, a fieldstore input selector 135 connected to select which fieldstore 32a, 32b the input data stream is sent to, and an output fieldstore selector 136 connected to select from which fieldstore 32a, 32b the output data stream is supplied. The selectors 135 and 136 are controlled to connect the framestore input and output to different fieldstores. When the framestore input is connected to fieldstore 32a, the framestore output is connected to fieldstore 32b, and vice versa. The input fieldstore selector can be controlled to feed complete fields of video alternately to the fieldstores 32a, 32b. This ensures that when video is being input into one of the framestores, there is always a complete field of video available for display in the other fieldstore. With an interlaced video input signal, even fields may thus be stored in one fieldstore 32a, and odd fields in the other fieldstore 32b.

Referring to FIG. 5, each fieldstore 32a, 32b is made up from three FIFO shift registers 500a, 500b, 500c. Each shift register has a capacity of 256K.times.4. A suitable FIFO chip device for each shift register is a Texas TMS4C1050-3. By using shift registers for the fieldstore instead of RAM, the cost and physical size of each fieldstore can be kept small. Control logic 502 is provided to step through the output from the shift registers, to provide a degree of random access to data in the registers.

The framestore input is fed in at a rate determined by the horizontal and vertical scalers that sub-sample the video signal, as described hereinbefore. The framestore has a clock input line 137 that is connected to the clock inputs of the fieldstore outputs, to control the rate at which video data is read out from the fieldstores. As described hereinafter, the clock is a 24/27 MHz signal Synchronised to the computer graphics generator 24. This achieves asynchronous conversion of the video input signal to a signal synchronised to the computer graphics generator. The selectors 135, 136 and the framestores 32a, 32b are controlled by means of a control bus 138 from the main controller 38. The digital output from the converter 30 is in the form of a 7-bit luma signal at the line doubled rate of 27 MHz and 4:1:1 multiplexed signed 7-bit color difference chroma signals at the rate of 6.75 MHz.

Referring to FIGS. 12, 13 and 14, the digital output is passed to input lines of the fading/mixing circuit 34, which combines the video signals with the computer graphics signals. The luma signal is fed through a luma Y input 144 to a luma D/A converter 150 that produces a luma signal current output. A suitable chip device for the D/A converter 150 is the Philips video chip SAA9060 D/A converter. The chip has dedicated control inputs (not shown) that enable a resolution improvement filter contained in the chip, to improve quantisation noise behavior in areas with small variation and to increase the input data width to 8-bits before D/A conversion.

The chroma data is supplied to chroma U and V input lines 140,142, respectively, and fed to chroma interpolators 146, 148, respectively. The interpolators over-sample the input chroma signals to increase the chroma data rate to 27 MHz, the same rate as the luma data. The outputs from the chroma interpolators 146,148 are fed to chroma D/A converters 154,152, respectively, that produce chroma signal current outputs. The chroma D/A converters 154,152 are similar to the luma D/A converter, and may use the same D/A converter chip device, e.g. the Philips SAA9060.

The output from the luma D/A converter 150 is connected through a first load resistor 160 to a voltage reference. The load resistor 160 converts the output current from the D/A converter 150 to a voltage signal. The outputs from the chroma D/A converters 154,152 are similarly connected through second and third load resistors, 156,158 respectively, to the voltage reference. The second and third load resistors 156,158 convert the chroma output current signals to corresponding voltage signals.

The luma and chroma analogue voltage signals are A.C. coupled to a YUV/RGB converter 164. Referring to FIG. 16, the U chroma input to the YUV/RGB converter 164 is fed to a first D.C. clamp circuit 260. The V chroma input is fed to a second D.C. clamp circuit 262, similar to the first clamp circuit 260. The D.C. clamp circuits 260,262 restore the correct D.C. voltage levels to the chroma signals. The chroma output from the first D.C. clamp 260 is fed to a first quadrant multiplier 261, and the chroma output from the second D.C. clamp 262 is fed to a second quadrant multiplier 263.

An analogue color saturation control voltage is supplied to the first and second quadrant multipliers 261,263 by means of a saturation control line 166. The gain of the multipliers 261,263 is controlled by the saturation control voltage, the larger the output from the multipliers the larger the chroma color difference signals, and hence the greater the color saturation in the video signal.

The luma input to the YUV/RGB converter 164 is connected to a third D.C. clamp circuit 226 that restores the correct D.C. voltage to the luma signal. The output from the third D.C. clamp circuit 226 is fed through a brightness selector 264 that can add an analogue brightness voltage offset to the luma signal. The brightness offset voltage is supplied through a brightness control line 168, as is described hereinafter. The selector 264 has a control input fed from the blanking line of the synchroniser 25, as described hereinafter. During the blanking interval of the input luma signal, the brightness offset is turned off by the selector 264 to allow D.C. clamps that are in the monitor 14 to be correctly set.

The luma output from the selector 264, and the gain adjusted chroma outputs from the first and second quadrant multipliers 261 and 263 are fed to respective inputs of an analogue YUV/RGB encoding matrix 270. The encoding matrix has YUV input lines and RGB output lines and performs the following conventional video matrix operations:

and

where -(R-Y)=U and -(B-Y)=V

The RGB video signal matrix output lines 170,172 and 174, respectively, are fed to respective video inputs of a fading matrix 268, described hereinafter, where they are faded and combined with computer graphics signals.

The computer graphics signals are encoded in RGB format within the computer graphics generator 24. However, the video/graphics system 20 functionally duplicates the RGB encoder within the computer graphics generator 24 so that it can process the computer graphics signals. By way of explanation, within the computer graphics generator 24, the computer graphics are sent to the RGB encoder as digital logical colors. Each logical color has a physical (i.e. real) color assigned to it, for example, the logical color 1, may be red, the logical color 2 may be white, etc. A palette is then used to convert the logical colors into analogue video signals in the three components R, G and B.

In the video/graphics system 20, a conventional graphics palette 176 is used to duplicate the palette of the graphics generator.

Referring to FIG. 17, a conventional palette 300 includes three conventional RAMDACs 302, 304 and 306. Each RAMDAC is an addressable device, the output of which is fed to a D/A converter to obtain an analogue output signal. A conventional palette is a standard Bt471 chip, which contains three 6-bit resolution D/A converters, and three main RAMs of 256.times.6-bit words. Another palette is the Bt478 chip, which contains three 8-bit resolution D/A converters, and three main RAMs of 256.times.8-bit words. For graphics use, each address in the RAMs corresponds to a logical color, and the RAMs contain the separate RGB component signal values of the physical colors assigned to the logical colors. The graphics logical colors are inputted through a common address input bus 308. The RAMs' contents can be updated by means of a load data input bus 310.

The palette 300 also includes an overlay palette, stored in an overlay section of the RAMs. For the RAMDAC 302, the RAM is partitioned into a main RAM 302a, and an overlay RAM 302b. Fifteen overlay palettes are available, and these can be selected by means of a four bit overlay input bus 312. When an overlay palette is selected, the outputs from the RAMDACs are set to predetermined values stored in the overlay palette, regardless of any addresses being selected through the address bus 308. When all of the input lines in the overlay input bus 312 are set to zero, the normal palette in the main RAM is in operation.

The palette also has a reference current input 313, used by the D/A convertors in the RAMDACs. The outputs from the RAMDACs are in the form of analogue current signals, dependent on the reference current. The reference current may be obtained by coupling the reference input through a reference resistor, to a reference voltage line.

Referring to FIG. 13, the graphics palette 176 is set up for normal palette use, the overlay palette being disabled by setting the overlay input 314 to zero.

The palette 176 is programmed with the same RGB palette values for the logical colors as the palette in the computer graphics generator 24. The palette 176 is connected to the system bus 27 of the computer system 22 by means of an input bus 186, so that the graphics palette 176 is updated whenever new palette data is written by the computer 22 to the palette in the graphics generator 24. The palette in the graphics generator 24 my also be disabled by a control command from the computer system 22. The graphics palette 176 is then also disabled from updating the RGB logical color values until the palette in the graphics generator 24 is re-enabled. This allows several alternative palettes to be installed in the computer 22. The computer system 22 cannot read data from the graphics palette 176. This is not necessary, as the palette in the graphics generator 24 contains the same palette data, and this may be read by the computer 22.

The logical color pixel data of the graphics display is inputted to the address input of the graphics palette 176 through an 8-bit graphics input bus 188, which is connected to the graphics extension bus 29. The logical color values are supplied for each pixel, line by line, to the palette 176, and the palette 176 outputs the appropriate analogue RGB color signals for the logical colors. The analogue outputs appear on RGB graphics output lines 190,192,194, respectively, which are fed to respective graphics inputs of the fading matrix 268, as described hereinafter.

The fading/mixing system also includes a control palette 195, containing three control RAMDACs. The control palette is preferably identical to the graphic palette 176, so as to match the rise and fall delay times of the palette 176. The control palette may also be connected to the same voltage reference that supplies the reference current for the graphics palette 300. The control palette 195 contains three RAMDACs for controlling the fading and mixing of the graphics and video signals at each pixel position on the display monitor, as well as the saturation of the video signal at each pixel position.

The first RAMDAC in the control palette 195 contains saturation data for the video signals at each pixel position, and has an analogue saturation output 166, connected to the YUV/RGB converter, as described hereinbefore. The second RAMDAC in the control palette 195 contains fading data for the graphics signals at each pixel position, and has an analogue Fg output 208. The third RAMDAC on the control palette 195 contains fading data for the video signals at each pixel position, and has an analogue Fv output 210. The fader outputs, Fg 208 and Fv 210, are fed to respective fader inputs of the fading matrix 268, as described hereinafter. For full graphics, no video, display at any pixel position, the Fg signal is set to MAX (i.e. its maximum current output value), and the Fv signal is set to zero. Similarly for full video, no graphics, display at any pixel position, the Fg signal is set to zero, and the Fv signal is set to MAX. Intermediate values of Fv and Fg cause fading and/or mixing of the graphics and video signals at the pixel position.

The address input to the control palette 195 is fed from the computer graphics input bus 188, although only the 6 least six significant bits of the 8-bit bus 188 are connected to the address input of the control palette 195. The remaining 2-bits of the bus 188 are connected to the overlay input of the palette 195, as described hereinafter. The logical colors supplied by the graphics generator 24 therefore refer both to physical graphics colors, as generated by the graphics palette 176, and to saturation, video fading and graphics fading combinations, as generated by the control palette 195. As will be described hereinafter, logical colors having values in the range 64 to 255 address overlay palettes in the control palette 195, and so only 64 logical colors in the range 0 to 63 can be used for fully programmable control of graphics colors and fading, and video saturation and fading. However, this is perfectly adequate for most applications.

Referring to FIGS. 13 and 18, the overlay input bus 200 comprises four input lines, labelled 0, 1, 2, 3, therefore allowing access to 15 overlay palettes. Only 5 overlay palettes are required by the fading/mixing circuit 34, and so a simple, highly redundant arrangement for addressing the overlay palette is used.

One overlay input line 320 is fed from a timing bus 187 connected to the graphics extension bus 29. The timing bus 187 includes a BLANK line that is active during the vertical blanking interval of the computer graphics. The BLANK line is coupled to the overlay input line 320. During the vertical blanking interval of the computer graphics, the input line 320 is active, and enables an overlay palette that forces the Fg and Fv signals to zero, i.e. the combined output display signal contains no video or graphics. During the vertical blanking interval, new control palette data can be read into the control palette 195 from the control microprocessor bus 36. The use of only 64 addresses in the control palette allows new tables to be loaded into the palette very rapidly. The ability to load a completely new control palette table during each vertical blanking interval allows effects such as time-varying fading of the screen display to be achieved. For example, referring to FIG. 19, say that a display 350 consists of a video window 352 comprising full video signals and no graphics signals, except for a graphics character T, and surrounded by a graphics display 354 comprising full graphics signals and no video signals. Then at an arbitrary pixel position 356 inside the video window 352, the fade values will be Fv=MAX and Fg=0. Where the graphics character appears at a pixel position 357, the fade values will be Fv=0, Fg=MAX. Similarly, at any arbitrary pixel position 358 outside the video window, the fade values will be, say Fv=0 and Fg=MAX. As will be described hereinafter, the fade values outside the video window do not matter because these are automatically set by the fading/mixing circuit 34. FIG. 29 shows the Fv and Fg values along a horizontal line 355 through the display.

Say that it is desired to fade the video window, gradually with time, into the graphics display. Referring to FIG. 20, this can be done by reloading new values for Fv and Fg into the control palette 195 during the vertical blanking interval between subsequent display frames. The new data for the control palette 195 is loaded at intervals T, the frame display time. Eventually, the Fv values in the palette will reach zero, and the Fg values in the palette will reach MAX. When this happens, the display window will have been completed faded into the graphics display, there being no video signals displayed. FIG. 30 shows the new Fg and Fv values along the line 355. The fading rate will be determined by the number of sets of new palette data that are loaded, and on whether new data is loaded into the palette on every vertical blanking interval, or whether the new data is loaded on, for example, every tenth vertical blanking interval.

Referring again to FIGS. 13 and 18, a second overlay input line 322 is connected to an "Out-Of-Window" signal line that is supplied by an In-Window generator 211, described hereinafter. The area on the display in which video signals are displayed is referred to as the video window. The purpose of the In-Window generator 211 is to generate a signal indicative of whether the current pixel position is inside or outside the video window. If the Out-Of-Window signal is active, the control palette 195 enables an overlay palette that forces the Fv signal to zero and the Fg signal to MAX. Thus the fading/mixing circuit 34 cannot attempt to display a video signal when the pixel position is outside the video-window, regardless of the logical color attributes at that pixel position. The third overlay input line 324 is connected to one of the two most significant lines in the graphics address bus 188. The fourth overlay input line 326 is connected to the other of the two most significant lines. Thus logical graphics colors in the range 64 to 255 will enable overlay palettes rather than allow pixel-by-pixel control of the saturation and fading levels.

Referring to FIG. 13, the In-Window Generator 211 is contained in a custom hardware chip, built from logic gates. The Generator 211 contains an interface 214 that is coupled to the microprocessor control bus 36, an In-Rectangle generator 218, an In-Window Routing Controller 220, and a memory controller 216. The memory controller accesses a 1 MBit FIFO RAM 224. The Generator 211 also includes a Sync and Timing Generator 212, which is fed from the timing bus 187. As explained hereinbefore, the In-Window Generator generates for each pixel position on the screen, a signal indicative of whether the pixel is inside or outside the video window. The In-Window calculation cannot be performed by a simple rectangle calculation on the screen co-ordinates, since the video window may be overlapped in arbitrary ways by other windows, and the adapter 20 may allow multiple video windows which may themselves overlap.

A bit-plane containing the in-window information for each pixel is built up in the RAM 224. A logical 1 bit in the bit-plane corresponds to the pixel position being inside the video window. The in-window data is read out sequentially from the RAM 224, at the same rate as the graphics pixel logical color information is supplied from the graphics generator 24. The BLANK signal in the timing bus 187 is used to synchronise the beginning of the In-Window bit-plane to the first graphics pixel position. The RAM 224 contains two In-Window bit-planes, one being used for display while the other is built up. The size of each bit plane is 640 pixels.times.480 lines=307 kbits. Thus the two bit-planes are easily stored in the 1 MBit RAM, which is double-buffered to allow independent access to the bit-planes. The memory controller 216 controls access to the bit-planes in the RAM 224, and the In-Window Routing Control 220 selects which bit-plane is used as the active bit-plane, and which bit-plane is to be built-up. The sync and Timing Generator 212 synchronises and controls the rate at which the In-Window data is read-out to the Control Palette 195. A new bit-plane is built up each time the video window layout is altered. To build up a new bit plane, the contents of the bit plane are cleaned, and new video window rectangle data for each asynchronous converter 30 is supplied by the control processor 36. For each set of rectangle data, the contents of the bit-plane are read out sequentially from the FIFO RAM 224, and, for each bit, if the bit is within the video window rectangle, the in-window generator sets the bit to a logical 1. If the bit is not within the video window, the bit value is no altered. The bit is then recycled into the FIFO bit plane. Thus overlapping windows will be correctly built up in the bit-plane, even if the resulting video window is not rectangular.

Time taken to build up a new rectangle in the bit plane is the same time as for a display field.

The fading/mixing circuit 34 also includes a brightness D/A converter 228 which generates the analogue brightness control signal on the brightness control line 168, referred to hereinbefore. The brightness D/A converter 228 is controlled from the microprocessor control bus 36. The brightness control signal introduces a brightness offset into the luma signal in the YUV/RGB converter 168, thereby increasing the brightness of the video signal to compensate for the normally bright graphics signals that are produced by the computer system.

Referring to FIG. 14, the video/graphics fading matrix 268 has RGB graphics inputs 190,192,194, respectively, and RGB video inputs 170,172,174 respectively. The R video input 170 and the R graphics input 190 are coupled to respective input of a first two-quadrant multiplier 240. The G video input 172 and the G graphics input 192 are coupled to respective inputs of a second two-quadrant multiplier 242. The B video input 174 and the B graphics input 194 are coupled to respective inputs of a third two-quadrant multiplier 242. The Fg and Fv fader line inputs 208,210 respectively, from the control palette 195 are supplied to respect inputs of each of the two-quadrant multipliers 244.

FIG. 15 shows in more detail the first two-quadrant multiplier 240. The second and third multipliers 242,244 respectively, are identical to the first multiplier 240. Referring to FIG. 15, the multiplier 240 contains a video operational transconductance amplifier 252 fed from the video input 170. The Fv video fade signal line 210 is coupled to a gain control input of the amplifier 252. For a fade value of Fv=MAX, the gain of the amplifier 252 is two. The multiplier 240 also contains a graphics operational transconductance amplifier 254, similar to the video operational transconductance amplifier 252, fed from the graphics input 190. The Fg graphics fade signal line 208 is coupled to a gain control input of the amplifier 254. The output from the video amplifier 252 is a current Iv, and the output from the graphics amplifier 254 is a current Ig. The currents Iv and Ig are summed at a node 255 and passed through a load resistance 256. The output from the two-quadrant multiplier 240 is a combined voltage signal line 258, providing an output voltage equivalent to the sum of the fade adjusted R graphics and fade adjusted R video signals. If the Fv signal is zero, the Iv output is zero and the combined output voltage is thus merely the graphics input signal. Similarly, if the Fg signal is zero, the Ig output is zero, and the combined output voltage is merely the video input signal. Intermediate values of Fg and Fv provide multiplicative ratio mixing of the R video and R graphics signals.

The second and third two-quadrant multipliers, 242,244, respectively, operate in the same manner as the first two-quadrant multiplier 240, to provide combined G and combined B signals.

The output from the first two-quadrant multiplier 240 is fed through a buffer amplifier 245 to drive an R display output line 246. The output line 246 is driven via a 75 Ohm resistor 340 and fed to the R input of the computer monitor 14. Similarly the output from the second two-quadrant multiplier 242 is fed through a buffer amplifier 247 to drive a G display output line 248. The output line 248 is driven via a 75 Ohm resistor 342, and fed to the G input of the computer monitor 14. The output from the third two-quadrant multiplier 244 is fed through a buffer amplifier 249 to drive a B display output line 250. The output line 250 is driven via a 75 Ohm resistor 344, and fed to the B input of the computer monitor 14. The video inputs to the monitor are conventionally terminated 75 Ohm resistors (not shown) inside the monitor 14.

Referring to FIG. 21, the timing synchroniser 25 has a 2 line input sync bus 360 connected to the video extension bus 29. The input sync bus 360 feeds separate horizontal and vertical sync pulses on lines 360a, 360b respectively from the computer graphics generator 24 in to the synchroniser 25. The input sync bus 360 passes through an input polarity selector 362. This is required because the sync pulses from the graphics generator 24 can be of either positive or negative polarity, depending on the number of visible lines on the display. The polarity of the sync signals is used to control the line spacing of the monitor 14 for 480 or 400 line modes. The polarity selector 362 can selectively invert the sync signals to form a positive polarity, under the control of a polarity control bus 364. Referring to FIG. 22, the horizontal and vertical graphics sync inputs 360a and 360b, respectively, are fed via resistors 416a, 416b respectively to inputs of exclusive-OR gates 418a, 418b respectively. The resistors 416a, 416b suppress ringing on the inputs 360. The polarity control bus 364 has horizontal and vertical control lines 364a, 364b respectively, and these are coupled to the other inputs of the exclusive-OR gates 418a, 418b, respectively. The exclusive-OR gates 418a, 418b have respective output lines 366a, 366b. When an active signal is applied by the control lines 364a, 364b, the exclusive-OR gates 418a, 418b invert the incoming horizontal and vertical sync signals, thereby inverting the sync polarity. The control lines 366 are fed from a vertical sync microprocessor 398 described hereinafter. This is, in turn, set up by the system CPU 21.

Referring to FIG. 21, the graphics horizontal sync output 366a from the selector 362 is fed to one input of a phase comparator 368 based on a conventional HC 4046A chip. A horizontal pulse generator (HPG) 370, described in more detail herein after, has a first output 372 connected to the other input of the phase comparator 368. The phase comparator 368 has an output which feeds to an input of a voltage controlled oscillator (VCO) 374. The output from the VCO 374 is fed back to clock input of the HPG 370, through a clock rate prescaler 375. The HPG 370, the phase comparator 368 and the VCO 374 thus form a conventional phase locked loop. The VCO may be a conventional Colpitt's oscillator, based on a single transistor design with a tuned circuit for feedback. The VCO 374 runs at approximately 40 to 65 MHz which when divided by the prescaler 375 give a clock rate in the region of 20 to 32 MHz. This range ensure that, despite tolerance variations, the required video clock frequency of 24 to 27 MHz can be met. The prescaler 375 divides the output from the VCO 374 by two, and provides an output signal having an equal mark/space ratio. This signal is fed by a video clock line 412 to the display bus 33, and to the video clock inputs of the D/A converters 154, 152, 150 (FIG. 12). The HPG has five other outputs. The second output 376 provides a horizontal sync output for the display bus 33. The fourth output 380 provides a vertical blank signal for the YUV/RGB encoding matrix, which controls the selector 264 (FIG. 16) to turn off the brightness offset signal introduced by the brightness control line 168 during vertical blanking. The fifth output 382 provides a clamp signal for the YUV/RGB encoder 164 (FIG. 16), which controls the clamps 260, 262 and 266 to clamp to the video D.C. level during blanking. The sixth output 384 provides a vertical blank signal for the D/A converters 154,152 and 150 (FIG. 12). When active, the sixth output 384 forces the D/A converter outputs to zero (i.e. no luma, no chroma) so that the signal can be A.C. coupled to the YUV/RGB converter 168, and clamped therein. The sixth output 384 also provides synchronisation for the chroma demultiplexing, and goes to the display bus 33 to enable output from the asynchronous converters 30. The durations and positions of the output signals from the HPG are independently programmable.

Referring to FIG. 24, the HPG 370 contains a gate array 386 for storing twelve 8-bit time durations, labelled t.sub.1 . . . t.sub.12, and for storing twelve 6-bit output state values, labelled x.sub.1 . . . x.sub.12. The time durations are fed sequentially through a duration-bus 388 to an 8-bit count down timer 390. The corresponding output states are fed sequentially through a output-state-bus 392 to a 6-bit output buffer 394. The output lines from the buffer form the six outputs 372,376,378,380,382 and 384 from the HPG 870. The values stored in the gate array 386 can be updated by means of a block load through an input bus 391. The HPG 370 also includes HPG control logic 393 that controls the operation of the count down timer by means of a control line 395. In operation, a time duration value t.sub.1 is loaded from the gate array 386 into the count down timer 390, and a corresponding output state value x.sub.i is loaded simultaneously into the output buffer 394.. The count down timer 390 then counts down from the value t.sub.i to zero, on clock pulses fed in through a clock input 396. When the count down timer reaches zero, the next time duration value t.sub.i+1 and its corresponding output state value x.sub.i+1 are loaded from the gate array 386 in to the time 390 and the output buffer 394, respectively, and the process is repeated. Once the values t.sub.12 and x.sub.12 have been loaded, the HPG 370 recycles to the first values t.sub.1 and x.sub.1. The HPG 370 can thus produce timing signals with an arbitrary period and phase timing relationship.

FIGS. 25 and 26 illustrate operation of the HPG 370, with reference to the three least significant output state lines, 384, 382 and 380, (in order of increasing significance). The output signals A, B and C refer to the signals appearing at the output lines 384, 382 and 380, respectively.

Referring to FIG. 21, the synchroniser 25 also includes a vertical sync microprocessor 398. The microprocessor 398 may, for example, be a mask programmed 80C51 chip device, which is controlled through a serial bus 400 from the control microprocessor 38 (FIG. 2). The vertical sync microprocessor 398 counts display bus horizontal syncs, and when the count exceeds a predetermined modulus value, a vertical sync pulse is generated. The vertical sync pulse appears as an output on a vertical pulse output line 402. The predetermined modulus value depends on the display field rate. The vertical pulse output 402 is fed through a multiplexer 404, where it is associated with the second output 376 from the HPG 370. The multiplexer 404 also has two other inputs, connected to the graphics vertical and horizontal sync outputs 366a, 366b, respectively from the input polarity selector 362. The multiplexer has a monitor sync output bus 406 comprising a horizontal sync line 406a and a vertical sync line 506b. Depending on a control input 408, the multiplexer 404 either routes the graphics horizontal and vertical sync lines 366a, 366b respectively, to the horizontal and vertical monitor sync lines 406a and 406b respectively, or it routes the horizontal pulse line 376 and the vertical pulse line 402 to the horizontal and vertical monitor sync lines 406a, 406b respectively. The control input 408 is connected via a port (not shown) to the computer system 22 (FIG. 2). This allows the computer 22 to bypass the synchroniser 25, and feed the sync signals from its graphics generator 24 directly to the monitor 14. This is normally used when the computer 22 is first switched on, as the vide/graphics adapter 20 will not at that time be properly initialised, and so would not itself enable a display to be generated on the monitor 14. Once the computer 22 has properly set up the vide/graphics adapter 20, the multiplexer 404 can be switched to use the output horizontal and vertical pulse sync signals from the synchroniser 25.

The monitor sync output bus 406 is fed to an output polarity selector 410, similar to the input polarity selector 362. The output polarity selector 410 has a control input coupled to the polarity control bus 364. Referring to FIG. 23, the horizontal and vertical output sync lines 406a, 406b, respectively, are connected to inputs of respective exclusive-OR gates 420a, 420b. The horizontal and vertical control lines 364a and 364b, respectively, are connected to the other inputs of the exclusive-OR gates 420a, 420b. The outputs from the exclusive-OR gates 420a, 420b are connected through respective buffer amplifiers 422a, 422b and output resistors 424a, 424b, respectively to drive the monitor 14. The vertical sync microprocessor 398 also has a graphics vertical sync input fed from the graphics vertical sync line 366b.

The signal lines supplied from the synchroniser 25 to the display bus 33 are buffered by a bus driver buffer 414, controlled by an enable line 426 from the vertical sync processor 398. The outputs from the buffer 414 are tristated, such that when the enable line 426 is inactive, the outputs from the buffer are set to high impedance. This allows several baseboards 18 to be connected together in the adapter 20, for example, for multiple video signal inputs, with only one synchroniser 25 being enabled to supply timing and synchronisation signals to the display bus 33.

In operation, the phase locked loop formed by the HPG370, the phase comparator 368, the VCO 374 and the prescaler 375, locks on to graphics horizontal sync signals fed in on the sync input bus 360. The timing signals produced by the HPG 370 are therefore synchronised to the incoming graphics sync signals. By altering the relative phases of the different horizontal, and vertical blanking, timing signals the video window can be moved both horizontally and vertically in the display screen relative to the computer graphics. The graphics position may also be finely adjusted, relative to the monitor sync signals.

As explained hereinbefore, the vertical sync processor 398 produces vertical sync signals by counting horizontal sync pulses generated by the HPG 370, and generating a vertical sync pulse whenever the count exceeds a predetermined modulus value. The vertical sync processor 398 also monitors the phase difference between the vertical sync pulses that it produces and the graphics vertical input pulses supplied through the graphics vertical sync line 366b. If the two vertical syncs are not in phase, the vertical sync processor 395 modifies the modulus value to bring the syncs into phase in a controlled manner. The modulus value can be increased or decreased by up to eleven lines from its predetermined value, set by the graphics mode. This has the effect of increasing or decreasing, respectively, the number of lines that occur during the vertical blanking interval of the output vertical sync signal. The two vertical sync signals are thereby brought into phase smoothly, without too great an effect on the display screen.

During graphics mode changes in the graphics generator 24, the vertical and horizontal syncs from the graphics generator 24 are halted until the new graphics mode is set up in the graphics generator 24, and then properly entered. To avoid any significant disruption in the display screen, the synchroniser 25 continues to generate horizontal and vertical sync pulses even though the graphics horizontal and vertical sync pulses to which the generated pulses are normally locked, are absent. Thus although the graphics display from the computer is inhibited, the video window can still be displayed on the monitor 14.

When the new graphics mode is entered, the graphics generator 24 restarts the graphics horizontal and vertical sync signals. However, these signals are restarted with random phase relative to the horizontal and vertical sync pulses generated by the synchroniser 25. The phase locked loop controlling the HPG 370 quickly brings the HPG 370 back into phase lock with the graphics horizontal sync signal. This causes negligible effect on the display screen. The phase difference between the graphics vertical sync signal and the output vertical sync signal is detected by the vertical sync microprocessor 398 and, as explained hereinbefore, the vertical sync microprocessor increases or decreases the number of lines in the vertical blanking portion of the output vertical sync, in order to bring the two vertical sync signals back into phase lock. Thus a stable display screen is maintained during mode changes, the mode change causing negligible visible disruption.

New data is written in to the HPG 370 if required during the graphics mode change. The new values are supplied from the vertical sync microprocessor 398 through a two bit serial data bus 430. The new data is block loaded into the gate array 386 of the HPG 370, the block load causing a momentary glitch in the display screen.

The vertical sync microprocessor 398 also includes a monitor identification input bus 432 connected to the monitor socket. The monitor identification bus 432 returns the monitor identification bits from the monitor socket to the vertical sync processor 398.

Referring to FIG. 27, the audio sub-system 9 is based on a conventional sound fader control circuit (SOFAC) 440, for example, a TEA6300T sound chip. The audio sub-system 9 includes three pairs of stereo inputs, 442, 444, 446, the left channel inputs being labelled 442a, 444a and 446a, and the right channel inputs being labelled 442b, 444b, 446b, respectively. The first input channel 442 is associated with the first video input, and the second audio input channel 444 is associated with the second video input. The third audio input channel 446 is coupled to the daughter board connector for future expansion.

The audio input channels are fed to respective inputs of a source selector 448 contained within the SOFAC 440. The SOFAC 440 also contains a volume, balance, base, treble control section 450 and two stereo faders 452 and 454 which are independent of the volume control. The stereo output from the source selector 448 is coupled through resistors 449 and passes through the volume, balance, base, treble control section 450, and in parallel to the inputs of the faders 452, 454. The output from the first fader 452, is passed through attenuation resistors 456 and to the inputs of a stereo power amplifier 458. The power amplifier 458 can typically drive 2.times.75 mW into 32 Ohm stereo headphones. A selector 460 can short circuit the stereo channels to convert the audio signals to mono. The selector 460 is controlled by a mono select control line 462 coupled to the control microprocessor 38. The stereo output from the second fader 454 is fed via two attentuation resistors 464 to a mono input of a mono buffer-amplifier 466. The output from the buffer-amplifier 466 is passed through a 1K2 impedance matching resistor to a pole of a selector matrix 470. Another pole of the selector matrix 470 is connected through a DC coupling capacitor 472 to the audio node of the audio system 8 in the computer 22. The selector matrix 470 has a third pole coupled through resistors 474 to the input of the volume, balance, base and treble control section 450 of the SOFAC 440, where it is mixed with the stereo output from the source selector 448. The selector matrix 470 is controlled by means of two control lines 476 coupled to the control microprocessor 38.

In operation, under the control of the control processor 38, stereo audio input signal from one of the input channels 442, 444, 446, is selected by the source selector 448 and processed by the volume, balance, base, treble control section 450. The stereo audio output from the control section 450 can be fed by the first fader 452 to the stereo amplifier 458 and to a stereo output socket. The output from the control section 450 can also be fed through the second fader 454 to the buffer 466 and to the selector matrix 470. The selector matrix 470 controls whether audio signals are sent from the audio sub-system 9 to the computer audio system 8 or whether audio signals from the computer audio system 8 are fed via the resistors 474 to the input of the SOFAC 440.

The video/graphics adapter 20 is also operable in a real-time switched input mode, in which video signals from two independent synchronised video sources may be displayed in two independent video windows on the display screen. FIG. 28 shows the parts of the adapter 20 that are switched when operating in the real-time switched input mode. Referring to FIG. 28, a first video source 480 is connected to a first video input for example, the first and composite video input 54a, and a second video source 482 is connected to a second video input, for example, the second RGB input 50B. The video inputs are fed to the video input selector 56 in the video input stage 26. The selector is controllable by means of a control line 484. For clarity, the remaining parts of the input stage 26 have been omitted.

The output from the video input stage 26 is fed to the asynchronous converter 30. As explained hereinbefore, the input fieldstore selector 135 selects which of the first and second fieldstores 32a, 32b respectively, the video input data is stored. The output fieldstore selector 136 selects from which fieldstore the video data is to be asynchronously read. The input and output fieldstore selectors 135,136 respectively, are controlled by respective control lines 486 and 488, such that complete video input fields are stored alternately in the fieldstores 32a, 32b and the output is always read from the fieldstore that is not being written into. In the real-time switched input mode, the video input selector 56 is controlled to switch alternately between the first and second video input sources 480, 482 respectively at the same time as the fieldstore input selector 135 is switched once a complete video field has been inputted. Thus, say the video input selector 56 selects the first video source 480, and the input fieldstore selector 135 selects the first fieldstore 32a. A complete video field from the first video source 480 will be stored in the first fieldstore 32a, and at the same time, the contents of the second fieldstore will be outputted for display on the display screen. Once a complete video field has been inputted and stored in the first fieldstore 32a, the video input selector 56, the fieldstore input selector 135 and the fieldstore output selector 136 will all be switched to their opposite sense. A complete video field from the second video source 482 will then be inputted and stored in the second fieldstore 32b, while at the same time the contents of the first fieldstore, i.e. the stored field from the first video source 480, will be displayed on the display screen. Once the complete field from the second video source 482 has been inputted, the selectors 56, 135 and 136 will again switch sense, allowing input from the first video source 480, while the stored video field from the second video source is displayed.

The real-time switching is performed on every complete field of input video data, thereby allowing apparent simultaneous display of two video signals.

Further features of the video/graphics adapter 20 described hereinbefore will now be described, which enable the video input signals to be manipulated, and mixed with the computer graphics generated by the host computer system 10.

FIG. 31 shows further detail of the asynchronous converter 30. In FIG. 31, the detail of the asynchronous converter 30 described hereinbefore with reference to FIGS. 4 and 11 is shown in the dotted circuit box 500. The horizontal and vertical filters and interpolaters shown in FIG. 4 are shown together as a horizontal and vertical scaler 502, and the scaler 502 is shown together with the first and second fieldstores 32a and 32b, respectively. Additional control logic is also provided in the asynchronous converter 30, which additional logic is under direct control of the control processor 38, and programmed by software in the host computer 10.

The video input 504 to the asynchronous converter 30 passes first through an input region cropper 506. Referring to FIG. 32, under the control of the control logic, the input cropper 506 has programmable X and Y co-ordinates defining a rectangular region 508 of an input video image 510. In FIG. 32, the video image 510 comprises a simple geometric shape in the form of a circle with a square in its top right quadrant. All of the video information 510a outside the input rectangle 508 is discarded, and all of the video information 510b within the input rectangle 508 is passed through to the circuit 500. Two input rectangles 508a and 508b are provided in the input cropper 506, the first rectangle 508a being associated with the first fieldstore 32a, and the second rectangle 508b being associated with the second field store 32b. As described in more detail hereinafter, the two input rectangles 508a and 508b enable the fieldstores 32a and 32b, respectively, to receive different parts of the same video image, or to receive identical parts of the image, depending on the operating mode of the apparatus.

The control logic includes a circuit for interrupting the input video signals to one or both of the framestores, so as to freeze the video image stored in the framestore.

The first fieldstore 32a has an output 507 which can be selectively fed back to the input of the input cropper 506 and used to supply an input signal to the cropper 506 instead of the video input 504. This configuration is used in the Freeze Frame Manipulation mode of operation described hereinafter, however, in other operating modes the input for the cropper 506 is taken from the video input 504.

The output from the circuit 500 is fed to an output field positioner 512. As explained hereinbefore, the video image from one or both of the framestores is displayed in a video window area defined in the display. The positioner includes control logic for defining for each of the first and second fieldstores 32a and 32b, respectively, a rectangular display window area in the visual display. In use, the video output from each fieldstore 32a, 32b is itself displayed in the respective display window, however, if the boundary of either display window lies outside the video window, only the portion of the video image that is inside the video window is visible on the monitor screen. As explained hereinbefore, the fading/mixing matrix 34 automatically crops any parts of the video images that would appear outside the video window. The positioner 512 controls the positioning of the display windows in the visual display, for the first and second fieldstores 32a and 32b, respectively.

The control logic acts as scaling control logic to determine from the relative sizes of the input region of the input video image and the display window, the appropriate horizontal and vertical scaling factors for each displayed video image. The control logic controls the horizontal and vertical scaler 502 to scale the video image stored in the framestores 32a and 32b, in both the horizontal and vertical directions, to fill the display windows. Separate scaling factors are determined for each of the first and second display windows.

The control logic also includes a circuit for selectively assigning a display priority to one of the display windows, or in other words to one of the fieldstores 32. The display priority determines which of the video display windows has display priority if the first and second display windows overlap in the display. Under control of the control logic, the positioner/cropper 512 displays the image from the fieldstore that has display priority, and crops the region of the display image from the other fieldstore that lies in the overlapping region of the display windows.

The output from the output cropper 512 is fed to a merge/display switch 516. The switch has a "merge" output 518 which is routed back to the input of the circuit 500, and a "display" output 521 which forms the video output from the asynchronous converter 30. In use, under the control of the control logic, the switch 516 routes video signals to the display output, and/or to the merge output. As described hereinafter in more detail, in the Picture Build-up mode of operation, the switch 516 acts as means for feeding the output signal from the asynchronous converter, representing a frame of the video images, to the fieldstore input, thereby enabling video information stored in one fieldstore 32 to be merged with the contents of the other fieldstore 32. In the other modes of operation, the switch 516 is set to send the signal to the display output 521.

A programmable background colour generator 519 is also coupled to an input of the output positioner/cropper 512. The background generator 519 generates a background colour to fill in any area in the video window which is not occupied by a display window. FIG. 33 illustrates a video image being displayed on the computer monitor 14, in a display window 522 (shown in solid line), which is smaller than the video window 520, (shown in dashed line), and offset relative to the video window 520. The portion 523a of the display window 522 within the display window 520 is visible on the screen, however, the portion 523b of the display window 522 outside the video window 520 is not visible. The area 524 in the video window 520 which is not occupied by the display window 522 is filled in with the background colour.

The output from the circuit 500 is also connected to downloading logic 526 which communicates via an interface 525 with the host computer 10 to enable video image data to be transferred from one or beth of the fieldstores 32a and 32b to the host computer, where it may be processed or saved on a disk. Uploading logic 527 also communicates with the interface 525 and provides an input 509 to the circuit 500 to enable video image data to be loaded from the host computer 10 into one or both of the fieldstores 32a and 32b.

The system therefore enables an input video image to be cropped, and selected regions of the input image to be stored in one or both of the fieldstores 32. A portion of the image stored in the fieldstore can be selected for output to the display monitor, and the output portion can be scaled in the horizontal and vertical directions by the scaler 502. The output image can be displayed in a randomly movable display window on the display monitor, and mixed or combined with computer graphics. The horizontal and vertical sync pulse processors 370 and 398 (FIG. 21) in the timing synchroniser 25 act as means for synchronising the computer display monitor 14 to the graphics generator 24, and the fading/keying matrix acts as means for combining the video and graphics images, pixel by pixel, according to at least one attribute of the graphics signal. In particular, the graphics processing portion of the matrix constitutes means for generating the graphics image in the display, and the video processig portion of the matrix together with the asynchronous converter constitute means for generating a video image in the display, in the display windows.

The control logic in the asynchronous converter enables the asynchronous converter 30 to be programmed from the host computer 10 by the use of the following command parameters:

INPUT REGION. This defines the input region of the video image that is inputted to the fieldstores, the part of the video signal outside the input region being cropped and discarded. An input region is definable for each fieldstore 32a, 32b.

DISPLAY WINDOW. This defines the size and position on the screen of the display window for each fieldstore 32a, 32b. Horizontal and vertical scaling factors for the scaler 502 are calculated automatically by comparing the display window size with the size of the input region. The position of the display window on the screen can be defined as an absolute position, ie. relative to the an origin fixed relative to the screen, or as a relative position, ie. relative to the origin of the video window.

VIDEO WINDOW. This defines the size and position on the screen of the video window. Any areas of the display windows that lie outside the video window are automatically cropped so that video images are only displayed within the video window. If a display window lies completely outside the video window, the contents of the display window will not be displayed.

VIDEO BACKGROUND COLOUR. This defines the colour that is generated by the background generator 519.

In use, the video/graphics adapter operates in one of five modes, as follows:

1. Normal Mode

2. Freeze Manipulation Mode

3. Parallel Mode

4. Dual Mode

5. Picture Build-up Mode

In the NORMAL MODE, a single video signal is input to the adapter, and a single live video image is displayed on the display monitor. As explained hereinbefore, with an interlaced video input signal, the even and odd fields of the signal can be stored alternately in the first and second fieldstores 32a and 32b, respectively. Alternatively, the input video image may routed through only one of the fieldstores, for example only through the first fieldstore 32a.

Thus the shift registers' output can be selected either alternately on consecutive display fields to display the images alternately on a field by field basis, or one of the shift registers can be selectively controlled to provide a continuous output.

Some of the video image manipulation effects that can be achieved with the adapter will now be described, with the adapter operating in the Normal Mode.

Referring to FIG. 34, the video image 510 is input to the adapter. The input region for each fieldstore is set to the maximum input field, and thus the video image 510 is uncropped. The image 510 is displayed at full size on the display 14 using a full size display window 522 for each fieldstore 32. The video window 520 is defined to cover the whole of the screen. The coincident video window and display window are shown by a dashed line.

Referring to FIGS. 35a and 35b, the display can be altered by reducing the size of the video window, while keeping the input region set to input the full image size. Thus, the whole of the video image is stored in the fieldstore, however, only the portion of the image inside the video window 520 is visible on the screen.

In FIG. 35a, the video window 520 is set to display the top right-hand quadrant of the screen. In FIG. 35b, the video window is the same size as that in FIG. 35a, however, the video window has been moved to the centre of the screen so that only the centre of the video image 510 is visible.

FIGS. 36a and 36b illustrate input cropping of the video image 510, by reducing the input region for the fieldstores 32, to select the top right-hand quadrant of the image 540 for input, and discard the remainder of the video image. In FIG. 36a, the display window is set to display the whole of the image stored in the fieldstore, in the top right hand corner of the screen. The display therefore looks identical to that in FIG. 35a. In FIG. 36b, the position of the display window has been moved relative to the screen, but without changing input region. In FIGS. 36a and 36b, the video window is set to the full screen size, so that the area 524 outside the display window is filled with the video background colour.

FIGS. 37a and 37b illustrate the effects of horizontal and vertical scaling of the video image 510. In FIG. 37a, the input region is set to full size so that the whole of the input video image 510 is input into the fieldstores. The size of the display window is set to be 50% of the full screen size. This causes the video image in the fieldstores to be scaled by a factor of 50% in both the horizontal and vertical directions, by the scaler 502. In FIG. 37b, the shape of the display window 522 on the screen has been distorted relative to the aspect ratio of the input video image. The whole of the video image is still displayed in the display window 522, however, the image is scaled by a factor of 50% in the vertical direction, and by a factor 180% in the horizontal direction, to produce a flattened image.

FIGS. 41a and 41b illustrate the effect of image magnification in the display window. In this FIG. 41, the input region is set to only the top right quadrant of the input image. In FIG. 41a and this image is displayed in a display window 522 which is defined to cover the whole display. Thus, the displayed image is magnified by a scaling factor of 2 in both the horizontal and vertical directions, so that the displayed image fills the display window. In FIG. 41b, the display window is defined to cover only the left hand side of the display, such that the shape of the display window is distorted relative to the aspect ratio of the selected region of the input video image. In this figure, the displayed image is magnified by a scaling factor of 2 in the vertical direction, but image is not altered by scaling in the horizontal direction, ie. the horizontal scaling factor is one.

In the NORMAL MODE, with both fieldstores operating, the input region and the display window size and position for the first and second fieldstores 32a and 32b, are set to be identical.

Referring to FIG. 31, in the FREEZE MANIPULATION MODE, the input 504 to the asynchronous converter 30 is interrupted, and a full size copy of the last frame of the video image is kept in the first fieldstore 32a. The output 507 from the first fieldstore 32a is fed through the input cropper 506 to the input of the second fieldstore 32b as a continuous video signal representing the frozen video image. The image can be manipulated as described above to alter its aspect ratio, size and positioning on the screen. In this mode, a full size copy of the original frozen image is kept in the first fieldstore 32a, and so the frozen image displayed on the screen from the second fieldstore 32b can be positioned and resized as if it were a live video input.

Referring to FIG. 38, in the PARALLEL MODE, two video images which originate from a common live video input can be displayed concurrently on the screen in different display windows. In this mode, the input regions and the display window sizes and positions for the first and second fieldstores 32a and 32b, respectively, can be varied individually, resulting in different horizontal and vertical scaling factors being allowable for each fieldstore. A priority is assigned to one of the fieldstores such that, if the display windows for the first and second fieldstores overlap in the video window, only the output signal from the fieldstore with the higher priority is displayed in the region of overlap. The Parallel Mode also allows the input to one or both of the fieldstores to be interupted to freeze the video image in that fieldstore.

In FIG. 38, the input region for the first fieldstore 32a is set to full size so that the video image 540 is input without cropping to the first fieldstore 32a. The display window 522a for the first fieldstore 32a is set to display the contents of the first fieldstore, but in a reduced size display window. The input region for the second fieldstore is set to select the top right hand quadrant 510b of the image 510, and this is displayed in a second window 522b, which is the same size as the input region and offset relative to the first display window 522a, but which overlaps a corner of the first display window 522a. The second fieldstore 32b is assigned a higher priority than the first fieldstore 32a, so that the second display window 522b overlies the corner the of first display window 522a. The are 524 outside the first and second display windows 522a and 522b, but inside the video window 520 is filled with the background colour.

In the DUAL MODE, two separate live video inputs are stored in the fieldstores 32a, and 32b, as explained hereinbefore with reference to FIG. 30, and the contents of the fieldstores are displayed in separate display windows on the monitor screen. Each video image can be sized, cropped and positioned, as explained hereinbefore. One of both of the video images can be frozen on the display.

In all of the modes described thus far, the merge/display switch has been in the display position, the output from the fieldstores being sent for display on the computer monitor. In the PICTURE BUILD-UP MODE, the merge/display switch is used to divert the video output from the asynchronous converter back to the input of one of the fieldstores, to enable an image stored in one fieldstore to be "pasted" or merged into the image stored in the other fieldstore. Referring to FIG. 39, the video window 520, and the display window 522a for the first fieldstore 32a, are set to cover the whole screen. The first fieldstore is initially filled with data corresponding to the video background colour, such that the first fieldstore contains a blank video image 528. A video signal is input to the second fieldstore 32b, and the video image 510 is displayed in the second display window 522b, which is of reduced size. The second display window 522b has a higher priority than the first display window 522a so that the video image 510 in the second display window 522b overlies the background colour 528 of the first display window 522a.

To actuate the Picture Build-up, the second display window 522b is moved to the position on the screen where it is desired to produce a permanent frozen copy of the video image. During Picture Build-up, the merge/display switch is set to send the video output data to the merge output 518, and back to the input of the first fieldstore 32a, and the video output that would otherwise have been displayed on the computer monitor is copied into the first fieldstore 32a. Thus the first fieldstore 32a now contains a copy of the contents of the video window, including a copy of the instantaeous video image 510 in the second display 522b window overlying the previous contents of the first display window 522a. Referring to FIG. 40, after Picture Build-up, the second display window can be moved away from the copied position, for example in the direction of the arrow 530, leaving behind a copy 532 of the video image that was displayed in the second display window 522b at the instant that Picture Build-up was commenced. By repeating this process several times, a "still" or frozen picture can be built up from several frozen images separated from, or overlying, one another. The size, as well as the positioning, of the second display window can be varied between each Picture Build-up process. The built-up picture is stored in the first fieldstore 32a, and it can be downloaded through the downloading logic 526 to the host computer 10, for example for storage on a disk.

If the switch 516 interrupts the video signal sent to the fading/mixing matrix (34 in FIG. 2), the actuation of Picture Build-up will cause a glitch to appear on the video display, as the video output signal from the asynchronous converter is blanked. This can be avoided if the video output to the fading/mixing matrix 34 is maintained during Picture Build-up by not blanking the video signal fed out from the asynchronous converter, and by posting a new image into the fieldstore at the same time as it is output from the asynchronous converter for display.

The video image in the second display window 522b is always visible on the screen 14, since the second display window 522b is assigned a display priority over the first display window 522a. The priorities can also be reversed, so that the movable display windop 522b lies under any built-up parts of the picture.

When in the Picture Build-up mode, the first fieldstore 32a contains a full size image of the video picture on the screen. The second fieldstore 32b is also set to the full screen size, and the video image in the second display window 522b is stored in the second fieldstore 32b in its reduced size, by scaling the video image before it is stored in the fieldstore. This simplifies the amount of horizontal and vertical procesing of the video images that is required when the new picture is built up.

Although the examples of video image and display window manipulation described above have not included details of combining the video image with a computer graphics image, it will be appreciated that these video images can be mixed, faded, or keyed with computer generated graphics, according to attributes of the generated graphics signals.

Under control of the control processor (38 in FIG. 2), the fading/mixing matrix (34 in FIG. 2) can combine video images and computer graphics on the display, pixel by pixel, in one of two ways, either by Keying or by Mixing.

In the keying mode, each graphics colour has associated with it a three state attribute, which determines whether the graphics colour is interpreted by the keying/fading matrix 34 as a solid colour, a transparent colour, or a translucent colour. For each pixel position, a solid colour denotes that only the graphics signal is to be displayed, a transparent colour denotes that only the video signal is to be displayed, and a translucent colour denotes a 50% mix, or "wash", of the video and graphics signals.

In the mixing mode, the three state attribute associated with each colour is turned off. Instead, each colour has associated with it attributes corresponding to the following: the fade graphics level (Fg), the fade video level (Fv), and the saturation level (Sat), explained hereinbefore. The attributes for each colour are stored in the control palette (195 in FIG. 13), and each attribute can be selectively varied under software control of the host computer 10. The control processor 38 further includes fade control logic (not shown) for selectively controlling the fade levels to generate a fade-up sequence, or a fade-down sequence, of the video and graphics images in the combined display, over a predetermined period of time. Such a fade has been described hereinbefore with reference to FIGS. 19 and 20. It will be appreciated that by including dedicated fade control logic to control the fade sequence, the host computer can is free to continue processing other data while the fade sequence is taking place in the visual display.

Although the video window and the display windows described hereinbefore are rectangular in shape, it will be appreciated that other shapes of window could also be used. The video window data is held in the 1 Mbit RAM (224 in FIG. 13), in the form of a bit plane. Thus by altering the data in the bit plane, other shapes of video window can be accomodated. For example, the window could comprise a grid pattern of spaced apart small square areas, in which the video image is displayed.