EP0276884B1 - Device for synthesizing images - Google Patents

Description

The invention relates to a digital image synthesizer device intended in particular for managing the overlaps of several image planes described line by line according to a television-type scanning system, associated with a microprocessor and at least one image generator which generates at least one image in real time from a pixel memory, device comprising in particular a controller for governing the overlap of a television image by an image plane or of an image plane by another plane image.

Such a device is used in particular in video processors of home computers, multi-image cathode ray tube controllers or else video processors for interactive digital audio-disc, called CD-I.

The object of the invention is to provide additional means for managing the overlap of one image by another. Means for this are known from the so-called "Antiope" teletext system. In this system a digitally transmitted character matrix can be superimposed on a television picture. To do this, this matrix is explored in synchronism with the television scan by an abscissa counter which successively points to the characters in a character memory, where one of the bits of the character description determines whether it is transparent or not. . According to this bit, the television picture is displayed or, on the contrary, replaced by the description of the character for the duration of this character. This system allows parts of images (Antiope characters) generated by the device to be inserted into an image (television image) which is not generated by the device. The format of the superimposed characters is fixed once and for all.

When the two images to be processed are all two generated by the device, another means is disclosed by patent application FR-A-2,569,020. According to this document, the objects to be represented are stored in memories each corresponding to a degree of priority, and one generates in a memory image a pixel map of the image. For this, we search for each pixel position if there is an object in the memory with maximum priority, in which case it is displayed, otherwise we search in the memory with the next priority and so on. It is also advisable during the filling of the image memory by an object, to search for each pixel if it does not correspond to the beginning of another object in a higher priority memory, in which case we will read from this pixel the contents of this last memory.

In this system, it is necessary to prepare the image memory in advance. The process of modifying it is too long to allow its content to be corrected between one scan line and the next, and the image is modified, to create moving objects, during the frame scan returns.

The device according to the invention makes it possible to vary the priorities of the different images during a line and to modify the priorities during the return between one line and the next, because very little data is sufficient to obtain this effect.

The device according to the invention is notably remarkable in that it comprises a so-called region register containing a series of words comprising in particular an abscissa value, means for reloading this register from the image generator during the returns of league scan, a current abscissa counter of the displayed pixels, a pointer counter of the words in the region register, an abscissa value comparator which compares the content of the current abscissa counter with the abscissa contained in the pointed word in the region register by the word counter, and produces, when there is identity, a signal which triggers the establishment of one of at least two predetermined types of overlapping of the image planes, and increments the word counter tough register of regions.

In addition, the device advantageously comprises a register known as types of recoveries, in which at least two of the above-mentioned predetermined types of recovery are defined and this register is divided into several parts, one part being constituted by a first set of bits contained in the register. of regions, another part being formed by a second set of controller control bits, the first set of bits being an operation code which can cause the change of a conditional bit, called of regions, to each state of which corresponds a type of overlap defined by the second set of bits.

Thanks to this arrangement which associates an operation code with each of the abscissae, it is possible to carry out from each abscissa defined in the register of regions a different type of recovery. Since the region register contains little data, it can be reloaded quickly, and the means for reloading the region register are advantageously active during the line scan returns.

By changing, for example, the abscissa between one line and the next, we can describe windows of overlap of any shape. If on the contrary the register is kept unchanged, the side edges of the windows are vertical lines.

In addition, it is possible that one of the types of overlapping of the image planes is an overlapping by a series of pixels whose color remains constant between two abscissa identity signals.

Such an overlap generates a range of colors whose contour is determined by the abscissas indicated on each line in the region register. Thus it is possible to create a new monochrome object in addition to the images generated from the pixel memories.

In the case where the image generator generates several images, or else if there are several image generators, it is possible to apply the device to the recou one of these two images by the other. Then the words in the region register advantageously include the indication of the image concerned, in order to allow the processing of two image planes with the same region register.

Instead of using a simple overlapping of one image part with another, more complex effects can be obtained if there are several images in the device, when one of the types of image overlapping consists of a weighted mixture of the colors of two images, and that the words of the region register include an area to indicate a weighting value.

The description which follows, with reference to the appended drawings, describing nonlimiting examples will make it clear how the invention can be implemented.

Figure 1 is a block diagram illustrating an application of the device according to the invention.

Figure 2 is a block diagram of a device according to the invention.

FIG. 3 schematically represents a video display screen with a window created by the device according to the invention.

Figure 4 illustrates the content of a word in the region register.

In Figure 1 the device according to the invention is indicated by the reference 1. It is used in combination with a microprocessor 24, itself connected by its bus with an image generator 2, 3, 4 which generates in real time that is to say in synchronism with a scanning by successive lines of the television type, successive words each of which defines a pixel of the image. This generator consists here of a master processor 2 which builds an image from descriptive elements of objects stored in a pixel memory 3 to which it is connected by a bus 11, and of an identical slave processor 2B associated to a 3B memory. The master processor is also associated with a graphics processor 4 which is connected to the main bus and also to the bus 11 of the memory, and subtracts certain repetitive operations to increase the power of the master processor. Such processors are described for example in the documents FR-A-2,569,020 and EP-A-0145046. The slave processor is so called only because its clock and its synchronization are imposed on it by the master processor. It generates images whose content is independent of those of the master processor.

The device 1 generates the red, green, blue components of a composite image, on three analog RGB outputs, to attack a television set or a monitor equipped with video inputs for the three colors. The device processes several images in parallel and finally combines them by making overlaps or additions with the pixel values, and this pixel by pixel. The device is controlled by control bytes passing through the pixel input ports during control sequences, during the scan back periods.

The two input streams can be processed in various ways to generate 1 to 3 images, depending on the mode chosen. The overlay of television images is controlled by the device.

The master processor 2 supplies the device 1 with bytes of data describing in particular the pixels of an image via an 8-wire connection P0-7 and the rhythm necessary for reading this data is transmitted by a PCLK1 connection. Similarly, the slave processor 2B supplies the data for another image via the connections P8-15 and PCLK2.

FIG. 2 represents in more detail the content of the device 1 of FIG. 1. It is divided into two roughly identical channels # 1 and # 2 each corresponding to at least one image. The elements of channel # 2 which correspond to those of channel # 1 have the same references with an additional index B.

The device comprises as input a multiplexer 5 to which the pixel bytes in P0-7 and P8-15 are brought, and the clocks PCKL1-2. In the simplest case, bytes P0-7 are used by channel # 1 and bytes P8-15 are operated by channel # 2. But we can also cross entries. You can also send 4 bits P8-11 to channel # 1, where they are associated with 4 bits P0-3 to constitute 8 bits. You can also use all 16 bits for a single channel. Other combinations can still be imagined easily. It is a word applied to the MODE 1-2 input which programs the multiplexer to choose one of these combinations.

The data from the multiplexer 5 pass for each channel by a "latch" circuit 8, 8b which maintains the value of the bits until the validation of the following ones.

The data is brought to the color decoders 15, 16, 15B, 16B, the outputs of which are applied to output multiplexers 18, 18B governed by a controller 19 for overlapping the image by the other, and a MODE register. The latter is constituted and connected in a known manner in computer science to most of the elements of the figure to set up the different modes of operation which will be described later. It is not shown in order not to complicate the figure. For the same reason, some secondary interconnections are not shown.

The outputs of the multiplexers 18, 18b are finally brought to an adder-converter 22 which digitally adds the images of two channels, then converts the digital data into red, green, blue analog values present on the connections R, G, B respectively which are directly applicable to the video stages of a color television or monitor. We could as well convert the images to analog values first with several converters, then add these values analogically.

Two weighting circuits 20, 20B also governed by the controller 19 each provide the adder 22 with a so-called weighting digital value by which the latter multiplies the color amplitude data of each channel before adding them.

In the simplest case, these weighting values are 1/0 or 0/1 to choose the image of channel # 1 or that of channel # 2 respectively. But others intermediate values, for example 0.5 / 0.5, provide a mixture of the two images at will, for example to create a crossfade effect.

In addition, CLK image line synchronization signals are supplied by the master processor of FIG. 1 to an internal clock circuit 6 which synchronizes the functions of the device.

The input data can define the color of a pixel according to several methods:
In a first method, each pixel is defined by a set of 8 bits representing a color code. This bit set is therefore used to point an address into a fast RAM which provides the color values. Such a system, which is called a palette, and bears the indication CLUT1 or CLUT2 (Color Look Up Table in English), is referenced 15, 15B in FIG. 2.

In a variant each byte of channel # 1 defines two pixels with 4 bits per pixel and the palette CLUT1 is divided under the command of the MODE register into two blocks each corresponding to 4 address bits which provides two images which can each present 16 different colors. This variant can be used to provide two different images at the output of channel # 1.

Another method for defining the color consists, as is done in television, in providing separately the so-called "Y" luminance and two color differences called U and V. In addition to reduce the amount of data required, we simply transmit the difference D between two successive pixels. The absolute value is given once at the start of each line to readjust the values in the event of an error during a line. This way is called DYUV. It requires only one byte per pixel, the data Y, U and V being coded each by means of four bits, and the data U and V being transmitted each on four bits, in only one byte in two only, in turn. A DYUV1 or DYUV2 decoder (reference 16, 16B) performs the decoding in this process, which is used when processing natural images with subtle colors.

A third method used in the case of ordinary synthetic images, consists in directly coding the colors red, green, blue (direct RGB) by means of 5 bits per color, plus 1 bit of transparency, which constitutes 16-bit words. For economic reasons, inputs B0-7 and B8-15 are used together and therefore only provide the input capacity for a single image. (The two processors 2 and 2B in FIG. 1 work in parallel to each supply half of the bits). This unique image is of course added to the image of outdoor television.

In this case, the color code does not pass through a decoder since it corresponds directly to the coding necessary for the input of the output multiplexers. However, the decodings used in the other methods require a certain time. To ensure the synchronism of the "direct RGB" images with the images obtained by the other methods, it is therefore necessary to delay them. To this end, two shift registers "FIFO" (for First In, First Out) are provided, referenced 14, 14B, the outputs of which are combined to provide the single image. To compensate for these delays, in the case where the image of the device is associated with a television image, an external circuit of the so-called PLL type, controlled by the synchronization of the television, provides the device with general synchronization which is ahead of that from television.

• hors service
• une image à partir de DYUV1,
• une image à partir de la palette CLUT1
• deux images à partir de la palette CLUT1
• une image à partir de la palette CLUT2

The outputs of the various color decoders are connected to inputs of the multiplexer 18 or 18B which transmits one or the other of the signals. In addition, an entry indicated with a zero permanently deletes the image of a channel. Various configurations can be programmed by loading the mode register. This last group of bits defining the mode of channel # 1 and channel # 2, and the data sources for each channel. The two data streams of each 8 bits in parallel can be combined or separated. We define successive planes as well as a background and a foreground. Channel # 1 is rather used to define the foreground image and can be programmed for trans put the data in one of the following ways:

• out of order
• an image from DYUV1,
• an image from the CLUT1 palette
• two images from the CLUT1 palette
• an image from the CLUT2 palette

• hors service
• une image à partir de DYUV2
• une image à partir de la palette CLUT2
• une image RVB directe, issue des registres FIFO

Channel # 2 is instead used to set the background and can be programmed to transmit data in one of the following ways:

• out of order
• an image from DYUV2
• an image from the CLUT2 palette
• a direct RGB image from FIFO registers

• le plan curseur
• un premier avant plan (canal #1)
• un deuxième avant plan (canal #1)
• un arrière plan (canal #2)
• le plan de l'image de télévision.

Since the CLUT1 decoder can provide two images, it is therefore up to three images which can be represented for example on the screen of a television superimposed on the image of the television. In addition, a cursor movable in the foreground can be provided. It is generated by a generator 23 which provides the adder 22 with a square of 16 × 16 pixels each defined by a single bit, and which is always superimposed on any other image when it is present. The images generated are therefore combined to define the final image by superimposing up to 5 different planes which are:

• the cursor map
• a foreground (channel # 1)
• a second foreground (channel # 1)
• a background (channel # 2)
• the plane of the television picture.

A first function of the device consists in defining what is the relative position of each intermediate plane.

A second function is to define transparent areas in the planes so that the planes behind them can be seen.

This is done by the controller 19. Figure 3 gives a very simple example with two planes. The front plan F is transparent at the level of a rectangle through which appears the rear plane B.

Several methods can be used to create such transparent windows. One of them was mentioned above about the direct RGB mode in which a particular bit defines transparency.

In a neighboring process that is more universal but more complicated to decode, a particular color has the meaning: "transparent". It is a color defined by its red green blue components. It is therefore at the exit of the palette that we can search if this color is present. This is the function of the three "Comp" comparators referenced 17, 170, 17B. These comparators check at each pixel whether the color corresponds to a predetermined color, in which case the pixel is transparent. The comparators then deliver a signal brought to the recovery controller 19 which programs the corresponding multiplexer 18, 18B so that it ceases to transmit the image during the display duration of the pixel in question.

In the case where the plane (s) generated by the device are superimposed on a television image, a VDS output of the recovery controller is connected to the adhoc pin of the peri-television socket of the television to ensure the switching between the television image and that (s) from the device.

In the case of the DYUV method, the precision of the color definition is insufficient to obtain a reliable comparison and the transparent color method is not used.

The foregoing description relates to the environment of the invention and is intended to allow a better understanding of the latter.

The transparency of the plans can also be controlled according to a coding mechanism whose implementation device constitutes the main object of the present invention. According to this mechanism, transition points are defined during each scan line and at these transition points the display mode changes. The horizontal position of these points, and the nature of the display change can also be defined during the sweep returns.

To implement this mechanism, a region register 13 is used. It is so named because it allows you to define regions in the image. This register contains for example 8 words of 24 bits. One of these words is illustrated in FIG. 4. It contains 4 bits CH0-3 which represent an action to be undertaken, 10 bits RL0-9 which represent an abscissa expressed in number of pixels and 7 optional bits PA0-6. Three X bits are unused. An abscissa counter 7 (FIG. 2) linked to the internal clock 6 makes it possible to know at all times what is in the line the number of the pixel being processed. In addition, a counter 10 points to a word in the region register. This counter is reset on each line feed. A comparator 12 receives on the one hand the current abscissa from the counter 7 and on the other hand the abscissa registered in the bits RL0-9 of the word pointed in the region register 13 by the counter-pointer 10. The bits CH0-3 of this word are also transmitted to the controller 19 by the connection 25.

The comparator 12 constantly compares the two abscissas and when there is identity of the abscissas, it delivers on the connection 26 a signal which is brought to the controller 19, which performs the action described by the bits CH0-3, and to the pointer 10 words from the region register to increment it. It is therefore a new abscissa which is henceforth compared by the comparator 12, with a new action to be undertaken during the identity of the abscissas and so on until the last word, or until the end of the line. Of course it would be possible in a variant to count the abscissae image by image instead of line by line, however this would unnecessarily increase the capacity necessary for the registers.

The recovery controller 19 comprises in a register at least one so-called "region" bit. In the present example, there are two which are each assigned to one of the channels. It could also assign these bits to images and have three since three images may be available. In general, it can include as many as there are images to be processed. Among the different possible modes governed by the controller 19, one mode is chosen by programming, for example using a set of 4 bits per plane, called bits T, loaded by the master processor during a time of scan return, and these four bits indicate the meaning of the region bit. An example indicating different possible programming is indicated by table I which relates to four bits T10-13 contained in a register of the controller 19 and relating to the foreground. There are of course two other groups of four bits each concerning one of the other planes.

We call :
BR: the region bit
BT: bit ^No. 16 showing transparency in RGB direct mode
CT: the bit delivered by the transparent color comparators. Table I Bit content Condition for the pixel to be transparent. T13 T12 T11 T10 0 0 0 0 none: all pixels are transparent 0 0 0 1 CT is high 0 0 1 0 BT is high 0 0 1 1 BR # 1 is high 0 1 0 0 BR # 2 is high 0 1 0 1 BR # 1 or CT is high 0 1 1 0 BR # 2 or CT is high 1 0 0 0 none: no transparent pixels 1 0 0 1 CT is low 1 0 1 0 BT is low 1 0 1 1 BR # 1 is low 1 1 0 0 BR # 2 is low 1 1 0 1 BR # 1 or CT is low 1 1 1 0 BR # 2 or CT is low.

We see that with such a method, the possibilities are very great since we can take into account or not each of the bits BR # 1, BR # 2, BT, CT available. In addition it is possible that two planes are together not transparent, but with a weighting which provides a mixture of the two images.

An example indicating various actions controlled by the bits CHO-3 of the word pointed in the register of regions is provided by the following table II: Content of the 4 bits: Action to take 0 0 0 0 End of changes for the current line 1 0 0 0 Region bit set to zero. 1 0 0 1 Region bit set to one. 0 1 0 0 Change channel weighting # 1 0 1 1 0 Change channel weighting # 2 1 1 0 0 Region bit set to zero + change channel weighting # 1 1 1 1 0 Region bit set to zero + change channel weighting # 2 1 1 0 1 Region bit set to + change channel weighting # 1 1 1 1 1 Region bit set to + change channel weighting # 2

When an action concerns one of the region bits of the controller, the one of the two which is concerned is defined by one of the optional bits of the words of the region register (PA6).

When the action concerns a weighting, the new weighting is indicated by other optional bits (PAO-5). Thus the set of bits CHO-3 of the region word and the bits T constitutes a register of the types of overlap in which at least two predetermined types of overlap of the planes are defined.

In a simplified variant, this register may not exist: it is then defined once for all in the controller 19 that, for example if the bit of regions is at zero, the pixel concerned is transparent, and vice versa.

In another intermediate variant, the action to be undertaken can be defined directly by a group of bits in the words of the region register. Then the type register is entirely contained in the region register.

Of course, the maximum possibilities are offered by the preferred variant described above, combining the bits CH of the region register and the bits T of the controller.
The channel # 1 being controlled by a single region bit, in the case where the CLUT1 palette provides two foreground images, the latter are affected together. This does not prevent that there are differences between the transparencies of these first two planes if provision is made for conditions of interpretation (table I) of the bit of region different for each of these two planes.

The 8 words of the region register are in practice grouped in 2 × 4, each of the groups of four addressing a channel. We can therefore define two windows (4 transitions) per plane. However, since the operation code contains the channel to be assigned, each word can also be addressed to one channel or the other. We can for example use the 8 words of the region register for the same channel, which allows to define 4 windows. In all cases the set of transitions of the set of planes, i.e. of the displayed image, is equal to the number of regions (at least as regards the action of the region controller ).

In the example presented by FIG. 3, a region relating to channel # 1 (foreground) is a rectangle in which there is transparency and outside of which there is for example the DYUV mode. At the abscissa X1 we enter the transparent region and the plane B corresponding to channel # 2 is visible. Arriving at the abscissa X2 registered in the second word of the region register, the region bit changes and we return to the display of the plane F. We could imagine by example that, in this rectangle, a second region bit concerning channel # 2, that is to say the background, defines a transparent circle through which we would see the television image. To the right of the figure is indicated an area which represents the line feed periods. During these periods, it is possible to reload the region register. As long as the words in the region register are not changed, the same actions are repeated on each line, which results in drawings with rectangular outlines. Thus to create the rectangular window represented on figure 3, it is only necessary to carry out a first loading (two words) during the period To with the line preceding that where the beginning of the window occurs and a second loading (two words ) during period T3. So four words are enough to create such a window. In addition, by changing the content of the region words during each line feed, it is possible to generate complex contour shapes. The corresponding words can be prepared in advance by the image generator 2, 3, 4 and read at the appropriate time (for example To, T3). In particular during frame returns, the software can update memory areas intended to be read then during line feeds, so as to obtain mobile transitions from one image to another. It is thus possible to generate sweeping effects or movable windows. The shape of the drawings, their movements, the transitions are entirely under the control of the software.

It is also possible to imagine other uses of the regions: by increasing the number of conditions for interpreting the bits BR, BT, CT (which requires 5 bits instead of 4 in table I) we open new ones possibilities such as for example "the pixel has a predetermined color, whatever its color code, if BR is high". Thus, the device with the region register makes it possible to cover a line part with a series of pixels whose color remains constant between two abscissa identity signals of the comparator 12, which creates additional monochrome objects of any shape and possibly mobile.

The data to reload the region register 13 and possibly reprogram the controller 19 are introduced into the device by the same inputs P0-7 and / or P8-15 as the pixels. These inputs are in fact unused during the scanning returns, since the pixels are transmitted in real time, that is to say during the forward periods of the scanning. The inputs WR1 and WR2 are used to indicate to the input multiplexer 5 that this is such data and no longer image pixels.

During the line scan returns it is possible to supply device 1 (at the same rate as that of the pixels during the outward journey) at least 64 bytes: we could therefore reload around twenty 24-bit words in the region register. In practice, the possibilities of the system are increased if we take advantage of the return periods to also reload other registers. This is why we voluntarily limited the capacity of the region register to 8 words.

• le registre de mode. Ceci permet par exemple d'avoir dans le haut d'une image une partie en couleurs subtiles obtenues via un décodeur DYUV, et dans le bas une partie d'image synthétique de sous-titrage obtenue en mode RVB directe.
• le contenu d'une palette. Selon la capacité de la palette qui dépend du mode choisi il est possible de recharger toute la palette ou seulement un fraction de palette. Dans tous les cas ceci permet d'obtenir qu'un même code de couleur produise une couleur qui evolue entre le haut et le bas de l'image.
• Tout ce qui définit le curseur dans le générateur 23, c'est-à-dire un groupe de 32 octets définissant 16×16 pixels, une couleur (4 bits) et des coordonnées X,Y. Ainsi il est possible d'avoir plusieurs curseurs différents dans une même image.

During line returns it is also planned to recharge:

• the mode register. This allows for example to have at the top of an image a part in subtle colors obtained via a DYUV decoder, and at the bottom a synthetic image part of subtitling obtained in direct RGB mode.
• the contents of a palette. Depending on the capacity of the pallet which depends on the mode chosen, it is possible to reload the entire pallet or only a fraction of a pallet. In all cases this makes it possible to obtain that the same color code produces a color which evolves between the top and the bottom of the image.
• Everything that defines the cursor in generator 23, i.e. a group of 32 bytes defining 16 × 16 pixels, a color (4 bits) and X, Y coordinates. Thus it is possible to have several different cursors in the same image.

The words from processor 2 have 32 bits. They are introduced into system 1 by halves, that is to say by 16 bits both on the inputs P0-7 and P8-15 together. In these words, four bits for example define the operation to be carried out, that is to say in general the register for receiving the information, and the other 28 bits represent said information itself.

• engendrer une interruption dans le processeur 24
• recharger l'adresse à laquelle le processeur 2, ou 2B lit une image dans la mémoire 3 ou 3B, ce qui permet de changer d'image entre deux lignes successives.
• changer des données semi-permanentes comme la couleur du cadre monochrome qui éventuellement entoure l'image.

It is also further planned to reload registers in the processors 2, 2B of FIG. 1 during the line feeds. We can for example:

• cause an interruption in processor 24
• reload the address at which processor 2, or 2B reads an image in memory 3 or 3B, which makes it possible to change the image between two successive lines.
• change semi-permanent data such as the color of the monochrome frame that possibly surrounds the image.

Of course, the example given here with two channels # 1 and # 2 can be easily extended to any number of channels, provided that several bits are provided instead of the single PA6 bit, or else several registers of regions.

Claims (8)

1. A digital picture synthesizing device intended particularly for controlling the overlays of a plurality of picture planes written line by line in accordance with a television-type scanning system, associated with a microprocessor (24) and with at least one picture generator (2) which generates at least one picture in real time from a pixel memory (3), the device in particular comprising a controller (19) for controlling the overlay of a television picture with a picture plane or of a picture plane with another picture plane, characterised in that it comprises a regions register (13) containing a series of words comprising in particular an abscissa value (RL09), means (5, 8) for reloading said register from the picture generator during the line flyback intervals, a counter (7) of the current abscissa of the displayed pixels, a pointer counter (10) for the words in the regions register, an abscissa value comparator (12) which compares the content of the current abscissa counter (7) with the abscissa contained in the word indicated in the regions register by the word counter, and which in the case of quality supplies the controller (19) with a signal controlling the set-up by the controller of one of at least two predetermined types of overlay of picture planes, and increments the word counter (10) of the regions register.
2. A device as claimed in Claim 1, characterised in that it comprises a types-of-overlay register, in which at least two of said predetermined types of overlay are defined.
3. A device as claimed in Claim 2, characterised in that the types-of-overlay register is divided into a plurality of sections, one section being constituted by a set of bits (CH) contained in the regions register (13), another section being constituted by a set of control bits (T) from the controller (19), the first set of bits (CH) being an operation code for causing a conditional bit, called the region bit, to be changed, each state of of said region bit corresponding to a type of overlay defined by the second set of bits (T).
4. A device as claimed in any one of the preceding claims, characterised in that the means for reloading the regions register are active during the line flyback intervals.
5. A device as claimed in any one of the preceding claims, characterised in that a demultiplexer (5) is arranged in a pixels input channel (P0-15) which normally conveys the colour codes of the successive pixels, which demultiplexer enables the data present on said input to be routed to the regions register (13) during at least a part of the flyback intervals.
6. A device as claimed in any one of the preceding Claims, characterised in that one of the types of picture plane overlay is an overlay with a series of pixels hose colour remains constant between two abscissa-equality signals.
7. A device as claimed in any one of the preceding Claims, whose picture generator system generates a plurality of pictures, characterised in that the words of the region register comprise the indication of the relevant picture (PA6) in order to allow two picture planes to be processed with the same regions register (13).
8. A device as claimed in Claim 7, in which one of the types of picture overlay consists of a weighted mixture of the colours of the two pictures, characterised in that the words of the region register comprise a field for indicating a weighting value.

Images