METHOD AND PRODUCT FOR SYNTHESIS OF PRINT DATA AND FOR PROVIDING THE SAME TO A PRINTER

Description

[0001] The present invention relates to a method and a product, respectively, for synthesizing print data from object-based data in the form of a page description language and for providing the same to a printer.

[0002] At this point, some terms used in this description should be defined in advance.

[0003] When this description refers to object-based data in the form of a page description language, it means objects and their properties that make up an image. The term "object" refers to a graphic primitive, a pixel array, a shade, a transparency, or text. In this application, a graphic primitive is defined as an elementary one- or two-dimensional geometric shape. A graphic primitive is described, among other things, by one or more vectors. A two-dimensional geometric shape is, for example, a circle, an ellipse, or a polygon, such as a rectangle, a square, or a triangle. Conversely, a one-dimensional geometric shape is, for example, a point (i.e., a pixel), a line, or a polynomial sequence.When this description refers to the properties of the objects, it means the object's position in the image, and, if applicable, its color, line thickness, fill pattern, and other data that determine its appearance.

[0004] In this description, the term programmable system-on-chip, hereinafter also referred to as PSoC, means that it is possible to program and reprogram the system-on-chip, in particular its functional modules, both at the hardware level and at the software level, i.e. to design and redesign it.

[0005] From US 6 091 506 A, a method is known for synthesizing print data for a printer using a processing unit comprising a master processor and several parallel processors, wherein the master processor is configured to create a display list and divide it into sublists which it distributes to the parallel processors, wherein the parallel processors are configured to interpret the sublists in such a way that an image is rendered as bitmap data.

[0006] US2009046105A1 discloses methods and systems for fast and efficient processing of data in a graphics processing unit pipeline.

[0007] Furthermore, DE69130132T2 discloses methods and devices for texture mapping of basic graphical elements in a graphics pipeline architecture system.

[0008] Methods for converting object-based vector graphics data into raster pixel data typically represent a first step towards synthesizing print data, which a printer can then use to print on a medium. Most of these methods nowadays utilize a raster graphics processor, which is generally designed to require a complete conversion of the vector graphic into a raster graphic before the resulting pixel graphic can be dithered using a predefined algorithm and subsequently delivered to a printer as a dithered print graphic. This is certainly the case when dealing with complex vector graphics.

[0009] Such raster graphics processors are inexpensive to purchase, but can only process incoming data inefficiently and therefore laboriously, so they are only suitable if a single image can and should be printed in such a large quantity that there is enough time available for rasterizing another vector graphic.

[0010] One problem causing the slow conversion speed, i.e., rasterization speed, is the fact that such methods use a raster graphics processor with a processor that can only perform the complex rasterization serially.

[0011] In order to allow for sufficient flexibility regarding short-term changes in the preparation of print jobs, it is of great importance that the conversion of object-based data from a vector graphic to a pixel graphic can be carried out as quickly as possible.

[0012] The reason why most images to be printed on modern single-pass printers with relatively high print resolutions must be rasterized well in advance is that the printer consumes this print image data faster than a typical raster graphics processor can process the underlying vector graphics. This lack of flexibility allows for short-term changes during print job preparation. This is especially true when print jobs with a high proportion of varying images need to be converted, where no two images are identical.

[0013] Typical high-performance printing presses, i.e., single-pass inkjet printers, now consume print data at rates of at least 100 Gbit / s, but often more than 300 Gbit / s or even 400 Gbit / s, with this trend continuing. For single-pass inkjet printers designed to print a two-meter-wide medium at a speed of five meters per second with a resolution of 1200 x 1200 DPI without interruption, data output rates of approximately 1600 Gbit / sec are required when converting vector graphics to raster graphics, especially if print jobs contain a high proportion of varying images and the conversion is to occur essentially simultaneously with the printing process.

[0014] An additional challenge with most common raster graphics processors is providing sufficient memory.

[0015] A single-pass printer with a width of 2.6 m and a resolution of 2400 DPI requires approximately 16 GB of storage capacity per m². If, for example, print data for 24-hour operation of the printer is to be prepared in advance, storage capacity for approximately 8 petabytes of print data per day must be provided. This currently necessitates several storage cabinets. Rasterizing such a large amount of data with most currently available serial raster graphics processors would take several days.

[0016] For the purposes of this description, a single-pass printer is understood to be a printing machine with a printing module for printing at least one printing medium, in which the printing medium or the respective printing media are continuously moved in the operating mode of the printing machine and the printing module sees an area of ​​the printing medium or the respective printing media only once and is arranged in a stationary position.

[0017] In a method for synthesizing print data and supplying it to a single-pass printer that has been extremely rare in recent times, a fast conversion of object-based data into 8-bit pixel data is achieved with a novel high-performance raster graphics processor that can output rasterized pixel data at an output rate of more than 800 Gbit / sec. This is accomplished by supplying the object-based data to a multitude of processors and processing it in parallel, i.e., simultaneously, by delegating one image from a group of several images to each processor of the raster graphics processor for conversion, so that multiple processors process different images in parallel but serially.

[0018] This previously known method comprises the following steps: (A) Inputting object-based data of multiple images in the form of a page description language into a raster graphics processor with at least one system-on-chip comprising multiple processors; (B) Converting the object-based data into 8-bit pixel data of each color channel of a first target color space with a specified image resolution by supplying the object-based data to the processors and processing it in parallel with the processors; (C) Converting the 8-bit pixel data of each color channel into 1-bit print data of each color channel of a second target color space using at least one dithering algorithm and storing the 1-bit print data in an output memory; and (D) Supplying the print data to the printer and printing on at least one print medium with the printer.

[0019] The raster graphics processor with a multitude of processors that enables such a process is suitable for overcoming the aforementioned disadvantage of slow raster speed. However, it requires a disproportionate amount of space, as several server racks in multiple server cabinets are needed, which, due to their considerable weight, can only be moved using a loading device, typically a forklift. Furthermore, it is prone to failure, meaning it is unreliable, as well as expensive, and therefore not particularly marketable. Currently, such high-performance raster graphics processors typically cost between four hundred thousand euros and one million euros to purchase, depending on the model.

[0020] Therefore, there is a need for a method and also for a product to convert object-based data into pixel data, which enables effective processing of object-based data with a raster graphics processor of simple and cost-effective design.

[0021] The present invention therefore aims to provide a method and a product for converting object-based data into pixel data, enabling the effective processing of object-based data with a simple and cost-effective raster graphics processor, thereby improving overall process efficiency and overall equipment effectiveness (OEE) with regard to availability, performance, and price. OEE is a key performance indicator used to measure the productivity and any losses of technical equipment.

[0022] According to the invention, the problem is solved by a method comprising the features of claim 1 and by a product comprising the features of claim 22. The respective dependent claims relate to further advantageous and, where applicable, additionally inventive embodiments.

[0023] The invention is based on the idea of ​​providing a raster graphics processor with at least one programmable system-on-chip comprising several single-instruction multiple-data-capable or multiple-instruction multiple-data-capable processors, each with at least one vector-based arithmetic logic unit, which vector-based arithmetic logic units are configured such that, when processing object-based data in parallel with the processors, several object-based data entries with a bit width of X bits, each received in parallel into the respective arithmetic logic unit, are converted into pixel data with a bit width of Y bits, which is larger than the bit width of X bits, using a predefined set of arithmetic operations, and the Y-bit pixel data is then buffered or stored in a pixel memory in the form of a pixel matrix consisting of an integer number of N-bit pixel data.

[0024] For the sake of clarity, it is noted that when the term "processor", in particular the term "vector processor", is used in connection with the invention or one of its preferred embodiments, it refers to a single-instruction multiple-data capable or multiple-instruction multiple-data capable processor, in particular a vector processor, of a programmable system-on-chip raster graphics processor.

[0025] A method, preferably a computer-implemented method, for synthesizing S-bit print data and supplying it to at least one printer, is known to comprise the following steps: (a) inputting object-based data of at least one image in the form of a page description language into a raster graphics processor with at least one system-on-chip comprising multiple processors; (b) converting the object-based data into N-bit pixel data of each color channel of a first target color space with a predetermined image resolution by supplying the object-based data to some or all of the multiple processors and processing them in parallel; (c) converting the N-bit pixel data of each color channel into S-bit print data of each color channel of a second target color space using at least one dithering algorithm and buffering or storing the S-bit print data in an output memory;(d) Providing the S-bit print data to the printer and printing at least one print medium with the printer.;

[0026] Accordingly, the inventive method is a method, preferably a computer-implemented method, for synthesizing S-bit print data and providing it to at least one printer, comprising: a) Inputting object-based data of at least one image, preferably several images, in the form of a page description language into a raster graphics processor with at least one system-on-chip comprising multiple processors; b) Converting the object-based data into N-bit pixel data of each color channel of a first target color space with a specified image resolution by providing the object-based data to some or all of the multiple processors and processing them in parallel; c) Converting the N-bit pixel data of each color channel into S-bit print data of each color channel of a second target color space using at least one dithering algorithm and buffering or storing the S-bit print data in an output memory; d) Providing the S-bit print data to the at least one printer and printing on at least one print medium with the printer.

[0027] According to the invention, the system-on-chip is provided as a programmable system-on-chip and the processors as single-instruction multiple-data capable or multiple-instruction multiple-data capable processors, each with at least one vector-based arithmetic logic unit, which vector-based arithmetic logic units are configured such that, during parallel processing according to step b), several object-based data sets with a bit width of X bits, each entering the respective arithmetic logic unit in parallel, i.e., simultaneously, are converted into pixel data with a bit width of Y bits, which is larger than the bit width of X bits, using a predefined set of arithmetic operations, and the Y-bit pixel data is then buffered or stored in a pixel memory in the form of a pixel matrix consisting of an integer number of N-bit pixel data.

[0028] This allows the PSoC to be configured so that its processors have parallel access to the PSoC's pixel memory, thus enabling a high data access rate. In addition to the advantages mentioned above, the inventive method offers the further benefit that, at least during the conversion phase, it can be operated with low energy consumption despite high data processing rates.

[0029] This also allows the raster graphics processor to be portable, preferably even when configured to generate N-bit pixel data at a rate of at least 800 gigabits per second, preferably at least 2500 gigabits per second, according to step b), by providing the raster graphics processor with a sufficiently high number of programmable system-on-chips, each comprising a sufficient number of single-instruction multiple data (SIMD) or multiple-instruction multiple data (MDM) capable processors, preferably vector processors. This is made possible because PSoCs are inherently extremely compact. For example, such a portable raster graphics processor can be implemented by providing at least eight Xilinx Inc. Premium Series Adaptive Computer Acceleration Platforms as PSoCs.The term "portable" as used here means that the raster graphics processor can be easily held and transported with one or both hands of a user. These further developments thus offer the advantage of facilitating the handling and mobility of the raster graphics processor according to the invention.

[0030] The single-instruction multiple-data capable or multiple-instruction multiple-data capable processor(s) can each be provided with multiple vector-based arithmetic logic units.

[0031] Single-instruction multiple-data or multiple-instruction multiple-data capable processors can be provided as single-instruction multiple-data or multiple-instruction multiple-data capable vector processors.

[0032] The object-based data consists of object-based PDF data and / or object-based PostScript data, each in the form of a vector-based page description language.

[0033] According to a preferred embodiment, the first target color space is a subtractive color space which preferably includes at least the color channels cyan (C), magenta (M), yellow (Y) and black (K), and the second target color space is a subtractive color space which preferably includes at least the color channels cyan (C), magenta (M), yellow (Y) and black (K), and in particular at least one of the color channels of a predetermined green and / or violet and / or orange.

[0034] According to another preferred embodiment, the N-bit pixel data has a bit width of 8- / 9- / 10- / 11- / 12- / 13- / 14- / 15- or 16-bit, preferably a bit width of 8- / 16- / 32- or 64-bit, in particular a bit width of 8-bit, and the S-bit print data has a bit width of 1- / 2- / 3- or 4-bit.

[0035] The forward slash " / " means "or". When this description refers to 8-bit pixel data of a color channel of a first target color space, it means pixels with a color depth of 8 bits for that single color channel, i.e., with a color gradation between 0 and 255.

[0036] When this description refers to 1-bit print data of a color channel from a second target color space, it means pixels with a color depth of 1 bit for that single color channel, i.e., with a color gradation of 0 and 1. In a corresponding binary printing mode, each pixel in a row of a halftone screen on the printing medium is either printed with no ink drop or with one drop of a color.

[0037] However, when referring to 2-bit print data of a color channel from a second target color space, this means pixels with a color depth of 2 bits for that one color channel, i.e., with a color gradation from 0 to 3. In a corresponding grayscale mode, each pixel in a row of a halftone screen on the printing medium is either printed with no droplet or with a combination droplet consisting of k drops of one color, where k can be 1-3.

[0038] According to another preferred embodiment, the X-bit pixel data has a bit width of 8- / 16- / 32- or 64-bit, preferably a bit width of 8-bit, and the Y-bit print data has a bit width of 16- / 32- / 64- / 128- / 256- / 512- / 1024- / 2048 or 4096-bit, preferably 128- / 256- / 512- / 1024- / 2048 or 4096-bit.

[0039] Accordingly, in this preferred embodiment, the X-bit pixel data of each vector-based arithmetic logic unit has a bit width of 8- / 16- / 32- or 64-bit, preferably a bit width of 8-bit, and the Y-bit print data of each vector-based arithmetic logic unit has a bit width of 16- / 32- / 64- / 128- / 256- / 512- / 1024- / 2048 or 4096-bit, preferably 128- / 256- / 512- / 1024- / 2048 or 4096-bit, such that the sum of the respective Y-bit print data of all vector-based arithmetic logic units corresponds to the sum of the respective bit widths of all vector-based arithmetic logic units, which sum preferably corresponds to a bit width of up to 1048576-bit.

[0040] According to a particularly preferred embodiment of the inventive method, the raster graphics processor was equipped with conversion modes R1 and R2, wherein the conversion according to step b) is carried out in one or more conversion cycles, wherein a conversion cycle comprises the steps: e) Determining a group of objects from the image to be converted based on the conversion progress; f) Determining the computational effort R for each object from the group of objects in order to predict whether the corresponding computational effort R exceeds or falls below a predetermined value S, and categorizing the respective objects into type A if the computational effort is below the value S, and into type B if the computational effort exceeds the value S; g) Starting the conversion by i. if the object has been categorized as type A, selecting mode R1 and decomposing the object according to mode R1 into several independent re-vectorized objects and providing the re-vectorized objects to a predetermined number of processors and converting them with the processors according to step b), which re-vectorized objects are sub-objects of the object that together form the object; ii.if the object was categorized as type B, mode R2 is selected, and the object is converted to mode R2 without being decomposed into several independent re-vectorized objects according to step b); .

[0041] These advanced training courses offer the advantage that, through this staged sequence of mathematical decisions regarding the type of conversion of diverse objects, the processing time of the object-based data of a single image can be immensely reduced.

[0042] Furthermore, if the object is categorized as type A, the object-based pixel data in step b) is buffered or stored in an internal pixel memory of the programmable system-on-chip (PSoC), and if the object is categorized as type B, the object-based pixel data is buffered or stored in an external pixel memory, preferably a cache memory, relative to the PSoC. This refinement has the advantage of reducing the wear and tear on the external pixel memory with respect to read and write operations. Consequently, the external memory may need to be replaced less frequently, saving time and costs. An internal pixel memory of the PSoC is designed to prevent wear and tear with respect to read and write operations, whereas a corresponding external pixel memory can wear and tear with respect to read and write operations.

[0043] According to a further preferred embodiment, for each of the graphic primitives from the group of graphic primitives rectangle, circle, triangle or line, at least one, preferably several, single-instruction multiple-data-capable or multiple-instruction multiple-data-capable processors are provided, each of which exclusively comprises a predetermined set of arithmetic operations tailored to the respective graphic primitive for controlling its at least one vector-based arithmetic logic unit, wherein each set of arithmetic operations from the group differs from every other set of arithmetic operations from the group, wherein, according to step b), the object-based data is provided, depending on its graphic primitive, only to the processor(s) that comprises the predetermined set of arithmetic operations tailored to the graphic primitive.This advanced training offers the advantage of significantly reducing the processing time of object-based data for an image that includes at least one or more rectangles and / or circles and / or triangles and / or lines as graphic primitives.

[0044] The relevant sets of arithmetic operations are known to those skilled in the art from the state of the art in the field of data processing.

[0045] According to a particularly preferred embodiment, the single-instruction multiple-data-capable or multiple-instruction multiple-data-capable processors are provided with at least one first and one second vector-based arithmetic logic unit, configured to be connected in series as needed. During parallel processing according to step b), several object-based data sets, each with a bit width of X bits, are first converted into several pixel data sets with a bit width of Y bits, which is greater than the bit width of X bits, by applying a predefined first set of arithmetic operations. These several Y-bit pixel data sets are then processed in parallel into the second arithmetic logic unit.Simultaneously, using a predefined second set of arithmetic operations, the data is converted into multiple pixel data sets with a bit width of Z-bits, which is larger than the bit width of X-bits and preferably also larger than the bit width of Y-bits. Subsequently, the Z-bit pixel data is buffered or stored in a pixel memory in the form of a pixel matrix consisting of an integer number of N-bit pixel data sets. This further development offers the particular advantage of intelligently reducing the processing time of object-based data, especially data consisting of multiple graphic primitives, particularly those including the graphic primitives rectangle, circle, triangle, and / or line.

[0046] The decomposition of the objects into re-vectorized objects according to step g[i]) and the provision of the re-vectorized objects to a predefined number of processors can be performed using a single- or multi-core main processor of the programmable system-on-a-chip. For clarity, it should be noted that the main processor is not the processor, and in particular not the vector processor, of the PSoC.

[0047] According to a preferred embodiment of the invention, for each object to be converted from the object group, the computational effort R according to step f) is predicted based on the size and / or orientation of the object to be converted in the image, which computational effort is preferably stored in a computational effort prediction table.

[0048] Furthermore, it is possible to specify either a constant value for the given value S or, after step e) and before step f), a variable value in each conversion cycle based on freely determined available computing power within the multiple processors, and preferably based on the size and / or orientation of a fictitious object, preferably taken from a predefined conversion mode table. This further development offers the advantage that, in the second case, the processing time for selected objects, for example, a relatively small but relatively complex object, can be further reduced.

[0049] According to a first preferred embodiment of the method, color maps and color profiles are created fully automatically by printing calibration fields of the respective color channels of the second target color space at predetermined time intervals, in particular on waste areas of the printing medium or at least some printing media, which are read by an image measurement system of the printer, wherein after reading, preferably by means of a programmable system-on-chip provided for this purpose, colorimetric calculations are carried out in order to create a new color profile of the first target color space compared to the color profile of the first target color space used by the printer in the method, on the basis of which the conversion according to step b) is carried out, such that after creation of the new color profile, printing is continued directly on its basis.

[0050] This advanced training offers the advantage of enabling the creation of multiple color charts daily without any operator intervention. Color charts are printer samples. They can be created for each color, especially for each color channel, of the second color space and serve as a visual reference. Typically, different ink batches of the same ink brand differ in their color and also in their printing properties. Therefore, if a color image is printed on a medium using different ink batches, multiple images of the same pattern will be produced, but with different colors. This can be due to unavoidable imperfections in the ink manufacturing process and also to the ink's properties changing over time.

[0051] According to an alternative second preferred embodiment of the method, predetermined print patterns of an image printed on a printing medium from S-bit print data are read in at predetermined time intervals, preferably by means of artificial intelligence, using an image measurement system of the printer, wherein after reading, preferably by means of a programmable system-on-chip provided for this purpose, colorimetric calculations are carried out in order to create a new color profile of the first target color space compared to the color profile of the first target color space used by the printer in the method, on the basis of which the conversion according to step b) is carried out, such that after creation of the new color profile, the printing is continued directly on its basis.

[0052] These training courses offer the advantage that color matching and the creation of new color profiles can be carried out in passing or on-the-fly, and printing ink can also be saved, since calibration fields no longer need to be printed.

[0053] According to a highly preferred embodiment of the method, the first preferred embodiment and the second preferred embodiment are carried out alternately at predetermined time intervals.

[0054] According to a further preferred embodiment of the inventive method, the programmable system-on-chip is provided with at least one programmable logic circuit configured to perform weighted addition operations such that, when multiple objects of the object-based data overlap in the image, the N-bit color values ​​of the overlapping pixels of each color channel of the multiple objects (converted to N-bit pixel data of each color channel) are weighted and added at the hardware level by the programmed logic circuit in step b) before buffering or storing in the pixel memory. These further developments offer the advantage of significantly reducing the storage time for the respective color values ​​of the overlapping pixels of each color channel of multiple overlapping objects.

[0055] Furthermore, in the inventive method, the programmable system-on-chip can be provided with a further programmable logic circuit which has been configured to carry out the process steps according to step c), in particular the computer-implemented process steps according to step c), such that all steps according to step c) are carried out on the further programmable logic circuit.

[0056] These training courses offer the advantage of enabling an additional improvement in overall process effectiveness and overall equipment effectiveness with regard to availability, performance and price.

[0057] Furthermore, in this preferred embodiment, it is possible that the additional programmable logic circuit for executing a method, in particular a computer-implemented method, for compensating for at least one failed or faulty nozzle is configured such that, if after printing a printing medium according to step d), the failed or faulty nozzle is detected by the printer's image measurement system, preferably by means of artificial intelligence, all steps of the method, in particular the computer-implemented method, for compensating for the failed or faulty nozzle are executed on the programmable logic circuit.

[0058] These training courses offer the advantage of enabling pixel-nozzle alignment in passing or on-the-fly, thus ensuring the continuity of high print quality.

[0059] Methods for compensating for a failed or defective nozzle are known to those skilled in the art. The basic principle on which such methods operate is to switch off the failed or defective nozzle and redistribute image data that was to be printed by the failed or defective nozzle to several nozzles located near the failed or defective nozzle.

[0060] The object-based data can be input into the raster graphics processor in the form of a main queue of multiple images, as described in step a). Before step b), a display list with multiple images in several rows intended for the printing medium is specified based on the images from the main queue. This makes it possible to increase the printer's productivity. The main queue can be created from several sub-queues, each containing at least one or more images, using an automated print job system provided in a cloud data processing environment, and preferably with the aid of predefined algorithms.

[0061] According to a further preferred embodiment, in step a), only object-based data is input into the raster graphics processor, the underlying image of which can be processed as a whole in all subsequent steps based on the available computing power of the processors in a temporally deterministic manner, preferably automatically, and particularly preferably by means of artificial intelligence, such that the time required to completely execute step b) and all subsequent steps except step d) is shorter in total than the time required to completely print the image onto at least one printing medium. This further development offers the advantage that the risk of a production stoppage can be significantly reduced or even completely eliminated.

[0062] According to a further preferred embodiment, prior to step a), the image or images are cleaned as needed to remove object-based data that is optically irrelevant to the human eye, such that the object-based data can be processed in a temporally deterministic manner in all subsequent steps. This cleaning is preferably carried out at least by filtering with a transparency filter. This further development offers the advantage that a larger number of customer images can generally be processed in a temporally deterministic manner.

[0063] Furthermore, it is possible to automate the cleaning of object-based data using artificial intelligence, preferably a continuously learning artificial intelligence that particularly preferably includes at least one adaptive decision algorithm on which the cleaning is based, and which decision algorithm is adapted by the artificial intelligence from time to time based on processing data. This further development has the advantage of enabling a further increase in overall equipment effectiveness in terms of availability, performance, and price.

[0064] The printer can be provided as an inkjet printer. Furthermore, the printer can be provided as a single-pass or multi-pass printer. Preferably, the printer is provided as a single-pass or multi-pass inkjet printer. Alternatively, the printer can be provided, for example, as a laser printer or offset printer. If the printer is provided as an inkjet printer, the printing of the at least one printing medium is carried out with ink droplets in the color of at least one color channel or in the colors of all color channels of the second target color space, based on the S-bit print data.

[0065] Step d) can be carried out by printing on a printing medium moving in a transport direction or on several printing media moving in a transport direction.

[0066] According to a particularly preferred embodiment of the inventive method, the vector-based arithmetic logic units were each configured such that, during parallel processing according to step b), several object-based data sets with a bit width of X bits entering the respective arithmetic logic unit in parallel are converted into pixel data with a bit width of Y bits, which is larger than the bit width of X bits, in a single clock cycle of the processor using a predetermined set of arithmetic operations, and the Y-bit pixel data is then buffered or stored in a pixel memory in the form of a pixel matrix consisting of an integer number of N-bit pixel data.

[0067] According to the invention, the problem is also solved by a product, preferably a computer program product, according to claim 22.

[0068] The product according to the invention, preferably the computer program product, for synthesizing S-bit print data and providing it to at least one printer, comprises: A. A raster graphics processor with at least one system-on-chip comprising multiple processors, configured such that object-based data of at least one image, input into the raster graphics processor in the form of a page description language, is converted into N-bit pixel data of each color channel of a first target color space with a predefined image resolution by providing the object-based data to some or all of the multiple processors and processing it in parallel with the processors; B. A dithering processing unit configured to convert the N-bit pixel data of each color channel into S-bit print data of each color channel of a second target color space using at least one dithering algorithm and to buffer or store it in an output memory; C. An output unit configured to provide the S-bit print data to the printer; D. The at least one printer for printing on at least one printing medium; E.wherein each system-on-chip comprises a single-core or a multi-core main processor for delegating predetermined tasks, preferably the conversion of object-based data into N-pixel data, to the processors.

[0069] According to the invention, the system-on-chip is designed as a programmable system-on-chip and the processors are designed as single-instruction multiple-data capable or multiple-instruction multiple-data capable processors, each with at least one vector-based arithmetic logic unit, which vector-based arithmetic logic units are configured such that, during the parallel processing of the object-based data, several object-based data sets with a bit width of X bits, each entering the respective arithmetic logic unit in parallel, i.e., simultaneously, are converted into pixel data with a bit width of Y bits, which is larger than the bit width of X bits, using a predefined set of arithmetic operations, and the Y-bit pixel data is then stored or buffered in a pixel memory in the form of a pixel matrix consisting of an integer number of N-bit pixel data.

[0070] According to a preferred embodiment, the single-instruction multiple-data capable or multiple-instruction multiple-data capable processors are designed as single-instruction multiple-data capable or multiple-instruction multiple-data capable vector processors.

[0071] One suitable programmable system-on-chip (SoC) is the "Adaptive Computer Acceleration Platform" recently launched by Xilinx Inc., which includes models such as the "Al Core Series," "Al Edge Series," "Prime Series," and "Premium Series." These SoCs comprise several hundred integrated vector processors.

[0072] According to another preferred embodiment, the programmable system-on-chip comprises at least one programmable logic circuit configured to perform weighted addition operations such that, when multiple objects of the object-based data overlap in the image, the N-bit color values ​​of the overlapping pixels of each color channel of the multiple objects (converted to N-bit pixel data of each color channel) are weighted and added at the hardware level by the programmed logic circuit in step b) before buffering or storing in pixel memory. These further developments offer the advantage of significantly reducing the storage time of the N-bit data of each color channel from multiple overlapping objects.

[0073] The programmable system-on-chip can include a single- or multi-core main processor configured to decompose an object into re-vectorized objects according to step g[i]) of the inventive method. The programmable system-on-chip can also include a high-speed interface for receiving and / or transmitting data.

[0074] The programmable system-on-chip can further include a data transmission bus capable of transferring data at a rate of at least 3 Tbit / s, preferably up to 4 Tbit / s, wherein at least the multiple processors, the pixel memory, the output memory, the at least one programmable logic circuit, the single-core or multi-core main processor and the high-speed interface are coupled to the data transmission bus.

[0075] Furthermore, it is possible that the raster graphics processor includes a master processor for performing certain tasks, preferably for delegating certain tasks to the at least one, in particular to the several, programmable SoCs.

[0076] The printer can be an inkjet printer. It can also be a single-pass or multi-pass printer. Preferably, the printer is a single-pass or multi-pass inkjet printer. Alternatively, the printer can be, for example, a laser printer or offset printer.

[0077] In a preferred embodiment, the raster graphics processor is designed in such a way that it can perform a method according to the invention.

[0078] According to a particularly preferred embodiment, the vector-based arithmetic logic units are each configured such that, during the parallel processing of the object-based data, several object-based data entries, each with a bit width of X bits, are converted into pixel data with a bit width of Y bits, which is larger than the bit width of X bits, in a single clock cycle of the processor using a predefined set of arithmetic operations, and the Y-bit pixel data is then stored in a pixel memory in the form of a pixel matrix consisting of an integer number of N-bit pixel data.

[0079] Furthermore, in a preferred embodiment, the dithering computing unit B can be configured as another programmable logic circuit.

[0080] Accordingly, the further programmable logic circuit is configured to convert the N-bit pixel data of each color channel into S-bit print data of each color channel of a second target color space using at least one dithering algorithm and to buffer or store them in an output memory.

[0081] This further training offers the advantage of enabling an additional improvement in overall process effectiveness and overall equipment effectiveness with regard to availability, performance and price.

[0082] Furthermore, in this preferred embodiment, it is possible that the additional programmable logic circuit for executing a method, in particular a computer-implemented method, for compensating for at least one failed or faulty nozzle is configured such that, if after printing a printing medium according to step d), the failed or faulty nozzle is detected, preferably by means of artificial intelligence, with the image measurement system of the printer, all steps of the method, in particular the computer-implemented method, for compensating for the failed or faulty nozzle can be executed on the programmable logic circuit.

[0083] This advanced training offers the advantage of enabling nozzle adjustment in passing or on-the-fly, thus ensuring the continuity of high print quality.

Claims

1. A method for synthesizing S-bit print data and providing it to at least one printer, comprising: a) entering object-based data of at least one image in the form of a page description language into a raster graphics processor with at least one system-on-chip comprising multiple processors; b) converting the object-based data into N-bit pixel data for each color channel of a first target color space with a predetermined image resolution by providing the object-based data to some or all of the plurality of processors and processing it in parallel with the processors; c) converting the N-bit pixel data of each color channel into S-bit print data for color channels of a second target color space using at least one dithering algorithm, and buffering or storing the S-bit print data in an output memory; d) providing the S-bit print data to the at least one printer, and printing at least one print medium with the printer; characterized in that the system-on-chip is provided as a programmable system-on-chip and the processors are provided as single-instruction multiple-data-capable or multiple-instruction multiple-data-capable processors, each with at least one vector-based arithmetic logic unit, wherein the vector-based arithmetic logic units are being configured such that, during parallel processing according to step b), multiple object-based data items each having a bit width of X bits and entering the respective arithmetic logic unit in parallel are converted into pixel data with a bit width of Y bits, which is greater than the bit width of X bits, using a predetermined set of arithmetic operations of the processor, whereupon the Y-bit pixel data is buffered or stored in a pixel memory in the form of a pixel matrix consisting of an integer number of N-bit pixel data, wherein the object-based data comprises object-based PDF data and / or object-based PostScript data in the form of a vector-based page description language, and wherein the programmable system-on-chip is designed in such a way that it is possible to program and reprogram the system-on-chip both at the hardware level and at the software level, i.e., to design and redesign it.

2. Method according to claim 1, characterized in that the single-instruction multiple-data-capable or multiple-instruction multiple-data-capable processors are provided as single-instruction multiple-data-capable or multiple-instruction multiple-data-capable vector processors.

3. Method according to at least one of the preceding claims, characterized in that the N-bit pixel data has a bit width of 8- / 9- / 10- / 11- / 12- / 13- / 14- / 15- or 16-bits, preferably a bit width of 8- / 16- / 32- or 64-bits, in particular a bit width of 8-bits, and the S-bit print data has a bit width of 1- / 2- / 3- or 4-bit.

4. Method according to at least one of the preceding claims, characterized in that the X-bit pixel data has a bit width of 8- / 16- / 32- or 64-bits, preferably a bit width of 8-bits, and the Y-bit print data has a bit width of 16- / 32- / 64- / 128- / 256- / 512- / 1024- / 2048 or 4096-bits, preferably 128- / 256- / 512- / 1024- / 2048- or 4096-bits.

5. Method according to at least one of the preceding claims, characterized in that the raster graphics processor has been equipped with conversion modes R1 and R2, wherein the conversion according to step b) is performed in one or more conversion cycles, wherein a conversion cycle comprises the steps: e) determining an object group of objects of the image to be converted based on the conversion progress; f) determining the computational effort R for each object from the object group in order to predict whether the corresponding required computational effort R exceeds a predetermined value S or falls below the value S, and categorizing the respective objects into type A if the computational effort falls below the value S, and into type B if the computational effort exceeds the value S; g) Starting the conversion by i. if the object has been categorized as type A, mode R1 is selected and the object is divided into several independent re-vectorized objects according to mode R1, and the re-vectorized objects are provided to a predeterminable number of processors and converted with the processors according to step b), which re-vectorized objects are sub-objects of the object that together form the object; i. if the object has been categorized as type B, mode R2 is selected and the object is converted according to mode R2 without being divided into several independent re-vectorized objects according to step b).

6. Method according to claim 5, characterized in that if the object has been categorized as type A, the object-based pixel data in step b) are buffered or stored in an internal pixel memory of the programmable system-on-chip, and if the object has been categorized as type B, the object-based pixel data are buffered or stored in a pixel memory external to the programmable system-on-chip, preferably a cache memory.

7. Method according to claim 5 or 6, characterized in that for the predetermined value S, either a constant value or, after step e) and before step f) in each conversion cycle, based on a freely determined available computing power within the plurality of processors, and preferably based on the size and / or orientation of a fictitious object, a variable value is specified, which is preferably taken from a specified conversion mode table.

8. Method according to one of claims 5 to 7, characterized in that the decomposition of the objects into re-vectorized objects according to step (g[i]) and the provision of the re-vectorized objects to a predeterminable number of processors is performed with a single- or multi-core main processor of the programmable system-on-chip.

9. Method according to one of claims 5 to 8, characterized in that for each object to be converted from the object group, the computational effort R is predicted according to step f) based on the size and / or orientation of the respective object to be converted in the image, which computational effort is preferably stored in a computational effort prediction table.

10. Method according to at least one of the preceding claims, characterized in that for each of the graphic primitives from the group of graphic primitives rectangle, circle, triangle, or line, at least one, preferably several single-instruction multiple-data-capable or multiple-instruction multiple-data-capable processors are provided, each of which exclusively comprises a predetermined set of arithmetic operations tailored to the respective graphic primitive for controlling its or their respective at least one vector-based arithmetic logic unit, wherein each set of arithmetic operations from the group differs from every other set of arithmetic operations from the group, wherein, according to step b), the object-based data are provided to only those processors that comprise the predetermined set of arithmetic operations tailored to the graphic primitive.

11. Method according to at least one of the preceding claims, characterized by the steps in that the single-instruction multiple-data-capable or multiple-instruction multiple-data-capable processors are provided with at least one first and second vector-based arithmetic logic unit, respectively, which have been configured in such a way that they are connected in series as required and, during parallel processing according to step b), first several object-based data items with a bit width of X-bits each, which are fed into the first arithmetic logic unit in parallel, are converted into several pixel data items with a bit width of Y-bits, which is greater than the bit width of X-bits, using a predetermined first set of arithmetic operations, which multiple Y-bit pixel data are then converted in the second arithmetic logic unit in parallel using a predetermined second set of arithmetic operations into multiple pixel data with a bit width of Z-bits, which is greater than the bit width of X-bits, and also preferably greater than the bit width of Y-bits, whereupon the Z-bit pixel data is buffered or stored in a pixel memory in the form of a pixel matrix consisting of an integer number of N-bit pixel data.

12. Method according to at least one of the preceding claims, further characterized by the steps of in a fully automated manner creating color charts and color profiles by printing calibration fields of the respective color channels of the second target color space at predetermined intervals, in particular on waste areas of the print medium or at least some print media, which are read in by an image measuring system of the printer, wherein, after reading, preferably by means of a programmable system-on-chip provided for this purpose, colorimetric calculations are performed in order to create a new color profile of the first target color space relative to the color profile of the first target color space used in the printer's process, on the basis of which the conversion according to step b) is performed, such that after creation of the new color profile, printing is continued immediately on the basis thereof.

13. Method according to at least one of the preceding claims, characterized in that the programmable system-on-chip is provided with at least one programmable logic circuit configured to perform weighted addition operations, such that when multiple objects of the object-based data overlap in the image, the N-bit color values of the overlapping pixels of each color channel of the correspondingly converted into N-bit pixel data of each color channel in step b) are weighted and added with the programmed logic circuit at the hardware level before being buffered or stored in the pixel memory.

14. Method according to claim 5 or at least one of the preceding claims 6 to 13 in relation to claim 5, characterized in that, according to step a), only object-based data are entered into the raster graphics processor, the underlying image of which can be processed as a whole in all subsequent steps based on the available computing power of the processors in a temporally deterministic manner, preferably automatically, and particularly preferably by means of artificial intelligence, such that the time required to completely execute step b) and all subsequent steps except step d) is shorter in total than the time required to completely print the image on at least the one print medium.

15. Method according to claim 14, characterized in that, before step a), if necessary, the image or images are cleaned up to remove object-based data of the image that is optically irrelevant to the human eye, such that the object-based data can be processed in a time-deterministic manner in all subsequent steps, wherein the cleaning up is preferably performed at least by filtering with a transparency filter.

16. Method according to claim 15, characterized in that the cleaning of the object-based data is performed automatically by means of artificial intelligence, preferably automatically by means of continuously learning artificial intelligence, which particularly preferably comprises at least one adaptive decision algorithm on the basis of which the cleaning is performed, and which decision algorithm is adjusted from time to time by the artificial intelligence on the basis of processing data.

17. Method according to at least one of the preceding claims, characterized in that the printer is provided as an inkjet printer, preferably as a single-pass inkjet printer or multi-pass inkjet printer.

18. Method according to at least one of the preceding claims, characterized in that the vector-based arithmetic logic units have each been configured in such a way that, during parallel processing according to step b), multiple object-based data items, each with a bit width of X-bits, entering the respective arithmetic logic unit in parallel are converted in a single clock cycle of the processor into pixel data with a bit width of Y-bits, which is greater than the bit width of X-bits, using a predetermined set of arithmetic operations whereupon the Y-bit pixel data is buffered or stored in a pixel memory in the form of a pixel matrix consisting of an integer number of N-bit pixel data.

19. Method according to at least one of the preceding claims, further characterized by the steps that, fully automatically at predetermined intervals, preferably by means of artificial intelligence, predetermined print patterns of an image printed on a print medium from S-bit print data are read in with an image measuring system of the printer, whereby after reading in, preferably by means of a programmable system-on-chip provided for this purpose, colorimetric calculations are performed in order to create a new color profile of the first target color space relative to the color profile of the first target color space used by the printer in the method, on the basis of which the conversion according to step b) is performed, such that after the new color profile has been created, printing is continued immediately on the basis of this profile.

20. Method according to at least one of the preceding claims, characterized in that the programmable system-on-chip is provided with a further programmable logic circuit which has been configured to execute the method steps according to step c), in particular the computer-implemented method steps according to step c), in such a way that all steps according to step c) are performed on the further programmable logic circuit.

21. Method according to claim 20, characterized in that the further programmable logic circuit has been configured to execute a method, in particular a computer-implemented method, for compensating for at least one failed or faulty nozzle, such that when, after printing a print medium according to step d), the failed or faulty nozzle, preferably by means of artificial intelligence, is detected by the printer's image measurement system, all steps of the method, in particular the computer-implemented method, for compensating for the failed or faulty nozzle are performed on the programmable logic circuit.

22. Product for synthesizing S-bit print data and for providing the same to at least one printer, comprising: A. a raster graphics processor with at least one system-on-chip comprising multiple processors, which is designed in such a way that object-based data of at least one image entered into the raster graphics processor in the form of a page description language is converted into N-bit pixel data for each color channel of a first target color space with a predetermined image resolution, by providing the object-based data to some or all of the plurality of processors and processing it in parallel with the processors; B. a dithering calculation unit configured to convert the N-bit pixel data of each color channel into S-bit print data of the color channels of a second target color space using at least one dithering algorithm and to store it in an output memory; C. an output unit configured to provide the S-bit print data to the printer; D. the at least one printer for printing on at least one print medium; E. wherein each system-on-chip comprises a single-core or multi-core main processor for delegating predetermined tasks, preferably the conversion of the object-based data into N-pixel data, to the processors, characterized in that the system-on-chip is designed as a programmable system-on-chip and the processors are designed as single-instruction multiple-data-capable or multiple-instruction multiple-data-capable processors, each with at least one vector-based arithmetic logic unit, wherein the vector-based arithmetic logic units are configured in such a way that, during parallel processing of the object-based data, several object-based data items with a bit width of X-bits each, which are fed into the respective arithmetic logic unit in parallel, are converted into pixel data with a bit width of Y-bits, which is greater than the bit width of X-bits , wherein the Y-bits pixel data are subsequently stored in a pixel memory in the form of a pixel matrix consisting of an integer number of N-bit pixel data, wherein the object-based data comprises object-based PDF data and / or object-based PostScript data in the form of a vector-based page description language, and wherein the programmable system -on-chip is designed in such a way that it is possible to program and reprogram the system-on-chip at both the hardware and software levels, i.e., to design and redesign it.

23. Product according to claim 22, characterized in that the single-instruction multiple-data-capable or multiple-instruction multiple-data-capable processors are designed as single-instruction multiple-data-capable or multiple-instruction multiple-data-capable vector processors.

24. Product according to claim 22 or 23, characterized in that the programmable system-on-chip comprises at least one programmable logic circuit configured to perform weighted addition operations such that when multiple objects of the object-based data overlap in the image, the N-bit color values of the overlapping pixels of each color channel of the multiple objects converted accordingly into N-bit pixel data of each color channel in step b) are weighted and added with the programmed logic circuit at the hardware level before buffering or storing in the pixel memory.

25. Product according to at least one of claims 22 to 24, characterized in that the printer is designed as an inkjet printer, preferably as a single-pass inkjet printer or multi-pass inkjet printer.

26. Product according to at least one of claims 22 to 25, characterized in that the raster graphics processor is designed in such a way that it can perform a method according to at least one of claims 1 to 21.

27. Product according to at least one of claims 22 to 26, characterized in that the vector-based arithmetic logic units are each configured in such a way that, during parallel processing of the object-based data, several object-based data items with a bit width of X-bits each, arriving in parallel at the respective arithmetic logic unit, are converted into pixel data in a single clock cycle of the processor using a predetermined set of arithmetic operations are converted in a single clock cycle of the processor into pixel data with a bit width of Y-bits, which is greater than the bit width of X-bits whereupon the Y-bit pixel data is subsequently stored in a pixel memory in the form of a pixel matrix consisting of an integer number of N-bit pixel data.

28. Product according to at least one of claims 22 to 27, characterized in that the dithering arithmetic unit B is configured as a further programmable logic circuit.

29. Product according to claim 28, characterized in that the further programmable logic circuit is configured to execute a method, in particular a computer-implemented method, for compensating for at least one failed or defective nozzle, such that when, after printing a print medium according to step d), the failed or defective nozzle preferably by means of artificial intelligence, is detected by the printer's image measurement system, all steps of the method, in particular the computer-implemented method, for compensating for the failed or faulty nozzle can be executed on the programmable logic circuit.