Methods for selling character goods and computer programs

Abstract

This technology enables the easy production and sale of multiple types of character goods based on character illustrations drawn by individuals. [Solution] The computer program causes the computer to perform the following steps: accept input of image data of a character drawn by an individual; input the image data into a machine learning model trained on multiple image data drawn by at least one designer and accept output of image data of the original character; acquire first manufacturing data that can be accepted by a first manufacturing device and second manufacturing data that can be accepted by a second manufacturing device based on the image data of the original character; acquire images of the first product and images of the second product before manufacturing the first product and the second product; and display images of each product in a manner that makes them accessible to third parties and suitable for sale.

Description

Technical Field

[0005] ,

[0001] The present invention relates to a method for selling character goods and a computer program.

Background Art

[0002] Conventionally, there has been a business where illustrators and designers draw original characters and develop and sell them on various products such as T-shirts, figures, and seals. In recent years, not only professional illustrators and designers but also ordinary people have been creating original characters and publishing them through SNS and the like.

[0003] Patent Document 1 discloses a technique for generating a character image by selecting an image from a group of pre-stored images based on a user's answer to a question, synthesizing it with a character original image, and printing it for sale as a character card.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, the invention described in Patent Document 1 merely generates a character image based on an image selected from a group of pre-stored images, and thus cannot produce character goods that make use of characters created based on an individual's free ideas. On the other hand, in order to sell original character goods, in addition to the character design, it is necessary to overcome many hurdles such as prototyping of products and construction of an EC site, which is not easily achievable by ordinary people who are not professionals. Therefore, even if an ordinary person creates an appealing character, it rarely gets commercialized.

[0006] Therefore, the inventors of this application aim to provide a method for selling character goods and a computer program that enables even ordinary people, who are not professionals, to sell characters they have created as character goods. The inventors of this application also focus on the fact that illustrations drawn by individuals carry the risk of infringing on the copyrights of existing characters, and aim to provide a method for selling character goods and a computer program that take copyright into consideration. [Means for solving the problem]

[0007] This application discloses a method for selling character goods. This sales method includes the steps of: receiving input of image data of a character drawn by an individual; inputting the image data into a machine learning model trained on a plurality of image data drawn by at least one designer and receiving output of image data of an original character from the machine learning model; generating first manufacturing data that can be accepted by a first manufacturing apparatus for manufacturing a first product of the original character using a first manufacturing apparatus based on the image data of the original character; generating second manufacturing data that can be accepted by a second manufacturing apparatus for manufacturing a second product of the original character using a second manufacturing apparatus based on the image data of the original character; generating an image of the first product before manufacturing the first product of the original character using the first manufacturing apparatus; generating an image of the second product before manufacturing the second product of the original character using the second manufacturing apparatus; and displaying images of the first product and images of the second product in a manner that allows third parties to access them, so that the first product and the second product can be sold.

[0008] The method for selling character goods according to this application includes a configuration that accepts output image data of original characters by inputting image data of characters drawn by individuals into a machine learning model that has been trained using image data drawn by designers. With the above configuration, even if a character drawn by an individual may infringe on copyright, it becomes possible to create an original character with reduced copyright issues by inputting it into a machine learning model trained on image data with minimal copyright issues created by a designer.

[0009] Furthermore, the method for selling character goods according to this application generates first manufacturing data that can be accepted by the first manufacturing apparatus for manufacturing the first product of the original character, generates an image of the first product, and displays it in a manner accessible to third parties, even before the first product of the original character is actually manufactured using the first manufacturing apparatus. With the above configuration, manufacturing data is generated for producing a first product (e.g., a figure) and a second product (e.g., a sticker formed to have a three-dimensionally raised surface, made of elastic resin or similar material, and having a glossy finish), based on an original character. Images of these first and second products are also generated. Therefore, even before the first and second products are actually manufactured, highly accurate images of the first and second products can be prepared and displayed in a way that is accessible to third parties. This technological approach, which generates highly accurate product images and makes them accessible to third parties even though the product hasn't actually been manufactured, makes it possible for ordinary people to sell their created characters without having to go through numerous hurdles such as prototyping, which are difficult for most.

[0010] Furthermore, inputting image data of the original character into the machine learning model includes inputting prompts into the machine learning model to output the image data of the original character so that the machine learning model can generate first manufacturing data and second manufacturing data that can be accepted by the first manufacturing apparatus and the second manufacturing apparatus, respectively. For example, if the outline of a character image drawn by an individual is not a continuous line but is interrupted in the middle, such image data is not suitable for manufacturing character goods (e.g., figurines) that require a defined outline. Therefore, by inputting prompts into the machine learning model (for example, an instruction to output the original character so that it forms a continuous line) that enable the machine learning model to generate first and second manufacturing data, it becomes possible to output image data of the original character that is suitable for manufacturing by the first and second manufacturing equipment.

[0011] Furthermore, the method of selling character goods relating to this application may include the step of training a machine learning model using image data drawn by a designer and with a low probability of copyright infringement as training data. For example, to reduce the likelihood of copyright infringement, it is possible to train a machine learning model using image data and image data drawn by a designer who has a contract to transfer the copyright (for example, a designer who has entered into a copyright transfer agreement) as training data.

[0012] Alternatively, or in addition to this, image data whose copyright has already expired (for example, portraits of people painted in the Middle Ages) may be used as training data to train the machine learning model. Instead of this, or in addition, on the condition that the copyright of the image data of the character drawn by an individual is transferred, by accepting the input of the image data of the character, the image data of the character drawn by an individual itself or the image data of the original character itself may be configured to be learned by a machine learning model as learning data.

Brief Description of Drawings

[0013] [Figure 1] It is a block diagram showing the overall configuration of a character goods sales system according to an embodiment of the present invention. [Figure 2] It is a flowchart showing the overall flow of the method for selling character goods according to this embodiment. [Figure 3] It is a diagram showing an example of an image data input screen according to this embodiment. [Figure 4] It is a diagram showing an example of an original character display screen according to this embodiment. [Figure 5] It is a diagram showing an example of a character correction screen according to this embodiment. [Figure 6] It is a diagram showing an example of an original character display screen after correction according to this embodiment. [Figure 7] It is a diagram showing an example of a product list display screen according to this embodiment.

Modes for Carrying Out the Invention

[0014] [First Embodiment] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Note that the following embodiments do not limit the present invention, and various modifications are possible within the scope of the technical idea of the present invention.

[0015] 〈System Configuration〉<FIG. 1 is a block diagram showing the overall configuration of a character goods sales system 1 according to the present embodiment. The character goods sales system 1 includes a server device 10, an information portable terminal 20, a first manufacturing device 30, a second manufacturing device 40, and an EC site 50. The server device 10, the information portable terminal 20, the first manufacturing device 30, the second manufacturing device 40, and the EC site 50 are connected to be mutually communicable via a network.

[0016] The server device 10 is a core device of the character goods sales system 1. The server device 10 includes an input reception unit 11, a machine learning model 12, a character generation unit 13, a manufacturing data generation unit 14, a product image generation unit 15, a display control unit 16, a contribution rate calculation unit 17, and a database 18. The server device 10 may be realized by, for example, one or more computers, a cloud server, or a combination thereof.

[0017] The input reception unit 11 receives the input of image data P1 of a character drawn by an individual. The input reception unit 11, for example, displays a screen for receiving the input of image data on the display 21 of the information portable terminal 20. An individual can input the image data P1 by photographing an illustration of a character drawn on paper such as a sketchbook with the camera 22 of the information portable terminal 20. Alternatively, the individual may input the image data P1 by directly drawing a character on the touch panel 23 of the information portable terminal 20.

[0018] The information portable terminal 20 is, for example, a smartphone, a tablet, or other portable information processing device. The information portable terminal 20 includes a display 21, a camera 22, and a touch panel 23. The information portable terminal 20 can access a web application provided by the server device 10 via a web browser. Note that, instead of the information portable terminal 20, a stationary information processing device such as a desktop computer may be used.

[0019] The machine learning model 12 is pre-trained using multiple image data drawn by at least one designer. The machine learning model 12 accepts image data P1 (Figure 3, etc.) of a character drawn by an individual as input and outputs image data P2 (Figure 4, etc.) of an original character that reflects the designer's style. Here, the designer may be a professional illustrator, graphic designer, or character designer, etc.

[0020] Through transformation by machine learning model 12, the characteristics of existing characters that may be included in illustrations drawn by individuals are transformed based on the designer's style. This reduces the risk of copyright infringement of existing characters. In this way, characters drawn by individuals are transformed via machine learning model 12, which has been trained on image data of designers with fewer copyright issues, resulting in original characters with a reduced possibility of copyright infringement.

[0021] The machine learning model 12 may be, for example, a generative adversarial network, a variational autoencoder, a diffusion model, or other machine learning model suitable for image generation. Furthermore, the machine learning model 12 may be used in combination with a large-scale language model.

[0022] The machine learning model 12 may be trained using multiple image data drawn by the same designer, or it may be trained using multiple image data drawn by different designers. When trained using image data from multiple designers, the machine learning model 12 can generate original characters that comprehensively reflect the characteristics of each designer's style. For example, if the machine learning model 12 is trained using multiple image data drawn by a popular designer, the machine learning model 12 can generate original characters that retain the characteristics of the characters drawn by the individual while reflecting the style of that popular designer. This makes it possible for even ordinary people to sell attractive character goods that reflect the style of popular designers.

[0023] The character generation unit 13 inputs the image data P1 received by the input reception unit 11 into the machine learning model 12 and receives the output of original character image data P2 from the machine learning model 12. The original character image data P2 is image data that retains the characteristics of the character drawn by the individual, while being refined by the designer's style.

[0024] Figure 3 shows an example of an image data input screen. As shown in Figure 3, the display 21 of the information mobile terminal 20 displays an area for displaying an illustration of a character drawn by the user, a shooting button for taking a picture with the camera 22, and a button for selecting a previously captured image data. The user inputs image data P1 by taking a picture of the character drawn in a sketchbook with the camera 22, or by selecting a previously captured image. Alternatively, the user may input image data P1 by directly drawing the character on the touch panel 23 of the information mobile terminal 20. In this way, by using the information mobile terminal 20, it becomes possible to easily input image data of a character created by the user, regardless of location.

[0025] Figure 4 shows an example of the original character display screen. As shown in Figure 4, the image data P2 of the original character generated by the machine learning model 12 is displayed. The user checks the generated original character and, if satisfied, presses the approve button. If the user wishes to modify the original character, they can input text data indicating the modifications. Furthermore, the original character generated via the machine learning model 12 has been trained using image data with minimal copyright issues drawn by designers, thus reducing the risk of copyright infringement. After visually confirming the original character with reduced copyright infringement risk, the user can proceed with the production of character goods.

[0026] Figure 5 shows an example of a character modification screen. As shown in Figure 5, the user can input text data in natural language, such as "make the eyes bigger." The input text data and the image data P2 of the original character are input to the machine learning model 12, and the machine learning model 12 outputs modified image data P3. This allows the user to modify the character using natural language instructions without having to redraw the picture. This makes it possible for even ordinary people who are not professionals to adjust the character design using natural language without having to redraw the picture, thus lowering the hurdles to commercializing character goods. When inputting the text data and the image data P2 of the original character into the machine learning model 12, prompts are input to the machine learning model 12 to output the image data of the original character so that the first manufacturing device 30 and the second manufacturing device 40 can each generate manufacturing data that can be accepted, thereby enabling the output of image data suitable for manufacturing even after modification.

[0027] Figure 6 shows an example of the display screen for the modified original character. As shown in Figure 6, the modified image data P3, in which the character's features have been corrected according to the user's text instructions, is displayed. In the example in Figure 6, the user entered the text data "Make the eyes bigger," and as a result, the modified image data P3 of the original character with larger eyes was generated, compared to the original character image data P2 shown in Figure 5. The user can check the modification results and, if satisfied, confirm the modified image data P3 by operating the approval button. On the other hand, if further modifications are desired, the user can input text data again to give modification instructions. In this way, the user can repeatedly input text data and check the modified image data until they are satisfied with the modification results, so even ordinary people who are not professionals can reach their desired character design through interactive instructions using natural language, without having to redraw the picture. The user can repeatedly input text data and make modifications until they are satisfied with the modification results.

[0028] <Generating manufacturing data> The manufacturing data generation unit 14 generates first manufacturing data that can be accepted by the first manufacturing apparatus 30 and second manufacturing data that can be accepted by the second manufacturing apparatus 40, based on the image data P2 (or modified image data P3) of the original character.

[0029] The first manufacturing apparatus 30 is, for example, a 3D printer. In this case, the first manufacturing data is 3D model data of the original character. The manufacturing data generation unit 14 automatically generates 3D model data from 2D image data P2 of the original character. For generating the 3D model data, for example, neural radiance field, Gaussian splatting, or other 3D reconstruction techniques may be used. The first product G1 is, for example, a figurine of the original character. The figurine is molded and colored by the 3D printer. Note that the first manufacturing apparatus 30 is not limited to a 3D printer, but may be a CNC machine, a laser processing machine, or other manufacturing apparatus. In this way, by generating a product image of the figurine based on the first manufacturing data before actually manufacturing the figurine using the first manufacturing apparatus 30, it becomes possible to prepare a highly accurate product image and display it in an accessible manner to third parties without going through the hurdle of product prototyping.

[0030] The second manufacturing apparatus 40 is, for example, a printing apparatus. In this case, the second manufacturing data is image data for printing. The second product G2 is, for example, a sticker displaying an original character. The sticker may be a regular flat sticker, or it may be a sticker made of an elastic material that has an outline following the contour of the original character and displays the original character on its surface. A sticker made of such an elastic material becomes a character product that can be enjoyed both tactilely and visually. Furthermore, by generating a product image of the sticker based on the second manufacturing data before actually manufacturing the sticker using the second manufacturing apparatus 40, a highly accurate product image of the sticker can be prepared and displayed in a marketable form accessible to third parties without actually manufacturing the sticker. In particular, since a sticker made of an elastic material is formed to have a three-dimensionally raised surface, generating a highly accurate product image before manufacturing allows buyers to grasp the texture of the product in advance.

[0031] Furthermore, the second manufacturing apparatus 40 is not limited to a printing apparatus, but may also be an embroidery apparatus, an engraving apparatus, or other manufacturing apparatus. Also, the second product G2 is not limited to a sticker, but may also be a T-shirt, acrylic keychain, smartphone case, mug, Christmas ornament, or other product. In other words, in this embodiment, multiple different types of products can be generated all at once from a single original character. Conventionally, in order to develop multiple types of character goods, it was necessary to adjust the design, manufacture prototypes, and take product photos for each product individually, which required considerable cost and effort. In contrast, according to this embodiment, by simply inputting image data P2 of a single original character, manufacturing data and product images for multiple products are automatically generated, making it extremely easy for individuals to develop multiple types of character goods simultaneously. As a result, even ordinary people who are not professionals can sell their own created characters as a variety of products all at once.

[0032] <Generating and displaying product images> The product image generation unit 15 generates an image of the first product G1 of the original character before manufacturing the first product G1 using the first manufacturing device 30. Similarly, the product image generation unit 15 generates an image of the second product G2 of the original character before manufacturing the second product G2 using the second manufacturing device 40. The product images generated by the product image generation unit 15 are not limited to just the first product G1 and the second product G2. As shown in Figure 7, product images for multiple products such as figures, stickers, acrylic keychains, smartphone cases, and T-shirts may be generated all at once before manufacturing. In this way, high-precision product images can be prepared all at once for multiple products without actually manufacturing them. This eliminates the traditional hurdle of manufacturing prototypes for each product and taking photos of them, making it possible for even ordinary people to simultaneously sell multiple products of their created characters.

[0033] The product image generation unit 15 may generate product images by modifying the original character design to the optimal size, placement, and design for each product, depending on the type of product. For example, the product image for a figurine may be a rendered image of a 3D model, and the product image for a T-shirt may be a mockup image of a T-shirt with the character printed on it. In this way, product images that are automatically adjusted to the optimal size and placement for each product are generated from a single original character image data P2, so users do not need to manually adjust the size and placement of the design for each product, making it easy for even ordinary people to develop multiple types of character goods.

[0034] Figure 7 shows an example of a product list display screen. The display control unit 16 displays images of the first product G1 and the second product G2 in a manner that makes them available for sale, and makes them accessible to third parties. Specifically, the display control unit 16 displays the product images generated by the product image generation unit 15 on the EC site 50, along with price information and a purchase button. As shown in Figure 7, multiple products such as figurines, stickers, acrylic keychains, smartphone cases, and T-shirts are displayed in a list.

[0035] Buyers can select and purchase their desired products on the e-commerce site 50. When a purchase order is placed, the ordered products are manufactured by the first manufacturing device 30 or the second manufacturing device 40 and delivered to the buyer. In this embodiment, since product images generated before manufacturing are displayed on the e-commerce site 50 and products are manufactured only after a purchase order is received, even ordinary people can start selling character goods without incurring initial investment or inventory risk.

[0036] Furthermore, users may set their own prices on the e-commerce site 50. This will allow individuals to sell their own original character goods at their own prices.

[0037] <Calculation of contribution rate> If the machine learning model 12 has been trained using image data from multiple designers, the contribution rate calculation unit 17 calculates the contribution rate of each designer's image data to the original character when the product is sold. The contribution rate may be calculated, for example, based on the similarity between the original character and each designer's image data. Alternatively, the contribution rate may be calculated based on the weighting of each designer's training data in the machine learning model 12. For example, royalty income may be distributed to each designer based on the calculated contribution rate. With this configuration, the intellectual property (IP) of designers can be properly managed and their motivation can be maintained. In addition, by obtaining the contribution rate, it is possible to understand to what extent the original character relies on the style of each designer, so it is possible to detect when the original character is excessively similar to the copyrighted work of a particular designer and reduce the risk of copyright infringement.

[0038] <Processing Flow> Figure 2 is a flowchart showing the overall flow of the sales method for character goods according to this embodiment.

[0039] First, the input receiving unit 11 receives image data P1 of a character drawn by an individual (S100). For example, the individual draws a character in a sketchbook, takes a picture of it with the camera 22 of the mobile information terminal 20, and uploads it to the web application. Alternatively, the individual may input the image data P1 by drawing the character directly on the touch panel 23 of the mobile information terminal 20.

[0040] Next, the character generation unit 13 inputs the image data P1 into the machine learning model 12 and receives the output of the original character image data P2 from the machine learning model 12 (S101). The generated original character image data P2 is displayed on the display 21 of the information mobile terminal 20. The user can check the original character and, if necessary, input text data to instruct modifications (S102).

[0041] Next, the manufacturing data generation unit 14 generates first manufacturing data and second manufacturing data based on the image data P2 of the original character (S103). In other words, manufacturing data for multiple different products is generated all at once from a single original character.

[0042] Next, the product image generation unit 15 generates images of the first product G1 and the second product G2 (S104). This step is performed before the actual products are manufactured.

[0043] Next, the display control unit 16 displays images of the first product G1 and the second product G2 on the e-commerce site 50 in a manner that allows them to be sold and accessed by third parties (S105). As described above, by having a computer program execute each of the above steps, the entire process from manufacturing to selling character goods can be automated. This makes it possible for even ordinary people who are not professionals to sell characters they have created as character goods, eliminating the need to overcome many hurdles such as designing characters, prototyping products, and building e-commerce sites.

[0044] <Variations> In the embodiments described above, two types of products, the first product G1 and the second product G2, were used as examples, but the invention is not limited to these. Three or more types of products may be produced simultaneously. For example, a variety of products such as figurines, stickers, acrylic keychains, T-shirts, smartphone cases, mugs, and Christmas ornaments may be produced all at once.

[0045] Furthermore, although the above-described embodiment includes a configuration in which the server device 10 includes each functional unit, the embodiment is not limited to this. For example, the machine learning model 12 may operate on a device separate from the server device 10. Also, some or all of the functions of the manufacturing data generation unit 14 may be executed on the information mobile terminal 20.

[0046] Furthermore, while the above-described embodiment involves accessing a web application via the web browser of the mobile information terminal 20, the system is not limited to this configuration. For example, a dedicated application may be installed and used on the mobile information terminal 20. Alternatively, a dedicated hardware device, such as a handheld camera device or a terminal specifically for workshops, may be used.

[0047] Furthermore, while the above-described embodiment involves selling products online via an e-commerce site 50, the system is not limited to this. For example, products may be sold offline using a vending machine. In this case, product images can be displayed on the vending machine's screen, allowing customers to select and purchase their desired products.

[0048] Although embodiments of the present invention have been described above, the present invention is not limited to the embodiments described above, and various modifications are possible without departing from the spirit of the invention as described in the claims. [Explanation of symbols]

[0049] 1...Character goods sales system, 10...Server device, 11...Input reception unit, 12...Machine learning model, 13...Character generation unit, 14...Manufacturing data generation unit, 15...Product image generation unit, 16...Display control unit, 17...Contribution rate calculation unit, 18...Database, 20...Information mobile terminal, 21...Display, 22...Camera, 23...Touch panel, 30...First manufacturing equipment, 40...Second manufacturing equipment, 50...EC site, G1...First product, G2...Second product, P1...Image data, P2...Image data of original character, P3...Modified image data

Claims

1. A step to accept input of image data of a character drawn by an individual, The steps include inputting the aforementioned image data into a machine learning model trained on multiple image data drawn by at least one designer, and receiving output of original character image data from the machine learning model, A step of generating first manufacturing data that can be accepted by the first manufacturing apparatus for manufacturing a first product of the original character using the first manufacturing apparatus based on the image data of the original character, A step of generating second manufacturing data that can be accepted by the second manufacturing apparatus, for manufacturing a second product of the original character using the second manufacturing apparatus based on the image data of the original character, Before manufacturing the first product of the original character using the first manufacturing apparatus, the process includes the step of generating an image of the first product, Before manufacturing the second product of the original character using the second manufacturing apparatus, the steps include generating an image of the second product, The steps include: displaying images of the first product and images of the second product in a manner that allows them to be sold, in a manner that makes them accessible to third parties; Sales methods for character goods, including those mentioned.

2. The aforementioned machine learning model has been trained using image data drawn by at least one first designer and one second designer, respectively. The process further includes the step of obtaining, when the first product is sold, the contribution rate of the image data drawn by the first designer to the original character and the contribution rate of the image data drawn by the second designer to the original character. A method for selling character goods as described in claim 1.

3. The step of receiving image data of a character drawn by the aforementioned individual is: This includes the step of displaying a screen on a mobile information terminal for receiving the aforementioned image data input. A method for selling character goods as described in claim 1.

4. The further step includes displaying the image data of the original character that was received, A method for selling character goods as described in claim 1.

5. The process further includes the step of displaying image data of the original character, as well as the step of receiving input text data for modifying the original character. A method for selling character goods as described in claim 4.

6. The process further includes inputting the text data and the image data of the original character into the machine learning model and receiving the output of corrected image data of the original character from the machine learning model. A method for selling character goods as described in claim 5.

7. The first manufacturing apparatus is a three-dimensional printer, and the first product is a figurine of the original character. A method for selling character goods as described in claim 1.

8. The second product is a sticker on which the original character is displayed. A method for selling character goods as described in claim 1.

9. The second product is a sticker made of an elastic material, having an outline that follows the contour of the original character, and displaying the original character on its surface. A method for selling character goods as described in claim 8.

10. On the computer, A step to accept input of image data of a character drawn by an individual, The steps include inputting the aforementioned image data into a machine learning model trained on multiple image data drawn by at least one designer, and receiving output of original character image data from the machine learning model, A step of acquiring first manufacturing data that can be accepted by the first manufacturing apparatus, for manufacturing a first product of the original character using the first manufacturing apparatus based on the image data of the original character, A step of acquiring second manufacturing data that can be accepted by the second manufacturing apparatus, for manufacturing a second product of the original character using the second manufacturing apparatus based on the image data of the original character, Before manufacturing the first product of the original character using the first manufacturing apparatus, the steps include: acquiring an image of the first product; Before manufacturing the second product of the original character using the second manufacturing apparatus, the steps include: acquiring an image of the second product; The steps include: displaying images of the first product and images of the second product in a manner that allows them to be sold, in a manner that makes them accessible to third parties; A computer program designed to execute something.

Images