Image generation device, image generation program, and image generation method

Abstract

To provide an image generation device suitable for apparel design.SOLUTION: The image generation device generates an image representing a state in which a model wears clothing. The image generation device receives an input of clothing shape information indicating a shape of clothing, receives an input of fabric information regarding a fabric, receives an input of a model image including a model, and generates a wearing image based on the clothing shape information, the fabric information, and the model image. The wearing image is an image representing a state in which the model wears virtual clothing, and the virtual clothing is clothing in a case where clothing represented by the clothing shape information is manufactured using the fabric represented by the fabric information.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to an image generating device, an image generating program, and an image generating method. [Background technology]

[0002] Techniques for calculating the state of clothing worn by an avatar using physical property data of the fabric have been publicly known. For example, Patent Document 1 describes a technique for generating 3D mesh data showing the state of clothing worn by an avatar, based on the physical property data of the clothing fabric and a 3D body mesh model of the avatar. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2020-113101 Summary of the Invention [Problem to be solved by the invention]

[0004] However, conventional techniques have had problems in applying them to general apparel design.

[0005] For example, the technology in Patent Document 1 requires the user to prepare the avatar body mesh data required for the thermal fluid simulation, making it difficult for users without specialized knowledge of thermal fluid simulation to use. Furthermore, it requires a high-performance computer and dedicated applications, which poses a cost problem when used for general apparel design.

[0006] Furthermore, because the technology in Patent Document 1 is aimed at thermal fluid simulation, it is suitable for supporting the design of clothing shapes, but it cannot handle clothing colors or patterns, making it unsuitable for general apparel design.

[0007] The present invention has been made to solve the above problems, and has an object to provide an image generation device, an image generation program, and an image generation method suitable for apparel design. [Means for solving the problem]

[0008] An example of an image generating device according to the present invention is an image generating device that generates an image showing a model wearing clothing, the image generating device comprising: The image generating device Accepting input of clothing shape information representing the shape of the clothing; Accepts input of fabric information about the fabric, Accepts input of a model image containing a model, generating a wearing image based on the clothing shape information, the fabric information, and the model image; the dressing image is an image showing the model wearing the virtual clothing, The virtual garment is the garment represented by the garment shape information, manufactured using the fabric represented by the fabric information.

[0009] In one example, the image generation device generates an image representing the virtual garment without the model.

[0010] In one example, the fabric information includes information representing a pattern of the fabric, The image generating device performs a tiling process based on the image representing the pattern, the tiling process including a process of generating a repeating unit of an image to be used in a process of generating an image representing the virtual garment.

[0011] In one example, the clothing shape information includes information indicating the type of clothing.

[0012] In one example, the fabric information includes information representing at least one of the physical properties of the fabric, the color of the fabric, and the pattern of the fabric.

[0013] In one example, the clothing shape information includes information representing a three-dimensional shape of clothing, and the virtual clothing is represented by the three-dimensional shape.

[0014] In one example, the image generation device generates the dressed image by placing the virtual clothing in an area of ​​the model image that is associated with clothing worn by the model.

[0015] An example of an image generating program according to the present invention causes a computer to function as the image generating device described above.

[0016] An example of an image generation method according to the present invention is an image generation method for an image generation device to generate an image showing a model wearing clothing, the method comprising: The image generation method includes: A step in which a computer receives input of clothing shape information representing a shape of clothing; A step in which a computer receives input of fabric information regarding a fabric; A step in which a computer receives an input of a model image including a model; a step in which a computer generates a wearing image based on the clothing shape information, the fabric information, and the model image; Including, the dressing image is an image showing the model wearing the virtual clothing, The virtual garment is the garment represented by the garment shape information, manufactured using the fabric represented by the fabric information. [Effects of the Invention]

[0017] The image generating device, the image generating program, and the image generating method according to the present invention are suitable for apparel design.

[0018] Effects other than those mentioned above will be explained by the embodiments and modifications described in this specification and the drawings. [Brief explanation of the drawings]

[0019] [Figure 1] 1 is a diagram showing the configuration of an image generation server 10 according to a first embodiment of the present invention. [Figure 2] 3 is a flowchart showing an outline of processing executed by the image generation server 10 of FIG. [Figure 3] An example of a GUI (Graphical User Interface) for inputting clothing shape information. [Figure 4] An example of a GUI for inputting physical property information from among fabric information. [Figure 5] An example of a GUI for entering color information from among fabric information. [Figure 6] An example of a GUI for inputting pattern information from among fabric information. [Figure 7] Another example of a GUI for inputting pattern information, which is part of fabric information. [Figure 8] Example of virtual clothing image 81. [Figure 9] Example of GUI for inputting model images. [Figure 10] Example of outfit image 101. [Figure 11] 3 shows an example of a GUI used in steps S4 to S5 of FIG. DETAILED DESCRIPTION OF THE INVENTION

[0020] Hereinafter, an embodiment of the present invention will be described with reference to the accompanying drawings. [Embodiment 1] 1 is a diagram showing the configuration of an image generation server 10 according to the first embodiment. The image generation server 10, as an image generation device, executes the image generation method described in this specification. As an example, the image generation server 10 can accept inputs such as the shape of clothing, the color and pattern of the fabric, and an image of a model, and output an image showing the model wearing virtual clothing (virtual clothing).

[0021] In this specification, a person wearing clothes (which may be a fashion model) may be simply referred to as a "model." When simply referring to a "model," it means a person wearing clothes, and is different from a learning model related to machine learning.

[0022] The image generation server 10 has a hardware configuration as a known computer, and includes, for example, a calculation means 11, a storage means 12, an input means 13, and an output means 14. The calculation means 11 includes, for example, a processor. The processor can be manufactured using an integrated circuit, an ASIC, an FPGA, or the like. The storage means 12 includes, for example, a storage medium such as a semiconductor memory device or a magnetic disk device. Some or all of the storage medium may be non-transitory storage media.

[0023] The input means 13 includes input devices such as a keyboard and a mouse, and a communication device such as a network interface. The output means 14 includes output devices such as a display device and a printer, and the above-mentioned communication device. The communication device can function as both the input means 13 and the output means 14. The communication device can perform wired communication and / or wireless communication.

[0024] The storage unit 12 may store an image generation program (not shown). The processor may execute the image generation program, causing the computer to perform the functions described in this embodiment. That is, the image generation program causes the computer to function as the image generation server 10 according to this embodiment.

[0025] 2 is a flowchart showing an outline of the processing executed by the image generation server 10. This flowchart shows an image generation method in which the image generation server 10 according to this embodiment generates an image showing a model wearing clothing. The image generation server 10 executes two main processes: a process for generating an image of virtual clothing (a virtual clothing step), and a process for generating an image of a model wearing the virtual clothing (a wearing step).

[0026] First, the virtual clothing step will be described. In the virtual clothing step, the image generation server 10 receives input of clothing shape information (step S1). The clothing shape information includes information representing the shape of the clothing.

[0027] FIG. 3 shows an example of a GUI (Graphical User Interface) for inputting clothing shape information. The image generation server 10 can accept the input of clothing shape information by outputting such a GUI. For example, in FIG. 3, an image 31 representing the shape of clothing is displayed in association with information identifying the clothing (in this example, name 32). In the example of FIG. 3, a selection mark 33 is displayed in association with the image of a dress shirt, indicating that the dress shirt has been selected. In other words, information representing the shape of the dress shirt is input as clothing shape information.

[0028] 3, the clothing shape information includes information indicating the type of clothing. The type may include one or more of a hoodie, a T-shirt, a short-sleeved shirt, a dress, a dress shirt, a camisole, shorts, and a jacket. This configuration allows the user to easily identify the three-dimensional shape of the clothing simply by selecting the type of clothing, improving user convenience.

[0029] The user can input clothing shape information of a desired piece of clothing by performing a selection operation (for example, tapping or clicking) on ​​the image of the desired piece of clothing. For example, the image generation server 10 can store the image and / or name of the clothing in association with the clothing shape information of the clothing, and can accept input of the corresponding clothing shape information in response to a selection operation on the image and / or name.

[0030] The clothing shape information includes, for example, information representing the three-dimensional shape of the clothing. This information may be called 3D mockup data, 3D mesh data, etc. The image displayed in the GUI of FIG. 3 is preferably one representing the clothing shape information (for example, a two-dimensional grayscale image representing the three-dimensional appearance of the clothing).

[0031] Prior to step S1, the image generation server 10 may receive and store desired clothing form information from the user. In this case, the clothing form information may be added as an option for selection in step S1. Alternatively, the image generation server 10 may be designed so that desired clothing form information is input to the image generation server 10 at the time of step S1, regardless of the selection operation.

[0032] Next, the image generation server 10 receives input of fabric information (step S2). The fabric information is information related to the fabric. The fabric information includes, for example, information representing at least one of the physical properties of the fabric, the color of the fabric, and the pattern of the fabric.

[0033] 4 to 7 show examples of GUIs for inputting texture information. The image generation server 10 can accept input of texture information by outputting such GUIs.

[0034] FIG. 4 is an example of a GUI for inputting information representing the physical properties of a fabric (physical property information) from among the fabric information. For example, in FIG. 4, an image 41 representing the appearance of the fabric and information identifying the fabric (in this example, identification number 42) are displayed in association with each other. In the example of FIG. 4, a selection mark 43 is displayed in association with fabric "FT-3000," thereby indicating that fabric "FT-3000" has been selected. In other words, information relating to the fabric having identification number "FT-3000" has been input as the fabric's physical property information.

[0035] The GUI in Fig. 4 may also display the result of the selection made in step S1. In the example in Fig. 4, an image 44 of a dress shirt is displayed at the bottom, indicating that a dress shirt was selected in step S1. This allows the user to select a fabric while checking the shape of the garment, improving convenience.

[0036] The user can input the physical property information of a desired fabric by performing a selection operation (for example, tapping or clicking) on ​​the image of the fabric. For example, the image generation server 10 can store the image and / or identification number of a fabric in association with the physical property information of the fabric, and can accept the input of the corresponding physical property information in response to a selection operation on the image and / or identification number.

[0037] The content and format of the physical property information can be designed arbitrarily. For example, the data format may be .dlb format. Furthermore, the physical property information may include information that represents the three-dimensional shape of the fabric. Furthermore, the physical property information may include information about the surface irregularities, information about expansion and contraction, information about thickness, information about firmness, etc. As a specific example, the physical property information may include information that can represent the degree of flare of a flared skirt.

[0038] The image displayed in the GUI of FIG. 4 can be, for example, a color image or a grayscale image, but even when a color image is used, the physical property information can be one that does not include information about color.

[0039] Prior to step S2, the image generation server 10 may receive and store desired physical property information input from the user. In that case, the physical property information may be added as a selection target in step S2. Furthermore, it is also possible to design the image generation server 10 so that desired physical property information is input to the image generation server 10 at the time of step S2, regardless of the selection operation.

[0040] Fig. 5 shows an example of a GUI for inputting information representing the color of a fabric (color information) from among the fabric information. For example, Fig. 5 displays images 51 representing the color (e.g., solid-colored areas) and information classifying the color (names 52 in this example). For example, 12 types of images 51 are displayed as colors belonging to the name 52 "Red." Note that for convenience of illustration, each image 51 is shown in grayscale, but in reality each image 51 is displayed in red or another color.

[0041] In FIG. 5, a scroll bar 53 is displayed, and the image generation server 10 scrolls the screen in response to an operation on this scroll bar 53, allowing a variety of colors to be selected.

[0042] The user can input color information of the fabric by selecting (for example, tapping or clicking) the image of the desired color. The color information can be expressed, for example, as RGB values, but is not limited to this.

[0043] It should be noted that it is also possible to design the image generation server 10 so that color information is input directly to the image generation server 10, without using a selection operation.

[0044] FIG. 6 is an example of a GUI for inputting information (pattern information) representing the pattern of a fabric, among the fabric information. The pattern may be what is called a "pattern." For example, in FIG. 6, an image 61 representing the pattern and information classifying the pattern (in this example, name 62) are displayed. For example, six types of images 61 are displayed as colors belonging to the name 62 "Flower." Note that for convenience of illustration, each image 61 is shown in grayscale, but in reality, each image 61 may represent a colored pattern. Note that the pattern does not need to be one that is specifically created for use on clothing, and any image, etc., can be used.

[0045] 6, a selection mark 63 is displayed in relation to image 61a among images 61, thereby indicating that the pattern associated with image 61a has been selected. That is, information about the pattern associated with image 61a has been entered as fabric pattern information.

[0046] The user can input pattern information by performing a selection operation (for example, tapping or clicking) on ​​the image of the desired pattern. For example, the image generation server 10 can accept input of corresponding pattern information in response to a selection operation on the pattern image. Here, the image 61 displayed on the GUI does not need to match the image representing the pattern that is actually input; for example, the image 61 may represent only a part of the entire pattern.

[0047] Prior to step S2, the image generation server 10 may receive and store desired pattern information input from the user. In that case, the pattern information may be added as an option for selection in step S2. Also, it is possible to design the image generation server 10 so that the desired pattern information (for example, in the form of an image) is input to the image generation server 10 at the time of step S2, without relying on a selection operation.

[0048] In FIG. 6, a scroll bar 69 is displayed, and the image generation server 10 scrolls the screen in response to an operation on this scroll bar 69, allowing a variety of patterns to be selected.

[0049] 4 to 6, the results of the selections made up to that point may be displayed. In the example of FIG. 6, an image 64 of a dress shirt, an image 65 showing the appearance of the fabric, an image 66 showing the color of the fabric, and an image 67 showing the pattern of the fabric are displayed at the bottom. This allows the user to select a pattern while checking the results of the selections made up to that point, which is more convenient. Also, in the example of FIG. 6, a proceed button 68 is displayed to confirm the input contents made up to that point and to instruct the end of step S2.

[0050] Such a configuration for showing the results of the selection can be provided in any of the GUIs shown in Figures 4 to 6, and if the GUIs shown in Figures 4 to 6 are configured to be displayed on a single screen (including when scrollable), one configuration can be provided for all of them.

[0051] 7 shows another example of a GUI for inputting pattern information from among fabric information. In this example, the GUI includes a search field 71, which allows a pattern to be searched for based on information entered in the search field (such as a pattern identification number or file name).

[0052] The GUI for inputting pattern information is not limited to those shown in Figures 6 and 7 and can be modified as appropriate. For example, a word representing a category shown in Figure 6 (e.g., "Flower") may be input, and pattern information relating to patterns belonging to that category may be displayed as selection candidates. In that case, a field for selecting or inputting a category may be displayed in the GUI of Figure 6.

[0053] Instead of categories, specific evaluation criteria may be entered, and patterns that match the criteria may be displayed as selection candidates. For example, the image generation server 10 may display a list of evaluation criteria using abstract expressions such as "cute," "chic," and "modern," and may accept input for selecting one or more evaluation criteria from among them. Images representing patterns may then be evaluated based on the selected evaluation criteria, and one or more images with high evaluation results may be extracted. The patterns associated with the extracted images may be displayed in a GUI such as that shown in FIG. 6 or FIG. 7 for selection by the user.

[0054] Furthermore, the image generation server 10 may automatically recommend a specific pattern(s) based on previously input information (e.g., at least one of clothing shape information, physical property information, and color information). If there are multiple recommended patterns, they may be displayed in a GUI such as that shown in FIG. 6 or FIG. 7 for selection by the user.

[0055] The fabric information may include information about the fabric other than the above. Furthermore, although it is preferable that the fabric information include all of the physical property information, color information, and pattern information, this is not required. For example, if either the color information or the pattern information is input, it is possible to execute subsequent processing. More specifically, if no color information is input, the color information included in the pattern information can be used (for example, the color of the pattern can be used as is without changing it). Furthermore, if the clothing shape information includes information about color, that information can be used. Furthermore, if no pattern information is input, it can be treated as if a plain fabric of a specific color has been input.

[0056] Thus, in step S2, the image generation server 10 accepts input of fabric information. Next, the image generation server 10 generates an image (virtual clothing image) representing virtual clothing based on the clothing shape information and fabric information (step S3). The virtual clothing is virtual clothing that results when clothing represented by the clothing shape information is manufactured using fabric represented by the fabric information. The virtual clothing may be represented by data including the three-dimensional shape of the clothing and information related to the fabric (which may include the fabric information described above).

[0057] FIG. 8 shows an example of a virtual clothing image 81. In this example, the background is displayed in black. The processing in step S3 can be designed as desired by those skilled in the art, and publicly known techniques may be used. For example, if the shape of the clothing represents the shape of one or more surfaces arranged in three-dimensional space, three-dimensional data for the virtual clothing can be generated by arranging fabric (which may have thickness) along the shapes of the surfaces, and a two-dimensional virtual clothing image 81 can be generated based on this. In particular, virtual clothing image 81 does not include a model, and in this respect it differs from the wearing image generated in step S5, described below.

[0058] Alternatively, machine learning may be used. For example, the image generation server 10 may store a trained model. This trained model is trained to input clothing shape information and fabric information and output an appropriate virtual clothing image. More specifically, such a trained model can be generated by preparing training data consisting of sets of clothing shape information, fabric information, and correct virtual clothing images, inputting the clothing shape information and fabric information into the training model, and training the model so that the difference between the output image and the correct virtual clothing image is small.

[0059] A known image generation AI (such as Stable Diffusion or DALL-E) may be used. For example, the image generation server 10 may transmit clothing shape information and fabric information to an image generation AI server and receive a virtual clothing image from the image generation AI server. The image generation AI may have an API (Application Programming Interface) for input and output, and may accept prompts in natural language. For example, the API may accept clothing shape information and fabric information as input, and generate a virtual clothing image as output.

[0060] The image generation server 10 may use a combination of a trained model stored in itself and a known image generation AI. For example, it may transmit clothing shape information and 3D data of virtual clothing generated by the trained model to the image generation AI server, and receive a virtual clothing image from the image generation AI server. In this case, the 3D data of the virtual clothing may be actual 3D data, or may be data encoded in a predetermined format suitable for the image generation AI.

[0061] The selected pattern information may be used as is for the fabric pattern, or a tiling process may be performed. Tiling refers to the process of generating an image that is actually used in the process of generating virtual clothing (or an image representing the virtual clothing) based on the pattern (or an image representing the pattern). The tiling process may include the process of generating a repeating unit of the image that is actually used (for example, but not limited to, a rectangular image). The tiling process may include the process of extracting a specific region of the pattern, the process of rotating part of the pattern (for example, a repeating unit) or the entire pattern, the enlargement process, the reduction process, etc. The tiling process can be performed automatically or manually, and when performed automatically, known techniques may be used as appropriate.

[0062] In this way, because the pattern is corrected by the tiling process, the image representing the pattern does not need to be one specifically created for use on clothing, and any image can be used. For example, if the pattern has an up and down direction, the pattern can be rotated depending on the position on the clothing where it will be used, generating a pattern that is suitable for the clothing. In this way, user convenience is improved.

[0063] The virtual clothing step is thus completed. At this point, the image generation server 10 may output information about the generated virtual clothing (including the clothing's three-dimensional shape, information about the fabric, etc.), or may output a virtual clothing image 81. In particular, by outputting the virtual clothing image 81, the user can easily confirm the appearance of a desired type of clothing when made with a desired fabric, color, and pattern. The virtual clothing information and virtual clothing image 81 may be output to the storage means 12 of the image generation server 10, to a display device, or to another computer via a communications network.

[0064] When the virtual clothing step is completed, the dressing step begins. In the dressing step, the image generation server 10 next accepts input of a model image (step S4). The model image is an image including a model, i.e., an image showing at least a part of the appearance of a fashion model. The model image may be an image representing a photograph or a part of it. As an example, the model image may be an image including the clothing as it was when photographed.

[0065] Fig. 9 is an example of a GUI for inputting a model image. In Fig. 9, a plurality of model images 91 are displayed. In the example of Fig. 9, a selection frame 92 is displayed around one of the model images 91, thereby indicating that the model image in question has been selected. In other words, the model image specified by the selection frame 92 has been input.

[0066] 9 also includes a display switching button 93. In response to the display switching button 93 being operated, the image generation server 10 switches the model image being displayed.

[0067] The user can input a desired model image 91 by performing a selection operation (for example, tapping or clicking) on ​​the model image 91. The image generation server 10 can accept the input of the model image in response to the selection operation on the model image 91.

[0068] Prior to step S4, the image generation server 10 may receive and store a desired model image from the user. In this case, the model image may be added as an option for selection in step S4. Alternatively, the image generation server 10 may be designed to input the desired model image to the image generation server 10 at the time of step S4, regardless of the selection operation.

[0069] Next, the image generation server 10 generates a dressing image (step S5). A dressing image is an image showing a state in which a model is wearing clothes (particularly virtual clothes), and more specifically, an image showing a state in which a model related to the model image input in step S4 is wearing the virtual clothes related to the virtual clothing image generated in step S3. The dressing image can be said to be an image that predicts or estimates the state in which a model is wearing the virtual clothes.

[0070] FIG. 10 shows an example of a dressing image 101. For comparison, a model image 91 is also shown. The dressing image 101 is generated by replacing clothing parts in the model image 91 with other clothing (virtual clothing). As mentioned above, the virtual clothing image 81 (FIG. 8) does not include a model, but the dressing image 101, as shown in FIG. 10, is an image that includes a model.

[0071] FIG. 11 is an example of a GUI used in steps S4 and S5. In this example, the GUI displays a virtual clothing image 81, a GUI 90 (similar to FIG. 9) for inputting a model image, and a wearing image 101. The GUI may also display the results of selections made up to that point. In the example of FIG. 11, an image 111 of a dress, an image 112 representing the appearance of the fabric, an image 113 representing the color of the fabric, and an image 114 representing the pattern of the fabric are displayed at the bottom. This allows the user to select a model image while checking the results of selections made up to that point, improving convenience. Also, in the example of FIG. 11, a cancel button 115 is displayed to discard the input contents made up to that point and return to step S1.

[0072] In step S5, the image generation server 10 generates a wearing image based on information representing the virtual garment and the model image 91. Here, the information representing the virtual garment may include information about the three-dimensional shape and fabric of the garment, or may be the virtual garment image 81. In either case, the image generation server 10 can be said to generate the wearing image 101 based on the clothing shape information input in step S1, the fabric information input in step S2, and the model image 91 input in step S4. More specifically, the image generation server 10 generates the wearing image 101 by placing the virtual garment in an area of ​​the model image 91 that is associated with the garment worn by the model.

[0073] The dressing image 101 may be an image in which the clothes in the model image 91 are simply replaced with virtual clothes, or may be an image in which further processing has been performed after the clothes have been replaced. For example, the image generation server 10 may preferably perform processing such as deforming the virtual clothes assuming the action of gravity, estimating the lighting conditions in the model image 91 and applying the estimated conditions to the virtual clothes, etc.

[0074] The image generation server 10 may output the generated wearing image 101. For example, the image may be output to the storage means 12 of the image generation server 10, to a display device or the like, or to another computer via a communication network.

[0075] The processing in step S5 can be designed arbitrarily by those skilled in the art and may use known techniques. For example, machine learning may be used. For example, the image generation server 10 may store a trained model. This trained model may be trained to generate a 3D shape of a human body based on a model image. This trained model is trained to input a model image and output an appropriate 3D shape of a human body without clothes. More specifically, such a trained model can be generated by preparing training data in which a model image and a correct human body shape are paired, inputting the model image into the training model, and performing training so as to reduce the difference between the output human body shape and the correct human body shape.

[0076] The image generation server 10 generates a three-dimensional shape of a human body using such a trained model, and then generates an image of the three-dimensional shape dressed in virtual clothing as a dressing image. For example, based on the three-dimensional shape of the human body and information about the virtual clothing (including the three-dimensional shape of the clothing, information about the fabric, etc.), the image generation server 10 can generate three-dimensional data (including information about the three-dimensional shape and fabric of the clothing) of the human body dressed in the virtual clothing by arranging the virtual clothing along the three-dimensional shape of the human body, and then generate a two-dimensional image representing the generated three-dimensional data as the dressing image.

[0077] As another example, the image generation server 10 may directly generate a wearing image using another trained model. This trained model is trained to input the virtual clothing image 81 and the model image 91 and output an appropriate wearing image. More specifically, training data consisting of pairs of virtual clothing images, model images, and correct wearing images is prepared, and the virtual clothing images and model images are input to the learning model, which can then be trained to reduce the difference between the output image and the correct wearing image.

[0078] A known image generation AI (such as Stable Diffusion or DALL-E) may be used. For example, the image generation server 10 may transmit a model image and a virtual clothing image to an image generation AI server and receive a wearing image from the image generation AI server. Here, information about the virtual clothing (including the three-dimensional shape of the clothing, information about the fabric, etc.) may be used instead of the virtual clothing image. The image generation AI may have an API for input and output, and may be able to accept prompts in natural language. For example, the API may accept a model image and virtual clothing as input, and generate a wearing image as output.

[0079] The image generation server 10 may use a combination of a trained model stored in itself and an image generation AI. For example, the image generation server 10 may generate 3D data (including information about the 3D shape and fabric of the clothing) of a human body wearing virtual clothing using the trained model stored in itself, send this data and a model image to the image generation AI server, and receive a wearing image from the image generation AI server. In this case, the 3D data of a human body wearing the virtual clothing may be actual 3D data, or may be data encoded in a predetermined format suitable for the image generation AI.

[0080] Thus, the dressing step is completed. The image generation server 10 outputs the dressing image 101, allowing the user to easily check the appearance of the desired model wearing the desired virtual garment. In this way, the image generation server 10 according to this embodiment is suitable for apparel design.

[0081] The above-described embodiment can be modified as follows.

[0082] In step S3, the image generation server 10 may generate only data including information about the three-dimensional shape and fabric of the virtual clothing, without generating a two-dimensional virtual clothing image. In such a case, the generation of the wearing image in step S5 can be performed using information about the three-dimensional shape and fabric of the virtual clothing, instead of the virtual clothing image.

[0083] In the above-described embodiment, the model image 91 is a two-dimensional image, and the three-dimensional shape of the human body is not required. Therefore, a dressing image can be generated without performing a three-dimensional scan of the model, which increases convenience. However, as a modified example, in steps S4 and S5, in addition to or instead of the model image 91, information including the three-dimensional shape of the model's human body can be input and used.

[0084] Those skilled in the art can add, change, or delete components as desired. For example, the image generation server 10 can be configured by multiple computers connected via a communication network. In this case, the image generation server may be called an image generation system.

[0085] The order of the steps in each figure can be changed as needed. For example, the order of steps S1 and S2 in Figure 2 can be reversed. Step S4 can be executed at any time before S5. [Explanation of symbols]

[0086] 10...Image generation server (image generation device) 31...Image showing the shape of clothing 41...Image showing the appearance of the fabric 51...Images that represent colors 61,61a...Images showing patterns 81...Virtual clothing image 91...Model image 101...wearing image

Claims

1. An image generating device that generates an image showing a model wearing clothing, The image generating device Accepting input of clothing shape information representing the shape of the clothing; Accepts input of fabric information about the fabric, Accepts input of a model image containing a model, generating a wearing image based on the clothing shape information, the fabric information, and the model image; the dressing image is an image showing the model wearing the virtual clothing, the virtual garment is a garment represented by the garment shape information manufactured using a fabric represented by the fabric information; Image generating device.

2. the image generating device generates an image representing the virtual garment without the model. The image generating device of claim 1 .

3. the fabric information includes information representing a pattern of the fabric, the image generation device performs a tiling process based on the image representing the pattern, the tiling process including a process of generating a repeating unit of an image to be used in a process of generating an image representing the virtual garment; The image generating device of claim 2 .

4. The clothing shape information includes information indicating the type of clothing. The image generating device of claim 1 .

5. The fabric information includes information representing at least one of the physical properties of the fabric, the color of the fabric, and the pattern of the fabric. The image generating device of claim 1 .

6. the clothing shape information includes information representing a three-dimensional shape of clothing, and the virtual clothing is represented by the three-dimensional shape; The image generating device of claim 1 .

7. the image generation device generates the dressed image by placing the virtual clothing in an area of ​​the model image that is associated with the clothing worn by the model; The image generating device of claim 1 .

8. An image generating program that causes a computer to function as the image generating device according to claim 1.

9. An image generation method for generating an image showing a model wearing clothing by an image generation device, comprising: The image generation method includes: A step in which a computer receives input of clothing shape information representing a shape of clothing; A step in which a computer receives input of fabric information regarding a fabric; A step in which a computer receives an input of a model image including a model; a step in which a computer generates a wearing image based on the clothing shape information, the fabric information, and the model image; Including, the dressing image is an image showing the model wearing the virtual clothing, the virtual garment is a garment represented by the garment shape information manufactured using a fabric represented by the fabric information; Image generation method.

Images