Recording device and its control method, program, and storage medium

Abstract

This invention provides a recording device that can suppress quality degradation when generating thumbnail images from prompts. [Solution] The system comprises an acquisition unit that acquires a first image, a first generation unit that generates a prompt, which is text information describing the content of the first image, based on the acquired first image, a second generation unit that generates a second image that reproduces the content described by the prompt from the prompt generated by the first generation unit, a third generation unit that reduces the size of the second image to generate a thumbnail image, and a recording unit that associates the prompt and the thumbnail image and records them on a recording medium.

Description

Technical Field

[0001] The present invention relates to a recording device that records information related to an image.

Background Art

[0002] Conventionally, digital image data captured and acquired by a digital camera, a smartphone, etc. has been used to share experiences and emotions.

[0003] In addition, as a technology related to digital images, remarkable progress has been made in the field of image generation AI that utilizes a high-performance CPU and cloud services.

[0004] For example, when inputting a combination of words written in a text-based form called a "prompt" as input data into a generation AI to perform image generation, an image that expresses the description described in the prompt can be generated.

[0005] In recent years, a technique (Non-Patent Document 1) has been proposed in which a generation AI model called a diffusion model is made to learn the process of removing noise from an image and the process is controlled by a prompt to generate a new image. This technique also contributes to the evolution of image generation technology by AI.

[0006] In addition to this, in recent years, like the research results of multimodal AI technology (Non-Patent Document 2) published, it has become possible to accurately realize image generation that reproduces the content of text and text generation that describes the content of an image with a single AI model. As a result, bidirectional conversion between an image and text has become possible.

[0007] Furthermore, as the performance of the image generation AI model improves and the situation becomes such that an image and a prompt are associated in a roughly one-to-one manner, in order to reduce the data amount of the image, it is assumed that the prompt will be treated as content instead of the image.

[0008] However, when treating prompts as content, it is difficult for humans to intuitively grasp the image that can be generated from a prompt simply by looking at the prompt itself.

[0009] Furthermore, as prompts are optimized to allow image generation AI models to more faithfully reproduce the content of the prompt as an image, even text-based prompts may not always be easily readable by humans. Therefore, it is conceivable to improve the readability of prompt data by associating and recording thumbnail images generated from prompts with the prompts themselves. [Prior art documents] [Non-patent literature]

[0010] [Non-Patent Document 1] Jonathan Ho, Ajay Jain, Pieter Abbeel “Denoising Diffusion Probabilistic Models”, in NeurIPS (2020) [Non-Patent Document 2] Meta AI Research, “Scaling Autoregressive Multi-Modal Models: Pretraining and Instruction Tuning”, https: / / ai.meta.com / research / publications / scaling-autoregressive-multi-modal-models-pretraining-and-instruction-tuning [Overview of the Initiative] [Problems that the invention aims to solve]

[0011] Incidentally, AI image generation models are generally optimized for generating images at a specific resolution. This is based on the resolution of the dataset used during model training. Therefore, if images are generated at a resolution significantly different from the optimized resolution, the quality of the generated images cannot be guaranteed.

[0012] It is particularly unsuitable for generating low-resolution images such as thumbnails, and since images are scaled down to various resolutions depending on their display purpose, there is a challenge in that it is not easy to design models individually for each resolution.

[0013] The present invention has been made in view of the above-mentioned problems, and its objective is to provide a recording device that can suppress the degradation of quality when generating thumbnail images from prompts. [Means for solving the problem]

[0014] A recording device according to the present invention is characterized by comprising: an acquisition means for acquiring a first image; a first generation means for generating a prompt, which is text information describing the content of the first image, based on the acquired first image; a second generation means for generating a second image that reproduces the content described by the prompt from the prompt generated by the first generation means; a third generation means for reducing the size of the second image to generate a thumbnail image; and a recording means for associating the prompt and the thumbnail image and recording them on a recording medium. [Effects of the Invention]

[0015] According to the present invention, it is possible to suppress the degradation of quality when generating thumbnail images from prompts. [Brief explanation of the drawing]

[0016] [Figure 1] A diagram showing the configuration of a recording device relating to one embodiment of the present invention. [Figure 2] A diagram showing the configuration of the thumbnail generation unit. [Figure 3]Flowchart showing the operation of the recording process in the recording device. [Figure 4] Diagram showing an input image, a prompt, and a thumbnail image. [Figure 5] Diagram explaining the relationship between the resolution, quality, and generation time of an image generated by an image generation AI model. [Figure 6] Flowchart showing the operation of the thumbnail generation process. [Figure 7] Diagram explaining the thumbnail generation process.

Mode for Carrying Out the Invention

[0017] Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. Note that the following embodiments do not limit the invention according to the claims. Although a plurality of features are described in the embodiments, not all of these plurality of features are essential for the invention, and the plurality of features may be arbitrarily combined. Further, in the accompanying drawings, the same or similar configurations are given the same reference numerals, and duplicate explanations are omitted.

[0018] <Functional Configuration Example of the Recording Device> FIG. 1 is a diagram showing the configuration of a recording device 100 according to an embodiment of the present invention.

[0019] FIG. 1(a) is a block diagram showing a functional configuration example of the recording device 100. Throughout the drawings, the functional blocks can be realized by software or a combination of software and hardware, excluding parts that can be realized only by hardware (e.g., a recording medium). For example, the functional blocks may be realized by dedicated hardware such as an ASIC. Also, the functional blocks may be realized by one or more processors capable of executing a program, such as a CPU, executing a program stored in a memory. Note that a plurality of functional blocks may be realized by a common configuration (e.g., one ASIC). Also, the hardware that realizes some functions of a certain functional block may be included in the hardware that realizes other functional blocks.

[0020] In FIG. 1(a), for the sake of easy explanation and understanding of the present embodiment, only the image input unit 101, the prompt generation unit 102, the thumbnail generation unit 103, and the recording unit 104 are shown as the functional blocks of the recording apparatus 100. However, the functions of the recording apparatus 100 are not limited to the functions realized by the functional blocks shown in FIG. 1(a).

[0021] The image input unit 101 acquires, as input image data, the image data recorded on a recording medium or the like mounted on the recording apparatus 100. Here, the image data is, for example, a captured image taken by a digital camera or a smartphone and recorded on the recording medium. The image input unit 101 supplies the acquired image data to the prompt generation unit 102. In addition to the image data, the image input unit 101 may supply to the prompt generation unit 102 a part of the attached information recorded in the data file in which the input image data is stored (for example, one or more tags regarding the shooting conditions and shooting situation of the captured image).

[0022] In the present embodiment, the case of acquiring image data from a recording medium mounted on the recording apparatus 100 will be described as an example. However, the method of acquiring image data is not limited to this. For example, the recording apparatus 100 may acquire image data recorded on an external server or in the cloud with which it can communicate. Alternatively, the recording apparatus 100 may acquire image data recorded on an external storage device such as a USB memory or an external HDD connected to the recording apparatus 100.

[0023] The prompt generation unit 102 has a function of converting image data into a prompt, which is text information explaining the scene represented by the image data, and generates a prompt based on the image data supplied from the image input unit 101 and various kinds of information.

[0024] The prompt generation unit 102 uses a multimodal AI learning model (generative AI model) to convert image data into text information that describes the scene represented by the image data. The multimodal AI learning model may be pre-stored in the recording device 100, for example, or it may reside in an external device that the recording device 100 can communicate with. The multimodal AI learning model in this embodiment is a neural network that takes image data as input and learns from text data such as captions and tags about the scene associated with the image data as training data.

[0025] A multimodal AI learning model outputs text data corresponding to input image data. This text data describes the scene represented by the image data. The multimodal AI learning model also outputs image data corresponding to input text data. This image data reproduces the content described by the text data. Such a multimodal AI learning model capable of bidirectional conversion between image data and text data can be realized using known technologies, such as those described in Non-Patent Document 2.

[0026] The prompt generation unit 102 may also obtain prompts from an external device with which the recording device 100 can communicate. In this case, the prompt generation unit 102 transmits image data for prompt generation (and other information as needed) to the external device. The prompt generation unit 102 then receives the prompt generated by the external device in response to the image data.

[0027] The thumbnail generation unit 103 generates a thumbnail image from the prompt supplied by the prompt generation unit 102. This thumbnail image reproduces the content described by the prompt and has a lower resolution than the input image used to generate the prompt.

[0028] Figure 2 is a block diagram showing an example of the functional configuration of the thumbnail generation unit 103. The thumbnail generation unit 103 includes at least a prompt / image conversion unit 201 and an image reduction unit 202.

[0029] The prompt / image conversion unit 201 uses a multimodal AI learning model to convert a prompt, which is text information, into an image that reproduces the content described by the prompt. The multimodal AI learning model in this embodiment is the same model as the multimodal AI learning model used in the prompt generation unit 102. However, the multimodal AI learning model used in the prompt / image conversion unit 201 is not limited to the same model as the multimodal AI learning model used in the prompt generation unit 102, and may be a different model. In that case, for example, it may be pre-stored in the recording device 100, or it may reside in an external device that the recording device 100 can communicate with.

[0030] Furthermore, when the prompt / image conversion unit 201 converts a prompt into an image that reproduces the content described by the prompt, it may refer to the input image data used to generate the prompt. By referring to the input image data, it becomes possible to convert the prompt into an image in which the elements constituting the image, such as composition and subject matter, are closer to the input image data.

[0031] The prompt / image conversion unit 201 may also acquire the converted image from an external device with which the recording device 100 can communicate. In this case, the prompt / image conversion unit 201 transmits prompt data for image generation (and other information as needed) to the external device. The prompt / image conversion unit then receives the image generated by the external device in response to the prompt.

[0032] The image reduction unit 202 reduces the image output by the prompt / image conversion unit 201 (including to 1:1 scale) to output an image of any resolution. The reduction process may be performed using, for example, a bicubic interpolation method.

[0033] The recording unit 104 records the prompt output by the prompt generation unit 102 and the thumbnail image output by the thumbnail generation unit 103 to the recording destination. The recording destination may be a recording medium installed in the recording device 100, a connected storage device, or a communication-capable external device.

[0034] Images recorded on the recording destination by the recording device 100 are played back on the playback device as needed by the user. When the playback device is instructed to display a screen for selecting the image the user wants to play back, a list of prompts recorded on the recording destination and their corresponding thumbnail images are displayed. Alternatively, only thumbnail images may be displayed without the prompts. The user can easily select the image they want to play back by looking at this list. If only the text data of the prompts is displayed in the list, the user cannot intuitively select the image they want from the prompts. In contrast, by displaying thumbnail images along with the prompts, or only thumbnail images, as in this embodiment, the user can intuitively understand the prompts and select an image.

[0035] Furthermore, the recording device 100 records not the original image data, but the prompt and thumbnail image corresponding to that image data to a recording medium or other recording location. Therefore, the amount of data to be recorded can be significantly reduced compared to when the original image data is recorded. The playback device uses a multimodal AI learning model to convert the prompt, which is text information, from the prompt recorded on the recording location into an image that reproduces the content that the prompt describes. In this way, the playback device can reproduce an image that is close to the original image, which has a large amount of data, from the prompt, which has a small amount of data. However, if there is sufficient recording capacity, the original image data may also be recorded on the recording medium, not just the prompt and thumbnail image.

[0036] <Example of recording device hardware configuration> Figure 1(b) is a block diagram showing an example of the hardware configuration of the recording device 100. Each block is connected to the others via the system bus 111 so that they can communicate with each other.

[0037] The CPU (Central Processing Unit) 112 and GPU (Graphics Processing Unit) 116 are each one or more processors capable of executing programs. The GPU 116 is configured to perform certain operations faster than the CPU 112, and in recent years it has often been used to perform neural network inference processing at high speed. Instead of the GPU 116, an NPU (Neural Processing Unit) that is more specialized for performing neural network training and inference processing may be used. Alternatively, instead of using separate CPU 112 and GPU 116, a System on Chip (SoC) that integrates the CPU and GPU (and possibly an NPU) may be used.

[0038] The CPU 112 works in cooperation with the GPU 116 to realize each functional block described in Figures 1 and 2 by, for example, loading a program stored in the ROM 113 into the RAM 114 and executing it. The CPU 112 achieves high-speed processing by utilizing the GPU 116 for processing that uses neural networks. Note that the GPU 116 may have its own dedicated RAM separate from the RAM 114.

[0039] ROM113 is, for example, an electrically rewritable non-volatile memory. ROM113 stores programs that the CPU112 can execute, configuration values, GUI (Graphical User Interface) data, and parameters for implementing a trained neural network (learning model).

[0040] RAM114 is used to load programs executed by CPU112 and to store values ​​necessary during program execution.

[0041] The recording medium 115 is, for example, a semiconductor memory card or an SSD (Solid State Drive), and is used as a recording location for the prompts generated by the prompt generation unit 102 and the thumbnail images generated by the thumbnail generation unit 103. It is also used as a recording location for the input image data that formed the basis of the prompts.

[0042] The input device 117 includes multiple operating elements such as buttons, touch panels, and switches that accept operational input to the recording device 100. The input device 117 may also include one or more devices (such as sensors) for acquiring additional information when prompts are generated.

[0043] This input device 117 includes, for example, • GPS receiver for acquiring location information of recording device 100 • Clock for obtaining the creation date and time These may include, but are not limited to, the above.

[0044] The communication interface 118 is a circuit for communicating with an external device in accordance with one or more communication standards. It includes a connector for wired communication, an antenna for wireless communication, and a transmitting / receiving circuit. The recording device 100 can transmit image data to an external device and receive data from an external device through the communication interface 118. Typical communication standards that the communication interface 118 conforms to include, but are not limited to, HDMI (registered trademark), USB, Bluetooth (registered publication), and wireless LAN (Wi-Fi).

[0045] The functional blocks shown in Figures 1(a) and 2 are implemented by one or more of the hardware components shown in Figure 1(b). For example, the prompt generation unit 102 is mainly implemented by the CPU 112 and GPU 116. The RAM 114 is used as a temporary storage location for data to be processed, data being processed, and processing result data, while the ROM 113 is used as a reference point for pre-stored settings and programs for various processes.

[0046] <Operation during recording process> Next, the recording operation of the recording device 100 will be explained with reference to Figures 1 to 4.

[0047] Figure 3 is a flowchart showing the operation of the recording process in the recording device 100. Each step in Figure 3 is performed by the CPU 112 or GPU 116 executing a program stored in the ROM 113 and controlling other hardware as needed. "S" indicates the step number.

[0048] Furthermore, the recording operation by the recording device 100 may be performed in response to instructions via the input device 117, or it may be performed according to predetermined conditions other than instructions. For example, the recording operation may be performed sequentially on multiple image data received from an external device via the communication interface 118.

[0049] In Figure 3, at S301, the image input unit 101 acquires input image data. The image input unit 101 outputs the acquired input image data and the associated information described above to the prompt generation unit 102. The image input unit 101 may also process the input image data so that it is suitable for use by the prompt generation unit 102 before outputting it.

[0050] In S302, the prompt generation unit 102 generates text information (prompt) that describes the scene of the image represented by the input image data from the input image data output by the image input unit 101.

[0051] Here, we will further explain the operation of the prompt generation unit 102.

[0052] The prompt generation unit 102 generates text information (prompts) from the input image data that describes the scene of the image represented by the input image data. The prompt generation unit 102 stores the generated prompts in the RAM 114.

[0053] As described above, the prompt generation unit 102 can obtain a prompt by inputting image data into the learning model stored in the ROM 113. Alternatively, the prompt generation unit 102 may transmit image data to an external device via the communication interface 118 and receive a prompt from the external device.

[0054] The level of detail of the prompts generated by the prompt generation unit 102 can be changed by setting. For example, if the level of detail is set lower than a predetermined value, a prompt describing only the gender of the person subject can be generated, while if the level of detail is set higher than a predetermined value, a prompt describing gender, age, hair color and length, etc. can be generated.

[0055] Furthermore, the prompt generation unit 102 may generate prompts that include not only descriptions of elements included in the image (positive prompts) but also descriptions of elements not included in the image (negative prompts).

[0056] Furthermore, the prompt generation unit 102 may generate a prompt that includes supplementary information of the input image data or information based on the supplementary information.

[0057] Figure 4(a) shows an example of an image represented by the input image data. Figure 4(b) shows an example of a prompt generated from the input image data. The prompt in Figure 4(b) includes not only the elements contained in the image in Figure 4(a), but also information based on the supplementary information of the input image data, such as the date and location of the image.

[0058] Returning to Figure 3, in S303, the thumbnail generation unit 103 generates a thumbnail image from the prompt output by the prompt generation unit 102, which is an image that reproduces the content described by the prompt and has a lower resolution than the input image used to generate the prompt, and stores it in the RAM 114. A detailed explanation of the operation of the thumbnail generation unit 103 will be given later.

[0059] Figure 4(c) shows an example of a thumbnail image generated from the prompt.

[0060] The thumbnail images generated by the thumbnail generation unit 103 have a lower resolution than the input image data. This is because thumbnail images are used as a simplified playback tool to display prompt data, which is text, in a visually appealing way, and also to reduce the amount of data recorded. For example, when displaying multiple prompt data in a list, thumbnail images can be displayed side by side instead of the prompt content.

[0061] In step S304, the recording unit 104 associates the prompt generated by the prompt generation unit 102 in step S302 with the thumbnail image generated by the thumbnail generation unit 103 in step S303 and records it on the recording medium 115.

[0062] The association between the prompt and the thumbnail image is recorded as a single container file, with both the prompt and the thumbnail image included in the same file container. However, other methods of association are also acceptable. For example, the text file containing the prompt and the image file containing the thumbnail image (recorded in the main image area) could be named the same. Alternatively, the prompt could be recorded in the metadata area of ​​the image file containing the thumbnail image.

[0063] Furthermore, the recording unit 104 may record a container file containing the prompt and thumbnail image, associating it with the input image data used to generate the prompt. For example, the container file may be recorded as a separate file with a filename that shares commonality with the input image data.

[0064] The recording unit 104 may also perform digital certification processing on the prompt and thumbnail image before recording. The digital certification processing can be any process that serves to guarantee the content of the prompt and thumbnail image. For example, it may be a process that assigns an NFT (Non-Fungible Token).

[0065] As described above, according to this embodiment, a thumbnail image generated from a prompt is recorded in association with the prompt. Furthermore, the thumbnail image is displayed simultaneously when the prompt is shown. This provides a recording device that improves the visibility of the prompt data.

[0066] Here, we will explain the operation of a typical image generation AI model by referring to Figure 5. Figure 5(a) shows the relationship between the resolution and quality of images generated by a certain model.

[0067] Image generation AI models have the characteristic that the quality of the generated images improves as the resolution increases, reaching a peak (maximum) at point P in the figure. Point P is based on the resolution of the dataset used during model training, and stems from the fact that the model is optimized for generating images at a specific resolution. Therefore, if images are generated at a resolution far outside the optimized resolution, the quality of the generated images cannot be guaranteed. This is especially true for generating low-resolution images such as thumbnail images, and since images are scaled down to various resolutions depending on their display purpose, it is not easy to design the model individually for each resolution. Considering this background, it is recommended to generate images at a resolution within the range Q where good quality is achieved (a predetermined range from the point of maximum resolution).

[0068] On the other hand, Figure 5(b) shows the relationship between the resolution and generation time of images generated by a certain model. Image generation time increases as the resolution increases, and although the generation time is short within the range R shown in the figure, efficiency deteriorates thereafter. Considering these factors, it is important to balance quality and generation time and generate images at an appropriate resolution.

[0069] Based on the above, the operation of the thumbnail generation unit 103 will be explained in detail with reference to Figures 6-7.

[0070] <How the thumbnail generation process works> Figure 6 is a flowchart showing the operation of the thumbnail generation process by the thumbnail generation unit 103.

[0071] The thumbnail generation process, like the recording process, is achieved by the CPU 112 or GPU 116 executing a program stored in the ROM 113 and controlling other hardware as needed. The thumbnail generation process starts when the process S303 in the recording process shown in Figure 3 is executed. As mentioned above, the thumbnail generation unit 103 includes at least a prompt / image conversion unit 201 and an image reduction unit 202.

[0072] Figure 7(a) shows an example of an image represented by the input image data. Figure 7(b) shows an example of a prompt generated from the input image data. The prompt in Figure 7(b) includes not only the elements contained in the image in Figure 7(a), but also information based on the supplementary information of the input image data, such as the date and location of the image.

[0073] In S601, the prompt / image conversion unit 201 generates an image from the prompt that reproduces the content described by the prompt.

[0074] As described above, the prompt / image conversion unit 201 can generate an image by inputting a prompt to the learning model stored in the ROM 113. Alternatively, the prompt / image conversion unit 201 may send a prompt to an external device via the communication interface 118 and receive a generated image from the external device.

[0075] Furthermore, the resolution of the image generated by the prompt / image conversion unit 201 is the resolution at which the image generation AI model produces images of good quality. Specifically, for example, it is the resolution within the range indicated by Q in Figure 5(a). On the other hand, if certain conditions are met, image generation may be performed at a resolution different from the resolution at which the image generation AI model produces images of good quality. The operation in that case will be described later.

[0076] Figure 7(c) shows an example of an image generated from a prompt. In this embodiment, the prompt / image conversion unit 201 generates an image by referring to the input image data used to generate the prompt. It also selects a resolution range (the range of resolutions indicated by Q in Figure 5(a)) in which the image generation AI model produces images of good quality, and generates an image with a resolution within that range. As a result, as shown in Figure 7(c), it is possible to generate an image with elements that constitute the image, such as composition and subject matter, that are closer to the input image data and of high quality. The image generated here has a resolution lower than or equal to the resolution of the image from which the prompt was generated, but generally, it is an image with a resolution lower than the resolution of the original image and a resolution higher than the resolution of the thumbnail image.

[0077] In S602, the image reduction unit 202 reduces the image generated by the prompt / image conversion unit 201 to output a thumbnail image of an arbitrary resolution and stores it in the RAM 114. Figure 7(d) shows an example of a thumbnail image output by the image reduction unit 202.

[0078] Thus, in the thumbnail generation process of this embodiment, a high-quality thumbnail image can be obtained by reducing the high-quality image generated by the prompt / image conversion unit 201 to an arbitrary magnification by the image reduction unit 202.

[0079] <Operation of thumbnail generation process under specific conditions> As mentioned above, it is important to determine the resolution of the image generated by the prompt / image conversion unit 201 while considering the balance between quality and generation time. In this regard, the prompt / image conversion unit 201 can also generate images at a resolution different from the resolution at which the generation model produces images of good quality, when certain conditions are met.

[0080] These specific conditions arise when generation speed is more important than image quality. Specifically, this includes cases where the number of images generated per unit time by the image generation AI model exceeds a predetermined number, or when the resolution of the thumbnail image falls below a predetermined value. The resolution can be set based on the resolution at which the image generation AI model produces images of good quality, according to the number of images generated per unit time by the image generation AI model. Alternatively, it can be set according to the resolution of the thumbnail image.

[0081] The following example illustrates the case where the resolution of the thumbnail image falls below a specified value, i.e., when the resolution is low.

[0082] In step S601 of Figure 6, the resolution of the image generated by the prompt / image conversion unit 201 is set to the resolution of the thumbnail image. An example of the generated image is shown in Figure 7(d).

[0083] In S602, the image reduction unit 202 processes the image generated by the prompt / image conversion unit 201 at 1:1 scale and stores it in the RAM 114. An example of a thumbnail image output by the image reduction unit 202 is shown in Figure 7(d).

[0084] Thus, in the thumbnail generation process under specific conditions in this embodiment, when generation speed is important, it is possible to efficiently obtain thumbnail images of any resolution by adjusting the resolution of the image generated by the prompt / image conversion unit 201.

[0085] As described above, according to the above embodiment, by inputting a prompt containing instructions for generating an image to the image generation AI model, and then generating a high-quality image and reducing its size, a thumbnail image of any resolution can be obtained. This makes it possible to prevent quality degradation or distortion when generating thumbnail images from a prompt. On the other hand, if generation speed is important, the resolution of the generated image can be adjusted to efficiently obtain a thumbnail image of any resolution. Thus, this embodiment allows for flexible adjustment of the balance between thumbnail image quality and generation speed, making it possible to accommodate various applications.

[0086] (Other embodiments) Furthermore, the present invention can also be realized by supplying a program that implements one or more of the functions of the above-described embodiments to a system or device via a network or storage medium, and by a process in which one or more processors in the computer of that system or device read and execute the program. It can also be realized by a circuit (e.g., an ASIC) that implements one or more functions.

[0087] The disclosures herein include the following recording devices and their control methods, programs, and storage media.

[0088] (Item 1) A means for acquiring a first image, A first generation means that generates a prompt, which is text information describing the content of the first image, based on the acquired first image, A second generation means generates a second image from a prompt generated by the first generation means that reproduces the content described by the prompt, A third generation means for reducing the size of the second image to generate a thumbnail image, A recording means for associating the prompt and the thumbnail image and recording them on a recording medium, A recording device characterized by comprising the following features.

[0089] (Item 2) The recording device according to item 1, characterized in that the second image has a resolution less than or equal to the resolution of the first image.

[0090] (Item 3) The recording device according to item 2, characterized in that the second image has a lower resolution than the first image and a higher resolution than the thumbnail image.

[0091] (Item 4) The recording device according to any one of items 1 to 3, characterized in that the second generating means selects a range of resolution for the second image based on the characteristics of the image quality that the second generating means can generate.

[0092] (Item 5) The recording device according to item 4, characterized in that the second generation means generates the second image such that the resolution is within a predetermined range from the first resolution which maximizes the quality of the image generated by the second generation means.

[0093] (Item 6) The recording device according to item 5, characterized in that the second generation means generates the second image with a second resolution different from the resolution of the predetermined range, based on the predetermined range and in response to the fulfillment of specific conditions.

[0094] (Item 7) The recording device according to item 6, characterized in that the specific condition is when the number of times an image is generated per unit time in the second generation means exceeds a predetermined number.

[0095] (Item 8) The recording device according to item 6, characterized in that the specific condition is when the resolution of the thumbnail image is less than a predetermined value.

[0096] (Item 9) The recording device according to item 6, characterized in that the second resolution is determined according to the number of times an image is generated per unit time in the second generation means.

[0097] (Item 10) The recording device according to item 6, characterized in that the second resolution is determined according to the resolution of the thumbnail image.

[0098] (Item 11) The acquisition process involves obtaining the first image, A first generation step of generating a prompt, which is text information describing the content of the first image, based on the acquired first image, A second generation step generates a second image from the prompt generated in the first generation step, which reproduces the content described by the prompt, A third generation step involves reducing the size of the second image to generate a thumbnail image, A recording step of associating the prompt and the thumbnail image and recording them on a recording medium, A control method for a recording device, characterized by comprising the following:

[0099] (Item 12) A program for causing a computer to function as one of the means of a recording device described in any one of items 1 through 10.

[0100] (Item 13) A computer-readable storage medium storing a program that causes the computer to function as one of the means of a recording device described in any one of items 1 to 10.

[0101] The invention is not limited to the embodiments described above, and various modifications and variations are possible without departing from the spirit and scope of the invention. Accordingly, claims are attached to disclose the scope of the invention. [Explanation of symbols]

[0102] 100: Recording device, 101: Image input unit, 102: Prompt generation unit, 103: Thumbnail generation unit, 104: Recording unit, 201: Prompt / image conversion unit, 202: Image reduction unit

Claims

1. A means for acquiring a first image, A first generation means generates a prompt, which is text information describing the content of the first image, based on the acquired first image. A second generation means generates a second image from a prompt generated by the first generation means that reproduces the content described by the prompt, A third generation means for reducing the size of the second image to generate a thumbnail image, A recording means for associating the prompt and the thumbnail image and recording them on a recording medium, A recording device characterized by comprising the following features.

2. The recording device according to claim 1, characterized in that the second image has a resolution less than or equal to the resolution of the first image.

3. The recording device according to claim 2, characterized in that the second image has a lower resolution than the first image and a higher resolution than the thumbnail image.

4. The recording device according to claim 1, characterized in that the second generation means selects a range of resolution for the second image based on the characteristics of the quality of the image that the second generation means can generate.

5. The recording apparatus according to claim 4, characterized in that the second generation means generates the second image such that the resolution is within a predetermined range from the first resolution which maximizes the quality of the image generated by the second generation means.

6. The recording device according to claim 5, characterized in that the second generation means generates the second image with a second resolution different from the resolution of the predetermined range, based on the predetermined range and in response to the fulfillment of specific conditions.

7. The recording device according to claim 6, characterized in that the specific condition is when the number of times an image is generated per unit time in the second generation means exceeds a predetermined number.

8. The recording device according to claim 6, characterized in that the specific condition is that the resolution of the thumbnail image is less than a predetermined value.

9. The recording device according to claim 6, characterized in that the second resolution is determined according to the number of times an image is generated per unit time in the second generation means.

10. The recording device according to claim 6, characterized in that the second resolution is determined according to the resolution of the thumbnail image.

11. The acquisition process involves obtaining the first image, A first generation step of generating a prompt, which is text information describing the content of the first image, based on the acquired first image, A second generation step generates a second image from the prompt generated in the first generation step, which reproduces the content described by the prompt, A third generation step involves reducing the size of the second image to generate a thumbnail image, A recording step of associating the prompt and the thumbnail image and recording them on a recording medium, A control method for a recording device, characterized by comprising the following features.

12. A program for causing a computer to function as one of the means of a recording device according to any one of claims 1 to 10.

13. A computer-readable storage medium storing a program for causing a computer to function as one of the means of a recording device according to any one of claims 1 to 10.

Images