Digital content generation method and digital content generation system

Abstract

The invention provides a digital content generation method and a digital content generation system. The method comprises the following steps: establishing a style description database; obtaining identity recognition data corresponding to the target user; obtaining style description data corresponding to a target user from a style description database according to the identity recognition data; and influencing the output of the artificial intelligence model by using the style description data, so that the artificial intelligence model generates the digital content with the specific style. Therefore, even if a plurality of manufacturers or a plurality of users commonly use the same artificial intelligence model, the artificial intelligence model can generate digital contents with own unique styles for individual manufacturers or individual users.

Description

Technical Field

[0001] This invention relates to an artificial intelligence application technology, and more particularly to a digital content generation method and a digital content generation system. Background Technology

[0002] With technological advancements, artificial intelligence (AI) models for automatically generating digital content such as images are widely used to improve people's daily lives and / or work efficiency. However, if a single vendor or user wants to use an AI model to generate digital content with its own unique style, they need to find ways to train and maintain their own AI model to ensure it can produce content with that distinctive style. This typically increases operating costs significantly, and obtaining training materials is also a challenging issue. On the other hand, using pre-trained AI models from others limits the universality of training materials, meaning the trained model may not be able to generate content that meets the specific vendor's or user's unique style. Summary of the Invention

[0003] In view of this, the present invention provides a digital content generation method and a digital content generation system, which can improve the above-mentioned problems.

[0004] An embodiment of the present invention provides a digital content generation method, comprising: establishing a style description database; obtaining identity recognition data corresponding to a target user; obtaining style description data corresponding to the target user from the style description database based on the identity recognition data; and using the style description data to influence the output of an artificial intelligence model, so that the artificial intelligence model generates digital content with a specific style.

[0005] An embodiment of the present invention also provides a digital content generation system, which includes a storage device and a processor. The storage device is used to store a style description database and an artificial intelligence model. The processor is connected to the storage device. The processor is used to: establish the style description database; obtain identification data corresponding to a target user; obtain style description data corresponding to the target user from the style description database based on the identification data; and use the style description data to influence the output of the artificial intelligence model, causing the artificial intelligence model to generate digital content with a specific style.

[0006] Based on the above, after obtaining the identity data corresponding to the target user, style description data corresponding to the target user can be obtained from the style description database based on the identity data. Subsequently, by using the style description data to influence the output of the artificial intelligence model, the artificial intelligence model can generate digital content with a specific style. Therefore, even if multiple manufacturers or users share the same artificial intelligence model, this artificial intelligence model can still generate digital content with its own unique style for individual manufacturers or individual users. Attached Figure Description

[0007] Figure 1 This is a schematic diagram of a digital content generation system according to an embodiment of the present invention;

[0008] Figure 2 This is a schematic diagram of the operation flow of a digital content generation method according to an embodiment of the present invention;

[0009] Figure 3 This is a schematic diagram of style description data stored in a style description database according to an embodiment of the present invention;

[0010] Figure 4 This is a schematic diagram of the operation flow of a digital content generation method according to an embodiment of the present invention;

[0011] Figure 5 This is a schematic diagram of the operation flow of a digital content generation method according to an embodiment of the present invention;

[0012] Figure 6 This is a schematic diagram illustrating digital content with different styles generated based on different style description data, according to an embodiment of the present invention;

[0013] Figure 7 This is a flowchart illustrating a digital content generation method according to an embodiment of the present invention. Detailed Implementation

[0014] Reference will now be made in detail to exemplary embodiments of the invention, examples of which are illustrated in the accompanying drawings. Wherever possible, the same element references are used in the drawings and description to denote the same or similar parts.

[0015] Figure 1 This is a schematic diagram of a digital content generation system according to an embodiment of the present invention. Please refer to... Figure 1 The digital content generation system 10 can be applied to or installed on various electronic devices that support image processing functions, such as smartphones, tablets, laptops, desktops, servers, game consoles, or in-vehicle computers, and the types of electronic devices are not limited thereto.

[0016] The digital content production system 10 includes a processor 11, a storage device 12, and an input / output (I / O) device 13. The processor 11 is responsible for the overall or partial operation of the digital content production system 10. For example, the processor 11 may include a central processing unit (CPU), a graphics processing unit (GPU), or other programmable general-purpose or special-purpose microprocessors, digital signal processors (DSPs), programmable controllers, application-specific integrated circuits (ASICs), programmable logic devices (PLDs), or other similar devices or combinations thereof.

[0017] In one embodiment, the processor 11 may further include processors dedicated to assisting in performing neural network operations and / or image processing, such as a vision processing unit (VPU), a neural network processing unit (NPU), and / or a tensor processing unit (TPU). Furthermore, the present invention does not limit the number or type of processor 11.

[0018] Storage device 12 is connected to processor 11 and used to store data. For example, storage device 12 may include volatile storage circuitry and non-volatile storage circuitry. Volatile storage circuitry is used to volatilely store data. For example, volatile storage circuitry may include random access memory (RAM) or similar volatile storage media. Non-volatile storage circuitry is used to non-volatilely store data. For example, non-volatile storage circuitry may include read-only memory (ROM), solid-state disk (SSD), hard disk drive (HDD), or similar non-volatile storage media. Furthermore, the present invention does not limit the number or type of storage device 12.

[0019] Input / output device 13 is connected to processor 11 and used to receive input signals or send output signals. For example, input / output device 13 may include power management circuitry, network interface card, mouse, keyboard, speaker, and microphone. Furthermore, the present invention does not limit the number or type of input / output device 13.

[0020] In one embodiment, the processor 11 can be used to establish a style description database 101. For example, the style description database 101 can be stored in the storage device 12. The style description database 101 is used to store multiple style description data corresponding to different users. For example, one user can correspond to one or more style description data. Different users can correspond to different style description data. In one embodiment, the style description data in the style description database 101 can also be classified by different projects. For example, one project can correspond to one or more style description data. Different projects can correspond to different style description data.

[0021] In one embodiment, the processor 11 is further configured to build an artificial intelligence model 102. For example, the artificial intelligence model 102 may be stored in a storage device 12. The artificial intelligence model 102 is used to generate digital content. For example, the artificial intelligence model 102 may be implemented using neural network architectures such as Deep Neural Network (DNN), Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), or other types of algorithmic architectures. In one embodiment, the artificial intelligence model 102 may also include a machine learning (ML) model and / or a deep learning (DL) model. The digital content may include images and / or sound. In one embodiment, the artificial intelligence model 102 may be trained to improve the quality of the digital content generated by the artificial intelligence model 102.

[0022] In one embodiment, the processor 11 can use style description data in the style description database 101 to influence the output of the artificial intelligence model 102, causing the artificial intelligence model 102 to generate digital content with a specific style. For example, suppose the theme of the digital content to be generated by the artificial intelligence model 102 is "runner". By using a certain style description data in the style description database 101 to influence the output of the artificial intelligence model 102, the artificial intelligence model 102 can generate digital content with various styles such as "Japanese", "American", "Graceful", "Hot-blooded", "Anime" or "Realistic" with the theme of "runner". However, the subject and style of the digital content that can be generated by the artificial intelligence model 102 can be adjusted according to practical needs, and the present invention does not impose any limitations.

[0023] Figure 2 This is a schematic diagram illustrating the operation flow of a digital content generation method according to an embodiment of the present invention. Please refer to... Figure 2In one embodiment, the processor 11 may obtain identification data 201 corresponding to a user (also referred to as a target user) 21. The identification data 201 can be used to identify the user 21. For example, the identification data 201 may include an identification code or other types of identification data, as long as it can be used to identify the user 21. Furthermore, each user can be identified through unique identification data.

[0024] In one embodiment, the processor 11 may obtain style description data 202 corresponding to the user 21 from the style description database 101 based on the identity recognition data 201. The style description data 202 can be used to control the style of the digital content 203 to be generated by the artificial intelligence model 102. It should be noted that the style description data 202 does not affect the theme of the digital content 203 generated by the artificial intelligence model 102. The processor 11 can then use the style description data 202 to influence the output of the artificial intelligence model 102, causing the artificial intelligence model 102 to generate digital content 203 with a specific style.

[0025] Figure 3 This is a schematic diagram illustrating style description data stored in a style description database according to an embodiment of the present invention. Please refer to... Figure 3 Assume that description table 31 is stored in style description database 101. Description table 31 records the mapping relationship (or correspondence) between users (1) to users (n) and style description data (1) to style description data (n). In one embodiment, user (i) can manage (e.g., upload or update) the style description data (i) recorded in description table 31. i can be any integer from 1 to n.

[0026] In one embodiment, based on the identity data corresponding to user (i), processor 11 can query style description database 101 (or description table 31) to obtain style description data (i) corresponding to user (i). For example, Figure 2 The style description data 202 may include style description data (i).

[0027] In one embodiment, it is assumed that processor 11 obtains identity data (also referred to as first identity data) corresponding to user (j) (also referred to as first user). Processor 11 can obtain style description data (j) (also referred to as first style description data) corresponding to user (j) from style description database 101 (or description table 31) based on the first identity data. Style description data (j) can be used to influence the output of artificial intelligence model 101, causing artificial intelligence model 101 to generate digital content (also referred to as first digital content) with a certain style (also referred to as first style).

[0028] In one embodiment, it is assumed that processor 11 obtains identity data (also called second identity data) corresponding to user (k) (also called second user). j and k are both arbitrary integers from 1 to n, and j is different from k. Processor 11 can obtain style description data (k) (also called second style description data) corresponding to user (k) from style description database 101 (or description table 31) based on the second identity data. Style description data (k) can be used to influence the output of artificial intelligence model 101, causing artificial intelligence model 101 to generate digital content (also called second digital content) with another style (also called second style).

[0029] It should be noted that the subject matter of the first digital content may be the same as or different from that of the second digital content. However, the style of the first digital content (i.e., the first style) may differ from the style of the second digital content (i.e., the second style).

[0030] For example, in one embodiment, suppose the subject of the first digital content and the theme of the second digital content are both "runners," the style of the first digital content (i.e., the first style) is "Japanese," and the style of the second digital content (i.e., the second style) is "American." In this example, the first digital content may present a "runner" with a "Japanese" style, while the second digital content may present a "runner" with an "American" style.

[0031] Alternatively, in one embodiment, suppose the subject of the first digital content is a "runner," the subject of the second digital content is a "child," the style of the first digital content (i.e., the first style) is "Japanese," and the style of the second digital content (i.e., the second style) is "American." In this example, the first digital content could present a "runner" with a "Japanese" style, while the second digital content could present a "child" with an "American" style. It should be noted that the subject and style of the digital content generated by the artificial intelligence model 102 can be adjusted according to practical needs, and the present invention does not impose any limitations on this.

[0032] Figure 4 This is a schematic diagram illustrating the operation flow of a digital content generation method according to an embodiment of the present invention. Please refer to... Figure 4 In one embodiment, the processor 11 can obtain an operation instruction (also referred to as a first operation instruction) 410 corresponding to the user 41 (i.e., the target user) via the input / output device 13. The operation instruction 410 can be used to control the theme of the digital content 403 to be generated by the artificial intelligence model 102. For example, the operation instruction 410 may reflect that the user 41 wants the theme of the digital content 403 generated by the artificial intelligence model 102 to be "runner" or other themes, which is not limited by the present invention.

[0033] On the other hand, the processor 11 can obtain identification data 401 corresponding to user 41. Identification data 401 can be used to identify user 41. Based on the identification data 401, the processor 11 can obtain style description data 402 corresponding to user 41 from the style description database 101. Style description data 402 can be used to control the style of digital content 403 generated by artificial intelligence model 102.

[0034] In one embodiment, the processor 11 can generate an operation instruction (also referred to as a second operation instruction) 420 based on the operation instruction 410 and the style description data 402. For example, the processor 11 can add the style description data 402 to at least a portion of the instruction content originally carried by the operation instruction 410 to generate the operation instruction 420. Then, the processor 11 can instruct the artificial intelligence model 102 to generate digital content 403 with a specific style based on the operation instruction 420. For example, the processor 11 can input the operation instruction 420 into the artificial intelligence model 102. The artificial intelligence model 102 can generate digital content 403 with a "Japanese" style and a "runner" theme based on the operation instruction 420. It should be noted that the theme and style of the digital content 403 can be set and adjusted according to practical needs, and the present invention does not impose any limitations.

[0035] In one embodiment, the processor 11 may also establish a large language model (LMM) 103. For example, the large language model 103 may be stored in the storage device 12. In one embodiment, the processor 11 may process the second operation instruction through the large language model 103 to normalize the instruction content of the second operation instruction. For example, the processor 11 may normalize the instruction content (e.g., colloquial instruction content) of the second operation instruction through the large language model 103 to process or convert the instruction content of the second operation instruction into a format or form that can be processed by the artificial intelligence model 102. Then, the processor 11 may input the processed (i.e., normalized) second operation instruction into the artificial intelligence model 102 to instruct the artificial intelligence model 102 to generate digital content with a specific style.

[0036] In one embodiment, the processor 11 may also establish an embedded model 104. For example, the embedded model 104 may be stored in the storage device 12. In one embodiment, the processor 11 can process the acquired identity recognition data through the embedded model 104 to normalize the data content of the identity recognition data. For example, the processor 11 can normalize the data content of the identity recognition data through the embedded model 104 to process or convert the data content of the identity recognition data into a format or form that can be processed by the style description database 101. Then, the processor 11 can query the style description database 101 based on the processed (i.e., normalized) identity recognition data to obtain the style description data.

[0037] Figure 5 This is a schematic diagram illustrating the operation flow of a digital content generation method according to an embodiment of the present invention. Please refer to... Figure 5 In one embodiment, the processor 11 can obtain an operation instruction 510 (i.e., a first operation instruction) corresponding to the user 51 (i.e., the target user) via the input / output device 13. The operation instruction 510 is used to control the theme of the digital content 503 generated by the artificial intelligence model 102. For example, the operation instruction 510 may reflect that the user 51 desires the theme of the digital content 503 generated by the artificial intelligence model 102 to be "runner" or other themes; this invention is not limited thereto. On the other hand, the processor 11 can obtain identification data 501 corresponding to the user 51. The identification data 501 can be used to identify the user 51.

[0038] In one embodiment, after obtaining the identity recognition data 501, the processor 11 can normalize the data content of the identity recognition data 501 through the embedded model 104, so as to process or convert the data content of the identity recognition data 501 into a format or form that can be processed by the style description database 101. Then, the processor 11 can query the style description database 101 based on the processed (i.e., normalized) identity recognition data 501 to obtain style description data 502. The style description data 502 can be used to control the style of the digital content 503 to be generated by the artificial intelligence model 102.

[0039] In one embodiment, the processor 11 can generate an operation instruction 520 (i.e., a second operation instruction) based on the operation instruction 510 and the style description data 502. For example, the processor 11 can add the style description data 502 to at least a portion of the instruction content originally carried by the operation instruction 510 to generate the operation instruction 520.

[0040] In one embodiment, after obtaining the operation instruction 520, the processor 11 can normalize the instruction content of the operation instruction 520 through a large language model 103, so as to process or convert the instruction content of the operation instruction 520 into a format or form that can be processed by the artificial intelligence model 102. Then, the processor 11 can input the processed (i.e., normalized) operation instruction 520 into the artificial intelligence model 102 to instruct the artificial intelligence model 102 to generate digital content 503 with a specific style. It should be noted that the theme and style of the digital content 503 can be set and adjusted according to practical needs, and the present invention does not impose any limitations.

[0041] Figure 6 This is a schematic diagram illustrating digital content with different styles generated based on different style description data, according to an embodiment of the present invention. Please refer to... Figure 6 In one embodiment, it is assumed that the style description data obtained for user 61 is style description data 611, and the style description data obtained for user 62 is style description data 621.

[0042] In one embodiment, the digital content 612 generated based on the style description data 611 can present a theme A with style B. For example, assuming style B is "Japanese" and theme A is "runner," the digital content 612 can present a runner wearing a kimono, with cherry blossom petals falling around the runner. Furthermore, the digital content 612 can also present any digital content with a "Japanese" concept to modify the "runner" in the digital content 612.

[0043] On the other hand, the digital content 622 generated based on the style description data 621 can present a theme A with style C. For example, assuming style C is "American" and theme A is "runner," then digital content 622 can present a runner wearing a loose T-shirt, and this runner may have large tattoos on his arms and calves. In addition, digital content 622 can also present any digital content with the concept of "American" to modify the "runner" in digital content 622.

[0044] It should be noted that, in Figure 6 In one embodiment, digital content 612 and 622 present the same theme (i.e., theme A). However, based on different style description data 611 and 621, the same theme (i.e., theme A) presented by digital content 612 and 622 may have drastically different styles (i.e., style B and style C).

[0045] In one embodiment, a user (i.e., the target user) can operate an electronic device (e.g., a smartphone or personal computer) to send a first operation command and / or identification data to the digital content generation system 10. In another embodiment, the user (i.e., the target user) can upload their own identification data to a blockchain network or online storage space beforehand.

[0046] In one embodiment, after receiving a first operation instruction sent by a user (i.e., the target user), the processor 11 can retrieve the identity data corresponding to the target user from the blockchain network or online storage space according to the first operation instruction. For example, the processor 11 can search for the location where information belonging to the target user is stored in the blockchain network or online storage space based on the source of the first operation instruction and / or at least part of the information carried by the first operation instruction. Then, the processor 11 can download the identity data corresponding to the target user from this information storage location. The specific usage of the blockchain network or online storage space is prior art and will not be elaborated here.

[0047] Figure 7 This is a flowchart illustrating a digital content generation method according to an embodiment of the present invention. Please refer to... Figure 7 In step S701, a style description database is established. In step S702, identity recognition data corresponding to the target user is obtained. In step S703, style description data corresponding to the target user is obtained from the style description database based on the identity recognition data. In step S704, the style description data is used to influence the output of the artificial intelligence model, causing the artificial intelligence model to generate digital content with a specific style.

[0048] However, Figure 7 Each step has been explained in detail above and will not be repeated here. It is worth noting that... Figure 7 Each step can be implemented as multiple program codes or circuits, and this invention is not limited thereto. Furthermore, Figure 7 The method can be used in conjunction with the above examples and embodiments, or it can be used alone. This invention does not impose any limitations.

[0049] In summary, the digital content generation method and system proposed in this invention allow different users to manage their own exclusive style description data in a style description database. Subsequently, based on the target user's identification data, the style description data corresponding to that target user can be retrieved from the style description database. In particular, this style description data can be used to influence the output of an artificial intelligence model, enabling the model to generate digital content with a specific style. Therefore, users do not need to actually participate in the maintenance (e.g., training) of the artificial intelligence model, yet can still share the same artificial intelligence model to generate digital content with their own unique style.

[0050] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for generating digital content, characterized in that, include: Establish a style description database; Obtain the identification data corresponding to the target user; Based on the identity recognition data, obtain style description data corresponding to the target user from the style description database; as well as The style description data is used to influence the output of the artificial intelligence model, enabling the artificial intelligence model to produce digital content with a specific style.

2. The digital content generation method according to claim 1, wherein the style description database is used to store multiple style description data corresponding to different users.

3. The digital content generation method according to claim 1, wherein the step of obtaining the style description data corresponding to the target user from the style description database based on the identity recognition data includes: Based on the first identity recognition data, first style description data corresponding to the first user is obtained from the style description database, wherein the first style description data is used to influence the output of the artificial intelligence model, so that the artificial intelligence model generates first digital content with the first style; as well as Based on the second identity recognition data, second style description data corresponding to the second user is obtained from the style description database, wherein the second style description data is used to influence the output of the artificial intelligence model, so that the artificial intelligence model generates second digital content with a second style, and the first style is different from the second style.

4. The digital content generation method according to claim 1, wherein the step of using the style description data to influence the output of the artificial intelligence model, so that the digital content generated by the artificial intelligence model has the specific style, includes: Obtain a first operation instruction corresponding to the target user, wherein the first operation instruction is used to control the theme of the digital content; Based on the first operation instruction and the style description data, a second operation instruction is generated, wherein the style description data is used to control the style of the digital content; as well as Based on the second operation instruction, the artificial intelligence model is instructed to generate the digital content with the specific style.

5. The digital content generation method according to claim 4, wherein the step of instructing the artificial intelligence model to generate digital content having the specific style based on the second operation instruction includes: The second operation instruction is processed by a large language model to normalize the instruction content of the second operation instruction; as well as The processed second operation instruction is input into the artificial intelligence model to instruct the artificial intelligence model to generate the digital content with the specific style.

6. The digital content generation method according to claim 1, wherein the step of obtaining the style description data corresponding to the target user from the style description database based on the identity recognition data includes: The identity recognition data is processed through an embedded model to normalize the data content of the identity recognition data; as well as The style description data is obtained by querying the style description database based on the processed identity recognition data.

7. The digital content generation method according to claim 1, wherein the step of obtaining the identity data corresponding to the target user includes: The identity data corresponding to the target user is obtained from the blockchain network or online storage space.

8. A digital content generation system, characterized in that, include: Storage devices used to store style description databases and artificial intelligence models; as well as The processor is connected to the storage device. The processor is used to: Establish the style description database; Obtain the identification data corresponding to the target user; Based on the identity recognition data, obtain style description data corresponding to the target user from the style description database; as well as The style description data is used to influence the output of the artificial intelligence model, causing the artificial intelligence model to produce digital content with a specific style.

9. The digital content generation system according to claim 8, wherein the style description database is used to store multiple style description data corresponding to different users.

10. The digital content generation system according to claim 8, wherein the operation of the processor obtaining the style description data corresponding to the target user from the style description database based on the identity recognition data includes: Based on the first identity recognition data, first style description data corresponding to the first user is obtained from the style description database, wherein the first style description data is used to influence the output of the artificial intelligence model, so that the artificial intelligence model generates first digital content with the first style; as well as Based on the second identity recognition data, second style description data corresponding to the second user is obtained from the style description database, wherein the second style description data is used to influence the output of the artificial intelligence model, so that the artificial intelligence model generates second digital content with a second style, and the first style is different from the second style.

11. The digital content generation system of claim 8, wherein the operation of the processor using the style description data to influence the output of the artificial intelligence model, causing the digital content generated by the artificial intelligence model to have the specific style, includes: Obtain a first operation instruction corresponding to the target user, wherein the first operation instruction is used to control the theme of the digital content; Based on the first operation instruction and the style description data, a second operation instruction is generated, wherein the style description data is used to control the style of the digital content; as well as Based on the second operation instruction, the artificial intelligence model is instructed to generate the digital content with the specific style.

12. The digital content generation system of claim 11, wherein the storage device is further configured to store a large language model, and the processor instructs the artificial intelligence model to generate the digital content having the specific style based on the second operation instruction, comprising: The second operation instruction is processed through the large language model to normalize the instruction content of the second operation instruction; as well as The processed second operation instruction is input into the artificial intelligence model to instruct the artificial intelligence model to generate the digital content with the specific style.

13. The digital content generation system of claim 8, wherein the storage device is further configured to store an embedded model, and the operation of the processor retrieving the style description data corresponding to the target user from the style description database based on the identity recognition data includes: The embedded model processes the identity recognition data to normalize the data content of the identity recognition data; as well as The style description data is obtained by querying the style description database based on the processed identity recognition data.

14. The digital content generation system of claim 8, wherein the operation of the processor acquiring the identity data corresponding to the target user includes: The identity data corresponding to the target user is obtained from the blockchain network or online storage space.

Images