Image generation method, image reproduction method, and electronic device applying the same

Abstract

This invention provides an image generation method, an image regeneration method, and an electronic device using the same. The image generation method includes the following steps: obtaining a generation prompt, a stable diffusion seed, and an out painting seed; obtaining a generated image based on the generation prompt, the diffusion seed, and the out painting seed using an image generation model; encoding the generation prompt, the diffusion seed, and the out painting seed into a hidden watermark; and adding the hidden watermark to the generated image.

Description

Technical Field

[0001] This invention relates to an image processing method and an electronic device, and more particularly to an image generation method, an image regeneration method, and an electronic device using the same. Background Technology

[0002] AI-powered image processing technology is developing rapidly and is frequently used in creative work. Currently, when users see an AI image they like, they can only find similar images by taking a picture or screenshot and then using image search, but they cannot obtain the exact same AI image. Therefore, enabling users to obtain the same AI image has become one of the current research directions in the industry. Summary of the Invention

[0003] This invention relates to an image generation method, an image regeneration method, and an electronic device using the same. By adding a hidden watermark to the image, the generation parameters and information of the AI ​​image are embedded in the image without affecting the image's aesthetics and quality, allowing users to quickly reproduce the same AI image.

[0004] According to one aspect of the present invention, an image generation method is proposed. The image generation method includes the following steps: obtaining a generation prompt, a stable diffusion seed, and an outpainting seed; obtaining a generated image based on the generation prompt, the stable diffusion seed, and the outpainting seed using an image generation model; encoding the generation prompt, the stable diffusion seed, and the outpainting seed into a hidden watermark; and adding the hidden watermark to the generated image.

[0005] According to another aspect of the present invention, an image regeneration method is proposed. The image regeneration method includes the following steps: A generated image is captured or screenshotted using an image capturing unit to obtain a captured image. The generated image has a hidden watermark. A decoding procedure for the hidden watermark on the captured image is performed to obtain a generation prompt, a stable diffusion seed, and an out painting seed for an image generation model. Based on the generation prompt, the stable diffusion seed, and the out painting seed, a regenerated image is obtained using the image generation model.

[0006] According to another aspect of the present invention, an electronic device is provided. The electronic device includes an image capturing unit, a decoding unit, and a processing unit. The image capturing unit is used to capture or screenshot a generated image to obtain a captured image. The generated image has a hidden watermark. The decoding unit is used to decode the hidden watermark on the captured image to obtain a generation prompt, a stable diffusion seed, and an out painting seed for an image generation model. The processing unit is used to obtain a regenerated image based on the generation prompt, the stable diffusion seed, and the out painting seed using the image generation model. Attached Figure Description

[0007] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein:

[0008] Figure 1 An image regeneration method according to an embodiment of the present invention is shown.

[0009] Figure 2 A block diagram of an electronic device for image generation according to an embodiment of the present invention is shown.

[0010] Figure 3 A block diagram of an electronic device for image reproduction according to an embodiment of the present invention is shown.

[0011] Figure 4 An example illustrates an image generation method and an image regeneration method according to an embodiment of the present invention.

[0012] Figure 5 Examples illustrate an image generation method and an image regeneration method according to another embodiment of the present invention.

[0013] Figure 6 Examples illustrate an image generation method and an image regeneration method according to another embodiment of the present invention.

[0014] Figure 7 Examples illustrate an image generation method and an image regeneration method according to another embodiment of the present invention.

[0015] Figure label:

[0016] 100, 200: Electronic devices

[0017] 110: Arithmetic Unit

[0018] 120: Coding Unit

[0019] 130: Synthesis Unit

[0020] 140: Output Unit

[0021] 210: Image Capture Unit

[0022] 220: Decoding Unit

[0023] 230: Arithmetic Unit

[0024] 240: Display Unit

[0025] SDSE1, SDSE3: Diffusion Seeds

[0026] SDST1: Diffusion Steps

[0027] SDMV1: Diffusion Model Version

[0028] IM11: Generate Image

[0029] IM12, IM22, IM42: Image capture

[0030] IM13, IM33: Regenerated Images

[0031] MD1, MD2: Image Generation Models

[0032] OBR4: Object Scope

[0033] OPSE1, OPSE3: Extended Seeds

[0034] OPMV1: Extended Model Version

[0035] PT1, PT3: Generate prompts

[0036] WM1, WM4: Hidden watermark

[0037] S110, S120, S130, S140, S210, S220, S220', S230: Steps Detailed Implementation

[0038] The technical terms used in this specification are based on common terminology in the field. Where this specification provides further explanation or definition of certain terms, the interpretation of those terms shall be based on the explanation or definition provided in this specification. Each embodiment of the present invention has one or more technical features. Where feasible, those skilled in the art may selectively implement some or all of the technical features in any embodiment, or selectively combine some or all of the technical features in these embodiments.

[0039] Please refer to Figure 1This illustrates an image regeneration method according to an embodiment of the present invention. In this embodiment, an AI model adds a hidden watermark WM1 containing parameters of the AI ​​model during the generation of a generated image IM11. The hidden watermark WM1 is scattered across several pixels of the generated image IM11, and its presence is imperceptible to the human eye. A user can use an image capturing unit 210 to capture an image IM12. The user may capture, for example, the outer frame, the wall, the generated image IM11, and the added hidden watermark WM1.

[0040] Next, the hidden watermark WM1 can be obtained from the captured image IM 12. After decoding the hidden watermark WM1, the parameters of the AI ​​model can be obtained, and the regenerated image IM13, which is the same as the generated image IM 11, can be obtained.

[0041] Please refer to Figure 2 This diagram illustrates a block diagram of an electronic device 100 for image generation according to an embodiment of the present invention. The electronic device 100 is, for example, a laptop computer, a desktop computer, a mobile phone, a server, or a cloud computing center. The electronic device 100 for image generation includes a processing unit 110, an encoding unit 120, a synthesis unit 130, and an output unit 140.

[0042] The arithmetic unit 110, encoding unit 120, and synthesis unit 130 are used to perform various arithmetic, encoding, and processing programs, such as a circuit, a circuit board, a storage device for storing program code, or a chip. The output unit 140 is used to output various information, such as a display, a transmission module, or a printer.

[0043] Please refer to Figure 3 The diagram illustrates a block diagram of an electronic device 200 for image regeneration according to an embodiment of the present invention. The electronic device 200 for image regeneration includes the aforementioned image capturing unit 210, a decoding unit 220, a processing unit 230, and a display unit 240. The image capturing unit 210 is used to capture images, such as a camera, webcam, monitor, video recorder, or a mouse, keyboard, etc., capable of taking screenshots.

[0044] The decoding unit 220 and the arithmetic unit 230 are used to perform various decoding and arithmetic programs, such as a circuit, a circuit board, a storage device for storing program code, or a chip. The aforementioned chip may be, for example, a central processing unit (CPU), or other programmable general-purpose or special-purpose microcontroller units (MCUs), microprocessors, digital signal processors (DSPs), programmable controllers, application-specific integrated circuits (ASICs), graphics processing units (GPUs), image signal processors (ISPs), image processing units (IPUs), arithmetic logic units (ALUs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other similar components or combinations thereof. The display unit 240 is used to display information, such as a liquid crystal display panel, an electronic paper display panel, or an OLED display panel.

[0045] In this embodiment, the electronic device 100 obtains a generated image IM11 using AI generation technology and then adds a hidden watermark WM1. After the user obtains a captured image IM12 by taking a picture or screenshot, the electronic device 200 can decode the hidden watermark WM1 to obtain a regenerated image IM13 that is identical to the generated image IM11. The operation of each component is explained in detail below with a flowchart.

[0046] Please refer to Figure 4 The example illustrates an image generation method and an image regeneration method according to an embodiment of the present invention. The image generation method and the image regeneration method include steps S110 to S140 and steps S210 to S230.

[0047] In step S110, as Figure 2 and Figure 4As shown, the computing unit 110 of the electronic device 100 obtains a generation prompt PT1, a stable diffusion seed SDSE1, an out painting seed OPSE1, a stable diffusion steps SDST1, a stable diffusion model version SDMV1, and an out painting model version OPMV1.

[0048] Next, in step S120, as Figure 2 and Figure 4 As shown, based on the generation prompt PT1, diffusion seed SDSE1, expansion seed OPSE1, diffusion step SDST1, diffusion model version SDMV1, and expansion model version OPMV1, the generated image IM11 is obtained using the image generation model MD1 of the electronic device 100. The generated image IM11 is, for example, an image containing a person, animal, or object.

[0049] Then, in step S130, as Figure 2 and Figure 4 As shown, the encoding unit 120 of the electronic device 100 encodes the generated prompt PT1, diffusion seed SDSE1 and extension seed OPSE1 into a hidden watermark WM1.

[0050] In this step, the encoding unit 120 may also encode the generation prompt PT1, diffusion seed SDSE1, extension seed OPSE1, diffusion step SDST1, diffusion model version SDMV1, and extension model version OPMV1 into a hidden watermark WM1.

[0051] Next, in step S140, as Figure 2 and Figure 4 As shown, the synthesis unit 130 of the electronic device 100 adds a hidden watermark WM1 to the generated image IM11. The hidden watermark WM1 is, for example, set over the entire area of ​​the generated image IM11. The generated image IM11 can be displayed on a tablet computer, an LCD screen, a computer screen, or printed out for display. Although the hidden watermark WM1 is added to the generated image IM11, it is not visible to the naked eye. Steps S110 to S140 complete the image generation method. Next, if it is desired to regenerate the generated image IM11, steps S210 to S230 will be performed.

[0052] In step S210, as Figure 3 and Figure 4As shown, the image capturing unit 210 of the electronic device 200 captures or screenshots the generated image IM11 to obtain the captured image IM12. Figure 4 In step S210, the image capturing unit 210 of the electronic device 200 captures the generated image IM 11 to obtain a captured image IM 12. Since the generated image IM 11 has a hidden watermark WM 1, the captured image IM 12 will also have a hidden watermark WM 1.

[0053] Next, in step S220, as Figure 3 and Figure 4 As shown, the decoding unit 220 of the electronic device 200 performs a decoding procedure on the captured image IM12 to obtain a generation hint PT1, a diffusion seed SDSE1, and an expansion seed OPSE1 for an image generation model MD2.

[0054] Then, in step S230, please refer to Figure 3 and Figure 4 The processing unit 230 of the electronic device 200 obtains the regenerated image IM13 using the image generation model MD2 based on the generation prompt PT1, diffusion seed SDSE1, and expansion seed OPSE1. In this step, the unchanged generation prompt PT1, diffusion seed SDSE1, and expansion seed OPSE1 are input into the image generation model MD2 to generate the regenerated image IM13, which is identical to the generated image IM11. The regenerated image IM13 does not contain the hidden watermark WM1. In one embodiment, the image generation model MD1 and the image generation model MD2 can be pre-defined to use the same diffusion step SDST1, diffusion model version SDMV1, and expansion model version OPMV1, without needing to obtain the diffusion step SDST1 and diffusion model version SDMV1 through the hidden watermark WM1.

[0055] In the above embodiments, since the hidden watermark WM 1 is first added to the generated image IM 11 in the electronic device 100, after the user takes a picture of the generated image IM 11, it can be decoded in the electronic device 200 to obtain a regenerated image IM 13 that is exactly the same as the generated image IM 11, and then completely reproduced on the display unit 240.

[0056] Please refer to Figure 5 Examples illustrate an image generation method and an image regeneration method according to another embodiment of the present invention. Figure 5 In step S210, the image capturing unit 210 of the electronic device 200 takes a screenshot of the generated image IM11 to obtain a captured image IM22. The image capturing unit 210 is, for example, a mouse, which moves the cursor to select the screen for taking a screenshot.

[0057] In this embodiment, please refer to Figure 5 The difference from the above embodiment is that, in step S210, when the generated image IM 11 is displayed on the screen of the electronic device 200, the image capturing unit 210 (e.g., a mouse) can be used to select the area to be captured to obtain the captured image IM 22. The generated image IM 11 has a hidden watermark WM 1, so the captured image IM 22 will also have a hidden watermark WM 1.

[0058] Next, in step S220, as Figure 3 and Figure 5 As shown, the decoding unit 220 of the electronic device 200 performs a decoding procedure for the hidden watermark WM 1 on the captured image IM22 to obtain the generation hint PT 1, diffusion seed SDSE1 and expansion seed OPSE1 of the image generation model MD2.

[0059] Then, in step S230, please refer to Figure 3 and Figure 5 The processing unit 230 of the electronic device 200 obtains the regenerated image IM13 using the image generation model MD2 based on the generation prompt PT1, diffusion seed SDSE1, and expansion seed OPSE1. In this step, the unchanged generation prompt PT1, diffusion seed SDSE1, and expansion seed OPSE1 are input into the image generation model MD2 to generate the regenerated image IM13, which is identical to the generated image IM11. The regenerated image IM13 does not contain the hidden watermark WM1. In one embodiment, the image generation model MD1 and the image generation model MD2 can be pre-defined to use the same diffusion step SDST1, diffusion model version SDMV1, and expansion model version OPMV1, without needing to obtain the diffusion step SDST1 and diffusion model version SDMV1 through the hidden watermark WM1.

[0060] In the above embodiments, since the hidden watermark WM 1 is first added to the generated image IM 11 in the electronic device 100, when the user takes a screenshot of the generated image IM 11, it can be decoded in the electronic device 200 to obtain a regenerated image IM 13 that is exactly the same as the generated image IM 11, and then completely reproduced on the display unit 240.

[0061] Please refer to Figure 6 Examples illustrate an image generation method and an image regeneration method according to another embodiment of the present invention. Figure 6In step S220, the decoding unit 220 of the electronic device 200 performs a decoding procedure for the hidden watermark WM1 on the captured image IM22 to obtain the generation hint PT1, diffusion seed SDSE1 and expansion seed OPSE1 of the image generation model MD2.

[0062] Then, in step S220, as Figure 6 As shown, according to the user's needs, the generation prompt PT 1, diffusion seed SDSE 1, and expansion seed OPSE 1 are changed to the generation prompt PT3, diffusion seed SDSE3, and expansion seed OPSE3. For example, the user can change the background style, foreground objects, colors, and styles as needed.

[0063] Next, in step S230, as Figure 6 As shown, the processing unit 230 of the electronic device 200 obtains the regenerated image IM33 based on the generation prompt PT3, diffusion seed SDSE3, and expansion seed OPSE3 using the image generation model MD2. In this step, the modified generation prompt PT3, diffusion seed SDSE3, and expansion seed OPSE3 are input into the image generation model MD2 to generate a regenerated image IM33 similar to the generated image IM11. The regenerated image IM33 does not contain the hidden watermark WM1.

[0064] In the above embodiments, since the hidden watermark WM 1 is first added to the generated image IM 11 in the electronic device 100, when the user takes a picture or screenshot of the generated image IM 11, it can be decoded in the electronic device 200 and the parameters can be modified. Finally, a regenerated image IM 33 similar to the generated image IM 11 can be obtained and completely reproduced on the display unit 240.

[0065] Please refer to Figure 7 Examples illustrate an image generation method and an image regeneration method according to another embodiment of the present invention. Figure 7 In step S130, the encoding unit 120 of the electronic device 100 encodes the generated prompt PT1, diffusion seed SDSE1, and extension seed OPSE1 into a hidden watermark WM4. In this step, the hidden watermark WM4 is set at a portion of the area or the center (e.g., the object area OBR4).

[0066] In this step, the encoding unit 120 may also encode the generation prompt PT1, diffusion seed SDSE1, extension seed OPSE1, diffusion step SDST1, diffusion model version SDMV1, and extension model version OPMV1 into a hidden watermark WM4.

[0067] Next, in step S140, as Figure 2 and Figure 7 As shown, the synthesis unit 130 of the electronic device 100 adds a hidden watermark WM4 to the generated image IM 11. The hidden watermark WM4 is, for example, set on the object area OBR4 of the generated image IM 11. The generated image IM 11 can be displayed on a tablet computer, an LCD screen, a computer screen, or printed out for display. Although the hidden watermark WM4 is added to the generated image IM 11, it is not visible to the naked eye. Steps S110 to S140 complete the image generation method. Next, if it is desired to regenerate the generated image IM 11, steps S210 to S230 will be performed.

[0068] In step S210, as Figure 3 and Figure 7 As shown, the image capturing unit 210 of the electronic device 200 captures or screenshots the generated image IM11 to obtain the captured image IM42. Figure 7 In step S210, the image capturing unit 210 of the electronic device 200 captures the generated image IM 11 to obtain a captured image IM 42. For example... Figure 7 As shown, although the image capturing unit 210 only captures a portion of the generated image IM 11, as long as the captured image IM42 covers the object range OBR4, the complete hidden watermark WM4 can be obtained from the captured image IM42.

[0069] Next, in step S220, as Figure 3 and Figure 7 As shown, the decoding unit 220 of the electronic device 200 performs a decoding procedure on the captured image IM42 to obtain the generation hint PT1, diffusion seed SDSE1 and expansion seed OPSE1 of the image generation model MD2.

[0070] Then, in step S230, please refer to Figure 3 and Figure 7The processing unit 230 of the electronic device 200 obtains the regenerated image IM13 using the image generation model MD2 based on the generation prompt PT1, diffusion seed SDSE1, and expansion seed OPSE1. In this step, for example, the unchanged generation prompt PT1, diffusion seed SDSE1, and expansion seed OPSE1 are input into the image generation model MD2 to generate the regenerated image IM13, which is identical to the generated image IM11. The regenerated image IM13 does not contain the hidden watermark WM1. In one embodiment, the image generation model MD1 and image generation model MD2 can be pre-defined to use the same diffusion step SDST1, diffusion model version SDMV1, and expansion model version OPMV1, without needing to obtain the diffusion step SDST1 and diffusion model version SDMV1 through the hidden watermark WM1.

[0071] according to Figure 7 In one embodiment, the hidden watermark WM4 is only added to a portion or the center of the generated image IM 11 (e.g., the object area OBR4). Even if the user does not capture the entire image, the hidden watermark WM4 can still be successfully decoded to obtain a regenerated image IM 13 that is exactly the same as the generated image IM 11, and then fully reproduced on the display unit 240.

[0072] Through the various embodiments described above, after obtaining the generated image IM 11 using AI generation technology, the electronic device 100 can add hidden watermarks WM1 and WM4. After obtaining captured images IM12, IM22, and IM42 through photography or screenshotting, the electronic device 200 can decode the hidden watermarks WM1 and WM4, allowing the user to obtain a regenerated image IM 13 that is the same as the generated image IM 11 or a similar regenerated image IM33.

[0073] Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications and improvements without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be defined by the claims.

Claims

1. An image generation method, comprising: Obtain a generation prompt, a stable diffusion seed, and an outpainting seed; Based on the generation prompt, the diffusion seed, and the expansion seed, an image generation model is used to obtain a generated image; The generation prompt, the diffusion seed, and the expansion seed are encoded as a hidden watermark; and Add the hidden watermark to the generated image.

2. The image generation method as described in claim 1, characterized in that, The hidden watermark is applied to the entire area of ​​the generated image.

3. The image generation method as described in claim 1, characterized in that, The hidden watermark is set on a portion of the generated image.

4. The image generation method as described in claim 1, characterized in that, The hidden watermark is placed in the center of the generated image.

5. The image generation method as described in claim 1, characterized in that, The hidden watermark is set within an object range of the generated image.

6. An image regeneration method, comprising: An image capturing unit captures or screenshots a generated image to obtain a captured image, which has a hidden watermark. A decoding procedure for the hidden watermark on the captured image is used to obtain a generation prompt, a stable diffusion seed, and an out painting seed for an image generation model; and Based on the generation prompt, the diffusion seed, and the expansion seed, a regenerated image is obtained using the image generation model.

7. The image regeneration method as described in claim 6, characterized in that, The hidden watermark is applied to the entire area of ​​the generated image.

8. The image regeneration method as described in claim 6, characterized in that, The hidden watermark is set on a portion of the generated image.

9. The image regeneration method as described in claim 6, characterized in that, The hidden watermark is placed in the center of the generated image.

10. The image regeneration method as described in claim 6, characterized in that, The hidden watermark is set within an object range of the generated image.

11. An electronic device comprising: An image capturing unit is used to capture or screenshot a generated image to obtain a captured image, the generated image having a hidden watermark; A decoding unit is used to decode the captured image using the hidden watermark to obtain a generation prompt, a stable diffusion seed, and an outpainting seed for an image generation model; and A processing unit is used to obtain a regenerated image based on the generation prompt, the diffusion seed, and the expansion seed, using the image generation model.

12. The electronic device as claimed in claim 11, characterized in that, The hidden watermark is applied to the entire area of ​​the generated image.

13. The electronic device as claimed in claim 11, characterized in that, The hidden watermark is set on a portion of the generated image.

14. The electronic device as claimed in claim 11, characterized in that, The hidden watermark is placed in the center of the generated image.

15. The electronic device as claimed in claim 11, characterized in that, The hidden watermark is set within an object range of the generated image.

Images