Click-based repetitive ai image generation tool

Abstract

To provide an image generation device, a method for generating an image, and program which allow a user to acquire an image only by inputting simple and short prompts and by clicking images close to an image that the user is picturing.SOLUTION: The method includes: receiving simple prompts from a user; generating a plurality of images on the basis of the prompts; displaying the images to the user; causing the user to select an image close to an image that the user is picturing from the displayed images; and causing a system to generate at least two new images by using a statistic method such as Bayes' statistics with reference to the direction indicated by the vector of words in the selected image on the basis of the selection. This process is repeated until the user can acquire an image that the user wants.SELECTED DRAWING: Figure 1

Description

[Technical field]

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

[0002] In recent years, AI image generation technology has been advancing. For example, there is technology that generates specific images based on specific prompts.

[0003] However, to obtain the desired image, the user must input specific and detailed prompts, and a certain level of knowledge is required to obtain the desired image.

[0004] To provide an image generating device, an image generating method, and a program that allow a user to obtain a desired image simply by clicking on a simple and short prompt and on an image that is close to the image the user has in mind.

[0005] Conventional AI image generation technology requires users to input detailed prompts, and the generated image is deeply related to the input prompt. For example, if you want to generate an image of a "dog sitting on a chair drinking coffee", you can input detailed prompts such as "best quality, masterpiece, ultra high res, (photo realistic: 1.4), 1 dog, sitting on a ((chair)), wearing a ((suit)), holding a ((coffee cup)), stylish glasses, confident posture, suave expression, (tie: 1.3), well-groomed fur, polished shoes, sipping ((steaming coffee: 1.5)), with a newspaper ((folded neatly on the table)), elegant surroundings", and an image will be generated based on that. However, this method has the problem that it is difficult for users to imagine a specific image. In order to create a specific prompt, the user needs to understand the mechanism of image generation and the meaning of words, which is a difficult challenge for general users.

[0006] In addition, in conventional methods, in order to adjust the attributes of the generated image (e.g., Steps, Scale, Seed, Sampler), the user had to understand the meaning and effect of each attribute and set appropriate values. However, this was difficult for users, and a lot of trial and error was required to obtain the desired image.

[0007] Given this background, there is a demand for the development of new image generation technology that allows users to obtain the image they want simply by entering a simple prompt, such as "chair dog coffee," and selecting an image that is closest to the image they have in mind. Summary of the Invention [Problem to be solved by the invention]

[0008] The problem that this invention aims to solve is the development of an image generation technology that allows a user to obtain a desired image by simply inputting a simple prompt and selecting an image that is close to the image that the user has in mind. Specifically, there is a need for a system that can generate a desired image without the user having to create a complex prompt or adjust each attribute of the image in detail. [Means for solving the problem]

[0009] The main feature of the present invention is that it receives simple prompts from the user, generates multiple images based on the prompts, and displays them to the user. The user selects the image that best resembles the image he or she has in mind. Based on the selection, the system generates two or more new images using statistical methods such as Bayesian statistics, based on the direction indicated by the vectors of the words in the selected image. This process is repeated until the user obtains the desired image. This allows the user to generate the desired image without creating complex prompts or fine-tuning each attribute of the images. Effect of the Invention

[0010] The AI ​​image generation system of the present invention has the advantage that users can obtain the image they want simply by inputting a simple prompt and selecting an image that is close to the image they have in mind. This allows users to generate the desired image without creating complex prompts or fine-tuning each attribute of the image. In addition, by selecting an image that is close to the image they have in mind, users can give more precise instructions to the system, which makes it possible to generate the desired image more accurately. These advantages can be applied in various fields such as e-commerce, advertising, comics, and video games. [Brief description of the drawings]

[0011] [Figure 1]FIG. 1 is a conceptual diagram of the AI ​​image generation system of the present invention. In the diagram, the user inputs a simple prompt (step 1), and the system generates multiple images based on the prompt (step 2) and displays these images to the user (step 3). The user selects from the displayed images the one that is closest to the image he or she has in mind (step 4). Based on the selection, the system generates a new image (step 5). This process is repeated until the user obtains the desired image. Note that this diagram is merely conceptual, and the design and configuration of the actual system can be modified without departing from the scope of the present invention.

Claims

1. An image generating device that generates a plurality of images based on simple, short prompts, and allows a user to select from the generated images, generating further images based on the selected image.

2. 2. The apparatus according to claim 1, wherein the image is generated using a statistical method based on the direction indicated by the vectors of the words of the selected image.

3. A method for generating multiple images based on simple, short prompts, allowing the user to select from the generated set of images, and generating further images based on the selected image.

4. 4. The method according to claim 3, further comprising the step of generating an image using a statistical method based on the direction in which the vectors of the words of the selected image point.

5. A program that generates multiple images based on simple, short prompts, allows the user to select from the generated images, and then generates further images based on the selected image.

6. 6. The program according to claim 5, further comprising: generating an image by using a statistical method to determine the direction indicated by the vector of the words in the selected image.

Images