Generation method

Abstract

The invention discloses a generation method, and the method comprises the steps: displaying to-be-confirmed first output information based on input information; in response to a target instruction, modifying or confirming the first output information to obtain second output information used for generating a target generation object of a preset type; and inputting the second output information into a target generation model, and outputting the target generation object.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of artificial intelligence, and are related to but not limited to a generation method. Background Art

[0002] Artificial Intelligence Generated Content (AIGC) is developing rapidly, but the quality of objects generated by generative models is unstable. Users use generative models to handle daily work, but do not know how to optimize; do not know how to express their own needs, etc. Especially for users who are not proficient in artificial intelligence (AI) generation tools, through simple prompt information (Prompt) input, the output results are prone to problems and often do not meet the needs of raw images. More than 80% of key target users will give up using the tool after trying 2 to 5 rounds and think that the raw image model is not capable enough. How to improve the quality of raw images has become a technical problem that needs to be solved urgently. Summary of the invention

[0003] In view of this, an embodiment of the present application provides a generation method.

[0004] The technical solution of the embodiment of the present application is implemented as follows:

[0005] In a first aspect, an embodiment of the present application provides a generation method, comprising:

[0006] Based on the input information, presenting first output information to be confirmed;

[0007] In response to the target instruction, modify or confirm the first output information to obtain second output information for generating a target generation object of a preset type;

[0008] The second output information is input into the target generation model, and the target generation object is output.

[0009] In a second aspect, an embodiment of the present application provides a generating device, including:

[0010] A display, for presenting first output information to be confirmed based on the input information;

[0011] A processor, configured to modify or confirm the first output information in response to the target instruction, and obtain second output information for generating a target generation object of a preset type;

[0012] The processor is further configured to input the second output information into the target generation model and present the target generation object on the display.

[0013] In a third aspect, an embodiment of the present application provides an electronic device, including a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and the above method is implemented when the processor executes the program.

[0014] In a fourth aspect, an embodiment of the present application provides a storage medium storing executable instructions for implementing the above method when executed by a processor.

[0015] In a fifth aspect, a computer program product includes a computer program or instructions, and when the computer program or instructions are executed by a processor, the steps in the above method are implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1A A schematic diagram of an implementation flow of a generation method provided in an embodiment of the present application;

[0017] Figure 1B A schematic diagram of an interactive window provided in an embodiment of the present application;

[0018] Figure 2 A schematic diagram of an implementation flow of presenting first output information provided in an embodiment of the present application;

[0019] Figure 3A A schematic diagram of an implementation flow of presenting first output information provided in an embodiment of the present application;

[0020] Figure 3B A visual page of a generation tool provided in an embodiment of the present application;

[0021] Figure 4 A schematic diagram of an implementation flow of a method for generating an image provided in an embodiment of the present application;

[0022] Figure 5 A schematic diagram of the composition structure of a generating device provided in an embodiment of the present application;

[0023] Figure 6 A hardware entity schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the specific technical solution of the embodiments of the present application will be further described in detail below in conjunction with the drawings in the embodiments of the present application. The following embodiments are used to illustrate the present application, but are not used to limit the scope of the present application.

[0025] In the following description, reference is made to “some embodiments”, which describe a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0026] In the following description, the terms "first\second\third" involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It can be understood that "first\second\third" can be interchanged with a specific order or sequence where permitted, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0027] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0028] The present application embodiment provides a generation method, as shown in FIG1 , the method comprising:

[0029] Step S110: presenting first output information to be confirmed based on the input information;

[0030] Here, the user can clearly describe the task information of the object to be generated to the generative model, that is, the input information, including the content type (such as text, image, audio, video, etc.), theme, style, length, etc. The generative model can understand the user's needs based on the input information, so as to generate a more satisfactory generated object. For example, the generative model can be a model for generating images, or a large language model for generating text.

[0031] The input information may be at least one of the following information: text information, image information, audio information, video information, etc. For example, the user may input a language description for generating a target generation object; or may input a basic picture for generating a target generation object.

[0032] In some embodiments, the first output information may include information extracted from the input information, may include information generated based on the input information and input information required by the generation model, and may include information expanded based on the input information of the generation model.

[0033] During implementation, the first output information may be presented in a preset interactive window so that the user can view the first output information.

[0034] Figure 1B A schematic diagram of an interactive window provided in an embodiment of the present application, such as Figure 1B As shown, the schematic diagram includes an information presentation window 11 and an information input window 12, wherein:

[0035] The information presentation window 11 presents the user's input information 111 as "company's International Women's Day activity illustration, a girl". The presented structured first output information 112 is:

[0036] "Based on your input, the system simulates the following keywords for you

[0037] Content: girls, festivals, celebrations, ice cream, drinks, happiness, publicity

[0038] Environment: Indoor

[0039] Light effect: ambient light

[0040] Style: Comic style

[0041] Have you made any adjustments to the above keywords? "

[0042] Here, part of the first output information 112, "girls" and "festival", is generated based on the input information 111 "company's International Women's Day event illustration, a girl", and part of the information is generated based on the input information and the input information required by the generation model, including "indoor" and "ambient light", and part of the information is expanded based on the input information of the generation model, such as "style: comic style".

[0043] Step S120, in response to the target instruction, modify or confirm the first output information to obtain second output information for generating a target generation object of a preset type;

[0044] Here, the target instruction is a modification and confirmation instruction triggered by the user based on the first output information. For example, the user can trigger the confirmation instruction when determining that the first output information meets the requirements for generating the target generation object; the user can also trigger the modification instruction when determining that the first output information needs to be modified or added.

[0045] In the implementation process, the first output information can be modified or confirmed based on the target instruction triggered by the user, and the second output information of the target generation object of the preset type generated by the user can be obtained. Here, the target generation object of the preset type can be a generated image or a generated text.

[0046] Figure 1B The information presentation window 11 shown presents the confirmation information 113 "No adjustment" input by the user through the information input window 12, realizing the confirmation of the first output information. Here, since there is no adjustment, that is, the first output information is the same as the second output information, in some embodiments, the first output information may be omitted from being presented again, so as to realize a simpler and clearer presentation of the output information.

[0047] In some embodiments, the user may also input an "adjustment" instruction to adjust the first output information to obtain second output information that better meets the user's generation intention. Figure 1B The first output information shown is adjusted to the second output information: "Content: girls, festival, celebration, ice cream, beverage, happy, publicity; Environment: outdoor; Light effect: ambient light; Style: oil painting style".

[0048] Step S130: input the second output information into a target generation model, and output the target generation object.

[0049] Here, the target generation model may be an image generation model, or may be a text generation model, a speech generation model, a video generation model, or the like, which can generate a target generation object based on the second output information.

[0050] During implementation, the second output information confirmed by the user may be used as input information of the target generation model, so that the target generation model can output the target generation object based on the second output information.

[0051] Figure 1B The information presentation window 11 shown presents target generation objects 114 and 115 generated by the image generation intelligent software based on input information 112.

[0052] In the embodiment of the present application, firstly, based on the input information, the first output information to be confirmed is presented; then, in response to the target instruction, the first output information is modified or confirmed to obtain the second output information for generating the target generation object of the preset type; finally, the second output information is input into the target generation model to output the target generation object. In this way, the intention understanding based on the user input information is realized, the first output information matching the processing capability of the target generation model is presented to the user, and the modification or confirmation function is further provided, so that the user can modify or confirm the first output information to obtain the second output information that is more in line with the user's intention, shortening the gap between natural language and model language, improving the quality of the target generation object, and the subjective feeling of the user using the target generation model.

[0053] In some embodiments, the above step S110 "presenting first output information to be confirmed based on the input information" is as follows: Figure 2 As shown, this can be achieved by following the steps below:

[0054] Step S210: parsing the input information to obtain at least one filled-in information matching a target generation template, wherein the target generation template is a generation template matching the input information;

[0055] Here, a corresponding target generation template can be obtained based on the input information. For example, if the input information is "company's Women's Day event illustration, a girl", the corresponding image generation template can be obtained as "content: XXX; environment: XXX; light effect: XXX; style: XXX".

[0056] A plurality of generation templates corresponding to different subject information may be preset, and other different types of generation templates may also be preset, and each generation template may include a plurality of prompt information items to be filled in.

[0057] During the implementation process, the user's input information can be parsed to obtain at least one filled-in information that matches the target generation template. For example, if the user input information is "company's International Women's Day event illustration, a girl", then the input information can be parsed to obtain the filled-in information that matches the picture generation module, including "girl, festival, celebration".

[0058] Step S220: fill in the target generation template based on the at least one filled-in information and the at least one generated information, and present the first output information after completing the information filling, wherein the generated information is generated based on the target generation template.

[0059] Here, the generated information may be generated based on the prompt information item in the target generation template. The generated information may be information recommended to the user for filling in the target generation template.

[0060] In some embodiments, the generated information may also be information generated based on user input information and matching the prompt information item in the target generation module.

[0061] During the implementation process, the target generation template may be filled in based on at least one filling information and at least one generation information.

[0062] In some embodiments, the first output information is presented in a structured form.

[0063] Here, structured presentation can refer to organizing and presenting information, data or content in a logical and hierarchical way so that it is easy to understand and process. This presentation method can effectively improve the readability of information. Figure 1B The first output information 112 shown in FIG. 1 is an exemplary structured form.

[0064] In the embodiment of the present application, the input information is first parsed to obtain at least one filled-in information that matches the target generation template; then the target generation template is filled in based on the at least one filled-in information and the at least one generated information, and the first output information after the information filling is completed is presented. In this way, a suitable target generation template can be matched based on the user's input information, and then the target generation template can be filled in using the filled-in information obtained based on the input information and the generated information generated based on the target generation template, so as to achieve the best matching for the user and generate the first output information with the best target generation object effect that can be presented to the user.

[0065] In some embodiments, the implementation of the above step S110 “presenting first output information to be confirmed based on the input information” further includes the following steps:

[0066] Step S230, parsing the input information to determine the subject information to be generated;

[0067] Here, the subject information is diverse, depending on the creator's intention, and can cover any visually expressible concept, scene, object or emotion. For example, it includes characters and portraits, landscapes and natural environments, product presentations, etc.

[0068] During the implementation process, the user's input information can be parsed to determine the user's creative intention, that is, to determine the subject information to be generated. For example, if the input information is "Please draw me an autumn forest, with sunlight shining through the leaves on the ground and fallen leaves covering the path", the subject information can be determined to be "Landscape and Natural Environment". If the input information is "I hope to have an anime image of a little boy with big eyes", the subject information can be determined to be "Characters and Portraits".

[0069] Step S240: acquiring the target generation template based on the subject information to be generated.

[0070] In the implementation process, the corresponding target generation template can be determined based on the subject information. For example, when the subject information is determined to be scenery and natural environment, the generation template corresponding to scenery and natural environment can be determined as the target generation template; when the subject information is determined to be people and portraits, the generation template corresponding to people and portraits can be determined as the target generation template.

[0071] In the embodiment of the present application, the input information is first parsed to determine the subject information to be generated, and then a target generation template is obtained based on the subject information to be generated. In this way, the determined target generation template can better realize the user's creative intention.

[0072] In some embodiments, the above step S230 "parsing the input information and determining the subject information to be generated" can be implemented by the following steps:

[0073] Step 231: Determine that the input information is text information;

[0074] Here, the user can enter a piece of text information as input information.

[0075] Step 232: Cut the text information into long sentences to obtain at least one target sentence;

[0076] Here, also known as sentence segmentation or fragmentation, it refers to the process of dividing a longer sentence into multiple shorter sentences or clauses that are easier to understand and process according to grammar, semantics or specific rules.

[0077] During the implementation process, when the text information input by the user is long, the text information can be cut into long sentences to obtain multiple target sentences.

[0078] Step 233: Perform structured element extraction on each target sentence to determine the topic information to be generated.

[0079] Sentence structured element extraction is a task in Natural Language Processing (NLP), which decomposes a sentence into a series of grammatically or semantically related components (i.e., structured elements) for deeper analysis or processing. These structured elements may include subject, predicate, object, attributive, adverbial, etc.

[0080] In the implementation process, the subject information can be determined based on the structured element extraction of the target sentence. For example, "Please draw me a picture of an autumn forest, with sunlight shining through the leaves on the ground and fallen leaves covering the trail" can extract structured elements including "forest", "ground", "trail", etc., and determine that the subject information is scenery and natural environment.

[0081] In the embodiment of the present application, the input information is determined to be text information; first, the text information is segmented into long sentences to obtain at least one target sentence; then, structured element extraction is performed on each target sentence to determine the topic information to be generated. In this way, the text information can be parsed by sequentially performing long sentence segmentation and structured element extraction, and the topic information can be determined based on the text information.

[0082] In some embodiments, the above step S230 "parsing the input information and determining the subject information to be generated" can be implemented by the following steps:

[0083] Step 234: Determine that the input information is image information;

[0084] Here, in addition to inputting text information, the user can also input an original image. Based on the original image, combined with the input text information, a new generated image is obtained.

[0085] During implementation, the user can provide an original image as image information.

[0086] Step 235: Identify the image information and obtain an image theme corresponding to the image information;

[0087] In some embodiments, image information may be identified, text information corresponding to the image information may be obtained, and then the image theme may be determined based on the text information.

[0088] In some embodiments, the extracted image features can be used to perform image recognition through machine learning or deep learning models to determine the image theme.

[0089] Step 236: Determine the subject information to be generated based on the image subject.

[0090] In the embodiment of the present application, the input information is determined to be image information; the image information is first identified to obtain the image theme corresponding to the image information; and then the theme information to be generated is determined based on the image theme. In this way, when the input information is image information, the image information can be identified to obtain the theme information.

[0091] In some embodiments, the target generation template includes multiple prompt information items. In the above step S220, "filling in the target generation template based on the at least one filling information and the at least one generation information" can be implemented by the following steps:

[0092] Step 221: Determine a target prompt information item that matches the at least one filled-in information from among the multiple prompt information items;

[0093] Here, there are target prompt information items that match the filled-in information among the multiple prompt information items in the target generation template.

[0094] During the implementation process, the fill-in information that can be filled in can be matched based on the project names of multiple prompt information items.

[0095] Step 222: determining at least one of the generation information from a target prompt information database based on at least one of the target prompt information items and the plurality of prompt information items, wherein the target prompt information database is constructed based on the target generation model;

[0096] Here, there are target prompt information items that cannot match the filled-in information among the multiple prompt information items in the target generation template.

[0097] During implementation, after determining the target prompt information item that matches the filled-in information, other prompt information items that cannot match the filled-in information are determined, and at least one generation information for filling in other information items is determined from the target prompt information database. The data in the target prompt information database is data adapted to the target generation model, that is, inputting the data can improve the generation effect of the target generation model.

[0098] Step 223: Fill in at least one of the filled-in information and at least one of the generated information into the multiple prompt information items to obtain the first output information.

[0099] In the embodiment of the present application, first, a target prompt information item matching at least one filled-in information is determined from the multiple prompt information items; then, based on the at least one target prompt information item and the multiple prompt information items, at least one generated information is determined from the target prompt information database; finally, at least one filled-in information and at least one generated information are filled into the multiple prompt information items to obtain the first output information. In this way, it is possible to fill at least one filled-in information and at least one generated information into the multiple prompt information items in the target generation template to obtain the first output information presented to the user, and the first output information can improve the generation effect of the target generation model due to the addition of the generated information.

[0100] In some embodiments, the above target prompt information database can be obtained by following the steps below:

[0101] Step 224, obtaining a source prompt information database;

[0102] Here, the source prompt information database may be an exclusive high-quality Prompt database corresponding to the generation model.

[0103] Step 225, classify each prompt information in the source prompt information database to obtain a classified source prompt information database;

[0104] During the implementation process, each prompt information in the source prompt information database may be labeled and classified to obtain a classified source prompt information database.

[0105] Step 226: Filter the classified source prompt information database based on the target generation model to obtain the target prompt information database.

[0106] In the implementation process, the target prompt information data can be screened out through the preset basic rules for screening data, and the target prompt information data with useful and good results generated by the screening generation model can be screened out. The target prompt information database is formed based on the screened target prompt information data.

[0107] In the embodiment of the present application, a source prompt information database is first obtained; each prompt information in the source prompt information database is then classified to obtain a classified source prompt information database; and finally, the classified source prompt information database is filtered based on the target generation model to obtain a target prompt information database. In this way, the target prompt information data in the target prompt information database is prompt information data that can improve the effect of generating the object.

[0108] In some embodiments, the implementation of "filling in the target generation template based on the at least one filling information and the at least one generation information" in the above step S220 further includes the following steps:

[0109] Step 227: determining a weight of each of the prompt information items based on the contribution of each of the prompt information items to the quality improvement of generating the target generation object;

[0110] Here, quality improvement may refer to the quality of the target generated object, including at least one of the following: improvement in clarity, color richness, detail expression, etc. A higher weight may be set for a prompt information item that contributes more to quality improvement, to represent the importance of the prompt information item to the generated object.

[0111] For example, the landscape template may be preset as: [*Image settings] + [*Theme content + Environment] + [*Composition / perspective] + [Lighting effect] + [*Style], where the prompt information items marked with * have a higher weight than the prompt information items not marked.

[0112] Step 228: determining the arrangement order of all the prompt information items based on the weight of each of the prompt information items;

[0113] Here, the arrangement order can represent the weight of each prompt information item. In the implementation process, the prompt information items with higher weights can be arranged in front. In this way, the prompt information items with higher weights play a greater role in generating objects.

[0114] Correspondingly, the above step 223 "filling at least one of the filled-in information and at least one of the generated information into the multiple prompt information items to obtain the first output information" can be implemented by the following process:

[0115] At least one of the filling information and at least one of the generating information are sequentially filled into the plurality of prompt information items based on the arrangement order to obtain the first output information.

[0116] In the embodiment of the present application, firstly, the weight of each prompt information item is determined based on the contribution of each prompt information item to the quality improvement of the generated target object; then, the arrangement order of all prompt information items is determined based on the weight of each prompt information item; finally, at least one filled-in information and at least one generated information are filled in to multiple prompt information items in sequence based on the arrangement order to obtain the first output information. In this way, the prompt information items with greater contribution to quality improvement are placed in front, which can effectively improve the quality of the generated object.

[0117] In some embodiments, the above step S110 “presenting first output information to be confirmed based on the input information” can be implemented by the following process:

[0118] Based on the input information, present a plurality of the first output information;

[0119] During implementation, a plurality of different first output information may be presented based on the input information, that is, the content filled in the prompt information item in each output information may be different, so as to provide the user with a visual choice.

[0120] Correspondingly, the above step S120 "modifying or confirming the first output information in response to the target instruction to obtain the second output information for generating the target generation object of the preset type" can be implemented by the following steps:

[0121] Step 121: Obtain a target instruction for determining the second output information, wherein the generation information in a plurality of the first output information is different;

[0122] Here, the target instruction may be triggered by a user, and an instruction for generating the second output information is determined from a plurality of first output information.

[0123] Step 122: In response to the target instruction, determine one second output information from a plurality of first output information.

[0124] During the implementation process, the user can select one first output information that can represent his / her generation intention as the second output information based on the multiple first output information seen.

[0125] In the embodiment of the present application, based on the input information, multiple first output information are presented; then a target instruction for determining the second output information is obtained; and finally, in response to the target instruction, one second output information is determined from the multiple first output information. In this way, the user can select a first output information that can represent his / her generation intention as the second output information based on the multiple first output information he / she sees, so as to effectively improve the quality of the generated object.

[0126] In some embodiments, the above step S110 "presenting first output information to be confirmed based on the input information" is as follows: Figure 3A As shown, this can be achieved by following the steps below:

[0127] Step S310, parsing the input information to obtain the background and target of the target generation object;

[0128] Here, context is used to provide background information. Providing detailed background information to the generative model (large language model) can help the generative model accurately understand the discussion scenario and ensure that the feedback provided is relevant. For example, when writing a prompt about a historical event, you can provide background information such as the time, place, and main characters of the event.

[0129] Objective: It can clearly define the task that the generative model is required to complete. Clearly defining the task objective allows the generative model to more focusedly adjust its response to achieve this specific goal. For example, you can ask the generative model to "generate an abstract of a paper on the development trends of artificial intelligence."

[0130] During the implementation process, when the input information is text information, the background and target can be identified through long sentence cutting, semantic recognition, etc.

[0131] Step S320: presenting the background and the target;

[0132] Figure 3B A visual page of a generation tool provided in an embodiment of the present application, such as Figure 3B As shown, the page may include a setting area 31 and a generated object 32, wherein:

[0133] In the setting area 31, the background and target obtained by analyzing the input information can be presented.

[0134] In some embodiments, the user may be allowed to directly add, delete, or modify the text corresponding to the background and target.

[0135] Step S330: present at least one of the following items to be adjusted: style, tone, audience, and response.

[0136] Here, Style is used to specify the desired writing style. You can specify the writing style of a particular famous person or industry expert, such as a business analyst or a writer. This can guide the generative model to respond in a manner and vocabulary selection that meets the requirements. For example, the style can be "write a speech about future technology products in the style of Steve Jobs."

[0137] Tone, which is used to set the emotional tone of the response. Setting an appropriate tone ensures that the generated model's response is consistent with the expected emotional or sentimental context. Possible tones include formal, humorous, sympathetic, etc. For example, you can ask the generated model to "write a welcome letter to new employees in an encouraging tone."

[0138] Audience, used to identify the target audience. Tailoring the generated model's responses to a specific audience, whether it is an expert in the field, a beginner, or a child, ensures that the content is appropriate and easy to understand in a specific context. For example, you can specify "write a popular science article about the solar system for elementary school students."

[0139] Response, which is used to specify the format of the output. The purpose of determining the output format is to ensure that the generated model outputs according to specific requirements and facilitates the execution of downstream tasks. Common formats include lists, JSON formatted data, professional reports, etc. For example, you can ask the generated model to "list 5 must-read books on artificial intelligence in the form of a list."

[0140] During the implementation process, a corresponding selection item may be set for each adjustment item so that the user can select the adjustment item that is adapted to the generated intention.

[0141] In some embodiments, at least one of the items to be adjusted: style, tone, audience, and response is presented in a graphical form.

[0142] Figure 3B In the setting area 31 shown in the figure, for the generation style, tone, etc., a graphical adjustment method can be provided, allowing the user to slide or select; for the audience and response, etc., a graphical and customized adjustment method is provided, allowing the user to select. The adjustment method includes at least the following two: completely manual conditions, that is, each item is manually clicked and selected; adaptive changes follow the selection of a certain dimension.

[0143] Different adjustment items can be set for text generation objects of different generation categories. For example, the style of the generated popular science articles will be concise, story-telling, in-depth analysis, etc.; the tone will be friendly and encouraging, authoritative and credible, humorous, serious and formal; the target audience is middle school students, college students and professionals, public readers, science enthusiasts, etc.; the response includes questions and answers, dialogues, etc. Another example is a project application, which can have completely different adjustment items.

[0144] During the implementation process, the adjustment items can be set in a flexible and diverse way, which not only supports manual modification of each dimension, but also can adapt to the unified modification of other dimensions according to a certain dimension. For example, when it is determined that the audience of the popular science article is elementary school students, the style, tone, response method, etc. are automatically suggested and can be fine-tuned.

[0145] In the embodiment of the present application, the input information is first parsed to obtain the background and target of the target generation object; then the background and the target are presented; and at least one of the following items to be adjusted is presented: style, tone, audience, and response. In this way, the user can make adaptive adjustment suggestions based on the input information and the requirements of the generated object to obtain a generated object that better meets the user's generation needs.

[0146] In some embodiments, the target generation object is a generated text. After the target generation object is generated, at least one of the following steps may be performed:

[0147] Step S340, in response to the adding instruction, adding new text to the generated text;

[0148] For example, at the beginning of a popular science article, a term is defined. To make it easy to read, the author needs to add examples. Figure 3B The generated object 32 shown shows [Example], inserting an example at a specific position. The optional operations include, but are not limited to: adding new content, that is, the user simply writes down the new content to be added, which can be a new topic, and after submission, the text can be added according to the user's requirements.

[0149] Step S350, in response to the deletion instruction, deleting the corresponding text in the generated text;

[0150] In some embodiments, the user can delete certain sentences, after which the content segment represented by the deleted sentences is also deleted.

[0151] Step S360: In response to the modification instruction, modify the generated text.

[0152] During the implementation process, users can write down modification requirements, and the statements for which modification requirements are written can be rewritten according to user requirements.

[0153] Figure 3B The generated object 32 shown shows [Examples, Explanations], and context can be inserted according to the user's selection to determine the content that needs examples and explanations, and generate a new paragraph.

[0154] In the embodiment of the present application, in response to the add instruction, new text is added to the generated text; in response to the delete instruction, the corresponding text is deleted from the generated text; in response to the modify instruction, the generated text is modified. In this way, the location where the rewriting or expansion occurs can be accurately controlled to meet the user's editing needs for the generated object, so that the generated object that meets the user's needs can be generated again.

[0155] Figure 4 A schematic diagram of a process of generating an image provided in an embodiment of the present application is shown in FIG. Figure 4 As shown, the method can be implemented by the following steps:

[0156] Step S410: User input;

[0157] Here, the user can enter a piece of text information or an original image.

[0158] Step S420, native prompt understanding;

[0159] When a user inputs a long text, the system can first segment the input text into long sentences and then extract structured elements to parse the prompt information. For example, the system can clarify the theme of the user's raw image based on the output text, such as portraits, landscapes, product presentations, etc.

[0160] Step S430: Pattern matching;

[0161] During the implementation process, the corresponding Pattern (scenario-based spell formula) can be matched according to the theme of the picture, and different raw picture patterns can be matched according to the theme. For example, the landscape pattern [*image setting] + [*theme content + environment] + [*composition / perspective] + [lighting effect] + [*style].

[0162] Step S440, related prompt retrieval;

[0163] Based on the topic content, similar content is retrieved from the scenario-based prompt library dedicated to the generated model.

[0164] During the implementation process, an exclusive high-quality prompt database can be obtained through the following steps: labeling and classification of the prompt database; high-quality prompt screening, retaining useful and effective prompts through basic rules and raw image results; high-quality prompt patterning, building an exclusive prompt database

[0165] Step S450, Prompt rewrite / expand;

[0166] According to the Pattern item, the keyworks corresponding to similar prompts are integrated to present a structured result.

[0167] Fill in the content according to the pattern. One part is described by the user, and the other part is to expand the topic-related content based on the Creatorzone's exclusive high-quality Prompt library, and other blank dimensions are automatically filled in.

[0168] Step S460: User modification / confirmation;

[0169] Step S470: Image generation.

[0170] In the embodiments of the present application, based on a large number of high-quality prompt words and scenario-based patterns, Prompt is intelligently expanded to effectively improve the quality of raw images; the Prompt library is selected according to the capabilities of the generation model, and content that is not good at generating is avoided as much as possible; the raw image pattern introduces a user confirmation process to narrow the gap between the blind box raw image results and the user's expected raw image results, thereby improving the experience of the raw image process.

[0171] Based on the foregoing embodiments, an embodiment of the present application provides a generating device, which includes the modules included, each module includes each sub-module, each sub-module includes a unit, which can be implemented by a processor in an electronic device; of course, it can also be implemented by a specific logic circuit; in the implementation process, the processor can be a central processing unit (CPU), a microprocessor (MPU), a digital signal processor (DSP) or a field programmable gate array (FPGA), etc.

[0172] Figure 5 A schematic diagram of the structure of the generating device provided in the embodiment of the present application is shown in FIG. Figure 5 As shown, the device 500 includes:

[0173] A presentation module 510, configured to present first output information to be confirmed based on the input information;

[0174] A modification or confirmation module 520, configured to modify or confirm the first output information in response to the target instruction, and obtain second output information for generating a target generation object of a preset type;

[0175] The information processing module 530 is used to input the second output information into the target generation model and output the target generation object.

[0176] In some embodiments, the presentation module 510 includes a first parsing submodule and a filling and presentation submodule, wherein the first parsing submodule is used to parse the input information to obtain at least one filling information matching the target generation template, wherein the target generation template is a generation template matching the input information; the filling and presentation submodule is used to fill in the target generation template based on the at least one filling information and at least one generation information, and present the first output information after the information is filled in, wherein the generation information is generated based on the target generation template.

[0177] In some embodiments, the presentation module 510 also includes a second parsing submodule and a first acquisition submodule, wherein the second parsing submodule is used to parse the input information and determine the subject information to be generated; and the first acquisition submodule is used to acquire the target generation template based on the subject information to be generated.

[0178] In some implementations, the second parsing submodule includes a first determination unit, a cutting unit, and an extraction unit, wherein the first determination unit is used to determine that the input information is text information; the cutting unit is used to perform long sentence cutting on the text information to obtain at least one target sentence; and the extraction unit is used to perform structured element extraction on each of the target sentences to determine the topic information to be generated.

[0179] In some embodiments, the second parsing submodule includes a second determination unit, an identification unit and a third determination unit, wherein the second determination unit is used to determine that the input information is image information; the identification unit is used to identify the image information and obtain an image theme corresponding to the image information; and the third determination unit is used to determine the theme information to be generated based on the image theme.

[0180] In some embodiments, the target generation template includes multiple prompt information items, and the filling and presentation sub-module includes a fourth determination unit, a fifth determination unit and a filling unit, wherein the fourth determination unit is used to determine a target prompt information item that matches the at least one filling information among the multiple prompt information items; the fifth determination unit is used to determine at least one of the generation information from a target prompt information database based on at least one of the target prompt information items and the multiple prompt information items, wherein the target prompt information database is constructed based on the target generation model; the filling unit is used to fill at least one of the filling information and at least one of the generation information into the multiple prompt information items to obtain the first output information.

[0181] In some embodiments, the generating device further includes an acquisition module, a classification module and a screening module, wherein the acquisition module is used to acquire a source prompt information database; the classification module is used to classify each prompt information in the source prompt information database to obtain a classified source prompt information database; the screening module is used to screen the classified source prompt information database based on the target generation model to obtain the target prompt information database.

[0182] In some embodiments, the generating device further includes a first determination module and a second determination module, wherein the first determination module is used to determine the weight of each of the prompt information items based on the contribution of each of the prompt information items to the quality improvement of generating the target generating object; the second determination module is used to determine the arrangement order of all of the prompt information items based on the weight of each of the prompt information items; the filling unit is also used to fill in at least one of the filling information and at least one of the generating information into the multiple prompt information items in sequence based on the arrangement order to obtain the first output information.

[0183] In some embodiments, the presentation module 510 is also used to present multiple first output information based on the input information; correspondingly, the modification or confirmation module 520 includes a second acquisition submodule and a determination submodule, wherein the second acquisition submodule is used to acquire a target instruction for determining the second output information, wherein the generation information in the multiple first output information is different; and the determination submodule is used to determine one second output information in the multiple first output information in response to the target instruction.

[0184] In some embodiments, the filling and presenting submodule is further used to present the first output information in a structured form.

[0185] In some embodiments, the presentation module 510 includes a third parsing submodule, a first presentation submodule and a second presentation submodule, wherein the third parsing submodule is used to parse the input information to obtain the background and target of the target generation object; the first presentation submodule is used to present the background and the target; and the second presentation submodule is used to present at least one of the following items to be adjusted: style, tone, audience and response.

[0186] In some embodiments, the second presentation submodule is further used to present at least one of the following items to be adjusted in a graphical form: style, tone, audience, and response.

[0187] In some embodiments, the target generation object is a generated text, and the generation device also includes an adding module, a deleting module and a modifying module, wherein the adding module is used to add new text to the generated text in response to an adding instruction; the deleting module is used to delete corresponding text in the generated text in response to a deleting instruction; and the modifying module is used to modify the generated text in response to a modifying instruction.

[0188] The description of the above device embodiment is similar to the description of the above method embodiment, and has similar beneficial effects as the method embodiment. For technical details not disclosed in the device embodiment of the present application, please refer to the description of the method embodiment of the present application for understanding.

[0189] It should be noted that in the embodiment of the present application, if the above method is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiment of the present application is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable an electronic device (which can be a mobile phone, a tablet computer, a laptop computer, a desktop computer, etc.) to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a U disk, a mobile hard disk, a read-only memory (ROM), a disk or an optical disk. In this way, the embodiment of the present application is not limited to any specific combination of hardware and software.

[0190] Correspondingly, an embodiment of the present application provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps in the generation method provided in the above embodiment are implemented.

[0191] Correspondingly, an embodiment of the present application provides an electronic device, Figure 6 A hardware entity diagram of an electronic device provided in an embodiment of the present application, such as Figure 6 As shown, the hardware entity of the device 600 includes: a memory 601 and a processor 602, wherein the memory 601 stores a computer program that can be run on the processor 602, and the processor 602 implements the steps in the generation method provided in the above embodiment when executing the program.

[0192] The memory 601 is configured to store instructions and applications executable by the processor 602, and can also cache data to be processed or processed by the processor 602 and various modules in the electronic device 600 (for example, image data, audio data, voice communication data, and video communication data), which can be implemented through flash memory (FLASH) or random access memory (Random Access Memory, RAM).

[0193] It should be noted here that the description of the above storage medium and device embodiments is similar to the description of the above method embodiments, and has similar beneficial effects as the method embodiments. For technical details not disclosed in the storage medium and device embodiments of this application, please refer to the description of the method embodiments of this application for understanding.

[0194] It should be understood that "one embodiment" or "an embodiment" mentioned throughout the specification means that specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present application, the size of the sequence number of the above-mentioned processes does not mean the order of execution, and the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application. The above-mentioned sequence numbers of the embodiments of the present application are only for description and do not represent the advantages and disadvantages of the embodiments.

[0195] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0196] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0197] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0198] In addition, all functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may be a separate unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0199] A person skilled in the art can understand that all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM), magnetic disks or optical disks, etc., various media that can store program codes.

[0200] Alternatively, if the above-mentioned integrated unit of the present application is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application can be essentially or partly reflected in the form of a software product that contributes to the relevant technology. The computer software product is stored in a storage medium, including several instructions to enable an electronic device (which can be a mobile phone, a tablet computer, a laptop computer, a desktop computer, etc.) to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0201] The methods disclosed in several method embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments.

[0202] The features disclosed in several product embodiments provided in this application can be arbitrarily combined without conflict to obtain new product embodiments.

[0203] The features disclosed in several method or device embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments or device embodiments.

[0204] The above is only an implementation method of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A generation method, the method comprising: Based on the input information, presenting first output information to be confirmed; In response to the target instruction, modify or confirm the first output information to obtain second output information for generating a target generation object of a preset type; The second output information is input into a target generation model, and the target generation object is output.

2. The method according to claim 1, wherein presenting first output information to be confirmed based on the input information comprises: Parsing the input information to obtain at least one filled-in information matching a target generation template, wherein the target generation template is a generation template matching the input information; The target generation template is filled in based on the at least one filling information and the at least one generation information, and the first output information after the information filling is completed is presented, wherein the generation information is generated based on the target generation template.

3. The method of claim 2, further comprising: Parsing the input information to determine the subject information to be generated; The target generation template is obtained based on the subject information to be generated.

4. The method according to claim 3, wherein the step of parsing the input information and determining the subject information to be generated comprises: Determining that the input information is text information; Performing long sentence segmentation on the text information to obtain at least one target sentence; Structural element extraction is performed on each of the target sentences to determine the topic information to be generated.

5. The method according to claim 2, wherein the target generation template comprises a plurality of prompt information items, and the step of filling in the target generation template based on the at least one filling information and the at least one generation information comprises: Determining a target prompt information item that matches the at least one filled-in information from among the multiple prompt information items; Determining at least one of the generation information from a target prompt information database based on at least one of the target prompt information items and the plurality of prompt information items, wherein the target prompt information database is constructed based on the target generation model; Fill in at least one of the filled-in information and at least one of the generated information into the multiple prompt information items to obtain the first output information.

6. The method of claim 5, further comprising: Obtain source prompt information database; Classify each prompt information in the source prompt information database to obtain a classified source prompt information database; The classified source prompt information database is screened based on the target generation model to obtain the target prompt information database.

7. The method of claim 5, further comprising: Determining a weight of each of the prompt information items based on a contribution of each of the prompt information items to improving the quality of generating the target generation object; Determining the arrangement order of all the prompt information items based on the weight of each of the prompt information items; Correspondingly, filling at least one of the filled-in information and at least one of the generated information into the multiple prompt information items to obtain the first output information includes: At least one of the filling information and at least one of the generating information are sequentially filled into the plurality of prompt information items based on the arrangement order to obtain the first output information.

8. The method according to any one of claims 2 to 7, wherein presenting first output information to be confirmed based on the input information comprises: Based on the input information, present a plurality of the first output information; Correspondingly, in response to the target instruction, modifying or confirming the first output information to obtain second output information for generating a target generation object of a preset type includes: acquiring a target instruction for determining the second output information, wherein the generation information in the plurality of first output information is different; In response to the target instruction, one second output information is determined from a plurality of first output information.

9. The method according to claim 1, wherein presenting first output information to be confirmed based on the input information comprises: Parsing the input information to obtain the background and target of the target generation object; presenting the background and the objectives; Present at least one of the following items to be adjusted: style, tone, audience, and response.

10. A generating device, comprising: A display, for presenting first output information to be confirmed based on the input information; A processor, configured to modify or confirm the first output information in response to the target instruction, and obtain second output information for generating a target generation object of a preset type; The processor is further configured to input the second output information into a target generation model and present the target generation object on the display.

Images