CN116091667B - Character artistic image generation system based on AIGC technology - Google Patents
Character artistic image generation system based on AIGC technology Download PDFInfo
- Publication number
- CN116091667B CN116091667B CN202310200611.XA CN202310200611A CN116091667B CN 116091667 B CN116091667 B CN 116091667B CN 202310200611 A CN202310200611 A CN 202310200611A CN 116091667 B CN116091667 B CN 116091667B
- Authority
- CN
- China
- Prior art keywords
- character
- generation
- image
- information
- target
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T13/00—Animation
- G06T13/20—3D [Three Dimensional] animation
- G06T13/40—3D [Three Dimensional] animation of characters, e.g. humans, animals or virtual beings
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
- G06F40/295—Named entity recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T19/00—Manipulating 3D models or images for computer graphics
- G06T19/20—Editing of 3D images, e.g. changing shapes or colours, aligning objects or positioning parts
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/174—Facial expression recognition
Abstract
The invention provides a character artistic image generating system based on AIGC technology; the generation system trains an AIGC core of the generation system with preset information comprising the target character image description, and marks at least one expected characteristic of the target character in the preset information; the generation system learns and trains the related characteristics of the generated target character by taking the reference information as a training material; and, including providing the relationship of two or more target persons to the generation system and marking at least one feature of the target persons as an countermeasure feature as a feature of maximum relevance or dissimilarity of the target persons; further, the generation system takes the countermeasure feature as a starting point of generation, performs the generation steps of the character images according to preset generation logic, finally efficiently generates a large number of character images with vivid personality and strong similarity or relativity, and is suitable for animation, films and other artistic creation requirements comprising a large number of character images.
Description
Technical Field
The invention relates to the field of artificial intelligence processing equipment. And more particularly, to a character art image generation system based on an AIGC technique.
Background
AIGC (Artificial Intelligence Generated Content) is artificial intelligence generation content, and the technology can be used for creating various artistic contents, including text contents, image pictures and music; meanwhile, based on the creativity of the character image, the character image can be performed by using the character image. Through machine learning and deep learning algorithms, the AIGC can generate character figures with realistic look and behavior. The technology can be used in the fields of movies, games, advertisements, animations and the like, and can save the manufacturing cost and time. Meanwhile, with the rapid increase of the requirements of the current film and game on the picture tricks, the AIGC is used for assisting in generating a plurality of character images based on specific targets, so that a great amount of working time and cost for manually designing the character images in the past are expected to be saved, the creation of scenes is more massive, and the content is richer or more creative and random character images and contents are expected to be created.
According to the disclosed technical scheme, the technical scheme with the publication number of CN111768334A provides that each part of the five sense organs is searched in a database and compared for multiple times, so that the five sense organs are continuously optimized and adjusted in the design process; the technical proposal with publication number of WO2013120453A1 provides a natural human digital image design system, which divides a character image into a plurality of classification libraries of subdivision characteristics, and a designer selects and adjusts each characteristic and then performs matching combination so as to form detailed image design; the technical solution disclosed in WO2011123802A1 proposes a system for three-dimensionally designing an animated character, wherein the character is three-dimensionally designed by inputting a plurality of designs of two-dimensional planes through an animator.
The above technical solutions all refer to a design and generation system of the character image, however, the design itself needs to consume a lot of manpower resource cost, and has extremely high requirements on originality; at present, under the high-speed development of artificial intelligence technology, the application expansion of the image design field is assisted by the technology of related aspects.
The foregoing discussion of the background art is intended to facilitate an understanding of the present invention only. This discussion is not an admission or admission that any of the material referred to was common general knowledge.
Disclosure of Invention
The invention aims to provide a character artistic image generating system based on AIGC technology; the generation system trains an AIGC core of the generation system with preset information comprising the target character image description, and marks at least one expected characteristic of the target character in the preset information; the generation system learns and trains the related characteristics of the generated target character by taking the reference information as a training material; and, including providing the relationship of two or more target persons to the generation system and marking at least one feature of the target persons as an countermeasure feature as a feature of maximum relevance or dissimilarity of the target persons; further, the generation system takes the countermeasure feature as a starting point of generation, performs the generation steps of the character images according to preset generation logic, finally efficiently generates a large number of character images with vivid personality and strong similarity or relativity, and is suitable for animation, films and other artistic creation requirements comprising a large number of character images.
The invention adopts the following technical scheme:
an AIGC technology-based character art image generation system, the generation system comprising a processor and a memory for storing instructions and information data; the instructions, when executed by the processor, cause the generation system to perform:
acquiring preset information and reference information, and transmitting the information to a memory;
analyzing the target person in the preset information, determining at least one characteristic of the target person, and determining a characteristic value of the at least one characteristic of the target person;
identifying information related to the target person in the reference information through a preset identification model and based on the characteristics of the target person, and extracting a plurality of sub-parameters based on the characteristic parameters in the related information to generate a sub-parameter set corresponding to each characteristic parameter;
generating a target character image;
verifying the rationality and creativity of the generated character image;
the generation system comprises a step of analyzing the connection of two or more target characters and determining the logical similarity relationship and the dissimilarity relationship of the two or more target characters;
selecting one or more characteristics of the two or more target character images as countermeasure characteristics, and setting characteristic values of the countermeasure characteristics so as to accord with the logical similarity or dissimilarity relation of the two or more target character images;
preferably, the generating system comprises one or more processors and a memory, and the processors and the memory are respectively configured to run the generating module and the judging module; wherein the method comprises the steps of
The generation module is used for generating the image of the target person;
the judging module is used for evaluating the similarity between the character image generated by the generating module and the existing character image, so that feedback information is provided to guide the generating module to generate the character image closer to the requirement;
preferably, the preset information and the reference information include: one or more of text information, video information, image information and audio information;
preferably, the generation system includes analyzing character features of a target character specified in the reference information of the character type based on a natural language processing technique, and includes analyzing at least one feature having a maximum weight ratio of the plurality of features of the target character from among the plurality of features of the target character, and setting the at least one feature having the maximum weight ratio as the countermeasure feature;
preferably, the generating system includes analyzing character features of a target character specified in the reference information of video and image types based on artificial intelligence vision techniques, and includes analyzing at least one feature having a maximum weight ratio of a plurality of features of the target character from among the plurality of features of the target character, and setting the at least one feature having the maximum weight ratio as the countermeasure feature;
preferably, the generating system generates the target character image based on preset generating logic in a specified data structure and sequence; the generation logic specifies a generation order of the individual avatar characteristics in generating the avatar; in the generation logic, each image feature is used as a node, and a plurality of image features form continuous generation logic by a net structure; and, in addition, the processing unit,
the generation logic comprises a step of allowing nodes with any one characteristic to serve as a character generation starting point and completing the generation of characters along the connection sequence of the network structures of the nodes;
preferably, in the generating logical mesh structure, one or more nodes are set as key nodes; the characteristic represented by the key node is a key characteristic; and is also provided with
The challenge feature may be only one of the key features of one or more; and is also provided with
Including setting up the generation logic may only start from any one of the critical nodes.
The beneficial effects obtained by the invention are as follows:
the generation system adopts AIGC technology, can quickly grasp the characteristics of the target character image through a large amount of training and feedback, and carries out related creation, thereby saving a large amount of time and labor cost;
the generation system can generate a large number of image groups with high identification degree and strong relevance aiming at images with obvious relevant or dissimilar characteristics by setting the countermeasure characteristics in the system;
the generation system of the invention executes the image generation sequence based on the set generation logic, so that after the key features are determined, other image features can be pertinently expanded and refined according to the key features, thereby being beneficial to generating the image with more vivid characteristics and higher integrity;
the generating system adopts modularized design for each software and hardware part, thereby being convenient for upgrading or replacing related software and hardware environments in the future and reducing the use cost.
Drawings
The invention will be further understood from the following description taken in conjunction with the accompanying drawings. The components in the figures are not necessarily to scale, emphasis instead being placed upon illustrating the principles of the embodiments. Like reference numerals designate corresponding parts throughout the different views.
FIG. 1 is an overall schematic of a generating system according to the present invention;
FIG. 2 is a schematic diagram of a character generation sequence according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a node network generating logic in an embodiment of the present invention;
FIG. 4 is a schematic diagram of another node network for generating logic in accordance with an embodiment of the present invention;
fig. 5 is a schematic diagram of a node network including generation logic for a node group in an embodiment of the present invention.
Reference numerals illustrate: 12-generating a computing device; 14-a processor; 16-volatile memory; 18-an input/output module; a 20-communication bus; 24-non-volatile memory; 26-deep learning neural network; 28-a generation module; 30-a judging module; 41-preset information; 42-reference information.
Detailed Description
In order to make the technical scheme and advantages of the present invention more apparent, the present invention will be further described in detail with reference to the following examples thereof; it should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention. Other systems, methods, and/or features of the present embodiments will be or become apparent to one with skill in the art upon examination of the following detailed description. It is intended that all such additional systems, methods, features and advantages be included within this description. Included within the scope of the invention and protected by the accompanying claims. Additional features of the disclosed embodiments are described in, and will be apparent from, the following detailed description.
The same or similar reference numbers in the drawings of embodiments of the invention correspond to the same or similar components; in the description of the present invention, it should be understood that, if any, the terms "upper," "lower," "left," "right," and the like indicate an orientation or a positional relationship based on the orientation or the positional relationship shown in the drawings, this is for convenience of description and simplification of the description, and does not indicate or imply that the apparatus or component to be referred to must have a specific orientation. The terms describing the positional relationship in the drawings are merely for illustrative purposes and are not to be construed as limiting the present patent, and specific meanings of the terms are understood by those of ordinary skill in the art according to specific circumstances.
Embodiment one: an AIGC technology-based character art image generation system, the generation system comprising a processor and a memory for storing instructions and information data; the instructions, when executed by the processor, cause the generation system to perform:
acquiring preset information and reference information, and transmitting the information to a memory;
analyzing the target person in the preset information, determining at least one characteristic of the target person, and determining a characteristic value of the at least one characteristic of the target person;
identifying information related to the target person in the reference information through a preset identification model and based on the characteristics of the target person, and extracting a plurality of sub-parameters based on the characteristic parameters in the related information to generate a sub-parameter set corresponding to each characteristic parameter;
generating a target character image;
verifying the rationality and creativity of the generated character image;
the generation system comprises a step of analyzing the connection of two or more target characters and determining the logical similarity relationship and the dissimilarity relationship of the two or more target characters;
selecting one or more characteristics of the two or more target character images as countermeasure characteristics, and setting characteristic values of the countermeasure characteristics so as to accord with the logical similarity or dissimilarity relation of the two or more target character images;
preferably, the generating system comprises one or more processors and a memory, and the processors and the memory are respectively configured in the operation work of the generating module and the judging module; wherein the method comprises the steps of
The generation module is used for generating the image of the target person;
the judging module is used for evaluating the similarity between the character image generated by the generating module and the existing character image, so that feedback information is provided to guide the generating module to generate the character image closer to the requirement;
preferably, the preset information and the reference information include: text information, video information, image information, and audio information;
preferably, the generation system includes analyzing character features of a target character specified in the reference information of the character type based on a natural language processing technique, and includes analyzing at least one feature having a maximum weight ratio of the plurality of features of the target character from among the plurality of features of the target character, and setting the at least one feature having the maximum weight ratio as the countermeasure feature;
preferably, the generating system includes analyzing character features of a target character specified in the reference information of video and image types based on artificial intelligence vision techniques, and includes analyzing at least one feature having a maximum weight ratio of a plurality of features of the target character from among the plurality of features of the target character, and setting the at least one feature having the maximum weight ratio as the countermeasure feature;
preferably, the generating system generates the target character image based on preset generating logic in a specified data structure and sequence; the generation logic specifies a generation order of the individual avatar characteristics in generating the avatar; in the generation logic, each image feature is used as a node, and a plurality of image features form continuous generation logic by a net structure; and, in addition, the processing unit,
the generation logic comprises a step of allowing nodes with any one characteristic to serve as a character generation starting point and completing the generation of characters along the connection sequence of the network structures of the nodes;
preferably, in the generating logical mesh structure, one or more nodes are set as key nodes; the characteristic represented by the key node is a key characteristic; and is also provided with
The challenge feature may be only one of the key features of one or more; and is also provided with
Including setting that the generation logic can only start from any one of the critical nodes;
as shown in fig. 1, an exemplary architecture diagram of the generation system is shown; the generation system includes a generation computing device 12, the generation computing device 12 including a processor 14, a volatile memory 16, an input/output module 18, and a non-volatile memory 24; the non-volatile memory 24 is used to store data, applications or other necessary information required by the generation system; wherein the application program comprises any algorithm or program used when the generation system generates the character image; and preferably, the application program includes a deep learning neural network 26 for running based on AIGC technology, and the generating module and the discriminating module perform the character image generating step of the generating system based on the deep learning neural network 26; and the deep learning neural network 26 operates as a hardware principal for generating the computing device 12 to operate;
in some embodiments, the learning, analyzing and identifying of the preset information 41 and the reference information 42 may also be applied to the deep learning neural network 26, which is not limited herein;
further, included in the generating computing device 12 is a communication bus 20 that may operatively couple the processor 14, the input/output module 18, and the volatile memory 16 to a non-volatile memory 24; while the deep learning neural network 26 is described as being hosted (i.e., executed) at one generating computing device 12, it should be understood that the deep learning neural network 26 may alternatively be hosted across multiple generating computing devices, the generating computing device 12 being communicatively coupled to other multiple generating computing devices through a network;
wherein processor 14 is a microprocessor, which may be one or more of a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), an Application Specific Integrated Circuit (ASIC), a system on a chip (SOC), a Field Programmable Gate Array (FPGA), logic circuitry, or other suitable type of microprocessor configured to perform the functions described herein;
further, the volatile memory 16 may be, for example, a Random Access Memory (RAM), a Static Random Access Memory (SRAM), a Dynamic Random Access Memory (DRAM), or the like, which temporarily stores data only during program execution and loses a memory function after stopping power supply support; in some examples, non-volatile random access memory (NVRAM) may be used;
preferably, the nonvolatile memory 24 is a memory capable of holding instruction storage data even in the absence of externally applied power, such as a flash memory, a hard disk, a Read Only Memory (ROM), an electrically erasable programmable memory (EEPROM), or the like; included in the non-volatile memory 24 are programs of instructions for the generating system to perform the operations described herein, as well as data used by these programs to perform the operations described herein, such as storing preset information, reference information, and identification models for performing analysis of the reference information described herein;
based on the above hardware and software settings, the generating computing device 12 performs artificial intelligence calculations to implement an computing platform for Artificial Intelligence Generated Content (AIGC);
in some embodiments, natural language processing techniques applied to artificial intelligence for analysis of literal materials; the method or principle comprises the following steps:
text analysis: analyzing a section of text material by the technologies of word segmentation, part-of-speech tagging, named entity recognition and the like, and extracting related information such as person names, behaviors, emotions and the like;
entity identification: and identifying the person entity in the text according to the entity information in the text, and extracting the entity relationship. Through analyzing the relationship among the people, the interaction relationship, character characteristics and the like among the people can be further known;
emotion analysis: through identifying and analyzing emotion vocabularies in the text, the emotion tendency and emotion state of the person are known, wherein the emotion tendency and emotion state comprise emotion positive and negative, emotion intensity, emotion polarity and the like;
feature extraction: extracting various characteristics of the character image according to character information and emotion characteristics extracted from the text, wherein the characteristics comprise appearance, character, experience, social status and the like;
and (3) data mining: and (5) summarizing and analyzing the extracted characteristics to obtain important characteristics and subjects of the character image. Through a data mining technology, the relation and the difference between different character images can be mined, and the internal characteristics of the character images can be further known;
in some embodiments, the characteristics of the character's image in the image material may be analyzed through artificial intelligence, wherein the techniques include object detection, face recognition, facial expression recognition, pose estimation, and the like; the target detection technology can identify people in the image and mark the positions of the people; the face detection technology can further detect the face of the person, locate and extract the face characteristics; the face recognition technology can match the characters in the image with a pre-marked character library to determine the identities of the characters; facial expression recognition techniques can analyze the expression of a person to infer their emotional state; pose estimation techniques may detect the body pose of a person, infer their motion or state;
the technology can be used singly or in combination, and the characteristics of the character image can be analyzed and identified according to different application scenes and requirements; for example, in video, the position of a target person may be determined from a plurality of persons in a screen using a face recognition technique and continued attention may be paid; alternatively, facial expression recognition techniques may be used to determine the emotional response of a person to a scene, event, or item; alternatively, gesture estimation techniques may be used to identify the actions of the target person for corresponding response and control;
further, computer vision techniques used primarily for artificial intelligence analysis of image material include, but are not limited to, the following:
object detection: by training a model, identifying objects in the image, commonly used algorithms are RCNN, fast RCNN, YOLO and the like;
face recognition: comparing the face in the image with the face in the database through the technologies of face detection, alignment, feature extraction and the like to determine the identity;
posture estimation: by analyzing the image, the gesture information of the person is deduced, and the method can be used for application such as action recognition, human body segmentation and the like;
human body segmentation: separating the person in the image from the background so as to better understand the information such as the position, the shape and the like of the person;
behavior recognition: information such as motion trail and gesture of the person is extracted through video analysis, and behavior of the person is identified, wherein common algorithms include CNN, LSTM and the like;
target tracking: deducing information such as the position, the speed and the like of the object by tracking the object in the image sequence so as to realize the function of target tracking;
image segmentation: dividing the image into a plurality of portions to better understand objects in the image, such as decorations, clothing, tattoos, etc. in the character image;
these techniques typically require extensive data and computational resources to train the model and analyze, and thereby form a model that is applied to identify the characters in the images of the characters in the various types of information, and quantify the characteristics to produce adjustments to the parameters and their weights in the deep neural network;
accordingly, when the reference information 42 of a large number of character information and the preset information 41 including the target information of the character to be generated are inputted to the generation system to be recognized, the generation system will perform analysis of the character, wherein the main features include:
sex, age, facial characteristics, hairstyle; and may also include apparel and apparel, and may include gestures and expressions, such as smiling, sad faces, and the like; these features may train the recognition model by providing the image dataset and tag information to the AIGC; the recognition model may learn the features and recognize the extent to which the features represent;
thus, when a desired feature is identified from the preset information including the target person, for example, a certain person, or an occupational image of the person, or a certain type of hairstyle of the person, etc., information related to the target person in the reference information can be identified based on the desired person and the feature thereof;
for example, in some embodiments, the target persona is a particular individual; the generating system marks a plurality of characteristics of the target person from the preset information, and then can identify information related to the person, such as facial features, from the reference information, so as to learn more decoration habit, dressing habit, behavior action habit and the like;
in some embodiments, the target character is a pupil and the facial features of the target character are ignored; the generating system marks a plurality of characteristics of the character of the pupil, such as height or clothes, from preset information, so that more information about the character of the pupil is identified from the reference information and a great deal of learning is performed;
in some embodiments, the target person is an original human; the generation system marks the shape, action and other characteristics of the original human from preset information, identifies the original human from different types of clans from reference information, and performs characteristic identification and difference identification on the original human from different types of clans; and generating a set of correlated sub-parameters from a plurality of features of the original human from the different types of clans; for example, for height features, the parameter values thereof may correspond to a plurality of height values; for hair color, various color values may be corresponded to, and so forth;
the deep learning neural network 26 learns potential patterns and laws by analyzing a large amount of sample data, and builds a generation model for generating a character image; the deep learning neural network 26 comprises two main components, namely a generating module 28 and a discriminating module 30; the generation module uses random noise or other input data as input to generate a new artwork; the judging module evaluates the similarity between the artwork generated by the generating module and the real artwork, so that feedback information is provided to guide the generating module to generate better artwork; in the training process, the two components are continuously and iteratively optimized until the generation module can generate vivid works of art;
in some embodiments, when it is desired to generate a large number of similar character images, such as a large number of people, passers-by, the large number of character images have similar features, such as wearing daily clothing, and no obvious expression; when a large number of people images with high similarity are generated, daily clothes and facial expressions can be used as countermeasure features, and feature values of the two features can be kept to have small difference values; when the image is generated, daily clothes or facial expressions can be used as a starting point of the generation logic, other parts are generated to form an integral image, so that the integral similarity in style is reflected while the multiple human image finally has certain difference;
in some embodiments, when it is desired to generate character images of two populations in a relative resistant state (e.g., opposing derivatives of forces, two different teams of comparison), one may use his clothing, body accessories or expressions as a challenge feature, set the polarity of the challenge feature in the two populations to be of opposite nature, and thereby generate other features; for example, one of the clothes is selected to be in a cool tone style, and the other clothes is selected to be in a warm tone style; or one face is more clean, the opposite face is more dark, and the like;
it should be appreciated that the challenge feature may be self-determined by the AIGC system after screening among a plurality of visual features; the AIGC system may also provide a plurality of countermeasure features, which may be determined by a second screening by the relevant executive technician.
Embodiment two: this embodiment should be understood to include at least all of the features of any one of the foregoing embodiments, and further improvements thereto:
in some embodiments, the generation sequence of each feature when the character image is generated needs to be set so as to unify the logic of the generation operation and reduce the conflict in the generation operation; here, the generation order is specified as generation logic; common generation logic may be:
(1) Setting basic information and characteristics of a person, such as gender, age, height, body type, skin color, facial characteristics and the like;
(2) Determining the clothing and accessories of the person, including the clothing style, color, pattern, and accessories such as hats, glasses, shoes, etc.;
(3) Selecting a hairstyle and a color of the character, taking into account coordination with facial features and clothing;
(4) Setting the expression and the gesture of the character, including eye spirit, mouth shape, arm gesture and the like, so as to increase the sense of reality and emotion expression of the character image;
as shown in fig. 2, in generating the character, the figure is in accordance with (a) the basic figure; (b) facial five sense organs, (c) hairstyle, clothing; (d) a garment pattern, a facial contour, a lower body garment; (e) overall color matching to complete the complete generation of the image;
however, the generation logic is not fixed and may vary depending on the use scenario, application purpose, etc.; overall, however, the various aspects of the character are gradually set according to the logic, so that the accuracy and the controllability of the generated result can be improved;
the order of generating the logic may use the features as nodes, and connect the nodes to form a network-like relationship graph; as shown in fig. 3, taking the node 31 as an example, its neighboring nodes include 32a, 32b, 32c, etc.; and these nodes are continuously connected and can eventually close to form a closed network; when the generation logic is adopted, any one node (namely, one image characteristic) in the network can be used as a starting point for image generation and is continuously expanded to other nodes;
in yet other embodiments, as shown in FIG. 4, not every node in the network that generates the logic has the same number of neighboring nodes; wherein the node 41 has 4 adjacent nodes 42a, 42b, 42c, 42d; whereas node 43 has only two adjacent nodes 42a, 42d; in particular, in the process of generating the image, certain features are not randomly connected with any subsequent features to generate, but have certain constraints.
Embodiment III: this embodiment should be understood to include at least all of the features of any one of the foregoing embodiments, and further improvements thereto:
in some embodiments, a number of key nodes are set by the relevant technician in the generation logic to represent the characteristics of a number of critical images, and only one of the selected key nodes is allowed to serve as a starting point for generation when the generation logic is required to run;
also taking fig. 4 as an example; the key node may be set as node 41 or 43, while the remaining nodes cannot serve as a starting point for generating logic;
the key characteristics represented by the key nodes can be determined according to the style characteristics of the image; for example, for a character image focusing on clothing as an important feature, clothing can be used as a key feature and a corresponding key node;
in some embodiments, a plurality of nodes may form a node group, and the plurality of nodes may form a complete network of generation logic; as shown in fig. 5, the node group 51 exists as one node that generates logic, and the node group 51 may include a network having sub-logic therein, the sub-logic network having a plurality of nodes connected to an external network to be connected to the nodes 52a, 52b, 52c, 52d; the node group 51 may be, for example, a jacket portion in the avatar, and the generation of the jacket portion serves as sub-logic to construct the inside of the node group 51; features of various parts of the coat part such as collar, sleeve, chest, back, etc. as nodes inside the node group 51 constitute sub-logics of the coat part;
through setting up the node group, can enclose more sub-characteristics in the characteristic and cover and accomplish the formation after, carry out subsequent image formation again to form reasonable image result.
In the foregoing embodiments, the descriptions of the embodiments are emphasized, and in part, not described or illustrated in any particular embodiment, reference is made to the related descriptions of other embodiments.
While the invention has been described above with reference to various embodiments, it should be understood that many changes and modifications can be made without departing from the scope of the invention. That is, the methods, systems and devices discussed above are examples. Various configurations may omit, replace, or add various procedures or components as appropriate. For example, in alternative configurations, the methods may be performed in a different order than described, and/or various components may be added, omitted, and/or combined. Moreover, features described with respect to certain configurations may be combined in various other configurations, such as different aspects and elements of the configurations may be combined in a similar manner. Furthermore, as the technology evolves, elements therein may be updated, i.e., many of the elements are examples, and do not limit the scope of the disclosure or the claims.
Specific details are given in the description to provide a thorough understanding of exemplary configurations involving implementations. However, configurations may be practiced without these specific details, e.g., well-known circuits, processes, algorithms, structures, and techniques have been shown without unnecessary detail in order to avoid obscuring configurations. This description provides only an example configuration and does not limit the scope, applicability, or configuration of the claims. Rather, the foregoing description of the configuration will provide those skilled in the art with an enabling description for implementing the described techniques. Various changes may be made in the function and arrangement of elements without departing from the spirit or scope of the disclosure.
It is intended that the foregoing detailed description be regarded as illustrative rather than limiting, and that it be understood that it is intended that it be regarded as illustrative rather than limiting. Various changes and modifications to the present invention may be made by one skilled in the art after reading the teachings herein, and such equivalent changes and modifications are intended to fall within the scope of the invention as defined in the appended claims.
Claims (3)
1. An AIGC technology-based character art image generation system, wherein the generation system comprises a processor and a memory for storing instructions and information data; the instructions, when executed by the processor, cause the generation system to perform:
acquiring preset information and reference information, and transmitting the information to a memory;
analyzing the target person in the preset information, determining at least one characteristic of the target person, and determining a characteristic value of the at least one characteristic of the target person;
identifying information related to the target person in the reference information through a preset identification model and based on the characteristics of the target person, and extracting a plurality of sub-parameters based on the characteristic parameters in the related information to generate a sub-parameter set corresponding to each characteristic parameter;
generating a target character image;
verifying the rationality and creativity of the generated character image;
the generation system is also used for analyzing the connection of more than two target characters and determining the logical similarity relationship and the logical dissimilarity relationship of the more than two target characters;
selecting one or more features of the two or more target character images as countermeasure features, and setting feature values of the countermeasure features so as to accord with the logical similarity or dissimilarity relation of the two or more target character images;
the processor and the memory are more than one, and the processor and the memory are used for the operation work of the generation module and the discrimination module;
the generation module is used for generating the image of the target person;
the judging module is used for evaluating the similarity between the character image generated by the generating module and the existing character image, so that feedback information is provided to guide the generating module to generate the character image closer to the requirement;
the generation system is further configured to analyze, based on artificial intelligence vision techniques, image features of a target character specified in the reference information of video and image types, analyze at least one feature having a largest feature weight ratio of the target character image from among a plurality of features of the target character image, and set the at least one feature having the largest weight ratio as the countermeasure feature;
the generation system is used for generating the target character image based on preset generation logic in a specified data structure and sequence; the generation logic specifies a generation order of the individual avatar characteristics in generating the avatar; in the generation logic, each image feature is used as a node, and a plurality of image features form continuous generation logic by a net structure; and the generation logic includes a step of allowing nodes of any one of the features to be used as a character generation starting point to complete the generation of characters along the connection order of the network structure of each node;
setting more than one node as a key node in the generating logic mesh structure; the characteristic represented by the key node is a key characteristic; and setting the generation logic may only start from any one of the key nodes.
2. The generating system of claim 1, wherein the preset information and the reference information include: text information, video information, image information, and audio information.
3. The generation system of claim 2, wherein the generation system analyzes character features of the target character specified in the reference information of the character type based on a natural language processing technique, analyzes at least one feature having a largest feature weight ratio of the target character from among a plurality of features of the target character, and sets the at least one feature having the largest weight ratio as the countermeasure feature.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310200611.XA CN116091667B (en) | 2023-03-06 | 2023-03-06 | Character artistic image generation system based on AIGC technology |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310200611.XA CN116091667B (en) | 2023-03-06 | 2023-03-06 | Character artistic image generation system based on AIGC technology |
Publications (2)
Publication Number | Publication Date |
---|---|
CN116091667A CN116091667A (en) | 2023-05-09 |
CN116091667B true CN116091667B (en) | 2023-06-20 |
Family
ID=86212165
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310200611.XA Active CN116091667B (en) | 2023-03-06 | 2023-03-06 | Character artistic image generation system based on AIGC technology |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116091667B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117150089B (en) * | 2023-10-26 | 2023-12-22 | 环球数科集团有限公司 | Character artistic image changing system based on AIGC technology |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115168745A (en) * | 2022-09-02 | 2022-10-11 | 环球数科集团有限公司 | Virtual character image recreating method and system based on image technology |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110706300A (en) * | 2019-09-19 | 2020-01-17 | 腾讯科技(深圳)有限公司 | Virtual image generation method and device |
CN111260763A (en) * | 2020-01-21 | 2020-06-09 | 厦门美图之家科技有限公司 | Cartoon image generation method, device, equipment and storage medium based on portrait |
CN113793256A (en) * | 2021-09-10 | 2021-12-14 | 未鲲(上海)科技服务有限公司 | Animation character generation method, device, equipment and medium based on user label |
CN114377405A (en) * | 2021-12-29 | 2022-04-22 | 完美世界(北京)软件科技发展有限公司 | Role image data-based generation method and device |
CN115423908A (en) * | 2022-08-19 | 2022-12-02 | 深圳市达旦数生科技有限公司 | Virtual face generation method, device, equipment and readable storage medium |
-
2023
- 2023-03-06 CN CN202310200611.XA patent/CN116091667B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115168745A (en) * | 2022-09-02 | 2022-10-11 | 环球数科集团有限公司 | Virtual character image recreating method and system based on image technology |
Also Published As
Publication number | Publication date |
---|---|
CN116091667A (en) | 2023-05-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zhang et al. | Fusing geometric features for skeleton-based action recognition using multilayer LSTM networks | |
CN110785767B (en) | Compact linguistics-free facial expression embedding and novel triple training scheme | |
Lassner et al. | A generative model of people in clothing | |
Liu et al. | Learning hierarchical cross-modal association for co-speech gesture generation | |
CN111626218B (en) | Image generation method, device, equipment and storage medium based on artificial intelligence | |
US20210174072A1 (en) | Microexpression-based image recognition method and apparatus, and related device | |
Ludl et al. | Enhancing data-driven algorithms for human pose estimation and action recognition through simulation | |
US20190164055A1 (en) | Training neural networks to detect similar three-dimensional objects using fuzzy identification | |
CN110909680A (en) | Facial expression recognition method and device, electronic equipment and storage medium | |
CN111222486A (en) | Training method, device and equipment for hand gesture recognition model and storage medium | |
CN116091667B (en) | Character artistic image generation system based on AIGC technology | |
Nojavanasghari et al. | Hand2face: Automatic synthesis and recognition of hand over face occlusions | |
Cui et al. | Facial expression recognition based on ensemble of mulitple CNNs | |
Roy et al. | Tips: Text-induced pose synthesis | |
RU2755396C1 (en) | Neural network transfer of the facial expression and position of the head using hidden position descriptors | |
CN116050284B (en) | Fashion redesign system utilizing AIGC technology | |
Gutiérrez et al. | An ontology of virtual humans: incorporating semantics into human shapes | |
Tripathi et al. | Facial expression recognition using data mining algorithm | |
Aishwarya et al. | The Analogy of Haar Cascade and HOG Approaches for Facial Emotion Recognition | |
Khan et al. | Real-Time American Sign Language Realization Using Transfer Learning With VGG Architecture | |
Brown | Faster upper body pose recognition and estimation using Compute Unified Device Architecture | |
Adholiya et al. | Visualizing Sentiments: Facial Expression Recognition in the World of Kathak Dance Using Deep Learning | |
Krishna et al. | A NOVE FACIAL EXPRESSION RECOGNITION SYSTEM USING CONVOLUTIONAL NEURAL NETWORKS | |
Pikula et al. | FlexComb: a facial landmark-based model for expression combination generation | |
Yang et al. | Interactive facial-geometric-feature animation for generating expressions of novel faces |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |