Special effect generation method and apparatus, device, and storage medium

Abstract

Embodiments of the present application disclose a special effect generation method, device and equipment and a storage medium. The method comprises: obtaining special effect description information; determining element description information corresponding to each visual element included in a target particle special effect according to the special effect description information; for each visual element, searching for a candidate particle emitter that satisfies a first semantic matching condition with the visual element in a target database according to the element description information corresponding to the visual element, as a target particle emitter corresponding to the visual element, and the target database stores a plurality of candidate particle emitters and emitter semantic data corresponding to each of the candidate particle emitters; and generating the target particle special effect based on the target particle emitters corresponding to each of the visual elements. The method can improve the generation efficiency of the particle special effect and reduce the professional requirements for special effect producers.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, apparatus, device and storage medium for generating special effects. Background Technology

[0002] Particle effects are a part of computer graphics that simulate the visual effects of natural or supernatural phenomena such as fire, smoke, explosions, and magical light effects by controlling the movement, color, lifecycle, and other attributes of a large number of tiny primitives (particles). In game development, film and television post-production, and digital twins, particle effects are an indispensable element in building immersive visual experiences.

[0003] In related technologies, particle effects are primarily generated manually using the particle effects editor provided by the game engine. For example, VFX artists manually add particle emitters in the particle effects editor (Cascade / Niagara) provided by Unreal Engine (UE), configure the parameters of the particle emitters, and then use the configured particle emitters to create the desired particle effects. However, generating particle effects in this way is extremely inefficient and requires a very high level of expertise from the VFX artists; the quality of the generated particle effects depends entirely on the individual skill level of the VFX artist. Summary of the Invention

[0004] This application provides a method, apparatus, device, and storage medium for generating special effects, which can improve the generation efficiency of particle special effects and reduce the professional requirements for special effects production personnel.

[0005] The first aspect of this application provides a method for generating special effects, the method comprising:

[0006] Obtain special effects description information, which is used to describe the target particle special effects to be generated;

[0007] Based on the special effect description information, determine the element description information corresponding to each visual element included in the target particle special effect;

[0008] For each visual element, according to the element description information corresponding to the visual element, a candidate particle emitter that satisfies the first semantic matching condition with the visual element is retrieved in the target database and used as the target particle emitter corresponding to the visual element. The target database stores multiple candidate particle emitters and their respective corresponding emitter semantic data.

[0009] The target particle effect is generated based on the target particle emitter corresponding to each of the visual elements.

[0010] A second aspect of this application provides a special effects generation apparatus, the apparatus comprising:

[0011] The information acquisition module is used to acquire special effect description information, which describes the target particle special effect to be generated;

[0012] The element determination module is used to determine the element description information corresponding to each visual element included in the target particle effect based on the effect description information.

[0013] The emitter retrieval module is used to, for each visual element, retrieve candidate particle emitters that satisfy the first semantic matching condition between the visual element and the visual element in the target database according to the element description information corresponding to the visual element, and use them as the target particle emitters corresponding to the visual element. The target database stores multiple candidate particle emitters and their respective corresponding emitter semantic data.

[0014] The special effects generation module is used to generate the target particle special effects based on the target particle emitter corresponding to each of the visual elements.

[0015] A third aspect of this application provides a computer device, the device comprising a processor and a memory:

[0016] The memory is used to store computer programs;

[0017] The processor is configured to perform the steps of the special effects generation method as described in the first aspect above, according to the computer program.

[0018] A fourth aspect of this application provides a computer-readable storage medium for storing a computer program for performing the steps of the special effects generation method described in the first aspect.

[0019] A fifth aspect of this application provides a computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps of the special effects generation method described in the first aspect.

[0020] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0021] This application provides a method for generating special effects, which innovatively proposes a scheme for automatically generating particle special effects based on user-provided natural language description information. The method includes: acquiring special effect description information to describe the target particle special effect to be generated; analyzing the visual elements that should be included in the target particle special effect and their corresponding element description information based on the special effect description information; then, searching a target database, which stores multiple candidate particle emitters and their corresponding emitter semantic data; for each visual element, searching the target database for candidate particle emitters that satisfy a first semantic matching condition with the visual element based on the element description information corresponding to that visual element, that is, searching the target database for candidate particle emitters whose corresponding emitter semantic data semantically matches the element description information corresponding to the visual element, and using these as the target particle emitters corresponding to that visual element; finally, combining the target particle emitters corresponding to each visual element to generate a complete target particle special effect. Therefore, using the above method, users only need to provide a description of the desired target particle effect in natural language. The system can then automatically retrieve target particle emitters that semantically match the visual elements in the target particle effect. Based on the target particle emitters corresponding to each visual element, the system generates target particle effects that meet the user's needs. This process eliminates the need for users to manually configure complex parameters, thus significantly improving the generation efficiency of particle effects and reducing the professional requirements for users. Furthermore, since the candidate particle emitters stored in the target database all have superior visual effects, constructing target particle effects based on candidate particle emitters retrieved from the target database can also, to a certain extent, ensure that the target particle effects have superior visual effects. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 An architecture diagram of a computer system provided in an embodiment of this application;

[0024] Figure 2 This is a schematic diagram illustrating an application scenario of the special effects generation method provided in the embodiments of this application;

[0025] Figure 3 A flowchart illustrating the special effects generation method provided in this application embodiment;

[0026] Figure 4A schematic diagram illustrating the construction of a target database provided in an embodiment of this application;

[0027] Figure 5 A schematic diagram illustrating another method for constructing a target database, as provided in an embodiment of this application;

[0028] Figure 6 A schematic diagram illustrating the determination of element evaluation parameters corresponding to visual elements in an embodiment of this application;

[0029] Figure 7 A schematic diagram illustrating the generation of target particle effects provided in an embodiment of this application;

[0030] Figure 8 A schematic diagram of the special effects creation panel provided in the embodiments of this application;

[0031] Figure 9 A schematic diagram of a special effects preview window provided for an embodiment of this application;

[0032] Figure 10 This is a schematic diagram of the structure of the special effects generation method provided in the embodiments of this application;

[0033] Figure 11 This is a schematic diagram of the special effects generation device provided in the embodiments of this application;

[0034] Figure 12 This is a schematic diagram of the structure of the terminal device provided in the embodiments of this application;

[0035] Figure 13 This is a schematic diagram of the server structure provided in an embodiment of this application. Detailed Implementation

[0036] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0037] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0038] It should be noted that this application may display prompt interfaces, pop-ups, or output voice prompts before and during the collection of user data. These prompt interfaces, pop-ups, or voice prompts are used to inform the user that their data is being collected. This ensures that the application only begins the steps for collecting user data after receiving confirmation from the user regarding the prompt interface or pop-up; otherwise (i.e., without user confirmation), the steps for collecting user data end, meaning no user data is collected. In other words, all user data collected in this application is collected with the user's consent and authorization, and the collection, use, and processing of related user data must comply with the relevant laws, regulations, and standards of the relevant countries and regions.

[0039] Currently, in related technologies, particle effects are typically generated manually using the particle effects editor provided by the game engine. For example, effects artists manually add particle emitters in the particle effects editor provided by UE and configure their parameters, such as the Spawn Rate, Lifetime, Velocity, and Color over Life. Then, the desired particle effects are created using the configured particle emitters.

[0040] However, generating particle effects in the above way is extremely inefficient. Special effects artists often need to configure hundreds of parameter options, which places extremely high demands on their professionalism. Furthermore, the quality of the generated particle effects depends entirely on the aesthetic sense and professional skills of the special effects artists. For non-professional special effects artists, it is impossible to generate particle effects smoothly through parameter configuration.

[0041] In related technologies, particle effects can also be generated by editing preset effect templates. For example, special effects creation tools typically provide preset effect templates (such as templates for generating explosion effects). Users can adjust the macroscopic parameters (such as size or color) in the preset effect templates. Correspondingly, the script automatically randomizes the underlying microscopic parameters. Based on the adjusted macroscopic parameters and the randomized microscopic parameters, particle effects can be generated. The parameters of the final generated particle effect can be represented as P. final =P base +Random( ), P final P represents the set of parameters for the final generated particle effects. base This represents the adjusted macroscopic parameter, Random( ) represents the randomized micro-parameters.

[0042] However, particle effects generated based on preset effect templates are prone to having a single form of expression and not meeting the actual needs of users. That is, the final generated particle effects are not much different from the particle effects corresponding to the preset effect templates. For example, it is impossible to generate a spark effect with lightning based on an effect template that indicates pure sparks.

[0043] To address the problems existing in the aforementioned related technologies, this application provides a special effects generation method. In this method, special effects description information describing the target particle special effects to be generated is obtained. Based on the special effects description information, the visual elements that should be included in the target particle special effects and their corresponding element description information are analyzed. Next, a search is performed based on a target database, which stores multiple candidate particle emitters and their corresponding emitter semantic data. For each visual element, based on the element description information corresponding to that visual element, a candidate particle emitter that satisfies a first semantic matching condition with that visual element is retrieved from the target database. That is, a candidate particle emitter whose corresponding emitter semantic data semantically matches the element description information corresponding to the visual element is retrieved from the target database and used as the target particle emitter corresponding to that visual element. Finally, the target particle emitters corresponding to each visual element are combined to generate a complete target particle special effect. Therefore, using the above method, users only need to provide a description of the desired target particle effect in natural language. The system can then automatically retrieve target particle emitters that semantically match the visual elements in the target particle effect. Based on the target particle emitters corresponding to each visual element, it generates target particle effects that meet the user's needs. This process eliminates the need for users to manually configure complex parameters, significantly improving the efficiency of particle effect generation and reducing the professional requirements on users. Furthermore, compared to particle effects generated using preset effect templates, the method provided in this application can generate richer particle effects that meet the user's actual needs. That is, accurate target particle effects can be generated based on the effect description information provided by the user. In addition, since the candidate particle emitters stored in the target database all have superior visual effects, constructing target particle effects based on candidate particle emitters retrieved from the target database can also, to a certain extent, ensure that the target particle effects have superior visual effects.

[0044] To facilitate understanding of the special effects generation method provided in the embodiments of this application, the computer system used to implement the special effects generation method will be described exemplarily below.

[0045] See Figure 1 , Figure 1 An architecture diagram of a computer system provided in an embodiment of this application. Figure 1 The computer system 101 described herein is a system architecture for implementing the special effects generation method in the embodiments of this application. The computer system 101 includes a terminal device 120, a server 140, and a database 160.

[0046] The terminal device 120 has a client application installed and running that supports generating target particle effects. Users can input effect description information through the client application, and then display the generated target particle effects. This target application can be, for example, a game engine application; however, this embodiment does not limit the specific target application. The client application can be, for example, an application (App) or mini-program on the terminal device 120, or it can be a webpage, or a tool embedded within a game engine (such as a UE engine). This embodiment does not limit the form of the client application on the terminal device 120.

[0047] Terminal device 120 can refer to one of a plurality of terminal devices, but this embodiment only uses terminal device 120 as an example. The device types of terminal device 120 include, but are not limited to, at least one of the following: smartphone, tablet computer, wearable device, laptop computer, desktop computer, smart home device, or smart vehicle device.

[0048] Those skilled in the art will understand that the number of terminal devices 120 described above can be more or less. For example, there may be only one terminal device 120, or there may be multiple or more terminal devices 120. This application does not limit the number or type of terminal devices 120.

[0049] Terminal device 120 can be connected to server 140 via wireless network or wired network.

[0050] Server 140 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud servers, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0051] For example, server 140 includes a processor and a memory, wherein the memory is used to store related computer programs, such as a computer program for executing the special effects generation method in the embodiments of this application, and the processor is used to read the computer program from the memory to execute the special effects generation method in the embodiments of this application.

[0052] Specifically, the processor can acquire effect description information, which describes the target particle effect to be generated. Based on the effect description information, the processor can determine the element description information corresponding to each visual element included in the target particle effect. For each visual element, the processor can retrieve candidate particle emitters that satisfy the first semantic matching condition with the visual element from the target database (i.e., database 160) based on the element description information corresponding to the visual element, and use these as the target particle emitters corresponding to the visual element. Finally, the processor can generate the target particle effect based on the target particle emitters corresponding to each visual element.

[0053] Optionally, server 140 undertakes the main computing work and terminal device 120 undertakes the secondary computing work; or, server 140 undertakes the secondary computing work and terminal device 120 undertakes the main computing work; or, server 140 and terminal device 120 adopt a distributed computing architecture for collaborative computing.

[0054] Database 160 is a storage device used to store and manage data according to preset rules. In this embodiment, database 160 is used to store multiple candidate particle emitters and their corresponding emitter semantic data. Server 140 can retrieve candidate particle emitters in database 160 that satisfy a first semantic matching condition with a visual element, and use them as target particle emitters corresponding to the visual element. This embodiment does not limit the content stored and managed by database 160 in any way.

[0055] Those skilled in the art will understand that the number of databases 160 described above can be more or less. For example, there may be only one database 160, or there may be multiple or more databases 160. This application embodiment does not limit the number of databases 160 or the type of device. Databases 160 can be connected to server 140 via a wireless network or a wired network, or databases 160 can be integrated inside server 140.

[0056] See Figure 2 , Figure 2 This is a schematic diagram illustrating an application scenario of the special effects generation method provided in this application embodiment. The method is executed by a computer device, which may be, for example, a computer device... Figure 1 The server 140 shown. The steps of the special effects generation method performed by server 140 are as follows:

[0057] Server 140 can obtain effect description information 201, which describes the target particle effect to be generated. For example, a user can input the effect description information in the client of the target application on terminal device 120. In response to a confirmation operation triggered by the input effect description information, the client can generate a particle effect generation request, which includes the effect description information. Then, terminal device 120 can send the particle effect generation request to server 140. After receiving the request, server 140 can obtain the effect description information from it.

[0058] Next, server 140 can determine the element description information corresponding to each visual element included in the target particle effect based on the effect description information 201. For example, server 140 can analyze the effect description information 201 using a Large Language Model (LLM) to obtain the visual elements included in the target particle effect and their corresponding element description information. For example, it can obtain the first visual element 202, the second visual element 203, ..., the Nth visual element 204 included in the target particle effect, as well as the element description information 205 corresponding to the first visual element, the element description information 206 corresponding to the second visual element, ..., the element description information 207 corresponding to the Nth visual element.

[0059] Then, for each visual element, server 140 can retrieve candidate particle emitters that satisfy the first semantic matching condition with the visual element from the target database based on the element description information corresponding to the visual element. These candidate emitters will be used as the target particle emitters corresponding to the visual element. The target database (i.e., database 160) stores multiple candidate particle emitters and their corresponding emitter semantic data. For example, taking the retrieval of the target particle emitter corresponding to the first visual element 202 as an example, the element description information 205 corresponding to the first visual element can be semantically encoded to obtain the corresponding semantic feature vector. Then, the similarity between the semantic feature vector and the emitter semantic data corresponding to each candidate particle emitter in the target database can be calculated to select k candidate particle emitters with high similarity as the target particle emitters 208 corresponding to the first visual element.

[0060] It should be noted that this application does not specifically limit the number of target particle emitters corresponding to a visual element; each visual element may correspond to one or more target particle emitters.

[0061] Finally, target particle effects are generated based on the target particle emitters corresponding to each visual element. For example, following the steps above, the target particle emitters corresponding to each visual element can be determined, such as the target particle emitter 208 corresponding to the first visual element, the target particle emitter 209 corresponding to the second visual element, ..., and the target particle emitter 210 corresponding to the Nth visual element. Furthermore, the emission source points of the target particle emitter 208 corresponding to the first visual element, the target particle emitter 209 corresponding to the second visual element, ..., and the target particle emitter 210 corresponding to the Nth visual element can all be set as the origin of the target coordinate system. Based on this, by combining the target particle emitters 208, 209, ..., and 210 corresponding to the first visual element and the Nth visual element from the same emission source point, a complete target particle effect 211 can be generated.

[0062] Therefore, using the above method, users only need to provide a description of the desired target particle effect in natural language. The system can then automatically retrieve target particle emitters that semantically match the visual elements in the target particle effect. Based on the target particle emitters corresponding to each visual element, the system generates target particle effects that meet the user's needs. This process eliminates the need for users to manually configure complex parameters, thus significantly improving the generation efficiency of particle effects and reducing the professional requirements for users. Furthermore, since the candidate particle emitters stored in the target database all have superior visual effects, constructing target particle effects based on candidate particle emitters retrieved from the target database can also, to a certain extent, ensure that the target particle effects have superior visual effects.

[0063] It should be understood that Figure 2 The application scenarios shown are merely examples. In practical applications, the special effects generation method provided in this application embodiment can also be applied to other scenarios. No limitation is made here on the application scenarios of the special effects generation method provided in this application embodiment.

[0064] The special effects generation method provided in this application will be described in detail below through method embodiments.

[0065] See Figure 3 , Figure 3 This is a flowchart illustrating the special effects generation method provided in an embodiment of this application. This special effects generation method can be executed by a computer device; for ease of description, the following description uses a server as the executing entity of this special effects generation method. Figure 3 As shown, the method for generating this special effect includes the following steps:

[0066] S301: Obtain special effects description information, which is used to describe the target particle special effects to be generated.

[0067] The target particle effect refers to the generated target in the embodiments of this application, which is a particle effect that meets the user's needs. This application does not specifically limit the target particle effect.

[0068] Special effects description information refers to natural language information provided by the user to describe the target particle special effects to be generated. For example, it can be description information entered by the user through a text input control, such as "yellow circular shock wave, orange flame burning, looping", etc. Alternatively, it can include description information entered by the user through a text input control and style information configured by the user (information used to indicate the style of the target particle special effects). This application does not specifically limit the special effects description information.

[0069] For example, a target application supporting particle effect generation runs on the terminal device. This target application provides an effects creation panel, which includes text input controls and style configuration controls. Users can input descriptive information in the text input controls and select style information for the target particle effect using the style configuration controls. Based on this descriptive information and style information, an effect description can be constructed. In response to a confirmation operation triggered by the input descriptive information and the selected style information, the terminal device can generate a particle effect generation request. This request includes the effect description information determined based on the descriptive information and style information. Subsequently, the particle effect generation request can be sent to a server. Upon receiving the request, the server can retrieve the effect description information from it.

[0070] S302: Based on the special effects description information, determine the element description information corresponding to each visual element included in the target particle special effects.

[0071] Visual elements refer to the components of a target particle effect, which can be distinguished and described separately. That is, a particle effect can include multiple visual elements. For example, the "explosion" effect includes visual elements such as "core light sphere", "sparks flying outward", and "residual smoke". In this regard, this application does not specifically limit the visual elements included in the target particle effect.

[0072] Element description information refers to information extracted from special effects description information that is used to describe the effects of a visual element itself. This application does not specifically limit the element description information.

[0073] For example, a large language model can be used to perform semantic understanding and visual element decomposition of special effect description information. Specifically, a prompt message can be constructed based on the special effect description information. This prompt message instructs the user to perform a task of decomposing visual elements and their corresponding element description information according to the special effect description information, so that the large language model can decompose the various visual elements included in the target particle special effect and their respective element description information through the prompt message. For example, the prompt message could be: decompose the following special effect description information input by the user into independent visual elements constituting this particle special effect and their respective element description information. The element description information is returned in the format of a list, and the special effect description information is [yellowish-brown circular shockwave, orange flame burning, loop]. This application does not specifically limit the prompt message.

[0074] After determining the prompt information, the prompt information is input into the large language model for processing. The large language model can decompose the visual elements included in the target particle effect and their corresponding element description information based on the prompt information. For example, taking the effect description information included in the prompt information as "yellow circular shock wave, orange flame burning, loop", the output result obtained by the large language model can be: q_elem[0]: "circular shock wave: yellowish-brown spreading outward, looping", q_elem[1]: "flame: orange burning upward, looping", where q_elem[0] represents the element description information corresponding to the 0th visual element in the target particle effect, that is, the 0th visual element is "circular shock wave", and its corresponding element description information is "yellowish-brown spreading outward, looping", and q_elem[1] represents the element description information corresponding to the 1st visual element in the target particle effect, that is, the 1st visual element is "flame", and its corresponding element description information is "orange burning upward, looping".

[0075] The large language model may include, but is not limited to, Gemini, Deepseek, and Generative Pre-trained Transformer (GPT) models. This application does not specifically limit the model structure of the large language model.

[0076] S303: For each visual element, based on the element description information corresponding to the visual element, retrieve candidate particle emitters that satisfy the first semantic matching condition with the visual element from the target database, and use them as the target particle emitters corresponding to the visual element. The target database stores multiple candidate particle emitters and their respective emitter semantic data.

[0077] Candidate particle emitters refer to particle emitters stored in the target database. They can be extracted from historical particle effects or artificially constructed. This application does not specifically limit the method of determining candidate particle emitters.

[0078] Among them, a particle emitter refers to the basic building block of a particle system, used to define the generation rate, initial position, speed, and material properties of particle effects. A complete particle effect is usually composed of multiple particle emitters. For example, an "explosion" effect is based on a combination of particle emitters that realize the "core light sphere", particle emitters that realize the "sparks flying outward", and particle emitters that realize the "residual smoke".

[0079] The emitter semantic data corresponding to the candidate particle emitter refers to the semantic feature vector corresponding to the candidate particle emitter. It can be obtained by semantically encoding the visual effect description text of the candidate particle emitter. In this regard, this application does not specifically limit the method of determining the emitter semantic data.

[0080] The target database refers to a pre-built database used to store multiple candidate particle emitters and their corresponding emitter semantic data.

[0081] The first semantic matching condition is the condition used when retrieving the target particle emitter corresponding to a visual element in the target database. Specifically, it is used to indicate the retrieval of candidate particle emitters with high semantic similarity to the visual element as the target particle emitter corresponding to that visual element. For example, the first semantic matching condition can specify selecting the k candidate particle emitters with the highest similarity between the corresponding emitter semantic data and the semantic feature vector of the element description information as the target particle emitters corresponding to the visual element. Here, k can be a fixed parameter, or k can be a dynamic parameter. For example, the value of k can be determined by a formula for determining the number of dynamic retrievals. In this regard, this application does not specifically limit the first semantic matching condition and the value of k.

[0082] The target particle emitter refers to the candidate particle emitter that is obtained through retrieval and is relatively matched with the semantic information of the currently judged visual element. All k candidate particle emitters obtained through retrieval can be used as the target particle emitter corresponding to the judged visual element. In this regard, this application does not specifically limit the target particle emitter or its number.

[0083] For example, for each visual element, semantic encoding can be performed based on the element description information corresponding to the visual element to obtain the semantic feature vector corresponding to the visual element. Then, based on the semantic feature vector corresponding to the visual element, candidate particle emitters that satisfy the first semantic matching condition with the semantic feature vector of the visual element are retrieved from the target database. For example, a preset similarity calculation method (such as the cosine similarity calculation method) can be used to calculate the similarity between the emitter semantic data corresponding to each candidate particle emitter stored in the target database and the semantic feature vector of the visual element, so as to retrieve the k candidate particle emitters with the highest similarity with the semantic feature vector of the visual element, and the retrieved k candidate particle emitters are used as the target particle emitters corresponding to the visual element.

[0084] For example, a pre-defined text encoder can be used to convert the element description information corresponding to a visual element into a corresponding semantic feature vector. Here, a text encoder refers to a model that performs semantic feature encoding on the element description information corresponding to a visual element. It is used to convert the element description information corresponding to a visual element into a vector that can represent its semantic information. For example, a text encoder can include, but is not limited to, embedding models (such as the OpenAI text-embedding-3 model), bidirectional and auto-regressive transformer models (BERT), and sentence-level BERT models (Sentence BidirectionalEncoder Representations from Transformers, Sentence-BERT), etc. This application does not specifically limit the text encoder.

[0085] It should be understood that the aforementioned transmitter semantic data can also be semantically encoded in the manner described above. Simply replace the object of the semantic encoding with the visual effect description text of the candidate particle transmitter. This application will not elaborate on the semantic encoding process of the transmitter semantic data here.

[0086] In practical applications, the target database can also store multiple candidate particle effects and their corresponding semantic data. The candidate particle emitters in the target database are extracted from the candidate particle effects. When retrieving the target particle emitter corresponding to a visual element, the search scope can be narrowed first based on the effect description information. For example, based on the semantic feature vector of the effect description information, a preliminary search can be conducted in the target database for candidate particle effects that semantically match the effect description information, which can then be used as reference particle effects. Furthermore, within the narrowed search scope, the target particle emitter corresponding to the visual element can be retrieved. For example, based on the semantic feature vector corresponding to the visual element, among the candidate particle emitters extracted from the reference particle effects stored in the target database, the k candidate particle emitters with the highest similarity to the semantic feature vector of the visual element can be retrieved, and these k candidate particle emitters can be used as the target particle emitters corresponding to the visual element. This application does not specifically limit the method for determining the target particle emitter corresponding to the visual element.

[0087] S304: Generate target particle effects based on the target particle emitters corresponding to each visual element.

[0088] Following step S303 above, the target particle emitter corresponding to each visual element can be determined. Then, the emission source point of the target particle emitter corresponding to each visual element can be set to a unified origin (such as the origin of the target coordinate system). Based on this unified origin, the target particle emitters corresponding to each visual element can be combined to generate a complete target particle effect.

[0089] In the special effects generation method provided in this application embodiment, the method innovatively proposes a scheme for automatically generating particle special effects based on user-provided natural language information description. Through this method, the user only needs to provide a natural language description of the desired target particle special effect, and the system can automatically retrieve target particle emitters that semantically match the visual elements in the target particle special effect. Then, based on the target particle emitters corresponding to each visual element, the system generates target particle special effects that meet the user's needs. This process eliminates the need for the user to manually configure complex parameters, thereby significantly improving the generation efficiency of particle special effects and reducing the professional requirements on the user. Furthermore, since the candidate particle emitters stored in the target database all have superior visual effects, constructing target particle special effects based on the candidate particle emitters retrieved from the target database can also, to a certain extent, ensure that the target particle special effect has superior visual effects.

[0090] How the target database is constructed:

[0091] In one possible implementation, the target database can be constructed through the following steps S11-S14 (not shown in the figure):

[0092] S11: Obtain multiple candidate particle effects.

[0093] Candidate particle effects are the particle effects used to determine candidate particle emitters, serving as the basis for subsequent retrieval and extraction of candidate particle emitters.

[0094] For example, all particle effect asset files in a historical effects project library (such as a game project effects database or a database of historically created effects files) can be traversed to obtain multiple candidate particle effects; that is, any particle effect included in the particle effect asset files can be used as a candidate particle effect. This application does not specifically limit the method of obtaining candidate particle effects.

[0095] S12: Extract candidate particle emitters from each candidate particle effect.

[0096] For example, a preset parser can be used to extract candidate particle emitters from each candidate particle effect. For example, for each candidate particle effect, a parser can be written to read the particle effect asset file corresponding to the candidate particle effect in binary or text format, so as to extract each candidate particle emitter included in the candidate particle effect as an independent unit.

[0097] S13: For each candidate particle emitter, render the emitter frame sequence corresponding to the candidate particle emitter, and determine the emitter description information corresponding to the candidate particle emitter based on the emitter frame sequence.

[0098] An emitter frame sequence refers to a sequence of multiple images captured in chronological order after a candidate particle emitter has been running for a period of time. It is used to display the visual effects of the candidate particle emitter, such as the particle's trajectory, color changes, and dissipation method.

[0099] The emitter description information is used to describe the visual effect of the candidate particle emitter. For example, the emitter description information can be "a particle emitter that sprays orange flames outward, with the flames spreading in a cone shape and accompanied by a small number of sparks." This application does not specifically limit the emitter description information.

[0100] For example, for each candidate particle emitter, the candidate particle emitter can be deployed in the game engine, and the emitter frame sequence corresponding to the candidate particle emitter can be rendered in a no-light scene. That is, the candidate particle emitter is started and runs a complete life cycle (or preset duration) in a no-light scene. Then, each frame can be captured at a fixed frame rate and saved as an image sequence, thus obtaining the emitter frame sequence corresponding to the candidate particle emitter.

[0101] Next, the Vision Language Model (VLM) can be used to determine the emitter description information corresponding to the candidate particle emitter based on the emitter frame sequence. For example, a cue message can be constructed based on the emitter frame sequence. This cue message instructs the emission of the emitter description information to be generated, enabling the Vision Language Model to generate the emission description information corresponding to the candidate particle emitter using the cue message. For example, the cue message could be "describe the visual effects of this candidate particle emitter, including the particle's shape, color, texture, trajectory, and lifespan," etc. This application does not specifically limit the cue message.

[0102] After determining the prompt information, the prompt information is input into the visual language large model for processing. This process yields the emitter description information corresponding to the candidate particle emitters generated by the visual language large model based on the prompt information. The visual language large model may include, but is not limited to, the Qwen3 Vision-Language (Qwen3-VL) model, Deepseek, and Gemini 2.5, etc. This application does not specifically limit the visual language large model.

[0103] S14: For each candidate particle emitter, determine the emitter semantic data corresponding to the candidate particle emitter based on the emitter description information, and store the candidate particle emitter and its corresponding emitter semantic data in the target database.

[0104] Emitter semantic data is used to characterize the semantic information of the emitter description information corresponding to the candidate particle emitter. It can be represented as a semantic feature vector. This application does not specifically limit the form of the emitter semantic data.

[0105] For example, for each candidate particle emitter, the semantic feature vector corresponding to the visual element can be determined by the text encoder according to the emitter description information corresponding to the candidate particle emitter, in the same way as the above method for determining the semantic feature vector corresponding to the visual element. This will not be elaborated further here.

[0106] After determining the emitter semantic data corresponding to each candidate particle emitter, the candidate particle emitter and its corresponding emitter semantic data can be stored in the target database. An index can then be constructed based on the mapping relationship between the candidate particle emitters and their corresponding emitter semantic data to enable subsequent retrieval. The target database can be a vector database (such as the open-source cloud-native vector database (Milvus)). This application does not specifically limit the target database.

[0107] For example, refer to Figure 4 , Figure 4 This is a schematic diagram illustrating the construction of a target database as provided in an embodiment of this application, such as... Figure 4 As shown, candidate particle effects 1, 2, ..., N can be obtained. At least one candidate particle emitter can be extracted from each candidate particle effect. Candidate particle emitter 11, ..., 1a can be extracted from candidate particle effect 1, candidate particle emitter 21, ..., 2b, ... can be extracted from candidate particle effect 2, and candidate particle emitter N1, ..., Nc can be extracted from candidate particle effect N.

[0108] Next, for each candidate particle emitter, the emitter frame sequence 11 corresponding to candidate particle emitter 11 can be rendered, and the emitter description information 11 corresponding to candidate particle emitter 11 can be determined based on the emitter frame sequence 11 using the visual language large model, ... The emitter frame sequence 1a corresponding to candidate particle emitter 1a can be rendered, and the emitter description information 1a corresponding to candidate particle emitter 1a can be determined based on the emitter frame sequence 1a using the visual language large model. Correspondingly, the emitter frame sequence 21 corresponding to candidate particle emitter 21 can be rendered, and the emitter description information 21 corresponding to candidate particle emitter 21 can be determined based on the emitter frame sequence 21 using the visual language large model, ... The emitter frame sequence 2b corresponding to candidate particle emitter 2b can be rendered, and the emitter description information 2b corresponding to candidate particle emitter 2b can be determined based on the emitter frame sequence 2b using the visual language large model. It can also render the emitter frame sequence N1 corresponding to the candidate particle emitter N1, and determine the emitter description information N1, ... based on the emitter frame sequence N1 using the visual language big model. It can also render the emitter frame sequence Nc corresponding to the candidate particle emitter Nc, and determine the emitter description information Nc corresponding to the candidate particle emitter Nc based on the emitter frame sequence Nc using the visual language big model.

[0109] Then, for each candidate particle emitter, the emitter semantic data 11 corresponding to candidate particle emitter 11 can be determined based on the emitter description information 11 corresponding to candidate particle emitter 11, ..., and the emitter semantic data 1a corresponding to candidate particle emitter 1a can be determined based on the emitter description information 1a corresponding to candidate particle emitter 1a. Correspondingly, the emitter semantic data 21 corresponding to candidate particle emitter 21 can be determined based on the emitter description information 21 corresponding to candidate particle emitter 21, ..., and the emitter semantic data 2b corresponding to candidate particle emitter 2b can be determined based on the emitter description information 2b corresponding to candidate particle emitter 2b. Furthermore, the emitter semantic data N1 corresponding to candidate particle emitter N1 can be determined based on the emitter description information N1 corresponding to candidate particle emitter N1, ..., and the emitter semantic data Nc corresponding to candidate particle emitter Nc can be determined based on the emitter description information Nc corresponding to candidate particle emitter Nc.

[0110] Finally, the candidate particle emitter 11 and its corresponding emitter semantic data 11, ..., candidate particle emitter 1a and its corresponding emitter semantic data 1a, candidate particle emitter 21 and its corresponding emitter semantic data 21, ..., candidate particle emitter 2b and its corresponding emitter semantic data 2b, ..., candidate particle emitter N1 and its corresponding emitter semantic data N1, ..., and candidate particle emitter Nc and its corresponding emitter semantic data Nc can be stored in the target database.

[0111] Therefore, using the above method, candidate particle emitters can be extracted from candidate particle effects, and their corresponding emitter descriptions and semantic data can be determined. Each candidate particle emitter and its corresponding semantic data are stored in a target database, enabling these candidate particle emitters to be retrieved and used subsequently. This also gives the candidate particle emitters semantic retrieval capabilities, allowing effects artists to combine and adjust existing particle emitters in the target database without having to recreate them, thus improving resource utilization efficiency. Furthermore, the target database supports retrieval based on user-provided natural language, ensuring the accuracy of semantic matching. Simultaneously, it solves the problem of unreusable historical particle effects; through semantic decoupling, particle emitters from historical particle effects can be recombined into new particle effects, maximizing the value of effects assets.

[0112] As an example, particle emitters with reference relationships in particle effects can be packaged into a group of candidate particle emitters, that is, the above S12 may include the following step S121 (not shown in the figure):

[0113] S121: For each candidate particle effect, extract multiple basic particle emitters from the candidate particle effect, and determine the candidate particle emitters based on the reference relationships between the multiple basic particle emitters.

[0114] The basic particle emitter refers to the particle emitter directly extracted from the candidate particle effects. Essentially, the basic particle emitter is a unit disassembled from the candidate particle effects on a particle emitter basis. This application does not specifically limit the basic particle emitter.

[0115] For example, for each candidate particle effect, multiple basic particle emitters can be extracted from the candidate particle effect using a preset parser. For instance, a parser can be written to read the particle effect asset file corresponding to the candidate particle effect in binary or text format to extract the multiple basic particle emitters included in the candidate particle effect.

[0116] Next, candidate particle emitters can be determined based on the reference relationships between these multiple basic particle emitters. For example, if there are basic particle emitters among these multiple basic particle emitters that meet the preset reference relationships, the basic particle emitters that meet the preset reference relationships can be combined into a candidate particle emitter (i.e., the candidate particle emitter can be a basic particle emitter group). Specifically, the preset reference relationships can include, but are not limited to, parent-child relationships, driving relationships, and lifecycle binding relationships (such as two basic particle emitters must be activated or terminated at the same time). This application does not specifically limit the preset reference relationships.

[0117] The reference relationship between multiple basic particle emitters can be determined by the information recorded in the event handler. For example, in a particle effect, when a particle element generated by basic particle emitter A dies, it triggers basic particle emitter B to generate the corresponding particle element. At this time, it can be considered that there is a driving relationship between basic particle emitter A and basic particle emitter B.

[0118] For example, taking the reference relationship between basic particle emitter A and basic particle emitter B as a parent-child relationship, with basic particle emitter A being the parent emitter and basic particle emitter B being the child emitter, basic particle emitter A can be used as a candidate particle emitter. Since basic particle emitter B needs to rely on basic particle emitter A to realize the parent-child relationship, basic particle emitter A and basic particle emitter B can be combined as a candidate particle emitter.

[0119] It should be noted that if the reference relationships between these multiple basic particle emitters do not conform to the preset reference relationships, then these multiple basic particle emitters can be regarded as an independent candidate particle emitter.

[0120] Therefore, by using the above method, based on the reference relationships between basic particle emitters, packaging basic particle emitters with strong dependencies into a single candidate particle emitter can ensure that these interconnected basic particle emitters are managed and stored as a whole, avoiding isolated candidate particle emitters or conflicts between candidate particle emitters in the target database, thereby ensuring the rationality and completeness of the candidate particle emitters stored in the target database.

[0121] As an example, preliminary emitter description information corresponding to candidate particle emitters can be generated through a large visual language model. Then, the initial emitter description information can be expanded to enrich the emitter description information corresponding to candidate particle emitters. That is, S13 above may include the following steps S131-S133 (not shown in the figure):

[0122] S131: Using a large visual language model, generate initial emitter description information corresponding to candidate particle emitters based on the emitter frame sequence.

[0123] Initial emitter description information refers to the processing result generated by the visual language large model based on the emitter frame sequence, which is used to indicate the visual effect of candidate particle emitters.

[0124] For example, a visual language big model can be used to generate initial emitter description information corresponding to the candidate particle emitter based on the emitter frame sequence. For instance, a cue message can be constructed based on the emitter frame sequence to instruct the initial emitter description information generation task to be performed, thus enabling the visual language big model to generate the initial emitter description information corresponding to the candidate particle emitter using the cue message.

[0125] After determining the prompt information, the prompt information is input into the visual language big model for processing, which can obtain the initial emitter description information corresponding to the candidate particle emitter generated by the visual language big model based on the prompt information.

[0126] S132: Perform synonym expansion processing on the initial emitter description information to obtain the extended emitter description information corresponding to the candidate particle emitter.

[0127] Extended transmitter description information refers to the result obtained after performing synonym expansion on the initial transmitter description information.

[0128] For example, the initial emitter description information can be extended using a large language model. For instance, a prompt can be constructed based on the initial emitter description information. This prompt is used to instruct the execution of a synonym extension task based on the initial emitter description information. In other words, the large language model extends the initial emitter description information using the prompt to generate extended emitter description information corresponding to the candidate particle emitter.

[0129] After determining the prompt information, the prompt information is input into the large language model for processing, which yields the extended emitter description information corresponding to the candidate particle emitter generated by the large language model based on the prompt information.

[0130] Alternatively, the initial transmitter description information can be expanded using synonyms through knowledge graphs or thesaurus. This application does not specifically limit the synonym expansion method.

[0131] S133: Use the initial emitter description information and the extended emitter description information corresponding to the candidate particle emitter as the emitter description information corresponding to the candidate particle emitter.

[0132] After determining the extended emitter description information corresponding to the candidate particle emitter, the initial emitter description information and the extended emitter description information corresponding to the candidate particle emitter can be combined to form the emitter description information corresponding to the candidate particle emitter.

[0133] Therefore, by using the above method, the initial emitter description information output by the visual language large model is expanded with synonyms, which enriches the emitter descriptions corresponding to the candidate particle emitters. This allows the emitter semantic data generated in this way to be adapted to more semantically similar search query information during subsequent retrieval, thereby improving the retrieval recall rate.

[0134] In one possible implementation, candidate particle effects and their corresponding semantic data can be stored in the target database. That is, the method provided in this application embodiment may further include the following steps S21-22 (not shown in the figure):

[0135] S21: For each candidate particle effect, render the corresponding effect frame sequence, and determine the candidate effect description information based on the effect frame sequence.

[0136] A special effects frame sequence refers to a sequence of multiple frames captured in chronological order after a candidate particle effect has been running for a period of time. It is used to display the visual effects of the candidate particle effect, such as the color distribution, change process, and visual elements contained in the particle effect.

[0137] The candidate effect description information is used to describe the visual effect of the candidate particle effect. For example, the candidate effect description information can be "blue explosion, accompanied by outward-spreading Martians". This application does not specifically limit the candidate effect description information.

[0138] For example, for each candidate particle effect, the candidate particle effect can be deployed in the game engine, and the effect frame sequence corresponding to the candidate particle effect can be rendered in a no-light scene. That is, the candidate particle effect is started and it runs for a complete life cycle (or preset duration) in a no-light scene. Then, each frame can be captured at a fixed frame rate and saved as an image sequence, thus obtaining the effect frame sequence corresponding to the candidate particle effect.

[0139] Next, the visual language big model can determine the candidate effect description information corresponding to the candidate particle effect based on the effect frame sequence. For example, a prompt message can be constructed based on the effect frame sequence to instruct the candidate effect description information generation task to be performed based on the effect frame sequence, so that the visual language big model can generate the candidate effect description information corresponding to the candidate particle effect through the prompt message.

[0140] After determining the prompt information, the prompt information is input into the visual language big model for processing, which can obtain the candidate effect description information corresponding to the candidate particle effect generated by the visual language big model based on the prompt information.

[0141] S22: For each candidate particle effect, determine the semantic data of the effect corresponding to the candidate particle effect based on the candidate effect description information, and store the candidate particle effect and its corresponding semantic data in the target database.

[0142] Special effects semantic data is used to characterize the semantic information of candidate special effects description information corresponding to candidate particle special effects. It can be represented as a semantic feature vector. In this regard, this application does not specifically limit the form of special effects semantic data.

[0143] For example, for each candidate particle effect, the semantic feature vector corresponding to the visual element can be determined by the text encoder according to the candidate effect description information corresponding to the candidate particle effect, in the same way as the above method of determining the semantic feature vector corresponding to the visual element. This will not be elaborated further here.

[0144] After determining the semantic data of each candidate particle effect, the candidate particle effects and their corresponding semantic data can be stored in the target database, and an index can be built based on the mapping relationship between the candidate particle effects and their corresponding semantic data to enable subsequent retrieval.

[0145] For example, refer to Figure 5 , Figure 5 This is another schematic diagram of constructing a target database provided in an embodiment of this application, such as... Figure 5 As shown, for each candidate particle effect, effect frame sequence 1 corresponding to candidate particle effect 1 can be rendered, and candidate effect description information 1 corresponding to candidate particle effect 1 can be determined based on effect frame sequence 1 using the visual language large model. Correspondingly, effect frame sequence 2 corresponding to candidate particle effect 2 can be rendered, and candidate effect description information 2 corresponding to candidate particle effect 2 can be determined based on effect frame sequence 2 using the visual language large model. Furthermore, effect frame sequence N corresponding to candidate particle effect N can be rendered, and candidate effect description information N corresponding to candidate particle effect N can be determined based on effect frame sequence N using the visual language large model.

[0146] Then, for each candidate particle effect, the semantic data 1 corresponding to candidate particle effect 1 can be determined based on the candidate effect description information 1. Correspondingly, the semantic data 2 corresponding to candidate particle effect 2 can be determined based on the candidate effect description information 2. Furthermore, the semantic data N corresponding to candidate particle effect N can be determined based on the candidate effect description information N.

[0147] Finally, candidate particle effect 1 and its corresponding semantic data 1, candidate particle effect 2 and its corresponding semantic data 2, ..., candidate particle effect N and its corresponding semantic data N can be stored in the target database.

[0148] Therefore, by using the above method, candidate particle effects and their corresponding semantic data are stored in the target database, providing a data foundation for the coarse-ranking stage in the subsequent retrieval process. This allows the coarse-ranking stage to quickly perform preliminary screening of a large number of candidate particle effects based on this semantic data, thereby reducing the amount of data that needs to be processed in the subsequent fine-ranking stage and improving retrieval efficiency.

[0149] Target particle emitter retrieval method:

[0150] Coarse-ranked search combined with fine-ranked search:

[0151] In one possible implementation, the target database also stores multiple candidate particle effects and their corresponding semantic data, with the candidate particle emitters extracted from the candidate particle effects. At this point, reference particle effects that semantically match the effect description information can be retrieved through coarse-sorting. Then, the candidate particle emitters extracted from the reference particle effects constitute the search range for fine-sorting retrieval. That is, the method provided in this application embodiment may further include the following step S31 (not shown in the figure):

[0152] S31: Based on the effect description information, retrieve candidate particle effects in the target database that satisfy the second semantic matching condition with the effect description information, and use them as reference particle effects.

[0153] The second semantic matching condition refers to the condition used when retrieving reference particle effects from the target database. Specifically, it indicates which candidate particle effects have a high semantic similarity to the effect description information and are selected as reference particle effects. For example, the second semantic matching condition can specify selecting the K particles with the highest similarity between the corresponding effect semantic data and the semantic feature vector of the effect description information. coarse K is one of the candidate particle effects, used as a reference particle effect. coarse It can be a preset fixed value; however, this application does not specifically limit the second semantic matching condition and K. coarse The value.

[0154] Reference particle effects refer to candidate particle effects that are semantically matched with the effect description information obtained through retrieval. All k candidate particle effects obtained through retrieval can be used as reference particle effects. This application does not specifically limit the reference particle effects or their number.

[0155] For example, it can be based on the special effects description information q overall Semantic encoding is performed to obtain the special effects description information q. overall The corresponding semantic feature vector. Furthermore, based on the effect description information q... overall The corresponding semantic feature vector is used to retrieve candidate particle effects from the target database that satisfy the second semantic matching condition with the semantic feature vector of the effect description information. For example, a preset similarity calculation method (such as cosine similarity calculation method) can be used to calculate the similarity between the semantic data of each candidate particle effect stored in the target database and the semantic feature vector corresponding to the effect description information, so as to retrieve the Kth particle effect with the highest similarity to the semantic feature vector corresponding to the effect description information. coarse A number of candidate particle effects, and retrieved K coarse One candidate particle effect is used as a reference particle effect.

[0156] It should be understood that the semantic feature vector corresponding to the visual element can be determined in the same way as the above method, and will not be elaborated further here.

[0157] Correspondingly, the step S303 above, "based on the element description information corresponding to the visual element, retrieving candidate particle emitters in the target database that satisfy the first semantic matching condition with the visual element, and using them as target particle emitters corresponding to the visual element," may include the following step S32 (not shown in the figure):

[0158] S32: Based on the element description information corresponding to the visual element, retrieve the candidate particle emitter that satisfies the first semantic matching condition with the visual element from the candidate particle emitters stored in the target database and extracted from the reference particle effects, and use it as the target particle emitter corresponding to the visual element.

[0159] For example, based on the semantic feature vector corresponding to the element description information of the visual element, the candidate particle emitters that satisfy the first semantic matching condition with the visual element can be retrieved from the candidate particle emitters extracted from the reference particle effects stored in the target database. For example, the similarity between the emitter semantic data corresponding to each candidate particle emitter extracted from the reference particle effects stored in the target database and the semantic feature vector corresponding to the element description information of the visual element can be calculated using a preset similarity calculation method (such as the cosine similarity calculation method). In order to retrieve the k candidate particle emitters with the highest similarity to the semantic feature vector corresponding to the element description information of the visual element, the retrieved k candidate particle emitters can be used as the target particle emitters corresponding to the visual element.

[0160] The extraction method described in step S12 above can be used to extract each candidate particle emitter from the reference particle effect, which will not be repeated here.

[0161] Therefore, by using the retrieval strategy of combining coarse-ranking retrieval with fine-ranking retrieval provided by the above method, in the coarse-ranking stage, candidate particle effects that meet the second semantic matching condition can be retrieved from the target database based on the effect description information as reference particle effects. This filters out candidate particle effects that do not semantically match the user's needs in advance, so that the subsequent fine-ranking stage only needs to search within the range of candidate particle emitters extracted from these reference particle effects, reducing the amount of data that needs to be processed and thus improving retrieval efficiency.

[0162] Dynamically set the number of transmitters to retrieve:

[0163] In one possible implementation, the number of candidate particle emitters recalled by the visual element is set according to the evaluation result of the visual element. That is, the method provided in this application embodiment may further include the following steps S41-S42 (not shown in the figure):

[0164] S41: For each visual element, determine the element evaluation parameters corresponding to the visual element based on the element description information corresponding to the visual element, and determine the number of emitters to be retrieved for the visual element based on the element evaluation parameters.

[0165] The element evaluation parameters are used to indicate the evaluation results obtained after evaluating the visual elements from the preset evaluation dimensions. For example, the preset evaluation dimensions may include, but are not limited to, complexity dimension, visual style dimension, and dependency dimension. In this regard, this application does not specifically limit the preset evaluation dimensions and element evaluation parameters.

[0166] For example, for each visual element, a large language model can be used to evaluate the visual element based on its corresponding element description information, according to a preset evaluation dimension, to obtain the element evaluation parameters for that visual element. Then, these element evaluation parameters can be substituted into a preset formula for calculating the number of transmitter retrievals to obtain the number of transmitter retrievals for that visual element.

[0167] S42: For each visual element, determine the first semantic matching condition corresponding to the visual element based on the number of emitters retrieved for each visual element. The first semantic matching condition is used to indicate the candidate particle emitters with a high semantic similarity to the visual element and a high number of emitters retrieved.

[0168] Following the steps described above, the number of emitters to be retrieved for each visual element can be determined. For each visual element, the first semantic matching condition can be determined based on the number of emitters to be retrieved for that visual element. The first semantic matching condition is used to indicate the number of candidate particle emitters with high semantic similarity to the visual element and the number of emitters to be retrieved. That is, when the first semantic matching condition specifies that k candidate particle emitters with high similarity to the semantic feature vector of the element description information of the visual element are selected as the target particle emitters for the visual element, k is the number of emitters to be retrieved for the visual element.

[0169] Therefore, by evaluating visual elements, the number of candidate particle emitters recalled for each visual element can be dynamically determined based on the evaluation results, enabling the allocation of computing resources according to the actual situation of the visual element. That is, for important or complex visual elements, more retrieval resources are allocated to ensure that highly matching candidate particle emitters are found; while for relatively minor or simple visual elements, the use of retrieval resources is reduced to avoid resource waste and improve the overall resource utilization efficiency.

[0170] As an example, the number of emitters retrieved corresponding to a visual element can be determined by the following steps, that is, S41 above may include the following steps S411-S412 (not shown in the figure):

[0171] S411: Based on the element description information corresponding to the visual element, determine the complexity evaluation parameter, visual style evaluation parameter, and dependency evaluation parameter corresponding to the visual element. The complexity evaluation parameter is used to indicate the complexity of the visual element, the visual style evaluation parameter is used to indicate the visual style of the visual element, and the dependency evaluation parameter is used to indicate the strength of the dependency between the visual element and other visual elements.

[0172] The complexity of an element refers to the evaluation result obtained when evaluating a visual element from the perspective of complexity. It is used to indicate the complexity of the visual element. The higher the complexity, the more complex the particle system configuration required for the visual element. For example, the visual element "smoke" involves complex materials and transparency sorting, indicating that the complexity of the visual element "smoke" is high, and its corresponding complexity evaluation parameter can be 0.8-1.0. The visual element "glow" is a simple two-dimensional planar image element (sprite), indicating that the complexity of the visual element "glow" is low, and its corresponding complexity evaluation parameter can be 0.2-0.4, etc. This application does not specifically limit the complexity evaluation parameter.

[0173] The Aesthetic Style Weight (ASH) is the evaluation result obtained when evaluating visual elements from the perspective of visual style. It is used to indicate the visual style of a visual element. The richer the visual style, the higher the ASH; the simpler the visual style, the lower the ASH. For example, if the description of a visual element includes information such as "grand," "stunning," or "epic," it indicates that the visual style of the visual element is relatively rich, and its corresponding ASH is larger to provide more diverse choices. If the description of a visual element includes information such as "minimalist" or "pixelated," it indicates that the visual style of the visual element is relatively simple, and its corresponding ASH is smaller. This application does not specifically limit the ASH.

[0174] The dependency strength parameter refers to the evaluation result obtained when evaluating visual elements from the dependency dimension. It is used to indicate the dependency strength between a visual element and other visual elements. The higher the dependency strength, the more the visual element needs to cooperate with other visual elements. In this case, searching more candidate particle emitters is beneficial to identifying candidate particle emitters that can match the time axis. For example, if the visual element "Shockwave" is synchronized with the other visual element "Core Burst", it indicates that the dependency strength between the two is high, and its corresponding dependency strength parameter is high. This application does not specifically limit the dependency strength parameter.

[0175] For example, a large language model can be used to evaluate visual elements based on the element description information corresponding to the visual elements, and the evaluation parameters for complexity, visual style, and dependency can be obtained.

[0176] S412: Determine the number of emitter retrievals corresponding to visual elements based on complexity evaluation parameters, visual style evaluation parameters, and dependency evaluation parameters.

[0177] For example, by substituting the complexity evaluation parameters, visual style evaluation parameters, and dependency evaluation parameters into the preset formula for calculating the number of emitter retrievals, the number of emitter retrievals corresponding to the visual element can be obtained.

[0178] For example, the formula for calculating the preset number of transmitters to be retrieved can be referred to the following formula (1).

[0179] Formula (1);

[0180] Where, x i Represents the i-th visual element The corresponding number of transmitters retrieved, for example, x i It can be a value greater than or equal to 1 and less than or equal to 10, C e Representing visual elements The corresponding complexity evaluation parameter, α, represents the adjustment coefficient corresponding to the complexity evaluation parameter, which can be a manually set hyperparameter, S. weight Representing visual elements The corresponding visual style evaluation parameter, β, represents the adjustment coefficient corresponding to the visual style evaluation parameter, which can be a manually set hyperparameter, D. strength Representing visual elements The corresponding dependency evaluation parameters include α and β, which can be adjusted as hyperparameters based on system load and response time requirements.

[0181] Therefore, by comprehensively considering the complexity evaluation parameters, visual style evaluation parameters, and dependency evaluation parameters corresponding to visual elements, the above method can more accurately assess the demand of each visual element for candidate particle emitters, thereby determining a more reasonable number of emitters to be retrieved. More candidates are provided for complex elements (such as fluid smoke), and fewer candidates are provided for simple elements (such as light), avoiding the deviation in the number of retrievals caused by considering a single factor. Furthermore, the number of emitters to be retrieved can be dynamically adjusted according to the actual situation of the visual elements, which can be adapted to the retrieval needs in different scenarios and balance visual richness and computational performance.

[0182] As an example, the complexity evaluation parameters, visual style evaluation parameters, and dependency evaluation parameters corresponding to visual elements can be determined through a large language model. That is, S411 above may include the following steps S411-1 to S411-2 (not shown in the figure):

[0183] S411-1: Based on the element description information corresponding to the visual element, construct evaluation prompt information. The evaluation prompt information is used to indicate the complexity evaluation task, visual style evaluation task, and dependency strength evaluation task of the visual element in combination with the element description information.

[0184] Complexity evaluation tasks refer to tasks that evaluate visual elements from the perspective of complexity, visual style evaluation tasks refer to tasks that evaluate visual elements from the perspective of visual style, and dependency strength evaluation tasks refer to tasks that evaluate visual elements from the perspective of dependency.

[0185] The evaluation prompt information is used to instruct the user to perform complexity evaluation, visual style evaluation, and dependency strength evaluation tasks on visual elements in conjunction with element description information. This allows the large language model to evaluate visual elements from the dimensions of complexity, visual style, and dependency. For example, the evaluation prompt information constructed by combining the element description information corresponding to the visual element can be: Please complete the complexity evaluation, visual style evaluation, and dependency strength evaluation tasks based on the element description information of the following visual elements. Complexity evaluation task: Evaluate the implementation complexity of the following visual elements, and the corresponding result can be represented by a value between 0 and 1. Visual style evaluation task: Evaluate the aesthetic weight of the following visual elements in the overall special effects style, and the corresponding result can be represented by a value between 0 and 1. Dependency strength evaluation task: Evaluate the dependency strength of the following visual elements on the context or other visual elements, and the corresponding result can be represented by a value between 0 and 1. The element description information of the visual element is [brown color spreading outwards, looping]. This application does not specifically limit the evaluation prompt information in this regard.

[0186] S411-2: Using a large language model, determine the complexity evaluation parameters, visual style evaluation parameters, and dependency evaluation parameters corresponding to visual elements based on the evaluation prompts.

[0187] After determining the evaluation prompt information, the evaluation prompt information is input into the large language model for processing. This yields the complexity evaluation parameters, visual style evaluation parameters, and dependency evaluation parameters of the visual elements output by the large language model based on the evaluation prompt information.

[0188] It should be understood that the determination process of the above-mentioned evaluation parameters is based on the evaluation prompt information containing the overall evaluation task. Correspondingly, the evaluation parameters corresponding to each individual evaluation task can be determined based on the evaluation prompt information containing the individual evaluation task. For example, complexity evaluation prompt information can be constructed based on the element description information corresponding to the visual element. This complexity evaluation prompt information is used to indicate the execution of a complexity evaluation task on the visual element in conjunction with the element description information. By inputting this complexity evaluation prompt information into a large language model for processing, the complexity evaluation parameters corresponding to the visual element output by the large language model can be obtained. Correspondingly, the determination methods of visual style evaluation parameters and dependency evaluation parameters can also refer to the above-mentioned process for determining complexity evaluation parameters, and will not be repeated here. This application embodiment does not specifically limit the determination methods of complexity evaluation parameters, visual style evaluation parameters, and dependency evaluation parameters corresponding to visual elements.

[0189] For example, refer to Figure 6 , Figure 6 This is a schematic diagram illustrating the determination of element evaluation parameters corresponding to visual elements provided in an embodiment of this application, such as... Figure 6 As shown, evaluation prompt information 602 can be constructed based on the element description information 601 corresponding to the visual element, so that the large language model 603 can perform complexity evaluation task, visual style evaluation task and dependency strength evaluation task accordingly, and then obtain the complexity evaluation parameter 604, visual style evaluation parameter 605 and dependency evaluation parameter 606 corresponding to the visual element output by the large language model 603.

[0190] Therefore, by deeply analyzing the element description information corresponding to visual elements through large language models, the complexity evaluation parameters, visual style evaluation parameters, and dependency evaluation parameters corresponding to visual elements can be determined efficiently and intelligently based on the evaluation prompt information, thereby improving the evaluation efficiency and accuracy.

[0191] How target particle effects are generated:

[0192] In one possible implementation, a main particle emitter and an auxiliary particle emitter can be determined in the target particle emitter corresponding to each visual element. Then, each auxiliary particle emitter is adjusted based on the main particle emitter to ensure that a coordinated target particle effect is finally obtained. That is, the above S304 may include the following steps S51-S53 (not shown in the figure):

[0193] S51: Determine the primary particle emitter among the target particle emitters corresponding to each visual element, and determine the remaining target particle emitters as secondary particle emitters.

[0194] The main emitter refers to the target particle emitter selected from multiple target particle emitters as a benchmark, which determines the visual core of the target particle effect. For example, the target particle emitter that meets the preset semantic matching conditions can be selected from the target particle emitters corresponding to each visual element as the main particle emitter. For example, the preset semantic matching conditions can specify that the target particle emitter with the highest semantic similarity to the effect description information is selected as the main particle emitter. In this regard, this application does not specifically limit the main particle emitter and its corresponding determination method.

[0195] Auxiliary emitters refer to the target particle emitters other than the main particle emitter among the target particle emitters corresponding to each visual element. The auxiliary particle emitters can be adjusted based on the main particle emitter to ensure the coordination of the target particle effects obtained by combining them.

[0196] S52: Based on the main particle emitter, each auxiliary particle emitter is adjusted to obtain the adjusted auxiliary particle emitter.

[0197] After determining the primary and secondary particle emitters, adjustments can be made to each secondary emitter based on the primary emitter. For example, for each secondary emitter, a style difference vector between the primary emitter and that secondary emitter can be calculated. This involves calculating the hue deviation and scale ratio between the primary particle emitter and the secondary particle emitter. Based on the calculated differences, the secondary particle emitter is adjusted to obtain an adjusted secondary particle emitter whose visual style is as close as possible to that of the primary particle emitter. Following this method, each secondary particle emitter can be adjusted individually to obtain its own adjusted version.

[0198] S53: Combine the main particle emitter and each adjusted auxiliary particle emitter to obtain the target particle effect.

[0199] For example, the emission source points of the main particle emitter and each adjusted auxiliary particle emitter are all set to the same origin, so that the main particle emitter and each adjusted auxiliary particle emitter are internally aligned. After combining the main particle emitter and each adjusted auxiliary particle emitter, the target particle effect can be obtained.

[0200] For example, refer to Figure 7 , Figure 7 This is a schematic diagram illustrating the generation of target particle effects provided in an embodiment of this application, as shown below. Figure 7As shown, taking target particle emitters A, B, and C corresponding to each visual element as an example, given that target particle emitter A is determined to be the primary emitter, target particle emitters B and C can be identified as secondary emitters. Furthermore, target particle emitter B can be adjusted based on the style difference vector between target particle emitters A and B to obtain an adjusted target particle emitter B. Similarly, target particle emitter C can be adjusted based on the style difference vector between target particle emitters A and C to obtain an adjusted target particle emitter C. Finally, combining target particle emitter A, the adjusted target particle emitter B, and the adjusted target particle emitter C yields the target particle effect.

[0201] Therefore, using the above method, the main particle emitter and auxiliary particle emitters can be clearly distinguished in the target particle emitters corresponding to each visual element. This allows for the adjustment of each auxiliary particle emitter based on the main particle emitter. Since the main particle emitter better aligns with the semantics of the effect's description, using it as a reference for targeted adjustments to the auxiliary particle emitters ensures that the visual styles of each target particle emitter match, thereby guaranteeing a visually harmonious and unified final assembled target particle effect and improving the overall quality of the target particle effect.

[0202] As an example, the determination of the main particle emitter can refer to the following steps, that is, the above S51 may include the following steps S511-S512 (not shown in the figure):

[0203] S511: For each target particle emitter, determine the semantic similarity between the target particle emitter and the effect description information based on the effect description information and the corresponding emitter semantic data.

[0204] For example, the special effects description information can be semantically encoded to obtain the semantic feature vector corresponding to the special effects description information. Then, for each target particle emitter, the similarity between the semantic feature vector corresponding to the special effects description information and the emitter semantic data corresponding to the target particle emitter can be calculated using a preset similarity calculation method (such as the cosine similarity calculation method), and this similarity is used as the semantic similarity between the target particle emitter and the special effects description information.

[0205] It should be understood that the semantic feature vector corresponding to the visual element can be determined in the same way as the above method, and will not be elaborated further here.

[0206] S512: Among the target particle emitters corresponding to each visual element, determine the target particle emitter with the highest semantic similarity to the special effect description information, and use it as the main particle emitter.

[0207] After determining the semantic similarity between each target particle emitter and the effect description information, the target particle emitter with the highest semantic similarity to the effect description information is selected from the target particle emitters corresponding to each visual element and used as the main particle emitter. Specifically, refer to formula (2).

[0208] E main = Formula (2);

[0209] Among them, E main E represents the main particle emitter. i V(E) represents the i-th target particle emitter among the target particle emitters corresponding to each visual element, C represents the candidate set composed of the target particle emitters corresponding to each visual element, and V(E) represents the i-th target particle emitter among the target particle emitters corresponding to each visual element. i ) represents the emitter semantic data corresponding to the i-th target particle emitter, q overall V(q) represents the description information of the special effects. overall ) represents the semantic feature vector corresponding to the special effects description information.

[0210] Therefore, by using the above method, the semantic similarity between each target particle emitter and the effect description information is calculated, and the target particle emitter with the highest similarity is selected as the main particle emitter. This allows for a deeper understanding of the semantic information in the effect description information, enabling the precise identification of the target particle emitter that best meets the semantic requirements of the effect from multiple target particle emitters. This ensures that the main particle emitter lays an accurate foundation for the final target particle effect, making the generated target particle effect more in line with user needs.

[0211] As an example, the auxiliary particle emitter can be adjusted based on the scale ratio and hue deviation between the auxiliary particle emitter and the main particle emitter. That is, S52 above may include the following steps S521-S522 (not shown in the figure):

[0212] S521: For each auxiliary particle emitter, adjust the scale parameters of the auxiliary particle emitter according to the scale ratio between the auxiliary particle emitter and the main particle emitter to obtain the adjusted scale parameters of the auxiliary particle emitter.

[0213] The scale ratio between the auxiliary particle emitter and the main particle emitter refers to the ratio of the scale parameters between the two emitters, used to quantify their relationship in terms of size, range, and particle size. For example, the scale ratio between the auxiliary particle emitter and the main particle emitter can be determined based on the scale parameters that indicate the range of the emitters. Specifically, refer to the following formula (3).

[0214] R scale =Bounds(E main ) / Bounds (E avx ) formula (3);

[0215] Among them, R scale Bounds (E) represents the scale ratio between the secondary particle emitter and the primary particle emitter. main Bounds(E) represents the scale parameter corresponding to the main particle emitter, used to indicate the emitter's range. avx ) represents the scale parameter corresponding to the auxiliary particle emitter, used to indicate the emitter range.

[0216] The scale parameter refers to the set of parameters that control the geometry or motion of the particle emitter. For example, it may include, but is not limited to, the initial radius of the particle, the emitter range, and the particle lifetime. This application does not specifically limit the scale parameter.

[0217] For each auxiliary particle emitter, the scale parameter of the auxiliary particle emitter can be adjusted according to the scale ratio between the auxiliary particle emitter and the main particle emitter. For example, the scale parameter of the auxiliary particle emitter can be adjusted by a preset formula (such as the damping alignment formula) based on the scale ratio between the auxiliary particle emitter and the main particle emitter, so that it can adapt to the magnitude of the main particle emitter while retaining its own characteristics, rather than directly forcibly scaling the auxiliary particle emitter to be completely consistent with the main particle emitter. Specifically, refer to the following formula (4).

[0218] Formula (4);

[0219] in, This represents the adjusted scale parameters of the auxiliary particle emitter. R represents the scale parameter of the auxiliary particle emitter before adjustment. scale This represents the scale ratio between the secondary particle emitter and the primary particle emitter. ∈ (0,1], The harmonic coefficient represents the harmonic coefficient, which ensures that the size of the auxiliary particle emitter is appropriate, neither too large nor too small. It can be set manually, but this application does not specifically limit the harmonic coefficient.

[0220] S522: For each auxiliary particle emitter, adjust the color parameters of the auxiliary particle emitter according to the hue deviation between the auxiliary particle emitter and the main particle emitter to obtain the adjusted color parameters of the auxiliary particle emitter.

[0221] The hue deviation between the auxiliary particle emitter and the main particle emitter refers to the difference in color patch characteristics between the two. For example, the hue deviation between the auxiliary particle emitter and the main particle emitter can be determined based on their color patch characteristics. Specifically, refer to the following formula (5).

[0222] =H(E) main )-H(E avx ) Formula (5);

[0223] in, H(E) represents the hue deviation between the secondary particle emitter and the primary particle emitter. main H(E) represents the color patch feature corresponding to the main particle emitter, which can be obtained by analyzing and encoding the colors in the main particle emitter using a color analysis model. avx ) represents the color block feature corresponding to the auxiliary particle emitter, which can be obtained by analyzing and encoding the colors in the auxiliary particle emitter through a color analysis model.

[0224] For example, for each auxiliary particle emitter, the color parameters of the auxiliary particle emitter can be adjusted according to the hue deviation between the auxiliary particle emitter and the main particle emitter. For example, the hue deviation can be directly applied to the color parameter control module (such as the Color Over Life module) of the auxiliary particle emitter to adjust the color parameters of the auxiliary particle emitter, so that the hue of the auxiliary particle emitter shifts towards the hue of the main particle emitter. The adjusted color parameters of the auxiliary particle emitter can then be obtained, thereby ensuring that the adjusted auxiliary particle emitter resonates with the main particle emitter on the color spectrum.

[0225] The color parameters refer to the set of parameters that control the color performance of the particle emitter. For example, they may include, but are not limited to, the starting color, ending color, and hue offset of the particles. This application does not specifically limit the color parameters.

[0226] Therefore, by calculating the scale ratio between the auxiliary particle emitter and the main particle emitter using the above method, and adjusting the scale parameters of the auxiliary particle emitter according to this ratio, it is possible to ensure that the auxiliary particle emitter matches the main particle emitter in scale. This results in a unified effect of the target particle effects assembled in this way, enhancing visual comfort. Furthermore, the color parameters of the auxiliary particle emitter can be adjusted based on the hue deviation between the auxiliary and main particle emitters to achieve color harmony, create a unified color atmosphere, improve the overall effect of the target particle effects, and achieve adaptive style alignment.

[0227] As an example, the emission source points of both the main particle emitter and each adjusted auxiliary particle emitter can be set to the origin of the target coordinate system to achieve combined target particle effects based on the new emission source points. That is, S53 above may include the following steps S53-11 to S53-12 (not shown in the figure):

[0228] S53-11: Set the emission source point of the main particle emitter and the emission source point of each adjusted auxiliary particle emitter to the origin of the target coordinate system.

[0229] The source point of the main particle emitter refers to the source point from which the main particle emitter emits particles, while the source point of the adjusted auxiliary particle emitter refers to the source point from which the adjusted auxiliary particle emitter emits particles.

[0230] The target coordinate system refers to the unified spatial coordinate system used by the target particle effects. For example, it can be the world coordinate system, etc. This application does not specifically limit the target coordinate system. The origin of the target coordinate system refers to the zero point (0,0,0) of the target coordinate system.

[0231] For example, after determining the target coordinate system, the emission source point of the main particle emitter and the emission source points of each adjusted auxiliary particle emitter can be reset to the origin (0,0,0) of the target system to eliminate the spatial displacement deviation of each particle emitter.

[0232] S53-12: Based on the emission source point of the main particle emitter and the emission source points of each adjusted auxiliary particle emitter, the main particle emitter and each adjusted auxiliary particle emitter are combined to obtain the target particle effect.

[0233] After setting the emission source point of the main particle emitter and the emission source points of each adjusted auxiliary particle emitter as the origin of the target coordinate system, the main particle emitter and each adjusted auxiliary particle emitter can be combined based on the same emission source point to obtain the target particle effect. For example, the combined particle effect can be used as the target particle effect.

[0234] Therefore, by resetting the emission source points of the main particle emitter and each adjusted auxiliary particle emitter to the origin of the target coordinate system using the above method, spatial displacement deviations between the particle emitters can be effectively eliminated. By unifying the emission source points, it is ensured that all particles start emitting from the same reference position, making the target particle effects more accurate and reasonable in spatial positioning.

[0235] As an example, after obtaining the target particle effect, it is possible to detect whether the effect of the target particle effect matches the effect description information. If they do not match, the target particle effect is adjusted. That is, the above S53 may include the following steps S53-21 to S53-23 (not shown in the figure):

[0236] S53-21: Combine the main particle emitter and each adjusted auxiliary particle emitter to obtain the initial particle effects.

[0237] The initial particle effect refers to the particle effect obtained by directly combining the main particle emitter and each adjusted auxiliary particle emitter, and it needs to be tested later.

[0238] For example, after setting the emission source point of the main particle emitter and the emission source points of each adjusted auxiliary particle emitter as the origin of the target coordinate system, the particle effect obtained by directly combining the main particle emitter and each adjusted auxiliary particle emitter can be used as the initial particle effect.

[0239] S53-22: Detect whether the effect of the initial particle effect matches the effect description information.

[0240] The special effects of the initial particle effects refer to the visual appearance of the initial particle effects. For example, it can include features such as the shape, color, trajectory, and lifespan of the particles.

[0241] For example, the semantic data of the initial particle effect can be determined by referring to the method of determining the semantic data of the effect corresponding to the candidate particle effect in steps S21-22 above. The difference is that the candidate particle effect is replaced with the initial particle effect. The determination method is similar and will not be described again here.

[0242] After determining the semantic data corresponding to the initial particle effect, the matching between the effect's effect and its description information can be detected based on the semantic data and the semantic feature vector. For example, a preset similarity calculation method (such as cosine similarity) can be used to calculate the similarity between the semantic data and the feature vector. If the similarity is greater than a preset similarity threshold, the effect is considered to match the description information; if the similarity is less than or equal to the threshold, the effect is considered not to match.

[0243] S53-23: If a match is found, the initial particle effect is determined to be the target particle effect; if a mismatch is found, the initial particle effect is adjusted according to the effect description information to obtain the target particle effect.

[0244] If a match is found, the initial particle effect is directly used as the target particle effect; if a mismatch is found, the initial particle effect is adjusted according to the effect description information to use the adjusted initial particle effect as the target particle effect. For example, the effect of the initial particle effect can be made closer to the effect description information provided by the user by adjusting the parameters of the initial particle effect. If the effect description information includes "blue", the color parameters of the main particle emitter and the auxiliary particle emitter included in the initial particle effect can be adjusted to shift the overall hue of the initial particle effect towards blue. This application does not specifically limit the adjustment method of the initial particle effect.

[0245] Therefore, by detecting whether the effect of the initial particle effect matches the effect description information provided by the user, it can be ensured that the final generated target particle effect accurately meets the user's needs. By using the effect description information as a standard to verify the generated initial particle effect, the accuracy of the target particle effect can be ensured.

[0246] User-oriented product presentation formats:

[0247] Product presentation format where user-provided special effects description information:

[0248] In one possible implementation, the user can simultaneously input basic description information and configure effect styles in the effects creation panel, that is, the above S301 may include the following steps S61-S62 (not shown in the figure):

[0249] S61: Obtain basic description information input through the description information input control in the effects creation panel, and obtain the effects style information configured through the style configuration control in the effects creation panel.

[0250] The Smart Creation Panel refers to an interactive panel provided to users, which supports functions such as inputting basic description information, configuring special effects styles, and displaying the special effects of target particle effects. This application does not specifically limit the functions provided by the Smart Creation Panel.

[0251] Basic description information refers to the description information entered by the user through the description information input control. It is used to describe the target particle effect to be generated. Its form of expression can include, but is not limited to, natural language text, images, voice, and links. This application does not specifically limit the basic description information and its form of expression.

[0252] Special effects style information refers to the visual style configured by the user through the style configuration control. For example, the style configuration control includes pre-provided visual styles (such as realistic or cartoon, chaotic or orderly, etc.), which the user can select according to their desired visual style.

[0253] The description information input control is used to receive basic description information input by the user. Its form may be a text input box; however, this application does not specifically limit the form of the description information input control. The style configuration control is used to receive special effects style information configured by the user. Its form may be a slider, including pre-provided visual styles; however, this application does not specifically limit the form of the style configuration control.

[0254] For example, a user can enter basic descriptive information in the description information input control in the effects creation panel, and the terminal device can obtain the basic descriptive information in response to this input operation. Users can also select a visual style through the style configuration control in the effects creation panel, and the terminal device can obtain effects style information in response to this selection operation.

[0255] For example, you can refer to Figure 8 , Figure 8 A schematic diagram of the special effects creation panel provided in the embodiments of this application, as shown below. Figure 8 As shown, the special effects creation panel includes a description information input control 801 and a style configuration control 802. Users can enter basic description information in the description information input control 801, such as "a blue dragon with a frosty aura breathes out, accompanied by falling ice crystals". The style configuration control 802 includes pre-provided visual styles (realistic and cartoon). Users can select a visual style by sliding the style configuration control 802.

[0256] The terminal device can generate an effect description information generation request based on the obtained basic description information and effect style information, and send the effect description information generation request to the server so that the server can obtain the basic description information and effect style information from the effect description information generation request.

[0257] S62: Determine the special effects description information based on the basic description information and the special effects style information.

[0258] Finally, the obtained basic description information and special effects style information can be combined to form the special effects description information.

[0259] Therefore, basic descriptive information is obtained through the description information input controls in the effects creation panel, while effect style information is obtained through the style configuration controls, thus collecting information describing the target particle effects from different dimensions. This multi-dimensional information collection method ensures that the effect description information is more complete and richer, providing a comprehensive and detailed foundation for the accurate generation of the target particle effects.

[0260] Product presentation formats for showcasing target particle effects:

[0261] In one possible implementation, while displaying the target particle effect, a view of its corresponding emitter composition can be displayed, thereby revealing the components of the target particle effect and the source of the target particle emitter in the target particle effect. That is, the method provided in this application embodiment may further include the following step S71 (not shown in the figure):

[0262] S71: Displays the target particle effect in the effect preview window, and displays the emitter composition view corresponding to the target particle effect. The emitter composition view is used to indicate the original reference particle effect to which the target particle emitter in the target particle effect belonged.

[0263] The Multi-View Preview window displays the effects of the target particle effect and the corresponding emitter composition view. It allows users to rotate, zoom in, and scale the target particle effect in the Multi-View Preview window to dynamically display the effects.

[0264] The emitter composition view corresponding to the target particle effect is used to indicate the target particle emitters included in the target particle effect and their original reference particle effects. That is, the emitter composition view not only reveals the individual target particle emitters included in the target particle effect, but also the source of each target particle emitter. For example, if the target particle effect includes target particle emitter A and target particle emitter B, the emitter composition view can show that target particle emitter A originates from reference particle effect A, and target particle emitter B originates from reference particle effect B.

[0265] Among them, the reference particle effect is the candidate particle effect on which the target particle emitter is extracted, that is, the original particle effect to which the target particle emitter belongs.

[0266] For example, refer to Figure 9 , Figure 9 A schematic diagram of the special effects preview window provided in the embodiments of this application, as shown below. Figure 9 As shown, the effect preview window 901 can display the target particle effect 902, and can also display the emitter composition view 903 corresponding to the target particle effect. When the target particle effect 902 includes "ice crystal" and "aura", the emitter composition view 903 can show that "ice crystal" comes from reference particle effect A and "aura" comes from reference particle effect B.

[0267] Therefore, by using the above method, while displaying the target particle effect, the corresponding emitter composition view can be shown to the user intuitively, and the internal structure of the target particle effect can be clearly shown. It also clearly shows which original particle effects' particle emitters combine to form the target particle effect, which helps users to deeply understand the structure of the target particle effect, greatly reduces the understanding threshold, and helps to improve the user experience.

[0268] Supports exporting the target particle effect file:

[0269] In one possible implementation, exporting the target special effects file is supported; that is, the method provided in this application embodiment may further include the following step S81 (not shown in the figure):

[0270] S81: In response to the effect export operation triggered for the target particle effect, generate the target effect file corresponding to the target particle effect. The target effect file is in the effect file format supported by the game engine.

[0271] The special effects export operation is used to indicate the export of target particle special effects. For example, the special effects export operation can be triggered by clicking the export control or by a preset gesture (such as swiping left or right). This application does not specifically limit the triggering method of the special effects export operation.

[0272] A target effect file refers to the file obtained after triggering an effect export operation on a target particle effect. The format of the target effect file is an effect file format supported by the game engine, meaning that the game engine can directly recognize the target effect file to support the application and editing of the target particle effect within the game engine. The game engine may include, but is not limited to, game engines such as Unity, Unreal Engine (UE), Godot, and CocosCreator. This application does not specifically limit the game engine.

[0273] For example, the target particle effect includes a corresponding confirmation export control. Users can trigger the effect export operation by clicking the confirmation export control. In response to the effect export operation, the terminal device can generate the target effect file corresponding to the target particle effect. For example, the target particle effect can be serialized into an effect file format supported by the game engine (such as the effect file formats that the UE engine can recognize: copy-paste format, or script format (Python Script)) through the game engine's application programming interface (API) (such as scripts in Unreal Engine). The hierarchical structure can be automatically constructed, setting the main particle emitter as the root node or the main rendering layer, and mounting the adjusted auxiliary particle emitter as a secondary detail, finally generating a highly consistent and detailed target effect file (".uasset").

[0274] Therefore, the above method allows users to export editable target effect files, which can then be imported into the game engine for subsequent editing, optimization, and use of the target particle effects. This provides users with greater creative freedom and facilitates the achievement of more complex special effects.

[0275] An overall example illustrating the special effects generation method:

[0276] Finally, combining Figure 10 The special effects generation method provided in the embodiments of this application will be described in its entirety by way of example. Figure 10 This is a schematic diagram of the structure of the special effects generation method provided in the embodiments of this application, such as... Figure 10 As shown, a target database can be built offline to support subsequent retrieval. Specifically, this can include asset parsing, rendering sampling and statistical features, synonym expansion, and indexing. Asset parsing refers to acquiring multiple candidate particle effects and extracting candidate particle emitters from each effect.

[0277] Rendering sampling and statistical features refer to rendering the corresponding emitter frame sequence for each candidate particle emitter (e.g., rendering using Unreal Engine with a headless model (Headless UE)), determining the emitter description information corresponding to the candidate particle emitter based on the emitter frame sequence, and determining the corresponding emitter semantic data (which may include image embedding, motion trajectory data, color data, and performance data) based on this. At this time, the output data of this module may include the candidate particle emitter and its corresponding emitter semantic data.

[0278] The rendering sampling and statistical features also include rendering the effect frame sequence corresponding to each candidate particle effect (such as rendering via Headless UE), and determining the candidate effect description information corresponding to the candidate particle effect based on the effect frame sequence, and determining the effect semantic data corresponding to the candidate particle effect accordingly. At this time, the output data of this module can include the candidate particle effect and its corresponding effect semantic data.

[0279] Synonym expansion refers to the process of expanding the initial emitter description information of a candidate particle emitter based on elements, style words, and tags such as material or form, to obtain the expanded emitter description information of the candidate particle emitter. Then, the initial emitter description information and the expanded emitter description information of the candidate particle emitter are used as the emitter description information of the candidate particle emitter.

[0280] Indexing into the database means that the target database can not only store candidate particle effects and their corresponding effect semantic data (which can be represented as similarity search library 1 (faiss1), which belongs to the group level), but also store candidate particle emitters and their corresponding emitter semantic data (which can be represented as similarity search library 2 (faiss2), which belongs to the element level).

[0281] In the process of generating the target particle effect, firstly, effect description information can be input. For example, referring to step S61 above, the effect description information can be determined based on the basic description information and effect style information. The effect description information may include the target particle effect to be generated (smoke or sparks), performance budget (perf_budget), and duration, etc. Next, the search and parsing module can be entered, and referring to step S302 above, the visual elements included in the target particle effect and their corresponding element description information can be determined based on the effect description information.

[0282] Then, the vectorization encoding module can be used to perform semantic encoding on the effect description information and the element description information corresponding to the visual elements through a text encoder (such as the Qwen-Vision-Language model, qwen-vl) to obtain the semantic feature vector corresponding to the effect description information and the semantic feature vector corresponding to each visual element.

[0283] Furthermore, multi-index recall can be achieved based on the semantic feature vector corresponding to the effect description information and the semantic feature vector corresponding to each visual element, that is, hierarchical dynamic retrieval can be achieved accordingly. First, coarse recall can be performed: based on the semantic feature vector corresponding to the effect description information, the k1 candidate particle effects with the highest similarity to the semantic feature vector corresponding to the effect description information are coarsely recalled in the similarity search library 1, and used as reference particle effects (i.e., candidate groups). The output data at this time is the reference particle effect, and the output content includes the candidate particle effect identifier (group_id) of the reference particle effect, its similarity (sim) with the semantic feature vector corresponding to the effect description information, and tags.

[0284] After determining the reference particle effect, the candidate particle emitters extracted from the reference particle effect can be reordered and enter the multi-level reordering module (Rerank). Based on this, fine retrieval with dynamic K can be achieved. Based on the semantic feature vector corresponding to the i-th visual element, the x candidate particle emitters with the highest similarity to the semantic feature vector corresponding to the i-th visual element can be finely retrieved from the similarity search library 2. These x candidate particle emitters are used as the target particle emitters (candidate_emitters[i]) corresponding to the i-th visual element. The output data at this time is the target particle emitter corresponding to the i-th visual element. The output content includes the particle emitter identifier (emitter_id) of the target particle emitter corresponding to the i-th visual element, its similarity to the semantic feature vector corresponding to the i-th visual element, and the label.

[0285] The number of transmitters to be retrieved is determined by the retrieval control module, and the specific method for determining the number of transmitters to be retrieved can be referred to in steps S41-S42 above.

[0286] Finally, the retrieval output module can combine the target particle emitters corresponding to each visual element to generate target particle effects. The target particle emitter can include the particle emitter identifier, the similarity between its identifier and the semantic feature vector corresponding to the effect description information, elements (element, elem), dependencies (dependencies, deps), conflicts, and performance estimation (perf_est).

[0287] At the same time, the initial particle effects can be adjusted through the adaptive adjustment module to obtain the target particle effects, as detailed in steps S51-S53 above.

[0288] Based on the special effects generation method provided in the preceding embodiments, this application also provides a special effects generation apparatus. The following, in conjunction with... Figure 11 To explain, Figure 11 This is a schematic diagram of the structure of the special effects generation device 1100 provided in the embodiments of this application. The device includes:

[0289] Information acquisition module 1101 is used to acquire special effect description information, which is used to describe the target particle special effect to be generated;

[0290] The element determination module 1102 is used to determine the element description information corresponding to each visual element included in the target particle effect according to the effect description information.

[0291] The emitter retrieval module 1103 is used to, for each visual element, retrieve candidate particle emitters that satisfy the first semantic matching condition between the visual element and the visual element in the target database according to the element description information corresponding to the visual element, and use them as the target particle emitters corresponding to the visual element. The target database stores multiple candidate particle emitters and their respective corresponding emitter semantic data.

[0292] The special effects generation module 1104 is used to generate the target particle special effects based on the target particle emitter corresponding to each of the visual elements.

[0293] Optionally, the target database also stores multiple candidate particle effects and their corresponding semantic data, and the candidate particle emitter is extracted from the candidate particle effects; the device further includes:

[0294] The first retrieval module is used to retrieve, based on the effect description information, the candidate particle effects that satisfy the second semantic matching condition with the effect description information in the target database, and use them as reference particle effects.

[0295] Correspondingly, the transmitter retrieval module 1103 is specifically used for:

[0296] Based on the element description information corresponding to the visual element, among the candidate particle emitters extracted from the reference particle effects stored in the target database, the candidate particle emitter that satisfies the first semantic matching condition with the visual element is retrieved and used as the target particle emitter corresponding to the visual element.

[0297] Optionally, the device further includes:

[0298] The retrieval quantity determination module is used to determine the element evaluation parameter corresponding to each visual element based on the element description information corresponding to the visual element, and to determine the emitter retrieval quantity corresponding to the visual element based on the element evaluation parameter.

[0299] The semantic matching condition determination module is used to determine the first semantic matching condition corresponding to each visual element based on the number of emitters retrieved for each visual element. The first semantic matching condition is used to indicate the retrieval of candidate particle emitters with high semantic similarity to the visual element and the number of emitters retrieved.

[0300] Optionally, the retrieval quantity determination module is specifically used for:

[0301] Based on the element description information corresponding to the visual element, a complexity evaluation parameter, a visual style evaluation parameter, and a dependency evaluation parameter corresponding to the visual element are determined. The complexity evaluation parameter is used to indicate the complexity of the visual element, the visual style evaluation parameter is used to indicate the visual style of the visual element, and the dependency evaluation parameter is used to indicate the strength of the dependency between the visual element and other visual elements.

[0302] The number of emitter retrievals corresponding to the visual element is determined based on the complexity evaluation parameter, the visual style evaluation parameter, and the dependency evaluation parameter.

[0303] Optionally, the retrieval quantity determination module is specifically used for:

[0304] Based on the element description information corresponding to the visual element, an evaluation prompt information is constructed. The evaluation prompt information is used to instruct the visual element to perform a complexity evaluation task, a visual style evaluation task, and a dependency strength evaluation task in conjunction with the element description information.

[0305] Using a large language model, the complexity evaluation parameter, the visual style evaluation parameter, and the dependency evaluation parameter corresponding to the visual element are determined based on the evaluation prompt information.

[0306] Optionally, the special effects generation module 1104 is specifically used for:

[0307] In each of the target particle emitters corresponding to each of the visual elements, a main particle emitter is determined, and the remaining target particle emitters are determined as auxiliary particle emitters.

[0308] Based on the main particle emitter, each of the auxiliary particle emitters is adjusted to obtain each adjusted auxiliary particle emitter.

[0309] The main particle emitter and each of the adjusted auxiliary particle emitters are combined to obtain the target particle effect.

[0310] Optionally, the special effects generation module 1104 is specifically used for:

[0311] For each target particle emitter, the semantic similarity between the target particle emitter and the special effect description information is determined based on the special effect description information and the emitter semantic data corresponding to the target particle emitter.

[0312] Among the target particle emitters corresponding to each of the visual elements, the target particle emitter with the highest semantic similarity to the special effect description information is determined as the main particle emitter.

[0313] Optionally, the special effects generation module 1104 is specifically used for:

[0314] For each of the auxiliary particle emitters, the scale parameters of the auxiliary particle emitter are adjusted according to the scale ratio between the auxiliary particle emitter and the main particle emitter to obtain the adjusted scale parameters of the auxiliary particle emitter.

[0315] For each of the auxiliary particle emitters, the color parameters of the auxiliary particle emitter are adjusted according to the hue deviation between the auxiliary particle emitter and the main particle emitter to obtain the adjusted color parameters of the auxiliary particle emitter.

[0316] Optionally, the special effects generation module 1104 is specifically used for:

[0317] The emission source point of the main particle emitter and the emission source point of each of the adjusted auxiliary particle emitters are all set as the origin of the target coordinate system.

[0318] Based on the emission source point of the main particle emitter and the emission source points of each of the adjusted auxiliary particle emitters, the main particle emitter and each of the adjusted auxiliary particle emitters are combined to obtain the target particle effect.

[0319] Optionally, the special effects generation module 1104 is specifically used for:

[0320] The initial particle effect is obtained by combining the main particle emitter and each of the adjusted auxiliary particle emitters.

[0321] Detect whether the effect of the initial particle effect matches the effect description information;

[0322] If they match, the initial particle effect is determined to be the target particle effect; if they do not match, the initial particle effect is adjusted according to the effect description information to obtain the target particle effect.

[0323] Optionally, the device further includes the following modules for constructing the target database:

[0324] The candidate particle effect acquisition module is used to acquire multiple candidate particle effects;

[0325] An extraction module is used to extract the candidate particle emitter from each of the candidate particle effects;

[0326] The emitter description information determination module is used to render the emitter frame sequence corresponding to each candidate particle emitter, and determine the emitter description information corresponding to the candidate particle emitter based on the emitter frame sequence.

[0327] The first storage module is used to determine the transmitter semantic data corresponding to each candidate particle emitter based on the emitter description information corresponding to the candidate particle emitter, and store the candidate particle emitter and its corresponding transmitter semantic data in the target database.

[0328] Optionally, the extraction module is specifically used for:

[0329] For each candidate particle effect, multiple basic particle emitters are extracted from the candidate particle effect, and the candidate particle emitter is determined based on the reference relationship between the multiple basic particle emitters.

[0330] Optionally, the transmitter description information determination module is specifically used for:

[0331] Using a large visual language model, initial emitter description information corresponding to the candidate particle emitter is generated based on the emitter frame sequence.

[0332] The initial emitter description information is expanded using synonyms to obtain the expanded emitter description information corresponding to the candidate particle emitter.

[0333] The initial emitter description information and the extended emitter description information corresponding to the candidate particle emitter are used as the emitter description information corresponding to the candidate particle emitter.

[0334] Optionally, the device further includes:

[0335] The candidate effect description information determination module is used to render the effect frame sequence corresponding to each candidate particle effect, and determine the candidate effect description information corresponding to the candidate particle effect based on the effect frame sequence.

[0336] The second storage module is used to determine the semantic data of the candidate particle effect corresponding to each candidate particle effect based on the candidate effect description information corresponding to the candidate particle effect, and store the candidate particle effect and its corresponding semantic data in the target database.

[0337] Optionally, the information acquisition module 1101 is specifically used for:

[0338] Obtain basic description information input through the description information input control in the special effects creation panel, and obtain special effects style information configured through the style configuration control in the special effects creation panel;

[0339] The special effects description information is determined based on the basic description information and the special effects style information.

[0340] Optionally, the device further includes:

[0341] The display module is used to display the target particle effect in the effect preview window and to display the emitter composition view corresponding to the target particle effect. The emitter composition view is used to indicate the reference particle effect to which the target particle emitter originally belonged in the target particle effect.

[0342] Optionally, the device further includes:

[0343] The export module is used to generate a target effect file corresponding to the target particle effect in response to the effect export operation triggered for the target particle effect. The format of the target effect file is an effect file format that the game engine supports.

[0344] This application also provides a computer device, which may specifically be a terminal device or a server. The terminal device and server provided in this application will be described below from the perspective of hardware implementation.

[0345] See Figure 12 , Figure 12 This is a schematic diagram of the structure of a terminal device provided in an embodiment of this application. Figure 12 As shown, for ease of explanation, only the parts relevant to the embodiments of this application are illustrated. For specific technical details not disclosed, please refer to the method section of the embodiments of this application.

[0346] refer to Figure 12The terminal device includes: a radio frequency (RF) circuit 1210, a memory 1220, an input unit 1230 (including a touch panel 1231 and other input devices 1232), a display unit 1240 (including a display panel 1241), a sensor 1250, an audio circuit 1260 (connected to a speaker 1261 and a microphone 1262), a wireless fidelity (WiFi) module 1270, a processor 1280, and a power supply 1290, etc. Those skilled in the art will understand that... Figure 12 The terminal device structure shown does not constitute a limitation on the computer and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0347] The memory 1220 can be used to store software programs and modules. The processor 1280 executes various computer functions and data processing by running the software programs and modules stored in the memory 1220. The memory 1220 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the computer (such as audio data, telephone directory, etc.). In addition, the memory 1220 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0348] The processor 1280 is the control center of the computer, connecting various parts of the computer through various interfaces and lines. It performs various computer functions and processes data by running or executing software programs and / or modules stored in the memory 1220, and by calling data stored in the memory 1220. Optionally, the processor 1280 may include one or more processing units; preferably, the processor 1280 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may also not be integrated into the processor 1280.

[0349] In this embodiment of the application, the processor 1280 included in the terminal is used to execute the steps in the special effects generation method described in the foregoing embodiments.

[0350] See Figure 13 , Figure 13This is a schematic diagram of a server structure provided in an embodiment of this application. The server can vary significantly due to different configurations or performance, and may include one or more central processing units (CPUs) 1322 (e.g., one or more processors) and memory 1332, and one or more storage media 1330 (e.g., one or more mass storage devices) for storing application programs 1342 or data 1344. The memory 1332 and storage media 1330 can be temporary or persistent storage. The program stored in the storage media 1330 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the server. Furthermore, the CPU 1322 may be configured to communicate with the storage media 1330 and execute the series of instruction operations in the storage media 1330 on the server.

[0351] The server may also include one or more power supplies 1326, one or more wired or wireless network interfaces 1350, one or more input / output interfaces 1358, and / or one or more operating systems, such as Windows Server. TM Mac OS X TM Unix TM Linux TM FreeBSD TM etc.

[0352] The steps performed by the server in the above embodiments can be based on this Figure 13 The server structure shown is as follows. The CPU 1322 is used to execute the steps in the special effects generation methods described in the foregoing embodiments.

[0353] This application also provides a computer-readable storage medium for storing a computer program that performs the steps in the special effects generation methods described in the foregoing embodiments.

[0354] This application also provides a computer program product or computer program that includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps in the special effects generation methods described in the foregoing embodiments.

[0355] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0356] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0357] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0358] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0359] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing computer programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0360] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0361] In the embodiments of this application, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0362] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for generating special effects, characterized in that, The method includes: Obtain special effects description information, which is used to describe the target particle special effects to be generated; Based on the special effect description information, determine the element description information corresponding to each visual element included in the target particle special effect; Based on the effect description information, candidate particle effects that satisfy the second semantic matching condition with the effect description information are retrieved from the target database and used as reference particle effects; wherein, the target database stores multiple candidate particle effects and their corresponding effect semantic data, and also stores multiple candidate particle emitters and their corresponding emitter semantic data, wherein the candidate particle emitters are extracted from the candidate particle effects; For each visual element, based on the element description information corresponding to the visual element, among the candidate particle emitters extracted from the reference particle effects stored in the target database, the candidate particle emitter that satisfies the first semantic matching condition with the visual element is retrieved and used as the target particle emitter corresponding to the visual element. The target particle effect is generated based on the target particle emitter corresponding to each of the visual elements.

2. The method according to claim 1, characterized in that, The method further includes: For each visual element, the element evaluation parameter corresponding to the visual element is determined based on the element description information corresponding to the visual element, and the emitter retrieval quantity corresponding to the visual element is determined based on the element evaluation parameter. For each visual element, the first semantic matching condition corresponding to the visual element is determined based on the number of emitters retrieved for each visual element. The first semantic matching condition is used to indicate the retrieval of candidate particle emitters with high semantic similarity to the visual element and the number of emitters retrieved.

3. The method according to claim 2, characterized in that, The step of determining the element evaluation parameter corresponding to the visual element based on the element description information corresponding to the visual element, and determining the emitter retrieval quantity corresponding to the visual element based on the element evaluation parameter, includes: Based on the element description information corresponding to the visual element, a complexity evaluation parameter, a visual style evaluation parameter, and a dependency evaluation parameter corresponding to the visual element are determined. The complexity evaluation parameter is used to indicate the complexity of the visual element, the visual style evaluation parameter is used to indicate the visual style of the visual element, and the dependency evaluation parameter is used to indicate the strength of the dependency between the visual element and other visual elements. The number of emitter retrievals corresponding to the visual element is determined based on the complexity evaluation parameter, the visual style evaluation parameter, and the dependency evaluation parameter.

4. The method according to claim 3, characterized in that, The step of determining the complexity evaluation parameter, visual style evaluation parameter, and dependency evaluation parameter corresponding to the visual element based on the element description information corresponding to the visual element includes: Based on the element description information corresponding to the visual element, an evaluation prompt information is constructed. The evaluation prompt information is used to instruct the visual element to perform a complexity evaluation task, a visual style evaluation task, and a dependency strength evaluation task in conjunction with the element description information. Using a large language model, the complexity evaluation parameter, the visual style evaluation parameter, and the dependency evaluation parameter corresponding to the visual element are determined based on the evaluation prompt information.

5. The method according to claim 1, characterized in that, The generation of the target particle effect based on the target particle emitter corresponding to each of the visual elements includes: In each of the target particle emitters corresponding to each of the visual elements, a main particle emitter is determined, and the remaining target particle emitters are determined as auxiliary particle emitters. Based on the main particle emitter, each of the auxiliary particle emitters is adjusted to obtain each adjusted auxiliary particle emitter. The main particle emitter and each of the adjusted auxiliary particle emitters are combined to obtain the target particle effect.

6. The method according to claim 5, characterized in that, The step of determining the master particle emitter among the target particle emitters corresponding to each of the visual elements includes: For each target particle emitter, the semantic similarity between the target particle emitter and the special effect description information is determined based on the special effect description information and the emitter semantic data corresponding to the target particle emitter. Among the target particle emitters corresponding to each of the visual elements, the target particle emitter with the highest semantic similarity to the special effect description information is determined as the main particle emitter.

7. The method according to claim 5, characterized in that, The step of adjusting each of the auxiliary particle emitters based on the main particle emitter to obtain each adjusted auxiliary particle emitter includes: For each of the auxiliary particle emitters, the scale parameters of the auxiliary particle emitter are adjusted according to the scale ratio between the auxiliary particle emitter and the main particle emitter to obtain the adjusted scale parameters of the auxiliary particle emitter. For each of the auxiliary particle emitters, the color parameters of the auxiliary particle emitter are adjusted according to the hue deviation between the auxiliary particle emitter and the main particle emitter to obtain the adjusted color parameters of the auxiliary particle emitter.

8. The method according to claim 5, characterized in that, The combination of the main particle emitter and each of the adjusted auxiliary particle emitters to obtain the target particle effect includes: The emission source point of the main particle emitter and the emission source point of each of the adjusted auxiliary particle emitters are all set as the origin of the target coordinate system. Based on the emission source point of the main particle emitter and the emission source points of each of the adjusted auxiliary particle emitters, the main particle emitter and each of the adjusted auxiliary particle emitters are combined to obtain the target particle effect.

9. The method according to claim 5, characterized in that, The combination of the main particle emitter and each of the adjusted auxiliary particle emitters to obtain the target particle effect includes: The initial particle effect is obtained by combining the main particle emitter and each of the adjusted auxiliary particle emitters. Detect whether the effect of the initial particle effect matches the effect description information; If they match, the initial particle effect is determined to be the target particle effect; if they do not match, the initial particle effect is adjusted according to the effect description information to obtain the target particle effect.

10. The method according to claim 1, characterized in that, The target database is constructed in the following manner: Get multiple candidate particle effects; Extract the candidate particle emitter from each of the candidate particle effects; For each candidate particle emitter, render the emitter frame sequence corresponding to the candidate particle emitter, and determine the emitter description information corresponding to the candidate particle emitter based on the emitter frame sequence; For each candidate particle emitter, the emitter semantic data corresponding to the candidate particle emitter is determined based on the emitter description information corresponding to the candidate particle emitter, and the candidate particle emitter and its corresponding emitter semantic data are stored in the target database.

11. The method according to claim 10, characterized in that, The step of extracting the candidate particle emitter from each of the candidate particle effects includes: For each candidate particle effect, multiple basic particle emitters are extracted from the candidate particle effect, and the candidate particle emitter is determined based on the reference relationship between the multiple basic particle emitters.

12. The method according to claim 10, characterized in that, The step of determining the emitter description information corresponding to the candidate particle emitter based on the emitter frame sequence includes: Using a large visual language model, initial emitter description information corresponding to the candidate particle emitter is generated based on the emitter frame sequence. The initial emitter description information is expanded using synonyms to obtain the expanded emitter description information corresponding to the candidate particle emitter. The initial emitter description information and the extended emitter description information corresponding to the candidate particle emitter are used as the emitter description information corresponding to the candidate particle emitter.

13. The method according to claim 10, characterized in that, The method further includes: For each candidate particle effect, render the effect frame sequence corresponding to the candidate particle effect, and determine the candidate effect description information corresponding to the candidate particle effect based on the effect frame sequence. For each candidate particle effect, based on the candidate effect description information corresponding to the candidate particle effect, the effect semantic data corresponding to the candidate particle effect is determined, and the candidate particle effect and its corresponding effect semantic data are stored in the target database.

14. The method according to claim 1, characterized in that, The acquisition of special effects description information includes: Obtain basic description information input through the description information input control in the special effects creation panel, and obtain special effects style information configured through the style configuration control in the special effects creation panel; The special effects description information is determined based on the basic description information and the special effects style information.

15. The method according to claim 1, characterized in that, The method further includes: The target particle effect is displayed in the effect preview window, and the emitter composition view corresponding to the target particle effect is also displayed. The emitter composition view is used to indicate the reference particle effect to which the target particle emitter originally belonged in the target particle effect.

16. The method according to claim 1, characterized in that, The method further includes: In response to the effect export operation triggered for the target particle effect, a target effect file corresponding to the target particle effect is generated, wherein the format of the target effect file is an effect file format supported by the game engine.

17. A special effects generation device, characterized in that, The device includes: The information acquisition module is used to acquire special effect description information, which describes the target particle special effect to be generated; The element determination module is used to determine the element description information corresponding to each visual element included in the target particle effect based on the effect description information. The first retrieval module is used to retrieve candidate particle effects that satisfy the second semantic matching condition with the effect description information in the target database, based on the effect description information, as reference particle effects; wherein, the target database stores multiple candidate particle effects and their corresponding effect semantic data, and also stores multiple candidate particle emitters and their corresponding emitter semantic data, wherein the candidate particle emitters are extracted from the candidate particle effects. The emitter retrieval module is used to, for each visual element, according to the element description information corresponding to the visual element, retrieve from each of the candidate particle emitters stored in the target database and extracted from the reference particle effects the candidate particle emitter that satisfies the first semantic matching condition with the visual element, and use it as the target particle emitter corresponding to the visual element. The special effects generation module is used to generate the target particle special effects based on the target particle emitter corresponding to each of the visual elements.

18. A computer device, characterized in that, The device includes a processor and a memory; The memory is used to store computer programs; The processor is configured to execute the special effects generation method according to any one of claims 1 to 16 according to the computer program.

19. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a computer program that, when executed by an electronic device, implements the special effects generation method according to any one of claims 1 to 16.

Images