CN-122023584-A - Brand image generation method and related equipment

CN122023584ACN 122023584 ACN122023584 ACN 122023584ACN-122023584-A

Abstract

The embodiment of the application discloses a brand image generation method and related equipment, wherein the method comprises the steps of obtaining a target raw image large model; the target raw image large model is obtained through sample image training of different types of target brand styles, and current image descriptive contents input by a user are input into the target raw image large model to generate brand images conforming to the target brand styles and the current image descriptive contents. The target generating map large model establishes a technical mechanism of organically integrating brand styles and user personalized demands, visual information of brand-specific styles can be extracted through a model algorithm, an image generation model conforming to the brand-specific styles is trained and generated, accurate adaptation of user personalized originality and the brand styles is achieved, and uniformity of the generated brand images about brand vision is guaranteed.

Inventors

YANG HAORAN
LIU CAIJUN
LUO YI
ZHANG YUJIN
WU ZEPENG
LIU YULONG
LIU YIXIN
ZHONG HONGBIN
HUO YONGHONG
LIU WEI

Assignees

喜茶(深圳)企业管理有限责任公司

Dates

Publication Date: 20260512
Application Date: 20260129

Claims (10)

1. A brand image generation method, the method comprising: The method comprises the steps of obtaining a target raw image large model, wherein the target raw image large model is obtained through sample image training of different types of target brand styles; The current image descriptive content input by the user is input into the target raw graph big model to generate a brand image conforming to the target brand style and the current image descriptive content.
2. The brand image generation method of claim 1, wherein the step of obtaining the target raw image large model comprises: semantic extraction is carried out on the sample image by using a prompt word back-pushing function of a graphical interface tool so as to generate an initial prompt word set for describing the target brand style; The initial prompt word set is subjected to content correction and/or supplementation to obtain a target prompt word set for identifying an image, wherein the supplemented prompt word meets the target brand style or is used for limiting the contradiction between the image content and the target brand style; Marking the sample image through the target prompt word set, and inputting the marked sample image and the target prompt word set into the graphical interface tool so that the graphical interface tool carries out training workflow based on an initial graph generation large model and an introduced low-rank matrix to obtain a trained style guiding model, wherein the low-rank matrix comprises network structure parameters needing training and learning; And guiding the initial raw graph large model to be adjusted into a target raw graph large model fused with the target brand style through the style guiding model.
3. The brand image generation method of claim 2, wherein the step of obtaining the target raw graphic large model further comprises, prior to inputting the marked sample image and the target hint vocabulary into the graphical interface tool: and carrying out size unification treatment on the sample image, wherein the sample image with the unified size is at least used for carrying out marking treatment.
4. The brand image generation method of claim 1, wherein the step of obtaining the target raw image large model comprises: the style control prompt word stock is at least extracted or summarized based on the visual content of the sample image; and performing iterative training on the initial large graph generation model through the style control prompt word stock to obtain a target large graph generation model.
5. The brand image generation method of claim 1, wherein after generating a brand image conforming to the target brand style and the current image descriptive content, the method further comprises: detecting sensitive content of the brand image; If sensitive content is detected, prohibiting the brand image from being output to a downstream node, wherein the downstream node comprises an interactive terminal and/or printing equipment; And if the sensitive content is not detected, allowing the brand image to be output to the downstream node.
6. The brand image generation method of claim 1 or 5, wherein after generating a brand image conforming to the target brand style and the current image descriptive content, the method further comprises: and adjusting printing parameters of the brand image so that the brand image is matched with the printing requirements of printing equipment.
7. A brand image generating apparatus, comprising: The system comprises an acquisition unit, a target raw image generation large model, a target brand style generation large model generation unit and a target brand style generation large model generation unit, wherein the target raw image generation large model is obtained through sample image training of different types of target brand styles; And the processing unit is used for inputting the current image descriptive content input by the user into the target generating graph big model so as to generate a brand image conforming to the target brand style and the current image descriptive content.
8. A computer program product comprising computer readable instructions that, when run on an electronic device, cause the electronic device to implement the brand image generation method of any of claims 1 to 6.
9. An electronic device comprising at least one processor and a memory coupled to the processor, wherein: the memory is used for storing a computer program; The processor is configured to execute the computer program to enable the electronic device to implement the brand image generation method of any one of claims 1 to 6.
10. A computer storage medium carrying one or more computer programs which, when executed by an electronic device, enable the electronic device to implement a brand image generation method of any one of claims 1 to 6.

Description

Brand image generation method and related equipment Technical Field The embodiment of the application relates to the technical field of image generation, in particular to a brand image generation method and related equipment. Background Along with the consumption upgrading of the tea industry, the personalized experience demand of users on tea products is continuously increased, brand images (which can be called cup paste images or style images) are used as important carriers for brand interaction with users, so that the brand identity transmission function is born, and the demands of personalized expression, scene adaptation (such as holiday souvenir and gift giving) and the like of the users are met. Meanwhile, the management and control requirements of enterprises on brand visual uniformity are increasingly improved, and the brand image is required to be strengthened to be specific in style identification (such as unique lines, logo graphics, customized fonts and the like) through the brand image, so that the brand image fragmentation is avoided. At present, an existing brand image generation method comprises the steps that an enterprise design team makes a plurality of sets of fixed cup paste templates (comprising brand mark elements such as Logo, exclusive lines and standard formats) according to brand styles and stores the fixed cup paste templates in a template database, a system provides a template selection interface, a user selects a target template from a template library through an online order platform or a store terminal, the user can adjust preset modifiable fields (such as text contents of a nickname, blessing and the like of the user and selection of part of decoration elements) in the template, the core visual structure and the brand elements of the template cannot be changed, and after receiving a user adjustment instruction, the system synthesizes a final cup paste image and outputs the final cup paste image to a tea cup through printing equipment. Therefore, the existing brand image generation method adopts a technical architecture of 'manual preset fixed template and limited field modification', but the core technology is limited to template library management and simple text replacement, so that a user can only make a small amount of adjustment in a preset frame, can not freely express personalized originality, and can not adapt to diversified and customized requirements of the user on cup pastes. Meanwhile, the adaptation degree of the brand style and the user creative cannot be dynamically adjusted by the immobilized template, so that the user is in personalized adjustment and the enterprise brand style are disjointed, the personalized and brand adjustability organic unification cannot be realized, and the consistency of brand visual transmission is difficult to guarantee. Disclosure of Invention The embodiment of the application provides a brand image generation method and related equipment, which are used for realizing accurate adaptation of user personalized creative and brand styles and guaranteeing uniformity of a generated brand image on brand vision. An embodiment of the present application provides a brand image generating method, including: The method comprises the steps of obtaining a target raw image large model, wherein the target raw image large model is obtained through sample image training of different types of target brand styles; The current image descriptive content input by the user is input into the target raw graph big model to generate a brand image conforming to the target brand style and the current image descriptive content. Optionally, the step of obtaining the target raw graph large model includes: semantic extraction is carried out on the sample image by using a prompt word back-pushing function of a graphical interface tool so as to generate an initial prompt word set for describing the target brand style; The initial prompt word set is subjected to content correction and/or supplementation to obtain a target prompt word set for identifying an image, wherein the supplemented prompt word meets the target brand style or is used for limiting the contradiction between the image content and the target brand style; Marking the sample image through the target prompt word set, and inputting the marked sample image and the target prompt word set into the graphical interface tool so that the graphical interface tool carries out training workflow based on an initial graph generation large model and an introduced low-rank matrix to obtain a trained style guiding model, wherein the low-rank matrix comprises network structure parameters needing training and learning; And guiding the initial raw graph large model to be adjusted into a target raw graph large model fused with the target brand style through the style guiding model. Optionally, before inputting the marked sample image and the target prompt word set into the graphical interface tool, the step of obtaining