片段级双通道多模态约束的广告视频生成方法以及系统
By introducing a knowledge base of subjects, templates, and generation tools, and combining a multimodal model and tool scoring mechanism, the problems of unstable script planning, poor multimodal consistency, and insufficient device adaptation in advertising video generation are solved. This achieves high-quality advertising video generation with clear logic and visual consistency, adapting to the application needs of multiple industries and multiple terminals.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHENZHEN ZHIXIANG FUTURE TECHNOLOGY CO LTD
- Filing Date
- 2025-11-06
- Publication Date
- 2026-07-17
AI Technical Summary
Existing advertising video generation methods suffer from problems such as unstable script planning, poor multimodal consistency, uncontrollable tool selection, and insufficient device adaptation, making it difficult for the generated videos to meet commercial-grade requirements in terms of logical coherence, visual consistency, and device adaptability.
An advertising video generation method with segment-level dual-channel multimodal constraints is adopted. By introducing a subject knowledge base, a template knowledge base, and a generation tool knowledge base, a structured script is generated. The multimodal model is combined to fuse image and text information, and a tool scoring and rollback mechanism is introduced to optimize the generation effect and display quality.
It achieves clear logic and unified visual style in advertising video generation, improves generation stability and efficiency, adapts to differentiated expansion across multiple industries and terminals, and has good engineering feasibility and commercial application value.
Smart Images
Figure CN121691835B_ABST