Image Generation Apparatus Using Feature Quantity Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image generation tools struggle to automatically create output images that match user desires, especially when multiple scenes with aligned orders are required, as they often lack consideration for material image composition and storyline coherence.
Innovation Solution
An image generation apparatus that includes a storage unit for selection conditions, a first material image selection unit, a feature quantity obtaining unit, a selection condition row selection unit, a second material image selection unit, and an output image generation unit, which selects and arranges material images based on feature quantities and selection conditions to generate output images with aligned scenes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If a template is designated by theme matching, then the automation level is improved, but the manufacturing precision of output image quality deteriorates because feature quantities like composition are not considered
Solution Approach 1:
The patent changes the selection parameters from simple theme matching to comprehensive feature quantity comparison including composition analysis. By evaluating multiple parameters (theme, composition, color, etc.) simultaneously, the system achieves both automation and high output image quality that matches user intentions.
Solution Approach 2:
The patent adds a new dimension of composition analysis to the traditional theme-based template selection. By incorporating spatial arrangement and compositional features as additional selection criteria, the system transitions from one-dimensional theme matching to multi-dimensional image characteristic evaluation.
2Ease of operation
If only correlation of feature quantities between adjacent material images is considered, then the ease of operation is improved, but the manufacturing precision of storyline coherence deteriorates because scenes with multiple cuts cannot be automatically generated
Solution Approach 1:
The patent performs preliminary analysis of selection condition rows that define multi-cut scene structures before image selection. By pre-establishing the relationship between selection conditions and scene compositions, the system can automatically generate coherent multi-cut scenes without requiring users to manually plan storyline sequences.
Solution Approach 2:
The patent introduces selection condition rows as an intermediary layer between user input and material image selection. These condition rows encode storyline requirements and scene structures, acting as mediators that translate high-level narrative intentions into specific image selection and arrangement decisions.
3Manufacturing precision
If manual selection and arrangement of material images is performed, then the manufacturing precision of output image quality is improved, but the productivity deteriorates due to challenging processing requirements
Solution Approach 1:
The patent enables the system to automatically perform image selection and arrangement by comparing material image feature quantities against selection condition rows. The system serves itself by autonomously evaluating multiple images against multiple criteria and generating optimized sequences without manual intervention, achieving both high quality and efficiency.
Solution Approach 2:
The patent implements a feedback mechanism where selection condition rows define desired image characteristics and the system evaluates material images against these conditions. The comparison results feed back into the selection process, allowing iterative optimization of output image quality while maintaining automation.
Data Source
AI summary
An image generation apparatus stores a plurality of selection condition rows including a plurality of selection conditions used for selecting an image from a plurality of material images, selects at least one material image from the plurality of material images as a first material image, obtains a feature quantity of the first material image, selects a selection condition row which is stored in the storage unit and includes a selection condition including the obtained feature quantity, selects a second material image from the plurality of material images based on a selection condition row which has been selected, and generates an image based on the first and the second material images which have been selected and the selection condition row which has been selected.