AI Scene Composition Using Control Elements and Sub-Prompts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional digital art creation systems face challenges in generating structured scenes with consistent quality, as they often require significant time and effort, and existing automatic generation methods fail to preserve the composition and structure of user descriptions.
Innovation Solution
A target scene generation system that utilizes natural language processing to decompose textual descriptions into sub-prompts, leveraging generative AI to create structured scenes with desired styles and visual elements, preserving the composition by identifying control language and deriving groupings of sub-prompts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If artists manually create each object or adapt previously created objects to new scenes, then the scene can be customized and reused, but the time and effort required increases significantly
Solution Approach 1:
The system segments the scene creation process by automatically generating individual visual elements (objects, backgrounds, characters) as separate components that can be independently created and then composed together. This allows artists to reuse pre-generated elements across multiple scenes without manual recreation, significantly reducing time while maintaining customization through selective assembly of segmented components.
2Manufacturing precision
If artists with varying skill levels create scenes manually, then creative control is maintained, but the quality of scenes becomes inconsistent
Solution Approach 1:
The system introduces an intermediary AI-based scene composition system that acts as a mediator between the artist's intent and the final scene output. This intermediary automatically handles the complex task of assembling and refining visual elements according to best practices, ensuring consistent quality while preserving the artist's creative control through prompt-based direction and selection of generated options.
3Ease of manufacture
If conventional tools are used for scene creation, then the process is straightforward, but computing resources are inefficiently utilized
Solution Approach 1:
The system performs preliminary actions by pre-generating a library of visual elements, backgrounds, and scene components that can be reused across multiple scene creation tasks. This preliminary generation reduces the need for repeated full-scene rendering and algorithm execution, improving computing resource efficiency while maintaining ease of use through simple assembly of pre-prepared components.
Data Source
AI summary
A method includes receiving a natural language description of an image to be generated using a machine learning model. The method further includes extracting, from the natural language description of the image to be generated, a control element and a sub-prompt. The method further includes identifying a relationship between the control element and the sub-prompt based on the natural language description of the image to be generated. The method further includes generating, by the machine learning model, an image based on the control element, the sub-prompt, and the relationship. The image includes visual elements corresponding to the control element and the sub-prompt.