Entity-Tagged Prompt Augmentation for Contextual Media Generation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The quality of media items generated or retrieved based on text prompts is often limited by the user's ability to provide detailed and specific prompts, as effective prompt writing is a learned skill, and unskilled users may struggle to create high-quality media content.

Innovation Solution

A media processing system that uses a language generation model to replace semantic entities in a text prompt with more descriptive phrases, allowing the model to generate or retrieve media items based on an augmented prompt that better fits the intended context.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If users provide simple text prompts, then the system is easy to operate, but the quality and detail of generated media items deteriorates

Engineering Contradiction:

Improveease of prompt inputVSAvoidmedia quality

Core Design Contradiction:

Ease of operationVSManufacturing precision

Solution Approach 1:

The system automatically enhances prompt details through AI processing. The entity enhancement module identifies semantic entities in user prompts and generates enhanced prompts with additional descriptive attributes, allowing the system to serve itself by improving prompt quality without requiring user intervention or expertise in prompt engineering

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system performs preliminary enhancement of prompts before media generation. By pre-processing the prompt through entity identification and enhancement, the system prepares enriched prompts that guide high-quality media generation, ensuring detailed and accurate media output is achieved before the actual generation process

Inventive Principle:

Principle #10Preliminary action

2Manufacturing precision

If users write detailed prompts, then the quality of generated media items improves, but the difficulty of operation increases

Engineering Contradiction:

Improvemedia qualityVSAvoidease of prompt writing

Core Design Contradiction:

Manufacturing precisionVSEase of operation

Solution Approach 1:

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The AI processing system acts as an intermediary between the user's simple prompt and the final detailed prompt used for media generation. This intermediary layer translates basic user intentions into comprehensive, detailed prompts automatically, bridging the gap between ease of operation and media quality

Inventive Principle:

Principle #24Intermediary (Mediator)

3Device complexity

If conventional language models are used, then the system is simple, but the contextual accuracy and relevance of generated media deteriorates

Engineering Contradiction:

Improvesystem complexityVSAvoidcontextual accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The system segments the prompt processing into distinct functional modules: entity identification, entity enhancement, and prompt generation. This segmentation allows each module to specialize in specific tasks, improving overall contextual accuracy while maintaining manageable system complexity through modular architecture

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system performs preliminary analysis of the input prompt to identify semantic entities and their contexts before generating the final enhanced prompt. This preliminary action ensures that the subsequent prompt generation is grounded in accurate contextual understanding, improving media relevance and alignment with user intentions

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS20260017298A1Prompt augmentation based on entity tagging

Publication Date: 2026.01.15 ADOBE INC

AI summary

A method, apparatus, non-transitory computer readable medium, and system for media processing include receiving a text prompt including an entity phrase, marking the entity phrase within the text prompt to obtain a revised prompt, generating a replacement phrase by performing autoregressive token generation based on a sequence of tokens from the revised prompt, where the replacement phrase comprises a variant of the entity phrase, and generating an augmented prompt that includes the replacement phrase.