Latent Diffusion Asset Creation for Rapid Character Variations

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The development of video game characters is time-consuming, delaying the release of the game, as detailed representations require extensive manual adjustments, prolonging the development cycle.

Innovation Solution

A method using generative artificial intelligence to iteratively generate and edit attributes of a target object, allowing dynamic creation of visual assets through phases of input, decomposition, selection, editing, and merging, utilizing latent diffusion models for efficient character development.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If manual creation of detailed character representations is used, then manufacturing precision is improved, but productivity deteriorates

Engineering Contradiction:
Improvecharacter representation detailVSAvoidcharacter development speed
Core Design Contradiction:
Manufacturing precisionVSProductivity

Solution Approach 1:

The patent replaces manual mechanical creation processes with an AI-based system that automatically generates character representations. The system uses generative models to create detailed character images from text prompts, eliminating the need for manual drawing while maintaining high detail quality. This substitution of mechanical manual work with automated AI processing directly resolves the contradiction between precision and productivity.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent changes the fundamental parameter of character creation from manual artistic rendering to AI-generated synthesis. By using generative models that can produce multiple variations from a single prompt, the system achieves both high detail precision and rapid generation speed. The ability to iterate through multiple versions quickly allows for precise detail control without sacrificing development time.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If iterative generation of multiple character versions is used, then productivity is improved, but device complexity deteriorates

Engineering Contradiction:
Improvecharacter development speedVSAvoidAI system complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the character creation process into distinct functional stages: prompt generation, image generation, attribute decomposition, variation selection, and blending. This segmentation allows each component to be optimized independently while working together as an integrated system. The prompt generator handles text processing, the generative model handles image synthesis, and the attribute system handles analysis and selection, dividing the overall complexity into manageable functional modules.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary attribute decomposition layer between the generative model and the final character selection. This intermediary system analyzes generated images, extracts meaningful attributes, and creates variations that can be systematically combined. The attribute system acts as a mediator that translates raw AI-generated images into structured character options, simplifying the interaction between the complex generative model and the user selection process.

Inventive Principle:
Principle #24Intermediary (Mediator)

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

Facilitates rapid creation of multiple character variations, reducing development time and enhancing creativity in video game character design.

Implementation Method 1

generating a plurality of images of the target object using an image generation artificial intelligence system configured for implementing latent diffusion

Methodology Applied
Scientific EffectLatent diffusion: Diffusion

Data Source

PatentEP4592897A1Asset creation using generative artificial intelligence
Publication Date: 2025.07.30 SONY INTERACTIVE ENTERTAINMENT LLC
  • EP4592897A1 patent drawingFigure 1A
  • EP4592897A1 patent drawingFigure 1B
  • EP4592897A1 patent drawingFigure 2A

AI summary

A method including collecting one or more inputs, each of which describes a target object. The method including generating a plurality of images of the target object using an image generation artificial intelligence system configured for implementing latent diffusion based on the one or more inputs. The method including decomposing the target object into a first plurality of attributes based on the plurality of images of the target object, wherein each of the plurality of attributes includes one or more variations. The method including receiving selection of one or more of a plurality of variations of the plurality of attributes. The method including blending the one or more of the plurality of variations of the plurality of attributes that have been selected into one or more options of the target object.