AI Character Generation for Digital Platforms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital interactive platforms in the sports and iGaming industry limit user experience by restricting avatars to predefined activities, failing to provide dynamic and personalized interactions.
Innovation Solution
A computer-implemented system that generates characters with face expressions based on user actions, using diffusion models, GAN models, and pix2pix models, to optimize user experience in digital interactive platforms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If avatars are restricted to predefined activities, then system complexity is reduced and ease of manufacture is improved, but user experience and adaptability deteriorate
Solution Approach 1:
The system uses image-to-image translation models (pix2pix, cycle GAN) to copy and transform user-uploaded images into animated character representations. This allows the system to generate customizable avatars without manual creation processes, maintaining ease of manufacture while enabling diverse character designs that can potentially perform various activities.
Solution Approach 2:
The system employs diffusion models and generative adversarial networks to dynamically generate and transform character parameters (visual appearance, expressions, animations) based on user inputs and platform context. This enables avatars to adapt their appearance and behavior dynamically while maintaining a relatively simple underlying system architecture.
2Productivity
If dynamic and personalized character interactions are implemented, then user experience and engagement are improved, but system complexity and computational resources increase
Solution Approach 1:
The system pre-trains diffusion models, GANs, and pix2pix models with large datasets of character images, expressions, and animations before deployment. This preliminary training enables the models to generate personalized character interactions in real-time without requiring complex runtime processing, thus improving user engagement while managing system complexity.
Solution Approach 2:
The patent introduces AI models (diffusion models, GANs, pix2pix) as intermediaries between user inputs and character generation. These intermediary models handle the complex computational transformations, allowing the rest of the system to remain relatively simple while still delivering personalized and dynamic character interactions that enhance user engagement.
3Adaptability or versatility
If AI models are used to generate characters in real-time, then adaptability and personalization are improved, but processing time and computational energy increase
Solution Approach 1:
The system performs extensive model training and parameter optimization in advance, storing pre-computed model weights and configurations. When generating characters in real-time, the system only needs to execute inference with these pre-trained models, significantly reducing generation time while maintaining high adaptability and personalization capabilities.
Solution Approach 2:
The patent implements dynamic character generation where the same base models can rapidly adapt to different user inputs by adjusting parameters and conditions during inference. This dynamic approach allows the system to maintain high customization capability while keeping generation time acceptable through efficient model architecture and optimization techniques.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A computer-implemented system (106) for generating characters associated with digital interactive platforms to optimize user experience, is disclosed. The computer-implemented system (106) is configured to: pre-create characters for users (102) based on first inputs from user devices (104) of users (102), by diffusion models, cycle (GAN) model, and pix2pix model; select at least one character with face expressions during playing of video contents associated with digital interactive platforms based on historical data and first inputs; obtain second inputs from the user devices (104) of users (102) upon actions performed by the users (102) during digital interactive platforms, and current state of digital interactive platforms; and determine corresponding characters providing the face expressions to be played as animated media contents with the video contents, based on second inputs from the user devices (104) of the users (102), to optimize the user experience in digital interactive platforms, by a machine learning model.