Avatar Image Generation Using Gaze Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image generation methods for displaying avatars in real-time communication require continuous data transfer of real facial videos, leading to increased communication bandwidth and reduced speed, which compromises the realism and smoothness of communication.
Innovation Solution
An image generation method where a first terminal apparatus displays a model image of a second user and determines if the gaze of the first user is directed towards it. Based on this determination, the second terminal apparatus generates a model image of the first user either using a captured image if the gaze is directed, or using pre-acquired data if the gaze is not directed, thereby reducing unnecessary data transfer.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If real facial video data is continuously transferred between terminal apparatus to paste onto avatars, then realistic communication is achieved, but communication bandwidth increases and communication speed decreases
Solution Approach 1:
The patent extracts only the necessary visual information (facial video data) from the complete user data, and further extracts only the essential features needed for avatar generation. This selective extraction reduces the volume of data that needs to be continuously transmitted, thereby maintaining communication speed while still achieving realistic communication quality through avatar display.
Solution Approach 2:
The patent performs preliminary actions by pre-processing and storing essential user data (such as facial features, avatar models, and communication history) before actual communication occurs. This preparation allows the system to generate avatars quickly during communication without requiring continuous transmission of large amounts of raw video data, thus maintaining both realism and communication speed.
2Reliability
If real facial video data is continuously received and processed, then avatar realism is maintained, but communication bandwidth consumption increases
Solution Approach 1:
The patent creates simplified copies of the user's appearance through avatar generation rather than transmitting and processing complete real-time video streams. These avatar copies maintain the essential visual characteristics needed for realistic communication while consuming significantly less bandwidth. The system generates avatar images based on pre-acquired user data and necessary real-time inputs, reducing bandwidth consumption while preserving avatar realism.
Solution Approach 2:
The patent changes the parameters of data transmission by converting continuous high-resolution video streams into discrete, optimized avatar image parameters. This includes adjusting resolution, color depth, and frame rate to match the minimum requirements for realistic avatar display, thereby reducing bandwidth consumption while maintaining sufficient realism for effective communication.
Data Source
AI summary
An image generation method includes generating, by a second terminal apparatus, in a case in which it is determined by a first terminal apparatus that the gaze of a first user is directed toward a displayed model image of a second user, a model image of the first user based on a captured image of the first user received from the first terminal apparatus. The image generation method includes generating, by the second terminal apparatus, in a case in which it is determined by the first terminal apparatus that the gaze of the first user is not directed toward the displayed model image of the second user, a model image of the first user based on data of the first user acquired in advance.