Avatar Creation Resource Allocation for Low-Latency Video Viewing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing avatar creation technologies face challenges with processing power, bandwidth constraints, and latency issues, particularly in virtual reality applications, leading to suboptimal user experiences.
Innovation Solution
A method utilizing machine learning to distribute avatar creation processes across cloud-based, edge-based, and local processing resources, prioritizing edge servers for reduced latency and efficient bandwidth usage, while integrating 3D modeling algorithms to generate high-quality avatars during video consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If avatar creation processing is performed using traditional centralized cloud processing, then processing power and modeling accuracy are improved, but latency and bandwidth consumption increase
Solution Approach 1:
The patent segments the avatar creation processing into multiple components and distributes them across different processing locations: local processing resources (for initial processing and reducing latency), edge processing resources (for intermediate processing), and cloud processing resources (for final high-quality rendering). This segmentation allows each component to be processed at the most appropriate location, balancing accuracy and latency requirements.
2Manufacturing precision
If high-quality 3D modeling algorithms are used to create realistic avatars, then avatar representation quality is improved, but processing power requirements and computational complexity increase
Solution Approach 1:
The patent applies local quality by using different processing resource qualities at different locations: local devices use standard processing resources for basic avatar creation, edge devices use enhanced processing resources for improved quality, and cloud devices use high-performance computing resources for maximum quality. Each location uses the quality level appropriate to its capabilities and requirements.
3Loss of time
If avatar processing operations are performed locally on user devices, then latency is reduced and user experience is improved, but processing power constraints and bandwidth usage increase
Solution Approach 1:
The patent adds a spatial dimension to the processing architecture by introducing edge processing resources as an intermediate layer between local and cloud processing. This creates a three-tier processing dimension (local-edge-cloud) that allows the system to balance latency and processing power requirements by routing operations to the appropriate tier based on real-time conditions.
4Loss of energy
If distributed processing resources are used to create avatars, then bandwidth consumption is optimized and latency is reduced, but system complexity and resource coordination difficulty increase
Solution Approach 1:
The patent implements feedback mechanisms where the system continuously monitors processing performance, latency, bandwidth usage, and resource availability across the distributed architecture. Based on this feedback, the system dynamically adjusts and redirects processing operations to optimize performance and balance the load across local, edge, and cloud resources.
Data Source
AI summary
According to aspects disclosed herein, a method of using machine learning as part of creating avatars while watching video content is provided. According to an aspect, a method is configured for receiving a request to create an avatar based on the video content and, in response to receiving the request to create the avatar based on the video content, determining one or more processing resources to use to create the avatar based on an output of a machine learning engine, wherein the one or more processing resources include a cloud-based processing resource, an edge-based processing resource, and a local processing resource. The method includes allocating avatar processing operations to the one or more processing resources to create the avatar based on the output from the machine learning engine. The method further includes creating the avatar using the one or more processing resources, and storing the avatar in an avatar database.