Image Frame Prediction Using Motion Vectors to Cut GPU Load
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Game performance on terminal devices, particularly mobile phones, is hindered by high resource consumption and frame drops due to the processing demands of heavy-load games, leading to issues like fast power consumption and heating.
Innovation Solution
An image prediction method that calculates motion vectors between successive frames to generate a prediction frame, reducing the need for real-time rendering by combining fixed objects directly onto predicted frames.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If real-time rendering is performed for all objects in heavy-load games, then image quality and game effects are improved, but GPU resource consumption and device load increase significantly
Solution Approach 1:
The patent applies preliminary action by predicting the positions of dynamic objects in advance based on historical motion data from previous frames. Instead of rendering all objects in real-time, the system pre-calculates predicted positions and only renders objects that have actually moved to their predicted positions, significantly reducing GPU workload while maintaining visual quality.
2Reliability
If all drawing instructions are executed on terminal device, then complete game scenes are rendered, but frame drops and performance issues occur
Solution Approach 1:
The patent extracts and identifies only the necessary drawing instructions related to dynamic objects that have actually moved. By separating dynamic object rendering from static object rendering, and only processing changed elements, the system maintains rendering completeness while dramatically improving frame rendering efficiency and eliminating frame drops.
3Measurement precision
If motion prediction is performed using multiple historical frames, then prediction accuracy is improved, but calculation complexity increases
Solution Approach 1:
The patent segments the motion prediction process into distinct stages: extracting motion vectors from multiple historical frames, calculating predicted positions based on these vectors, and validating predictions against actual object positions. This segmentation allows the system to achieve high prediction accuracy through multi-frame analysis while managing calculation complexity through structured processing steps.
Data Source
AI summary
This application provides an image prediction method, an electronic device, and a storage medium. The prediction method includes: obtaining a drawing instruction stream of a first image frame, wherein the first image frame contains one or more first objects; calculating a first motion vector between the one or more first objects and the one or more third objects, wherein the one or more third objects match the one or more first objects; obtaining a second motion vector, wherein the second motion vector is a motion vector between the one or more third objects and one or more fourth objects in a third image frame; calculating a third motion vector based on a first motion vector and a second motion vector; obtaining a first prediction image frame based on the first motion vector, the third motion vector, and the one or more first objects.