Variable Video Chunk Duration Based on Points of Interest
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing streaming technologies face inefficiencies due to fixed chunk sizes in video streaming, leading to wasted bandwidth and reduced video quality, as they do not account for interesting points in the video that users are likely to seek, resulting in suboptimal chunk duration and keyframe placement.
Innovation Solution
A system and method that utilize analytics to determine variable chunk durations based on points of interest in a video file, ensuring keyframes occur at or near these points, thereby optimizing chunk lengths and minimizing waste while maintaining random access capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If fixed chunk sizes are used in video streaming, then implementation simplicity is maintained, but bandwidth efficiency deteriorates and video quality suffers
Solution Approach 1:
The patent implements dynamic chunk duration adjustment by analyzing points of interest in the video content and adapting chunk lengths accordingly. Instead of using fixed chunk sizes, the system varies chunk durations to align with semantically meaningful boundaries, thereby reducing wasted bandwidth while maintaining implementation feasibility through automated analysis algorithms.
Solution Approach 2:
The system changes the parameter of chunk duration from a fixed value to a variable determined by content analysis. By computing optimal chunk durations based on points of interest and keyframe locations, the patent transforms the static parameter into a dynamic one that adapts to video content characteristics, improving bandwidth efficiency without excessive complexity.
2Ease of operation
If keyframes are placed at fixed intervals, then decoding simplicity is maintained, but alignment with user-interest points deteriorates
Solution Approach 1:
The patent performs preliminary analysis of the video content to identify points of interest before generating the streaming chunks. By pre-computing the optimal locations for keyframes based on content semantics and user behavior patterns, the system ensures that keyframes are strategically placed to align with user-interest points while maintaining decoding efficiency through controlled intervals.
Solution Approach 2:
The system uses feedback from points of interest analysis to adjust keyframe placement. By continuously referencing the identified semantically important moments in the video, the system optimizes keyframe locations to coincide with user-interest points, thereby improving precision without completely abandoning regular intervals for decoding simplicity.
3Loss of energy
If chunk duration is increased, then network load is reduced, but random access capability deteriorates
Solution Approach 1:
The patent segments the video stream into variable-length chunks based on points of interest rather than using uniform segmentation. This intelligent segmentation creates larger chunks in less critical sections (reducing network load) while ensuring that chunks are bounded by keyframes at meaningful locations, thereby preserving random access capability when needed.
Solution Approach 2:
The system dynamically adjusts chunk duration to balance network load reduction with random access requirements. By varying chunk lengths based on content importance and user behavior patterns, the patent achieves longer chunks for load reduction while maintaining the ability to access specific portions of the video efficiently through strategically placed keyframes.
Data Source
AI summary
A method is provided in one example embodiment and includes receiving analytics information for a video file at a transcoder, the analytics information includes a plurality of points of interest. The method also includes determining a plurality of durations associated with a plurality of chunks of video data associated with the video file, the determining includes identifying a minimal distance between certain times associated with the plurality of points of interest and keyframe times. The method also includes transcoding the video file using the plurality of durations. The video file can be received from an analytics extractor that is to generate the plurality of points of interest.