Just in Time Video Transcoding Context Computation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video transcoding systems face significant delays due to a linear relationship between transcoding time and video length, making it impractical for users to access content in a timely manner, especially for longer videos.
Innovation Solution
Implementing a Just in Time (JIT) transcoding method that computes context for each segment independently, allowing for predictable media segment sizes, playback timestamps, and generating primed and real audio segments to enable serial transcoding, thereby decoupling transcoding time from video length.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional batch transcoding is used, then video quality and encoding accuracy are improved, but transcoding time increases linearly with video length
Solution Approach 1:
The patent divides the video into multiple segments and processes them independently using JIT transcoding. Each segment is transcoded separately with pre-computed context information, allowing parallel processing and eliminating the linear time dependency on total video length. This segmentation enables the system to maintain quality while reducing overall transcoding time.
Solution Approach 2:
The patent performs preliminary actions by pre-computing context information (such as scene boundaries, audio cues, and metadata) before the actual transcoding process. This pre-computation allows each segment to be processed independently without requiring sequential analysis of previous segments, thereby reducing the transcoding time from linear to constant relative to video length.
2Reliability
If batch transcoding is used to ensure complete video processing, then all video segments are available for playback, but user waiting time increases significantly
Solution Approach 1:
The video is segmented into multiple independent units that can be processed and delivered separately. The JIT transcoding system processes segments as they are requested rather than waiting for complete batch processing, allowing users to start playback of available segments while others are still being processed in the background.
Solution Approach 2:
Context information and segment metadata are pre-computed and stored before playback requests. This preliminary preparation enables the system to quickly process and deliver video segments on-demand without requiring time-consuming analysis during the actual playback request, ensuring both reliability and reduced waiting time.
3Manufacturing precision
If context from prior segments is collected for accurate transcoding, then encoding accuracy is improved, but processing complexity and time increase
Solution Approach 1:
The patent segments the video processing into independent units with pre-computed context. Each segment contains or has access to its own context information (scene boundaries, audio synchronization points, metadata) that was prepared in advance, eliminating the need for complex sequential context collection from previous segments during processing.
Solution Approach 2:
Context information is pre-computed and stored before the actual transcoding process. This includes scene boundaries, audio cues, timing information, and other metadata that would otherwise need to be collected during processing. By performing this work preliminarily, the system maintains encoding accuracy while significantly reducing processing complexity and time.
Data Source
AI summary
Systems and methods for just in time (JIT) video on demand (VoD) transcoding using a computed context are provided. The context for each segment is computed rather than collected from the prior segment, thereby allowing for very short playback timing compared to batch transcoding techniques. Computed context requires the setting of a predictable media segment size, computing of a predictable playback time stamp (PTS) as a function of the predictable media segment size and a derived time offset value, and generating a prior audio segment including a priming portion and discard samples, and a “real” audio segments including playback samples. Then generating another primed segment for the time period of the “real” audio segment, and a subsequent “real” audio segment. This doubles the computational resources required over batch audio transcoding.