Video Encoding Ladder from Partial Decode Proxy Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video transcoding methods are computationally intensive and inefficient, requiring full decoding and re-encoding of video files to derive bitrate and resolution settings, leading to high latency and resource wastage.
Innovation Solution
A method for just-in-time transcoding that involves partial decoding of video files to generate proxy representations, extracting entropy and quantization characteristics, and using models to derive target bitrates and resolutions based on viewer characteristics and historical data, thereby generating an encoding ladder for efficient streaming.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If full decoding and re-encoding of video files is performed to derive bitrate and resolution settings, then accurate bitrate and resolution settings can be obtained, but computational intensity and processing time increase significantly
Solution Approach 1:
The patent extracts only the essential bitrate and resolution settings from the video file metadata without performing full decoding and re-encoding. By taking out only the necessary information (bitrate, resolution, duration) from the file properties, the system avoids computationally intensive processing while still obtaining accurate settings for transcoding operations.
Solution Approach 2:
The patent performs preliminary extraction of bitrate and resolution settings from video file metadata before actual transcoding operations. This preliminary action allows the system to prepare encoding parameters in advance without requiring full video decoding, thereby improving processing efficiency while maintaining setting accuracy.
2Measurement precision
If full decoding and re-encoding is performed, then accurate video representation is obtained, but processing time and latency increase
Solution Approach 1:
The patent extracts only the necessary time-related parameters (duration, bitrate) from video file metadata without performing full decoding. This selective extraction provides sufficient information for accurate video representation and encoding parameter derivation while avoiding the time-consuming process of complete video decoding and re-encoding.
3Measurement precision
If full decoding and re-encoding is performed, then complete video analysis is achieved, but computational resources are wasted
Solution Approach 1:
The patent extracts only the essential computational requirements (bitrate, resolution, duration parameters) from video file metadata without performing complete decoding and re-encoding operations. This selective extraction achieves necessary video analysis for encoding parameter derivation while significantly reducing computational resource consumption compared to full processing.
Solution Approach 2:
The patent applies partial action by processing only the necessary portions of video data (metadata parameters) rather than performing complete decoding and re-encoding. This partial processing approach provides sufficient video analysis for encoding purposes while avoiding excessive computational resource consumption associated with full video processing.
Data Source
AI summary
A method includes: receiving a video file from a first publisher; partially decoding the video file based on visual characteristics in the video file to generate a proxy video representation of the video file. The method further includes, for a first resolution: accessing a first model associated with the first resolution, the first model configured to derive target bitrates based on resolutions and target viewing qualities; passing the proxy video representation and the source resolution to the first model; and receiving a first target bitrate for the first resolution returned by the first model. The method further includes: defining a first rendition, for the video file, characterized by the first resolution and the first target bitrate; generating an encoding ladder identifying the first rendition; and publishing the encoding ladder for access by a video player for streaming playback segments of the video file.