Adaptive Video Encoding Profile Optimization

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Streaming video services face challenges in determining optimal sets of resolution and bit rate levels for encoding profiles across diverse content and playback conditions, especially with codecs having numerous encoding parameters, which has traditionally relied on manual trial and error or limited computational optimization.

Innovation Solution

The techniques involve adaptive sampling of video content, prioritization of encoding parameters, rate-quality data sampling, data space pruning, and optimizing cost functions based on system constraints to generate tailored encoding ladders that accommodate varied content and playback conditions, allowing for scalable optimization of multiple encoding parameters.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If manual trial and error analysis is used to determine encoding profiles, then encoding parameter optimization can be performed, but the process is time-consuming and cannot scale to large datasets with diverse content characteristics

Engineering Contradiction:

Improveencoding parameter optimizationVSAvoidtime consumption

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent replaces manual trial-and-error analysis with an automated machine learning system that uses computational models to predict encoding parameters. The system automatically analyzes diverse content characteristics and determines optimal encoding profiles without human intervention, resolving the contradiction between optimization precision and time consumption.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system dynamically adjusts encoding parameters based on content characteristics and playback conditions. By changing parameters adaptively rather than using fixed manual settings, the system achieves optimal encoding efficiency while reducing the time required for analysis through automated parameter selection algorithms.

Inventive Principle:

Principle #35Parameter changes

2Extent of automation

If computational modules are used for search/optimization processes, then manual processes are replaced, but the optimization is limited to one or two encoding parameters

Engineering Contradiction:

Improveautomation of encoding optimizationVSAvoidnumber of encoding parameters

Core Design Contradiction:

Extent of automationVSDevice complexity

Solution Approach 1:

The patent segments the encoding parameter space into multiple independent dimensions, allowing the system to optimize each parameter separately while maintaining overall automation. This segmentation enables comprehensive optimization of 30-50 encoding parameters through coordinated computational modules without overwhelming complexity.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system transitions from optimizing single parameters to multi-dimensional parameter spaces by using machine learning models that simultaneously consider multiple encoding parameters. This dimensional expansion enables automated optimization across all encoding parameters while managing complexity through intelligent algorithms.

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

3Adaptability or versatility

If a large media content catalog with diverse content characteristics is processed, then comprehensive encoding optimization can be achieved, but determining optimal resolution and bit rate levels becomes challenging

Engineering Contradiction:

Improveencoding optimization for diverse contentVSAvoiddetermination of optimal encoding levels

Core Design Contradiction:

Adaptability or versatilityVSDifficulty of detecting and measuring

Solution Approach 1:

The patent uses machine learning models to create virtual copies of content characteristics and encoding outcomes. By training on existing data, the system can predict optimal encoding levels for new diverse content without actually encoding everything, thus achieving comprehensive optimization while reducing the measurement and detection complexity.

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The system introduces machine learning models as intermediary components between content analysis and encoding parameter determination. These models mediate the complex relationship between diverse content characteristics and optimal encoding levels, making the determination process more manageable and scalable.

Inventive Principle:

Principle #24Intermediary (Mediator)

4Manufacturing precision

If numerous encoding parameters with direct or indirect impact on bit rate and quality are optimized, then encoding profile precision is improved, but the complexity of parameter management increases

Engineering Contradiction:

Improveencoding profile precisionVSAvoidparameter management complexity

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent implements feedback mechanisms where the system continuously monitors encoding outcomes and adjusts parameters accordingly. By using feedback from quality metrics and bit rate performance, the system achieves high precision in encoding profiles while managing complexity through adaptive parameter adjustment rather than static management of all parameters.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS11729438B1Optimizing streaming video encoding profiles

Publication Date: 2023.08.15 AMAZON TECH INC

AI summary

Techniques are described for optimizing streaming video encoding profiles.