Per-Title Encoding in 6.2 Dimensions: Lesson of the Week
Per-title encoding customizes the encoding parameters of the encoding ladder for the characteristics of the video file. The original per-title technologies optimization technologies like Beamr/CRF that adjusted only the data rate of the file (1 dimension).
In December 2015, Netflix launched per-title encoding using a brute force technique that customized the number of rungs in the ladder, rung resolution, and data rate (3 dimensions)
In 2018, companies like Brightcove started considering connection and device data when creating encoding ladders (3.2 dimensions).
In 2020, Ateme added support for frame rate, dynamic range, and color gamut (6.2 dimensions).
This video tracks the progression of these technologies and details their impact.
00:00 - Introduction
1:02 - Netflix invents per-title encoding
1:58 - Optimization technologies
2:57 - Why changing the number of rungs and resolution is important
4:57 - How Netflix per-title works
6:41 - The advent of commercial per-title systems
7:43 - Netflix implements shot-based encoding
9:26 - Context-based encoding (Brightcove adds connection/device data)
11:38 - Expanding to frame rate, dynamic range, and color gamut
12:25 - The evolution of quality metrics (from PSNR to Quality Vector)
14:00 - Ateme Quality Vector Driven Content Adaptive Encoding
This updated lesson was recently added to my online course, Streaming Media 101: Technical Onboarding for Streaming Media Professionals. You can download the slide deck here:
https://streaminglearningcenter.com/w...
Click below for more information:
https://bit.ly/StreamingMedia101