AdaptiveStreamGen

A multi-purpose Python library to predict optimal video encoding settings (CRF) and generate an adaptive stream.

Why use AdaptiveStreamGen

ffmpeg and the Bento4 SDK are very powerful tools in themselves and in combination, can be used to create high-quality streaming media.

However, they are not exactly easy to use with video encoding settings, in particular, requiring a lot of tuning to achieve desirable results.

AdaptiveStreamGen is a library designed to make it easy for everyone to create standalone streams or full-fledged adaptive streams designed for streaming from a source video.

Features

• Optimal CRF prediction to encode video using Netflix's VMAF for evaluation.
• No pre-tuning needed for video encoding.
• Adaptive streaming implemented by automatically encoding videos to different resolutions based on a scaling factor.
• Easy conversion of unfragmented videos to an MPEG-DASH compatible format.
• Customizable encoder, presets and options selection for ffmpeg.

Dependencies

• Requires Python 3.11+ to be installed
• ffmpeg
• Bento4

Usage

Install the requirements listed in requirements.txt.

Sample usage of the library is demonstrated in main.py.