libvmaf ☆

« Back to VersTracker

Description:
Perceptual video quality assessment based on multi-method fusion

Type: Formula | Tracked Since: Dec 28, 2025

Links: Homepage | @netflixoss | formulae.brew.sh

Category: Multimedia

Tags: video quality vmaf multimedia library

Install: brew install libvmaf

About:
Libvmaf is the official library for Netflix's Video Multi-Method Assessment Fusion (VMAF) algorithm. It provides a perceptual video quality metric that combines multiple elementary models to predict human visual satisfaction. This library is designed for integration into video processing pipelines for accurate quality control and optimization.

Key Features:

Implements the VMAF perceptual video quality metric
C library for high-performance integration
Supports custom models and training datasets
Includes command-line tools for quality assessment

Use Cases:

Benchmarking video codec performance and compression efficiency
Automated quality control in video encoding pipelines
Comparing the visual quality of different video processing algorithms

Alternatives:

ffmpeg – FFmpeg is a general-purpose multimedia framework that can calculate PSNR and SSIM, whereas Libvmaf provides the specific, advanced VMAF metric.
vmaf – The 'vmaf' formula is the Python wrapper; this 'libvmaf' formula provides the core C library needed for performance-critical applications.

Version History

Detected	Version	Rev	Change	Commit
Oct 27, 2024 12:49pm		0	VERSION_BUMP	0338faec