Overview
The article discusses the advancements in the Video Multi-Method Assessment Fusion (VMAF) metric, which aims to improve video quality evaluation for the video community. It highlights recent developments, including speed optimizations, a new API, and the introduction of a 'No Enhancement Gain' mode to better assess codec performance.
What You'll Learn
1
How to integrate VMAF with existing open source projects using the new libvmaf API
2
Why VMAF is preferred over traditional metrics like PSNR and SSIM for video quality assessment
3
When to use VMAF NEG mode for codec evaluation to assess pure compression effects
Key Questions Answered
What is VMAF and how does it improve video quality assessment?
VMAF, or Video Multi-Method Assessment Fusion, is a video quality metric developed by Netflix that models human visual perception to evaluate video quality more accurately than traditional metrics like PSNR and SSIM. It combines evidence from human visual system modeling and machine learning to produce a score that reflects perceived quality across different resolutions and content types.
How has VMAF's speed performance improved recently?
VMAF's execution speed has improved by over 4x through low-level code optimization and vectorization, and recent collaborations with Facebook and Intel have yielded an additional 2x speed gain while maintaining numerical accuracy. This allows VMAF to run in real-time for 4K videos.
What is the purpose of the new libvmaf API?
The new libvmaf API, released as version 2.0.0, enhances compatibility with existing open source projects and simplifies integration with FFmpeg. It is designed to be extensible, allowing for easy addition of new feature extractors and facilitating frame-level calculations.
What is the VMAF NEG mode and when should it be used?
The VMAF NEG mode is designed to disregard enhancement gains from image processing, allowing for a pure assessment of codec performance. It is recommended for codec evaluations to measure the effects of compression without the influence of pre-processing enhancements.
Key Statistics & Figures
Speed improvement of VMAF execution
over 4x
Achieved through low-level code optimization and vectorization, with an additional 2x gain from collaborations with Facebook and Intel.
Numerical accuracy retention
first decimal digit
Maintained while achieving significant speed gains in VMAF's processing.
Technologies & Tools
Video Quality Metric
Vmaf
Used for evaluating video quality in various applications, including streaming and encoding.
C Library
Libvmaf
Facilitates integration of VMAF with FFmpeg and other projects.
Key Actionable Insights
1Utilize the new libvmaf API to enhance your video processing workflows.The revamped API simplifies integration with existing tools and allows for more flexible memory allocation and frame-level calculations, making it easier to implement VMAF in various applications.
2Adopt VMAF NEG mode for more accurate codec evaluations.This mode helps isolate the effects of compression from enhancement gains, providing clearer insights into codec performance and ensuring that evaluations reflect true compression efficiency.
3Leverage VMAF's speed optimizations for real-time video applications.With recent improvements allowing VMAF to process 4K videos in real time, developers can implement it in live streaming and other time-sensitive applications without sacrificing quality.
Common Pitfalls
1
Over-reliance on traditional metrics like PSNR and SSIM for video quality assessment.
These metrics often fail to predict quality across different resolutions and content types, leading to inaccurate evaluations. Transitioning to VMAF can provide a more reliable measure of perceived quality.
Related Concepts
Video Quality Assessment
Human Visual System Modeling
Machine Learning In Video Processing