Operation compare-video

The compare-video operation allows for the comparison of differences between two provided videos.
It utilizes the Video Multimethod Assessment Fusion (VMAF) video quality metric to evaluate the similarity in quality between the two videos.

VMAF is a machine learning-based video quality assessment algorithm that combines the strengths of several existing video quality metrics into a single, unified score.
It was designed to closely predict the Mean Opinion Score (MOS) of human viewers and has been shown to be highly correlated with subjective quality assessments.
By calculating the VMAF score for two videos, you can get an idea of how similar in quality the videos are to each other.

Example

{
    "conversion": [{
        "category": "operation",
        "target": "compare-video"
    }],
}
                                

Output Details

The output of this operation includes information about the quality of each frame in the video, as well as summary statistics for the entire video.
The metrics field contains a set of quality metrics for each frame, as in the following example:

{
    "version": "2.3.1",
    "fps": 24.03,
    "frames": [
    {
        "frameNum": 0,
        "metrics": {
            "integer_adm2": 0.842100,
            "integer_adm_scale0": 0.779905,
            "integer_adm_scale1": 0.749054,
            "integer_adm_scale2": 0.846122,
            "integer_adm_scale3": 0.901889,
            "integer_motion2": 0.000000,
            "integer_motion": 0.000000,
            "integer_vif_scale0": 0.198811,
            "integer_vif_scale1": 0.522634,
            "integer_vif_scale2": 0.660773,
            "integer_vif_scale3": 0.768340,
            "vmaf": 42.687407
        }
    },
    <...more frames...>
    ],
    "pooled_metrics": {
        "integer_adm2": {
        "min": 0.794626,
        "max": 0.863825,
        "mean": 0.830444,
        "harmonic_mean": 0.830332
    },
    "integer_adm_scale0": {
        "min": 0.698013,
        "max": 0.818805,
        "mean": 0.761654,
        "harmonic_mean": 0.761299
    },
    <...more metrics...>
    "vmaf": {
        "min": 40.664485,
        "max": 61.936634,
        "mean": 51.017497,
        "harmonic_mean": 50.703806
    }
}
                                
Field Name Description
integer_adm2 The Average Difference Modulus (ADM) metric, which measures the average difference between the two videos in terms of pixel intensity.
integer_adm_scale0
integer_adm_scale1
integer_adm_scale2
integer_adm_scale3
The ADM metric calculated at a lower resolution.
integer_motion2 The Motion2 metric, which measures the degree of motion in the video.
integer_motion The Motion metric, which measures the degree of motion in the video.
integer_vif_scale0
integer_vif_scale1
integer_vif_scale2
integer_vif_scale3
The Visual Information Fidelity (VIF) metric calculated at a lower resolution.
vmaf The VMAF score is calculated on a per-frame basis and the vmaf field in the metrics field contains the VMAF score for each individual frame.
pooled_metrics Contains summary statistics for each metric over the entire video. The statistics include the minimum, maximum, mean and harmonic mean values for each metric.
The vmaf field in the pooled_metrics field contains summary statistics for the VMAF scores over the entire video.

results matching ""

    No results matching ""