easyVmaf v2.0

• Cambi feature - Netflix banding detector, supported.

• Command line ussage updated according to libvmaf docs

• Cambi heatmap support added. The outputs may be visualized with ffplay

• Built-in VMAF models are only supported since they are included in FFmpeg >= v5.0.

• Docker image - better handling of dependencies and built instruccions

• 'HD Neg' and 'HD phone' models are computed by default

easyVmaf v1.3

• Fixed space character support on video file names when using -pogress flag.
• Docker image - Improved stdout log output in realtime when using -pogress flag.

easyVmaf v1.2

• Progress indicator added -progress. It shows the progress while doing vmaf computations.

• Added the option to explicilty set the number of threads to run -threads (int)

• With libvmaf v2.0.0, the vmaf models files were refactored. Basically, the models use now fixed-point data types instead of float-point data types. So depending on the libvmaf version you are using, you could need to modify the easyVmaf/config.py file:

  • libvmaf < v2.0.0 : Use the old vmaf_*.pkl model files with float-point data types: vmaf_4k_v0.6.1.pkl, vmaf_v0.6.1.pkl, etc.

  • libvmaf >= v2.0.0 : Use the new vmaf_*.json model files with fixed-point data types: vmaf_4k_v0.6.1.json, vmaf_v0.6.1.json, vmaf_v0.6.1neg.json

• New neg model is supported since libvmaf v2.0.0 only. So if you want to use it, be sure to update libvmaf first.

easyVmaf v1.1

• frame rate conversion improved and code cleaned

easyVmaf v1.0

• first version