Skip to content

easyVmaf v1.2

Compare
Choose a tag to compare
@gdavila gdavila released this 14 May 23:18
· 17 commits to master since this release
  • 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.