Skip to content

libvmaf-fast-v1.1

Latest
Latest

Choose a tag to compare

View all tags
@4KVCD 4KVCD released this 16 Oct 02:22
libvmaf-fast-v1.1
994b181

An unofficial fork of Netflix's libvmaf, not affiliated with or endorsed by Netflix. Built from fast at 994b1814: libvmaf-fast-v1 with one change to the Vulkan engine. See the README.

What's new

  • The Vulkan engine keeps 169 bytes a frame instead of 1,184 (VMAF and NEG) and 208 instead of 1,602 (VMAF v1): a 3-hour film at 60 fps takes about 110 MB instead of 770 MB. Every score is the same.

Changes that need attention

  • vv_sums returns a frame's raw sums for the first 256 frames only; for a later frame it returns -3.

In the archive (Windows x64)

  • libvmaf/libvmaf.dll: libvmaf with CUDA, its public C API, built-in models, static C runtime. Reports version 994b1814.
  • vmaf_vulkan/vmaf_vulkan.dll: VMAF, NEG and VMAF v1's features on any GPU with Vulkan 1.1.
  • Licences for each, BUILD.txt, and SHA256SUMS of every file.

Checked

With these DLLs, on an RTX 5090 (driver 616.92) and the Core Ultra 9 285K's integrated GPU (driver 32.0.101.8860):

  • Vulkan VMAF and NEG identical to CUDA in all 45 cases of compare_vmaf_vulkan.py --matrix; VMAF v1 identical to libvmaf's CPU code in all 73 of compare_vmaf_v1.py --matrix.
  • Every output of the engine, compared with libvmaf-fast-v1's bit for bit: 556,372 values, all the same.
Assets 4
Loading