Netflix / vmaf Public

7 Pull requests merged by 3 people

• libvmaf: avoid some compiler warnings

  #962 merged Dec 12, 2021

• libvmaf/predict: use vmaf_feature_name_from_options

  #961 merged Dec 11, 2021

• Update on CAMBI changes

  #956 merged Dec 10, 2021

• Introduce adm_csf_mode feature overload.

  #960 merged Dec 10, 2021

• Add more cases to vif_kernelscale.

  #958 merged Dec 9, 2021

• float_vif: increase max value of vif_kernelscale to 4.0.

  #957 merged Dec 9, 2021

• libvmaf/feature: add vmaf_feature_collector_append_with_dict()

  #954 merged Dec 8, 2021

2 Pull requests opened by 2 people

• vif_filter1d_rd_8_avx512: fix inf loop with j = j++

  #965 opened Dec 15, 2021

• y4m_input: Add support for 422p10 y4m files

  #967 opened Dec 15, 2021

5 Issues opened by 4 people

• [Question] Why is the vmaf score calculated by using vmaf.exe directly different from the vmaf score calculated by ffmpeg.exe?

  #966 opened Dec 15, 2021

• `vmaf_predict_score_at_index()` templated feature name generation could be more efficient.

  #964 opened Dec 13, 2021

• Does anyone run successfully on windows?

  #963 opened Dec 13, 2021

• Possible unefficient implementation of adm feature

  #959 opened Dec 9, 2021

• Weird rounding in i4_adm_cm

  #955 opened Dec 8, 2021

4 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• Possible unwanted float to double promotion

  #930 commented on Dec 8, 2021 • 1 new comment

• Lower precision in VIF filtering might make more efficient AVX2 code possible

  #946 commented on Dec 10, 2021 • 0 new comments

• PSNR-HVS: fix JPEG

  #921 commented on Dec 12, 2021 • 0 new comments

• CAMBI heatmap generation

  #947 commented on Dec 14, 2021 • 0 new comments