)]}'
{
  "commit": "0c90a5b39e993ca5a8195281da7ced1e477fca35",
  "tree": "43c52bdfd0d03f85a0049f3d138f2a5364154919",
  "parents": [
    "c7d4cd39f9c90a7077c688f388ef0dff7fe7634e"
  ],
  "author": {
    "name": "Ryan",
    "email": "ryan.z.lei@intel.com",
    "time": "Tue Jan 02 10:10:45 2018 -0800"
  },
  "committer": {
    "name": "Ryan Lei",
    "email": "ryan.z.lei@intel.com",
    "time": "Thu Jan 04 19:37:32 2018 +0000"
  },
  "message": "update filter mask calculation for 5 tap filter on chroma plane\n\nwhen 5 tap filter(filter6) was introduced for chroma plane, it reads 6 pixel\nlines/columns and modify 4 pixels. However, for the filter mask\ncalculation, it still use the original filter mask calculation function\nfor 7 tap filter(filter8), i.e, read 8 pixel lines/columns to check whether the\nedge sould be filtered or not. This change updates the filter mask\ncalculation to make it consistent with the filering operation to only\nread 6 pixels for filter mask calculation.\n\nAWCY low latency test result show no noticable quality impact as below\nRLDeblockUpdateFilterMaskBaselineLL@2018-01-02T18:42:02.782Z -\u003e \nRLDeblockUpdateFilterMaskNewLL@2018-01-02T18:43:46.627Z\n\n  PSNR | PSNR Cb | PSNR Cr | PSNR HVS |    SSIM | MS SSIM | CIEDE 2000\n0.0206 | -0.0226 | -0.0412 |  -0.0226 | -0.0671 | -0.0398 |     0.0370\n\nhttps://arewecompressedyet.com/?job\u003dRLDeblockUpdateFilterMaskBaselineLL%402018-01-02T18%3A42%3A02.782Z\u0026job\u003dRLDeblockUpdateFilterMaskNewLL%402018-01-02T18%3A43%3A46.627Z\n\nBUG\u003daomedia:1183\n\nChange-Id: Iec434e48161e1bc2e67b38ea33c46e9c73774ce2\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "be4bd9bcc6239f718d2060b55f40b856fa48718b",
      "old_mode": 33188,
      "old_path": "aom_dsp/loopfilter.c",
      "new_id": "fb4fb4db803b3d9dce165080b51bbd354302a546",
      "new_mode": 33188,
      "new_path": "aom_dsp/loopfilter.c"
    }
  ]
}