)]}'
{
  "commit": "e06c242baa46ead2fb3d9801b5197ee199f8c60d",
  "tree": "1c034b4a8f8ad3a623adba68c7178463b24bdb7c",
  "parents": [
    "c4aeff94b16a6f71d6bea91c442d8b94056d9a75"
  ],
  "author": {
    "name": "Yunqing Wang",
    "email": "yunqingwang@google.com",
    "time": "Wed Dec 14 15:35:04 2011 -0500"
  },
  "committer": {
    "name": "Yunqing Wang",
    "email": "yunqingwang@google.com",
    "time": "Thu Dec 15 11:19:18 2011 -0500"
  },
  "message": "Only call vp8_find_near_mvs() once for each macroblock\n\nWhile doing motion search on a macroblock, we usually call\nvp8_find_near_mvs once per reference frame. Actually, for\ndifferent reference frames, the only difference in calculating\nthese near_mvs is they may have different sign_bias, which\ncauses a sign change in resulting near_mvs. In this change, we\nonly do find_near_mvs for the first reference frame. For other\nreference frames, only need to adjust the near_mvs according to\nthat reference frame\u0027s sign_bias value.\n\nChange-Id: I661394b49c6ad79fed7d0f2eb2be239b9c56f149\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "d9c89750b82bc66cddbdec72f02066471738a5fc",
      "old_mode": 33188,
      "old_path": "vp8/encoder/pickinter.c",
      "new_id": "0537d6648b13565576ee78a35061a6ad2008bf2b",
      "new_mode": 33188,
      "new_path": "vp8/encoder/pickinter.c"
    },
    {
      "type": "modify",
      "old_id": "4f9a0ce804a67499824fe14fb3db37bf54fcba11",
      "old_mode": 33188,
      "old_path": "vp8/encoder/rdopt.c",
      "new_id": "79fe63c8b0b3652d969bd9ce8b43e9b1b99521c1",
      "new_mode": 33188,
      "new_path": "vp8/encoder/rdopt.c"
    }
  ]
}
