)]}'
{
  "commit": "1e2fb103ae4e07328d838eeddee9130342a885c2",
  "tree": "5d530634df18ba9c6faed136d52c9abc17377dcd",
  "parents": [
    "27e9029579e9de1d2e7d2b899f68fd67f19d92e7"
  ],
  "author": {
    "name": "David Barker",
    "email": "david.barker@argondesign.com",
    "time": "Fri Aug 18 13:18:16 2017 +0100"
  },
  "committer": {
    "name": "David Barker",
    "email": "david.barker@argondesign.com",
    "time": "Wed Oct 04 09:58:32 2017 +0000"
  },
  "message": "Harmonize OBMC prediction at tile edges\n\nWhen we have an OBMC block at the top of a tile, we don\u0027t generate\nan \u0027above\u0027 prediction, in order to preserve tile independence.\nHowever, the conditions for when to *generate* the above prediction\nand when to *use* it were different when dependent-horztiles is\nenabled.\n\nSince dependent-horztiles allows each tile to depend on the tile\nabove, it seems reasonable to unify the conditions to use\nxd-\u003eabove_available (and similar for left preds, for the sake\nof consistency). This allows OBMC across horizontal tile boundaries,\nbut not across tile group boundaries, in line with the rest of the\ndependent-horztiles experiment.\n\nChange-Id: I86d5de132c4429106a6025e7a7a2baec974e618c\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "0a6cb49cabaf36ed00cf2273fdb51650663ef908",
      "old_mode": 33188,
      "old_path": "av1/common/reconinter.c",
      "new_id": "5d79560b594cb2af1f2bcee318ffe169594d0d2b",
      "new_mode": 33188,
      "new_path": "av1/common/reconinter.c"
    }
  ]
}