1. 06 May, 2016 5 commits
  2. 05 May, 2016 4 commits
    • Jingning Han's avatar
      Clean up ext-interp experiment · 8b084b68
      Jingning Han authored
      Remove the unused sub-experiments within the ext-interp experiment.
      Change-Id: I716e3392412d02c56f9395a86c9cab02f580fa59
    • Jingning Han's avatar
      Remove a redundant variable definition from sub8x8 RD loop · cf512171
      Jingning Han authored
      Change-Id: I464cbb75fbd3872f66ca024dd803605542a9d887
    • Geza Lore's avatar
      Fix mismatch with ext-interp. · 1502d9e4
      Geza Lore authored
      The encoder signals the interp filter type in the frame header if all
      blocks use the same filter (see bitstream.c:fix_interp_filter). This
      decision is made based on the counts, but with ext-interp, the counts
      are actually only incremented for blocks that fail vp10_is_interp_needed
      (see for example encodeframe.c:update_state), otherwise a default value
      is used (EIGHTTAP_REGULAR). The decoder however first checks if the
      interp filter is signaled at the frame level, and uses that filter type
      for all blocks, even if the default value should have been used.
      This patch makes the decoder first check with vp10_is_interp_needed
      to see if the default value should be used and then checks the frame
      level signaling, which reconciles the difference between encoder and
      Change-Id: I87857ade42dea06b0d5ec2a029e9219268334dbb
    • Geza Lore's avatar
      Refactor supertx rd search · a905c45c
      Geza Lore authored
      General code cleanup, but also use the same supertx condition for
      ext-partition-types as for conventional partitions.
      Change-Id: If86eb18b3c07b9c60434eec2c98b97ce93665b67
  3. 04 May, 2016 6 commits
    • Yaowu Xu's avatar
      Change to call build_masked_compound_highbd() · f0c7e767
      Yaowu Xu authored
      from combine_interintra_highbd(). This fixes a crash in encoder in
      highbitdepth build.
      Change-Id: I0aa4cc30200703ff21e9990163bb26ace41aabbc
    • Jingning Han's avatar
      Refactor intra filter type context fetch function · 928d72f3
      Jingning Han authored
      Factor out common codes from vp10_get_pred_context_intra_interp().
      This prevents a potential invalid access of pointers xd->left_mbmi
      and xd->above_mbmi.
      The coding statistics are identical.
      Change-Id: I72dbf9380da7359b997bbe925010faab8e9e7f8d
    • Yaowu Xu's avatar
      Remove the use of non-declared "plane" · 357c5387
      Yaowu Xu authored
      The variable is not defined, it is not needed by the called function
      Change-Id: Ia601c03231afc0ae68a10ae1f35e8fc4121c3d28
    • Yaowu Xu's avatar
      Change to use proper type in vp10_token_state · 0d7dc0ca
      Yaowu Xu authored
      "qc" in vp10_token_state is used to save quantized coefficients, this
      commit changes the type from short to tran_low_t to properly reflect
      the value range for highbitdepth build.
      This fixes an out-of-range bug when optimize_b is used in highbitdepth
      Change-Id: I914c6fd3d3f4b9d061f9ed7cc5f08a883ab59dcd
    • Sarah Parker's avatar
      Add 1D tx set that corresponds to reduced ext tx inter sets · 3da61efe
      Sarah Parker authored
      This is the set of 1D transforms that are used in each
      ext_tx_used_inter set. The 1D sets will help speed up
      the ext tx pruning functions.
      Change-Id: Ib46ad26be2df60b3bfcd2f22d96e7f38ae286df5
    • Geza Lore's avatar
      Fix vp10_accumulate_frame_counts once and for all. · c959151f
      Geza Lore authored
      This ensures the multi-threaded and single-threaded encoder/decoder
      always uses the same probability contexts.
      Change-Id: I6f1e7c6bd8808c390c1dc0a628ae97db3acedf6d
  4. 03 May, 2016 4 commits
    • Geza Lore's avatar
      Compute end of frame data precisely with ext-tile. · cba70d29
      Geza Lore authored
      Decoding superframes correctly requires computing the end of the
      frame contents in the bitstream precisely. This patch enables
      ext-tile to do so.
      Also extended superframe_test to test with multiple tiles if using
      Change-Id: I04bb8cde8755a3d764ee3c36aa8b7a6c5c9db742
    • Geza Lore's avatar
      Test tile row independence. · 1982d677
      Geza Lore authored
      Tile rows should now be independent, so make pbi->inv_tile_order
      invert the decoding order of tile rows as well as tile columns.
      This should improve test coverage. Also added more tile configurations
      to the tile_independence_tests.
      Change-Id: I14b0f2fa9241c1acaf9e2a07071952cb33feca77
    • Jingning Han's avatar
      Replace hard coded values in mv_has_subpel · 113f8d87
      Jingning Han authored
      Change-Id: Id437740c2db1a3a56c1ad29d8b51bb763c044c1d
    • Debargha Mukherjee's avatar
      Refactoring and uv fix for wedge · 34077855
      Debargha Mukherjee authored
      lowres: -1.72%
      Change-Id: I4c883097caac72fab8e01945454579891617145e
  5. 30 Apr, 2016 3 commits
    • Yue Chen's avatar
      Bug fixes for obmc/ext-inter/ext-tile experiment · c1d47384
      Yue Chen authored
      Fix 1: in ext-inter + obmc config, properly identify if the left
      predictor used for obmc is a compound one in the case that the
      neighbor uses wedgeinterinter pred and we will dump the ALTREF part.
      This will fix the seg fault in unit test:
      Fix 2: in ext-tile + obmc experiment, handle the case that the
      above block does not fit in the same row tile with the current one,
      so as to prevent potential crashes.
      Change-Id: I1c177d4f4ad15e10d11d8756e146496437753eea
    • Jingning Han's avatar
      Fix encoder failure in segmentation mode · e729d28c
      Jingning Han authored
      This commit fixes an encoder segment fault in the codebase, when
      the segmentation feature is turned on. The issue was introduced in
      5cce322a Porting ext_partition experiment from nextgen
      Change-Id: Ifb4c06c5a6976114a8bd061d40d0338a136abaaf
    • Yi Luo's avatar
      HBD hybrid transform 8x8 SSE4.1 optimization · 299c5fc2
      Yi Luo authored
      - Update bit-exact unit test against current C version.
      - HBD encoder speed improves ~3.8%.
      Change-Id: Ie13925ba11214eef2b5326814940638507bf68ec
  6. 29 Apr, 2016 1 commit
  7. 28 Apr, 2016 3 commits
    • Debargha Mukherjee's avatar
      Make the backward updates work with bitshifts · e4bf50b9
      Debargha Mukherjee authored
      Removes integer divides from backward updates for VP10.
      Currently this is put in as part of the entropy experiment.
      Coding efficiency change is in the noise level.
      Change-Id: I5b3c0ab6169ee6d82d0ca1778e264fd4577cdd32
    • Debargha Mukherjee's avatar
      Brings back near-near compound mode into ext-inter · 7ff79434
      Debargha Mukherjee authored
      lowres: improves by 0.1%
      Change-Id: I245019916bf47c6e24bc8c3953b86715ab0193c9
    • Geza Lore's avatar
      Fix some mismatches when using ext-interp. · bf93b385
      Geza Lore authored
      With ext interp, write_switchable_interp_filter calls
      vp10_is_interp_needed, which needs access to the reference frame
      buffers to check if they are scaled, the ref frame buffer pointer
      at this point used to be uninitialized in the encoder resulting in
      bitstream syntax mismatch when the encoder/decoder did not read/write
      the interp filter element consistently.
      Change-Id: Ie0be2a19cbfcb5639a751aa857458e91c23b8fe3
  8. 27 Apr, 2016 7 commits
  9. 26 Apr, 2016 4 commits
    • Yue Chen's avatar
      Optimization for EXT_INTER + OBMC · 3ac12aec
      Yue Chen authored
      Remove the restriction that the neighboring predictor cannot be
      used in obmc prediction if it is an interintra or wedgeinterinter
      block. The inter predictor of the interintra block, or the first
      inter predictor(using LAST or GOLDEN frame) of the wedgeinterinter
      block will be exploited in obmc prediction.
      Coding gain: 0.248% (2.833%->3.081%) lowres
      Change-Id: I4ac0368b9d2f2956f266b30c1ac97db8bafa0742
    • Yue Chen's avatar
      Remove double counting for mv costs · 34177e67
      Yue Chen authored
      The bug is introduced by commit 1a0352d1, in which mv costs are
      counted twice in joint_motion_search() in ext_inter experiment.
      Change-Id: Ibace453df999d3c2e781d73f1f0912038fee2d4e
    • Jingning Han's avatar
      Rework motion vector precision limit · 8678ab4c
      Jingning Han authored
      This commit enables 1/8 luma component motion vector precision
      for all motion vector cases. It improves the compression performance
      of lowres by 0.13% and hdres by 0.49%.
      Change-Id: Iccfc85e8ee1c0154dfbd18f060344f1e3db5dc18
    • Debargha Mukherjee's avatar
      Reduce intra transform set · 8851acc5
      Debargha Mukherjee authored
      Reduce transform set for intra for 8x8 and smalller to 7 from 12.
      Also fixes an issue with prob updates.
      Enocder Speed-up about 8-10%
      Coding efficiency very little change.
      lowres: -2.996 (from -3.055 before)
      midres: -2.482 (from -2.552 before)
      Change-Id: I4ba50ff967521b33c748fe423bd92f7cf4105ebc
  10. 25 Apr, 2016 3 commits