1. 06 Dec, 2016 1 commit
    • David Barker's avatar
      Improve rdopt decisions for ext-inter · ac37fa3d
      David Barker authored
      Relative to previous ext-inter:
      lowres: -0.177%
           or -0.029% (with USE_RECT_INTERINTRA = 0)
      
      * When predicting interintra modes, the previous code did not provide
        the intra predictor with the correct context during rdopt. Add an
        explicit 'ctx' parameter to the relevant functions, to provide this
        context.
        This fixes a nondeterminism bug, which was causing test failures in
        *EncoderThreadTest*
      
      * For rectangular blocks, build_intra_predictors_for_interintra needs
        to overwrite part of the context buffer in order to set up the
        correct context for intra prediction. We now restore the original
        contents afterwards.
      
      * Add a flag to enable/disable rectangular interintra prediction;
        disabling improves encoding speed but reduces BDRATE improvement.
      
      Change-Id: I7458c036c7f94df9ab1ba0c7efa79aeaa7e17118
      ac37fa3d
  2. 01 Dec, 2016 2 commits
  3. 30 Nov, 2016 2 commits
  4. 29 Nov, 2016 1 commit
    • Yue Chen's avatar
      Add WARPED_MOTION experiment · 69f18e1a
      Yue Chen authored
      Performance gain (REF_MV on by default)
      WARPED_MOTION: 1.061%
      WARPED_MOTION+MOTION_VAR: 2.917%
      MOTION_VAR: 2.337%
      
      Change-Id: I43f742a02cdd43d13ef333a0a15087062ad020ab
      69f18e1a
  5. 17 Nov, 2016 1 commit
  6. 16 Nov, 2016 1 commit
    • Zoe Liu's avatar
      Fix the code to allow a 3rd compound mode · 00d5472a
      Zoe Liu authored
      It is possible down the road new compound modes are added in addition to
      COMPOUND_AVERAGE and COMPOUND_WEDGE.
      
      Change-Id: I85985d4c104effd34df13845344aac2a08d28019
      00d5472a
  7. 11 Nov, 2016 2 commits
  8. 10 Nov, 2016 1 commit
  9. 02 Nov, 2016 1 commit
    • Jingning Han's avatar
      Fix a merge bug between dual_filter and sub8x8mc · 46003149
      Jingning Han authored
      The function module in inter_predictor() has been changed to
      universally support arbitrary block size inter prediction. Hence
      sub8x8mc can be a standalone experiment now.
      
      Change-Id: Ie9d87f61fc317b1d114edb4e0bf5544f918ed08e
      46003149
  10. 28 Oct, 2016 1 commit
  11. 24 Oct, 2016 2 commits
  12. 22 Oct, 2016 1 commit
  13. 21 Oct, 2016 1 commit
    • Jingning Han's avatar
      Sub8x8 block chroma component inter prediction · e29ea12f
      Jingning Han authored
      Handle the sub8x8 chroma component at the unit of 2x2/4x2/2x4 level
      and use the motion vector inherited from the luma component. This
      improves the coding performance:
      
      lowres 0.4%
      midres 0.25%
      hdres  0.15%
      
      Change-Id: I34dff4218cfa3e5d55e7ed0341f36f4719389f7e
      e29ea12f
  14. 19 Oct, 2016 1 commit
    • Urvang Joshi's avatar
      Code cleanup: mainly rd_pick_partition and methods called from there. · 52648448
      Urvang Joshi authored
      - Const correctness
      - Refactoring
      - Make variables local when possible etc
      - Remove -Wcast-qual to allow explicitly casting away const.
      
      Cherry-picked from aomedia/master: c27fcccc
      And then a number of more const correctness changes to make sure other
      experiments build OK.
      
      Change-Id: I77c18d99d21218fbdc9b186d7ed3792dc401a0a0
      52648448
  15. 14 Oct, 2016 1 commit
    • Yue Chen's avatar
      Fix obmc chroma prediction involving neighboring sub8x8 blocks' mvs · 1c263e03
      Yue Chen authored
      When the neighbor uses a sub8x8 partition, to predict chroma pixels in
      the overlapping region, instead of averaging all mvs in the 8x8 block,
      only apply those belonging to neighboring sub8x8 parts to the
      corresponding sub-regions (no averaging).
      The change applies to both settings w/ and w/o SUB8X8_MC experiment.
      
      Change-Id: I033e9f218d02bc9879841b794931b575fac14f25
      1c263e03
  16. 13 Oct, 2016 1 commit
    • Yue Chen's avatar
      Renamings for OBMC experiment · cb60b185
      Yue Chen authored
      To get ready for pulling AV1 to nextgenv2
      Replace the experimental flag by MOTION_VAR. Rename major variables.
      
      Change-Id: If6cf4f37b9319c46d8f90df551cc7295d66ca205
      cb60b185
  17. 06 Oct, 2016 1 commit
    • Yue Chen's avatar
      Fix border check in MOTION_VAR · 79d8a07d
      Yue Chen authored
      Correct distance to right and bottom edge when predicting the overlapping
      region using neighbor predictors. The distance is more than that for the
      whole block, so the condition in clamp_mv will be looser.
      
      Change-Id: I40dfdd9c0a2a5e854cac12195b282493981e9d93
      79d8a07d
  18. 04 Oct, 2016 1 commit
  19. 15 Sep, 2016 1 commit
  20. 09 Sep, 2016 1 commit
    • James Zern's avatar
      s/INTERP_FILTER/InterpFilter/ · 7b9407a8
      James Zern authored
      this matches style guidelines and stabilizes successive runs of
      clang-format across the tree. remaining types should be address in
      successive commits.
      
      Change-Id: I6ad3f69cf0a22cb9a9b895b272195f891f71170f
      7b9407a8
  21. 02 Sep, 2016 1 commit
  22. 01 Sep, 2016 2 commits
  23. 12 Aug, 2016 1 commit
  24. 08 Aug, 2016 1 commit
    • Sarah Parker's avatar
      Add reconstruction using gm parameters · b659281e
      Sarah Parker authored
      This patch only includes inter frame reconstruction using gm
      parameters when GLOBAL_MOTION and/or VP9_HIGHBITDEPTH are enabled.
      GM is not currently used when EXT_INTER or DUAL_FILTER is enabled.
      This will be added in a followup patch. For now, these experiments
      will take precedence over GLOBAL_MOTION when they are all enabled.
      
      Change-Id: I930ddda529c44d7245dbb56db3c9c5524cf45473
      b659281e
  25. 18 Jul, 2016 1 commit
  26. 13 Jul, 2016 1 commit
    • Geza Lore's avatar
      Optimize and cleanup obmc predictor and rd search. · 4c4f04ac
      Geza Lore authored
      Use vpx_blend_a64_hmask and vpx_blend_a64_vmask to speed up
      computing the obmc predictor. Clean up calc_target_weighted_pred.
      
      Encoder speedup: 1.3%
      Decoder speedup: 6.5%
      
      Change-Id: I0c774fe53d22399e92a10d1daf3af0010d88d2c5
      4c4f04ac
  27. 11 Jul, 2016 2 commits
    • Geza Lore's avatar
      Optimize and cleanup supertx predictor. · cd489264
      Geza Lore authored
      Use vpx_blend_a64_hmask and vpx_blend_a64_vmask to speed up
      computing the supertx predictor.
      
      Decoder speedup of up to 4% has been observed.
      
      Change-Id: I255a5ba4cc24f78dc905d25b6e2f7fbafac13253
      cd489264
    • Geza Lore's avatar
      Improve vpx_blend_* functions. · bfa59b4a
      Geza Lore authored
      - Made source buffers pointers to const.
      - Renamed vpx_blend_mask6b to vpx_blend_a64_mask. This is more
        indicative that the function does alpha blending. The 6, or 6b
        suffix was misleading, as the max mask value (64) does not fit into
        6 bits.
      - Added VPX_BLEND_* macros to use when needing to blend scalars.
      - Use VPX_BLEND_A256 in combine_interintra to be more explicit about
        the operation being done.
      - Added versions of vpx_blend_a64_* which take 1D horizontal/vertical
        masks directly and apply them to all rows/columns
        (vpx_blend_a64_hmask and vpx_blend_a64_vmask). The SSE4.1 optimzied
        horizontal version now falls back on the 2D version. This can be
        improved upon if it show up high enough in a profile.
      - All vpx_blend_a64_* functions now support block sizes down to 1x1
        (ie: a single pixel). This is for usage convenience. The SSE4.1
        optimized versions fall back on the C implementation if
        w <= 2 or h <= 2. This can again be improved if it becomes hot code.
      
      Change-Id: I13ab3835146ffafe3e1d74d8e9cf64a5abe4144d
      bfa59b4a
  28. 08 Jul, 2016 1 commit
  29. 07 Jul, 2016 1 commit
  30. 21 Jun, 2016 1 commit
  31. 14 Jun, 2016 1 commit
    • Jingning Han's avatar
      Fix enc/dec mismatch in non-420 settings · a4ea8fd8
      Jingning Han authored
      This commit makes the dual filter experiment work with non-420
      settings. It fixes unit test failure in EndToEndTestLarge.
      
      Change-Id: I04f7afdee78f91389d9ff72947efa152098af930
      a4ea8fd8
  32. 13 Jun, 2016 1 commit
  33. 10 Jun, 2016 2 commits