1. 28 Feb, 2013 1 commit
  2. 05 Feb, 2013 2 commits
    • John Koleszar's avatar
      Convert subpixel filters to use convolve framework · 7a07eea1
      John Koleszar authored
      Update the code to call the new convolution functions to do subpixel
      prediction rather than the existing functions. Remove the old C and
      assembly code, since it is unused. This causes a 50% performance
      reduction on the decoder, but that will be resolved when the asm for
      the new functions is available.
      
      There is no consensus for whether 6-tap or 2-tap predictors will be
      supported in the final codec, so these filters are implemented in
      terms of the 8-tap code, so that quality testing of these modes
      can continue. Implementing the lower complexity algorithms is a
      simple exercise, should it be necessary.
      
      This code produces slightly better results in the EIGHTTAP_SMOOTH
      case, since the filter is now applied in only one direction when
      the subpel motion is only in one direction. Like the previous code,
      the filtering is skipped entirely on full-pel MVs. This combination
      seems to give the best quality gains, but this may be indicative of a
      bug in the encoder's filter selection, since the encoder could
      achieve the result of skipping the filtering on full-pel by selecting
      one of the other filters. This should be revisited.
      
      Quality gains on derf positive on almost all clips. The only clip
      that seemed to be hurt at all datarates was football
      (-0.115% PSNR average, -0.587% min). Overall averages 0.375% PSNR,
      0.347% SSIM.
      
      Change-Id: I7d469716091b1d89b4b08adde5863999319d69ff
      7a07eea1
    • Paul Wilkins's avatar
      Change definition of NearestMV. · 81043e8d
      Paul Wilkins authored
      This commit makes the NearestMV match the chosen
      best reference MV. It can be a 0,0 or non zero vector
      which means the the compound nearest mv mode can
      combine a 0,0 and a non zero vector.
      
      Change-Id: I2213d09996ae2916e53e6458d7d110350dcffd7a
      81043e8d
  3. 24 Jan, 2013 1 commit
    • Deb Mukherjee's avatar
      Adds an error-resilient mode with test · 01cafaab
      Deb Mukherjee authored
      Adds an error-resilient mode where frames can be continued
      to be decoded even when there are errors (due to network losses)
      on a prior frame. Specifically, backward updates are turned off
      and probabilities of various symbols are reset to defaults at
      the beginning of each frame. Further, the last frame's mvs are
      not used for the mv reference list, and the sorting of the
      initial list based on search on previous frames is turned off
      as well.
      
      Also adds a test where an arbitrary set of frames are skipped
      from decoding to simulate errors. The test verifies (1) that if
      the error frames are droppable - i.e. frame buffer updates have
      been turned off - there are no mismatch errors for the remaining
      frames after the error frames; and (2) if the error-frames are non
      droppable, there are not only no decoding errors but the mismatch
      PSNR between the decoder's version of the post-error frames and the
      encoder's version is at least 20 dB.
      
      Change-Id: Ie6e2bcd436b1e8643270356d3a930e8989ff52a5
      01cafaab
  4. 14 Jan, 2013 2 commits
  5. 10 Jan, 2013 1 commit
  6. 08 Jan, 2013 2 commits
  7. 06 Jan, 2013 1 commit
  8. 18 Dec, 2012 1 commit
  9. 10 Dec, 2012 1 commit
    • Paul Wilkins's avatar
      Further changes to mv reference code. · d1244659
      Paul Wilkins authored
      Some further changes and refactoring of mv
      reference code and selection of center point for
      searches. Mainly relates to not passing so many
      different local copies of things around.
      
      Some place holder comments.
      
      Change-Id: I309f10ffe9a9cde7663e7eae19eb594371c8d055
      d1244659
  10. 05 Dec, 2012 1 commit
    • Paul Wilkins's avatar
      Change to MV reference search. · 4cc657ec
      Paul Wilkins authored
      This patch reduces the cpu cost of the MV ref
      search by only allowing insert for candidates
      that would be in the current top 4.
      
      This could alter the outcome and slightly favors
      near candidates which are tested first but also
      limits the worst case loop count to 4 and means in
      many cases it will drop out and not happen.
      
      Change-Id: Idd795a825f9fd681f30f4fcd550c34c38939e113
      4cc657ec
  11. 29 Nov, 2012 1 commit
  12. 28 Nov, 2012 1 commit
  13. 27 Nov, 2012 1 commit
    • John Koleszar's avatar
      Add vp9_ prefix to all vp9 files · fcccbcbb
      John Koleszar authored
      Support for gyp which doesn't support multiple objects in the same
      static library having the same basename.
      
      Change-Id: Ib947eefbaf68f8b177a796d23f875ccdfa6bc9dc
      fcccbcbb
  14. 26 Nov, 2012 1 commit
    • Yunqing Wang's avatar
      Improve sad3x16 SSE2 function · e7cd8071
      Yunqing Wang authored
      Vp9_sad3x16_sse2() is heavily called in decoder, in which the
      unaligned reads consume lots of cpu cycles. When CONFIG_SUBPELREFMV
      is off, the unaligned offset is 1. In this situation,
      we can adjust the src_ptr to be 4-byte aligned, and then do the
      aligned reads. This reduced the reading time significantly. Tests
      on 1080p clip showed over 2% decoder performance gain with
      CONFIG_SUBPELREFM off.
      
      Change-Id: I953afe3ac5406107933ef49d0b695eafba9a6507
      e7cd8071
  15. 15 Nov, 2012 1 commit
    • Yaowu Xu's avatar
      subpelrefmv for superblocks · 61416aed
      Yaowu Xu authored
      duplicate code clean-up and variable name corrections
      
      Change-Id: Ibc4703228e652ec425125de5e7bc038fa46595c5
      61416aed
  16. 13 Nov, 2012 2 commits
  17. 12 Nov, 2012 1 commit
    • Paul Wilkins's avatar
      New inter mode context · 2669f42b
      Paul Wilkins authored
      This change is a fix / extension of the newbestrefmv
      experiment. As such it is presented without IFDEF.
      
      The change creates a new context for coding inter modes
      in vp9_find_mv_refs(). This replaces the context that
      was previously calculated in vp9_find_near_mvs().
      The new context is unoptimized and not necessarily
      any better at this stage (results pending), but eliminates
      the need for a legacy call to vp9_find_near_mvs().
      Based on numbers from Scott, this could help decode
      speed by several %.
      
      In a later patch I will add support for forward update of
      context (assuming this helps) and refine the context as
      necessary.
      
      Change-Id: I1cd991b82c8df86cc02237a34185e6d67510698a
      2669f42b
  18. 10 Nov, 2012 1 commit
    • Deb Mukherjee's avatar
      New b-intra mode where direction is contextual · d01357bb
      Deb Mukherjee authored
      Preliminary patch on a new 4x4 intra mode B_CONTEXT_PRED where the
      dominant direction from the context is used to encode. Various decoder
      changes are needed to support decoding of B_CONTEXT_PRED in conjunction
      with hybrid transforms since the scan order and tokenization depends on
      the actual direction of prediction obtained from the context. Currently
      the traditional directional modes are used in conjunction with the
      B_CONTEXT_PRED, which also seems to provide the best results.
      
      The gains are small - in the 0.1% range.
      
      Change-Id: I5a7ea80b5218f42a9c0dfb42d3f79a68c7f0cdc2
      d01357bb
  19. 07 Nov, 2012 2 commits
    • Yaowu Xu's avatar
      merge full pixel refmv experiment · 0cedaa36
      Yaowu Xu authored
      Change-Id: Ib39ad47a7d188f3b45416937b7eeb28c3e79b74c
      0cedaa36
    • James Zern's avatar
      Fix variance (signed integer) overflow · 98473443
      James Zern authored
      In the variance calculations the difference is summed and later squared.
      When the sum exceeds sqrt(2^31) the value is treated as a negative when
      it is shifted which gives incorrect results.
      
      To fix this we force the multiplication to be unsigned.
      
      The alternative fix is to shift sum down by 4 before multiplying.
      However that will reduce precision.
      
      For 16x16 blocks the maximum sum is 65280 and sqrt(2^31) is 46340 (and
      change).
      
      This change is based on:
      16982342 Missed some variance casts
      fea3556e Fix variance overflow
      
      Change-Id: I2c61856cca9db54b9b81de83b4505ea81a050a0f
      98473443
  20. 06 Nov, 2012 1 commit
  21. 01 Nov, 2012 2 commits
  22. 31 Oct, 2012 3 commits
  23. 30 Oct, 2012 1 commit
    • Yaowu Xu's avatar
      Separated experiment for subpel refmv selection · 24265e88
      Yaowu Xu authored
      As suggested by Paul, this commit separate the subpel refmv selection
      into a separate experiment. It also changed a couple variable names
      to better reflect the nature of the variables.
      
      Change-Id: Id951c3cadc61a982dd15afe641000f60213b8995
      24265e88
  24. 29 Oct, 2012 1 commit
    • Yaowu Xu's avatar
      Improves subpixel reference mv evaluation · cc9c637d
      Yaowu Xu authored
      Previously, in evaluating reference motion vectors, MVs are always
      rounded to integer pixel position and SADs are calculated.  This
      commit takes into account the subpixel portion of the mvs, and uses
      bilinear interpolation to produce reference pixel values in subpixel
      postions. In addition, SSE is used in place of SAD. Pixels used are
      16x2 above and 2x16 to the left.
      
      This commmit intends to test the potential of this line of work in
      term of compression improvement, obviously, the change would increase
      decoder complexity significantly.
      
      Test results
      std-hd: 1.738%(avg) 1.779%(glb), 1.663%(ssim)
      derf: 0.472%(avg) 0.477%(glb), 0.418%(ssim)
      
      Change-Id: I3ae1b098f6289df78891134d9a5e4bb2fde87a0b
      cc9c637d
  25. 26 Oct, 2012 1 commit
    • Paul Wilkins's avatar
      Explicit MV reference experiment. · ba30e774
      Paul Wilkins authored
      Coding and costing of mv reference signal.
      
      Issues in updating MV ref with COMPANDED_MVREF_THRESH
      to be resolved. Ideally the MV precision should be defined based
      on absolute MV magnitude not as now the MV ref magnitude.
      
      Update to mv counts moved into bitstream.c because otherwise
      if the motion reference is changed at the last minute the encoder
      and decoder get out of step in terms of the counts used to update
      entropy probs.
      
      Code working on a few test clips but no results yet re benefit vs
      signaling cost and no tuning of red loop to test lower cost alternatives
      based on the available reference values.
      
      Patch 3. Added check to make sure we don't pick a reference
      that would give rise to an uncodeable / out of range residual.
      
      Patch 6-7: Attempt to rebase. OK to submit but best to leave flag off for now.
      
      Patch 9. Remove print no longer needed.
      
      Change-Id: I1938c2ffe41afe6d3cf6ccc0cb2c5d404809a712
      ba30e774
  26. 22 Oct, 2012 3 commits
  27. 15 Oct, 2012 1 commit
  28. 14 Sep, 2012 1 commit
    • Yaowu Xu's avatar
      Use 3 rows and cols of pixels for ref_mv scoring · 6d8538e5
      Yaowu Xu authored
      The commit changed to use 3 rows above and 3 cols from left for SAD
      scoring for selecting the best reference motion vector. The change
      helped std-hd set by >.2% on psnr/ssim metrics.
      
      Change-Id: Ifad3b528d0b4b6e3c22518af789d76eff23c1520
      6d8538e5
  29. 12 Sep, 2012 1 commit
    • Deb Mukherjee's avatar
      Adds feature for companded MV encoding · 2b26cf17
      Deb Mukherjee authored
      The high-precision (1/8) pel bit is turned off if the reference
      MV is larger than a threshold. The motivation for this patch is
      the intuition that if motion is likely large (as indicated by
      the reference), there is likley to be more motion blur, and as
      a result 1/8 pel precision would be wasteful both in rd sense
      as well as computationally.
      
      The feature is incorporated as part of the newmventropy experiment.
      There is a modest RD improvement with the patch. Overall the
      results with the newmventropy experiment with the threshold being
      16 integer pels are:
      
      derf: +0.279%
      std-hd: +0.617%
      hd: +1.299%
      yt: +0.822%
      
      With threshold 8 integer pels are:
      
      derf: +0.295%
      std-hd: +0.623%
      hd: +1.365%
      yt: +0.847%
      
      Patch: rebased
      Patch: rebase fixes
      
      Change-Id: I4ed14600df3c457944e6541ed407cb6e91fe428b
      2b26cf17
  30. 07 Sep, 2012 1 commit
    • Paul Wilkins's avatar
      Partial merge of NEWBESTREFMV and NEW_MVREF. · 778ec421
      Paul Wilkins authored
      This commit merges those parts of the CONFIG_NEW_MVREF
      that specifically relate to choosing a better set of candidate
      MV references into the NEWBESTREFMV experiment.
      
      CONFIG_NEW_MVREF will then be used for changes relating
      to the explicit coding of a cost optimized MV reference in the
      bitstream as part of MV coding.
      
      Change-Id: Ied982c0ad72093eab29e38b8cd74d5c3d7458b10
      778ec421