1. 17 Aug, 2017 1 commit
    • Yushin Cho's avatar
      Introduce runtime switch for dist_8x8 · 55104335
      Yushin Cho authored
      Even if 'dist-8x8' is enabled with configure,
      the dist-8x8 is not acutally enabled (so, no change in encoding behaviour)
      until the command line option, '--enable-dist-8x8=1" is used.
      
      The cdef-dist and daala-dist can not be enabled by a command line option yet.
      
      This commit is a part of prep-work to remove DIST_8X8, CDEF_DIST,
      and DAALA_DIST experimental flags.
      
      Change-Id: I5c2df90f837b32f44e756572a19272dfb4c3dff4
      55104335
  2. 10 Aug, 2017 2 commits
    • Jingning Han's avatar
      Refactor lv-map rate cost estimation · dfd72323
      Jingning Han authored
      Use the rate cost to replace the probability fetch. This improves
      the encoding speed of level map by 10%. No coding stats change.
      
      Change-Id: Iab0beac5da291eed50f657549667d4edffb4ead8
      dfd72323
    • Urvang Joshi's avatar
      Remove PALETTE flag · c6300aa1
      Urvang Joshi authored
      This experiment is now adopted as it was cleared by Tapas.
      
      Note: Palette use can still be controlled by command-line option
      "--tune-content=..." in 'aomenc'.
      
      Change-Id: I832f49f20f60c34bdef5b424755849c496687e87
      c6300aa1
  3. 03 Aug, 2017 1 commit
    • hui su's avatar
      Calculate coeff token cost from CDF · c0cf71df
      hui su authored
      AWCY results:
      PSNR	PSNR HVS  SSIM	CIEDE 2000
      -0.09	-0.04	  -0.02	  -0.03
      
      On Google testsets:
      lowres  -0.18%
      midres  -0.20%
      
      Above results are obtained with
      --disable-ext-refs --disable-dual-filter --disable-loop-restoration
      --disable-global-motion --disable-warped-motion
      
      Change-Id: Iba58d5e5ec9a65d0afba29609aa2e379a80d7236
      c0cf71df
  4. 31 Jul, 2017 2 commits
    • Yue Chen's avatar
      Move mode costs that can be updated inside a frame to MACROBLOCK · b23d00a0
      Yue Chen authored
      It is a refactoring patch, which aims to make the code ready for
      implementation of in-frame mode cost update in RDO.
      Also add mode cost update per sb row, but it won't affect coding
      results because cdf update in RDO is not there.
      Mode cost arrays are moved to MACROBLOCK because in multi-thread
      coding, threads share the same AV1_COMP.
      
      This patch does not have impact on coding results.
      
      Change-Id: I2e8f7d7d066b23ebfbfc998269023781f359a6ff
      b23d00a0
    • Yue Chen's avatar
      motion_var: computer motion_mode_cost from cdf · bdc8dab2
      Yue Chen authored
      Initialize mode cost using frame-level cdf.
      Also in rd selection stage, cdf is updated per 64x64.
      Performance gain 0.20%
      
      Still suboptimal since in real bitstream packing, cdf is updated
      per symbol. Per symbol update in RDO is work in progress.
      
      Change-Id: I5062af91d8b00e5bf4c08abd0a7bfb0e5b27a619
      bdc8dab2
  5. 20 Jul, 2017 1 commit
    • Yushin Cho's avatar
      New experiment DIST_8x8 · b7b60c57
      Yushin Cho authored
      A framework for computing a distortion at 8x8 luma block level
      during RDO-based mode decision search. New 8x8 distortion metric can
      be plugged in by way of this tool.
      
      Existing daala_dist now uses this experiment as well.
      Other possible applications that can make use of this experiment would be
      a distortion meric, which should apply at 8x8 pixels such as PSNR-HVS, SSIM, or etc.
      
      A rd_cost for final coding mode decision for a super block is
      computed for a partition size 8x8 or larger. For a block larger than 8x8,
      a distortion of each 8x8 block is independently computed then summed up.
      
      The rd_cost for 8x8 block with new 8x8 distortion metric is computed
      only when the mode decision of its sub8x8 blocks are completed.
      However, MSE distortion metric is used with sub8x8 mode decision. Thus,
      early termination is also determined with the MSE based rd_cost.
      Because the best rd_cost (i.e. the reference rd_cost) during sub8x8 prediction
      or sub8x8 tx is based on new 8x8 distortion while each sub8x8 uses MSE,
      the existing early termination cannot be used (And this can be the one of possible reason
      for the BD-Rate change with this revision).
      
      For a sub8x8 prediction, prediction mode for each sub8x8 block of a 8x8 block is
      decided with existing MSE and then av1_dist_8x8() is applied to the 8x8 pixels.
      (There is also av1_dist_8x8_diff, which can input diff signal directly)
      
      For a sub8x8 tx in a block larger than 8x8, instead of computing MSE distortion for
      each sub8x8 tx block, we wait until all sub8x8 tx blocks are encoded before av1_dist_8x8()
      is applied to 8x8 pixels.
      
      Sub8x8 prediction and transformas were most of tricky parts in this change.
      Two kind of distortions, for a) predicted pixels and b) decoded pixels
      (i.e. predicted + possible reconstructed residue), are always computed during RDO.
      In order to access those two signals a) and b) for a 8x8 block after
      its sub8x8 mode decision is finished, a) and b) need be properly stored for later retrieval.
      
      The CB4X4 makes the task of accessing a) and b) signals for sub8x8 block further difficult,
      since the intermediate data (i.e. a and/or b) for sub8x8 block
      are not easily accessible outside of current partition unless reconstruced
      with decided coding modes.
      
      Change-Id: If60301a890c0674a3de1d8206965bbd6a6495bb7
      b7b60c57
  6. 26 Jun, 2017 1 commit
  7. 15 Jun, 2017 1 commit
    • Urvang Joshi's avatar
      Remove 'rddiv' member from various structs. · 70006e46
      Urvang Joshi authored
      This was initialized from a const and never modified. But was still
      passed around and stored in multiple structs.
      
      Removed these 'rddiv' member variables and now RDOPT() and RDOPT_DBL()
      always use the const RDDIV_BITS directly.
      
      Change-Id: I1a8dfd2c8fa857d466ad1207b4f0dd6ec07eafb8
      70006e46
  8. 08 Jun, 2017 1 commit
    • Yushin Cho's avatar
      Refactor sub8x8 tx size RD for daala-dist · 30a2c5f2
      Yushin Cho authored
      For a tx size RD search with partition size >= 8x8 and tx size < 8x8,
      daala-dist function is applied to the whole partition after all tx blocks are encoded
      instead of each 8x8 sub block of the partition.
      
      Change-Id: I27d9e2960aa641f550096e32ebcdf8dfb4de79a6
      30a2c5f2
  9. 02 Jun, 2017 1 commit
    • Jingning Han's avatar
      Resolve extremely large stack alloc in rdopt · d064cf03
      Jingning Han authored
      Move the large stack allocation from stack initialization to
      dedicated mem space. This resolves the extremely large stack issue
      when ext-partition, motion-var, and high bit-depth are all turned
      on.
      
      BUG=aomedia:415
      
      Change-Id: I85b77bbc6429093fcb0152176d9e237087d6bbd8
      d064cf03
  10. 01 Jun, 2017 1 commit
    • Yushin Cho's avatar
      Fix daala-dist for cb4x4 · 63927c43
      Yushin Cho authored
      The place where av1_daala_dist() is applied for sub8x8 partition is
      moved from sub8x8 mode decision functions to rd_pick_partition().
      
      BD-Rate change by daala-dist with '--disable-var-tx' is:
      (AWCY, objective-1-fast, high delay mode)
      
         PSNR | PSNR Cb | PSNR Cr | PSNR HVS |    SSIM | MS SSIM | CIEDE 2000
      15.1558 | 12.9585 | 14.4662 |  -3.8651 | -1.7102 | -9.2956 |    10.8686
      
      In MSE probe mode:
      
        PSNR | PSNR Cb | PSNR Cr | PSNR HVS |   SSIM | MS SSIM | CIEDE 2000
      0.0429 |  0.0435 |  0.1651 |  -0.0415 | 0.0850 |  0.0122 |     0.0546
      
      Change-Id: I3b2ea916d41c48e433eb641adf44552e4725c198
      63927c43
  11. 31 May, 2017 2 commits
  12. 05 May, 2017 1 commit
  13. 27 Apr, 2017 1 commit
    • Luc Trudeau's avatar
      [CFL] Store luma as prediction for chroma · e3980281
      Luc Trudeau authored
      Stores the reconstructed luma pixels for each transform block inside a
      prediction block. Rectangular transform blocks are supported.
      
      As for RDO, after all the modes have been tested for luma, an extra
      encoding is perform in order to store the reconstructed pixel values of
      the best mode. These values are then used for RDO on the chromatic
      planes.
      
      Change-Id: I354d9827e32fd41065f1b2ce02832d943a6fa156
      e3980281
  14. 24 Apr, 2017 1 commit
    • Alex Converse's avatar
      Refactor mv limits. · 0fa0f42c
      Alex Converse authored
      cherry pick of libvpx commit 6554333b59cf773983c7cacbd832d754f75139e8
      
      BUG=aomedia:443
      
      Change-Id: Ifebdc9ef37850508eb4b8e572fd0f6026ab04987
      0fa0f42c
  15. 10 Apr, 2017 1 commit
    • Angie Chiang's avatar
      Use block_idx rather than block_raster_idx · 752ccce7
      Angie Chiang authored
      1) block_raster_idx is actually raster order only when tx_size
      is TX_4x8.
      It's very specific, so we should put it near to the place it's
      actually used.
      
      2) Sync the meaning of block_idx on encoder/decoder sides
      
      Change-Id: I7d37a992cb773503e29f9c0d9d2586e580aa6173
      752ccce7
  16. 27 Mar, 2017 1 commit
    • Angie Chiang's avatar
      Use txb_entropy_ctx to store entropy ctx of lv_map · 74e2307b
      Angie Chiang authored
      1) Add txb_entropy_ctx into MACROBLOCK_PLANE and PICK_MODE_CONTEXT
      
      2) Add av1_get_txb_entropy_context() to compute the entropy context
      
      3) Compute and sore the entropy context before av1_xform_quant()
      return
      
      Change-Id: Ia2170523af3163b9456f7c6a305c1e77ad2b23be
      74e2307b
  17. 21 Mar, 2017 1 commit
    • Angie Chiang's avatar
      Set mbmi_ext's tcoeff to size MAX_SB_SQUARE · c484abe7
      Angie Chiang authored
      This a simple implementation.
      We will use a more precise buffer size for tcoeff once the
      experiment functions correctly.
      
      Change-Id: Ib561974f21ee1b8d72ce407882ea2be3cf0b069f
      c484abe7
  18. 16 Mar, 2017 2 commits
    • Angie Chiang's avatar
      Draft of av1_update_txb_context · 0397eda0
      Angie Chiang authored
      This draft version only pass compiling check, it's not working yet.
      
      The following goal is to use new coding system when doing bitstream
      packing but keep old coding system in RD loop.
      
      Change-Id: I224a1581d1cc5c67d73e71558fb77d9faf9c2470
      0397eda0
    • Angie Chiang's avatar
      Add frame level buffer to store txfm coeffs · f0fbf9d9
      Angie Chiang authored
      Doing tokenize in the last step of RD loop and then doing packing
      tokens in bitstream packing phase is hard for debugging.
      
      Therefore, we create a frame-level buffer to store the txfm coeffs
      from the reconstruction in RD loop and then in bitstream packing
      phase, we can code the txfm coeffs directly.
      
      Change-Id: I999470eef6e038317a91585df2bdfc20aca3573e
      f0fbf9d9
  19. 23 Feb, 2017 1 commit
  20. 22 Feb, 2017 1 commit
    • Jingning Han's avatar
      Skip 2x2 uv block in the rd search loop · 8efdbc8b
      Jingning Han authored
      Skip the 2x2 uv block rate-distortion optimization process in the
      non-reference positions in key frame encoding.
      
      Change-Id: I51112492fd79f9a1c205fba151f8658cea9ee85f
      8efdbc8b
  21. 14 Feb, 2017 1 commit
    • James Zern's avatar
      av1/encoder/block.h: s/inline/INLINE/ · d90eefeb
      James Zern authored
      vs2013 is still supported, it still uses the old behavior which supports
      inline in c++ only.
      
      Change-Id: I74417c6c93774e55122bbbf1db0283e2c30fe63e
      d90eefeb
  22. 23 Jan, 2017 1 commit
    • Yushin Cho's avatar
      Add a new experiment, DAALA_DIST · 7a428ba2
      Yushin Cho authored
      This commit adds a new experiment, Daala's distortion function,
      which is designed to better approximate perceptual distortion
      in 8x8 pixel blocks.
      
      This experiment is expected to work best with PVQ.
      
      It measures the variance of overlapped 4x4 regions in the 8x8 area,
      then uses these variances to scale the MSE of weighted frequency domain
      distortion of 8x8 block.
      
      Since AV1 calculates distortion in blocks as small as 4x4, it is not possible to
      directly replace the existing distortion functions of AV1,
      such as dist_block() and block_rd_txf().
      Hence, there has been substantial changes in order to apply
      Daala's 8x8 distortion function.
      The daala distortion function is applied
      after all 4x4 tx blocks in a 8x8 block are encoded (during RDO),
      as in below two cases:
      1) intra/inter sub8x8 predictions and
      2) 4x4 transform with prediction size >= 8.
      
      To enable this experiment, add '--enable-daala-dist' with configure.
      
      TODO: Significant tuning of parameters is required since the function has
      originally came from Daala thus most parameters would not work
      correctly outside Daala.
      The fact that chroma distortion is added to the distortion of AV1's RDO is
      also critical since Daala's distortion function is applied to luma only
      and chroma continues to use MSE.
      
      Change-Id: If35fdd3aec7efe401f351ba1c99891ad57a3d957
      7a428ba2
  23. 18 Jan, 2017 1 commit
    • Urvang Joshi's avatar
      Enable rectangular transforms for Intra also. · feb925fe
      Urvang Joshi authored
      These are under EXT_TX + RECT_TX experiment combo.
      
      Results
      =======
      
      Derf Set:
      --------
      All Intra frames: 1.8% avg improvement (and 1.78% BD-rate improvement)
      Video: 0.230% avg improvement (and 0.262% BD-rate improvement)
      
      Objective-1-fast set
      --------------------
      Video: 0.52 PSNR improvement
      
      Change-Id: I1893465929858e38419f327752dc61c19b96b997
      feb925fe
  24. 06 Jan, 2017 1 commit
  25. 04 Jan, 2017 1 commit
  26. 07 Dec, 2016 1 commit
  27. 07 Nov, 2016 1 commit
    • Yushin Cho's avatar
      New experiment: Perceptual Vector Quantization from Daala · 77bba8d3
      Yushin Cho authored
      PVQ replaces the scalar quantizer and coefficient coding with a new
      design originally developed in Daala. It currently depends on the
      Daala entropy coder although it could be adapted to work with another
      entropy coder if needed:
      ./configure --enable-experimental --enable-daala_ec --enable-pvq
      
      The version of PVQ in this commit is adapted from the following
      revision of Daala:
      https://github.com/xiph/daala/commit/fb51c1ade6a31b668a0157d89de8f0a4493162a8
      
      More information about PVQ:
      - https://people.xiph.org/~jm/daala/pvq_demo/
      - https://jmvalin.ca/papers/spie_pvq.pdf
      
      The following files are copied as-is from Daala with minimal
      adaptations, therefore we disable clang-format on those files
      to make it easier to synchronize the AV1 and Daala codebases in the future:
       av1/common/generic_code.c
       av1/common/generic_code.h
       av1/common/laplace_tables.c
       av1/common/partition.c
       av1/common/partition.h
       av1/common/pvq.c
       av1/common/pvq.h
       av1/common/state.c
       av1/common/state.h
       av1/common/zigzag.h
       av1/common/zigzag16.c
       av1/common/zigzag32.c
       av1/common/zigzag4.c
       av1/common/zigzag64.c
       av1/common/zigzag8.c
       av1/decoder/decint.h
       av1/decoder/generic_decoder.c
       av1/decoder/laplace_decoder.c
       av1/decoder/pvq_decoder.c
       av1/decoder/pvq_decoder.h
       av1/encoder/daala_compat_enc.c
       av1/encoder/encint.h
       av1/encoder/generic_encoder.c
       av1/encoder/laplace_encoder.c
       av1/encoder/pvq_encoder.c
       av1/encoder/pvq_encoder.h
      
      Known issues:
      - Lossless mode is not supported, '--lossless=1' will give the same result as
      '--end-usage=q --cq-level=1'.
      - High bit depth is not supported by PVQ.
      
      Change-Id: I1ae0d6517b87f4c1ccea944b2e12dc906979f25e
      77bba8d3
  28. 04 Nov, 2016 1 commit
    • Yushin Cho's avatar
      New experiment: Perceptual Vector Quantization from Daala · 09705fe7
      Yushin Cho authored
      PVQ replaces the scalar quantizer and coefficient coding with a new
      design originally developed in Daala. It currently depends on the
      Daala entropy coder although it could be adapted to work with another
      entropy coder if needed:
      ./configure --enable-experimental --enable-daala_ec --enable-pvq
      
      The version of PVQ in this commit is adapted from the following
      revision of Daala:
      https://github.com/xiph/daala/commit/fb51c1ade6a31b668a0157d89de8f0a4493162a8
      
      More information about PVQ:
      - https://people.xiph.org/~jm/daala/pvq_demo/
      - https://jmvalin.ca/papers/spie_pvq.pdf
      
      The following files are copied as-is from Daala with minimal
      adaptations, therefore we disable clang-format on those files
      to make it easier to synchronize the AV1 and Daala codebases in the future:
       av1/common/generic_code.c
       av1/common/generic_code.h
       av1/common/laplace_tables.c
       av1/common/partition.c
       av1/common/partition.h
       av1/common/pvq.c
       av1/common/pvq.h
       av1/common/state.c
       av1/common/state.h
       av1/common/zigzag.h
       av1/common/zigzag16.c
       av1/common/zigzag32.c
       av1/common/zigzag4.c
       av1/common/zigzag64.c
       av1/common/zigzag8.c
       av1/decoder/decint.h
       av1/decoder/generic_decoder.c
       av1/decoder/laplace_decoder.c
       av1/decoder/pvq_decoder.c
       av1/decoder/pvq_decoder.h
       av1/encoder/daala_compat_enc.c
       av1/encoder/encint.h
       av1/encoder/generic_encoder.c
       av1/encoder/laplace_encoder.c
       av1/encoder/pvq_encoder.c
       av1/encoder/pvq_encoder.h
      
      Known issues:
      - Lossless mode is not supported, '--lossless=1' will give the same result as
      '--end-usage=q --cq-level=1'.
      - High bit depth is not supported by PVQ.
      
      Change-Id: I1ae0d6517b87f4c1ccea944b2e12dc906979f25e
      09705fe7
  29. 25 Oct, 2016 2 commits
  30. 24 Oct, 2016 1 commit
    • David Barker's avatar
      Fix a bug when combining new-quant + supertx · d7d78c83
      David Barker authored
      Previously, we assumed that av1_init_plane_quantizers is always called with
      segment_id == xd->mi[0]->mbmi.segment_id (and use the latter to derive the value
      of 'qindex' to use in the quantizer). But this is no longer true when supertx
      is enabled. This patch instead remembers the value of 'qindex' derived from
      the latest call to av1_init_plane_quantizers and uses that directly.
      
      Change-Id: Ifa1c5bf74cad29942ff79b88ca92c231bc07f336
      d7d78c83
  31. 21 Oct, 2016 1 commit
    • Jingning Han's avatar
      Allow frame level tx_mode switch · 9777afc3
      Jingning Han authored
      Check the encoding statistics. If all the coding blocks use the
      max transform size, skip transform size coding in the frame header.
      
      Change-Id: I31cb16314e87f945d7e95a34a90a5536b3ed82d5
      9777afc3
  32. 19 Oct, 2016 1 commit
  33. 18 Oct, 2016 1 commit
  34. 17 Oct, 2016 1 commit
    • Yue Chen's avatar
      Refactor motion estimation in MOTION_VAR experiment · e9638ccf
      Yue Chen authored
      To get ready for pulling AV1 to nextgenv2. Refactoring is done to
      make the code structures similar, especially for the motion search
      part.
      
      Change-Id: I5d7636394408d97de55394d668540f5627827983
      e9638ccf
  35. 14 Oct, 2016 1 commit
    • Urvang Joshi's avatar
      Wrap palette code inside CONFIG_PALETTE flag. · b100db7c
      Urvang Joshi authored
      This flag was already added to aomedia/master, so bringing it back to
      webm/nextgenv2, as part of an effort to get the two codebases in sync.
      
      Change-Id: I2b933a6a160e4210d1411a9e7978149eb8553205
      b100db7c