1. 03 Nov, 2017 2 commits
    • Alexander Bokov's avatar
      Add highbd support in predict_skip_flag · 80eedf2e
      Alexander Bokov authored
      Change-Id: I4270d1260854ac27b68c5694ca8102b92bee6faa
      80eedf2e
    • Alexander Bokov's avatar
      Introducing a model for pruning the TX size search · 79a37242
      Alexander Bokov authored
      Use a neural-network-based binary classifier to predict the first split
      decision on the highest level of the TX size RD search tree. Depending
      on how confident we are in the prediction we either keep full unmodified
      TX size search or use the largest possible TX size and stop any further
      search.
      
      Average speed-up: 3-4%
      Quality loss (lowres): 0.062%
      Quality loss (midres): 0.018%
      
      Change-Id: I64c0317db74cbeddfbdf772147c43e99e275891f
      79a37242
  2. 02 Nov, 2017 2 commits
  3. 01 Nov, 2017 6 commits
    • Sarah Parker's avatar
      Rename ZEROMV to GLOBALMV · 2b9ec2ea
      Sarah Parker authored
      This name better reflects the functionality of the mode.
      
      Change-Id: Idc2372e7c7b226f70f63fa4f272a88cb4d857089
      2b9ec2ea
    • Debargha Mukherjee's avatar
      Add speed feature to reduce tx size search depth · edc7346f
      Debargha Mukherjee authored
      The speed feature simply restricts the number of depths
      searched. Currently it is turned on by default for speeds>=1.
      The coding efficiency impact (tested on lowres 30 frames) seems
      to be ~0.15% and the speedup is in the order of 15%.
      
      Change-Id: I514832bd7df937292875f73d9c9026e49ac576f2
      edc7346f
    • Zoe Liu's avatar
      Remove redundant macro USE_UNI_COMP_REFS · c6b17d82
      Zoe Liu authored
      USE_UNI_COMP_REFS is always true when ext-comp-refs is enabled and is
      always false otherwise.
      
      Change-Id: Ic6061a4ab88a2ebfeb11f178872ab13909060c0b
      c6b17d82
    • Sarah Parker's avatar
      Use tx_size 1 level down for transform type search · 90024e44
      Sarah Parker authored
      This addresses an inconsistency between the set used
      to decode the tx_type in the bitstream and the set used
      for the tx_type search. Previously, the set used to
      read/write the tx_type was based on the smallest tx_size
      in the vartx partitioning, but the search uses a set
      based on the largest possible tx_size. This patch
      changes the tx_type search to use the transform type
      set associated with the tx_size 1 recursive level down from
      the max square tx_size to make the search more consistent
      with the bitstream syntax. If a tx_size is selected for an
      invalid tx_type, DCT_DCT is used for that partition instead.
      
      This patch also adds assertions to all exposed transform
      functions to ensure that no illegal transform type/size
      combinations occur.
      
      This currently gets a 0.1% drop in performance on lowres.
      The drop is due to the reduction of the tx_types available
      for 32x16 and 16x32 transform sizes. Before this patch,
      32x16 and 16x32 transforms were getting assigned a
      set of 12 tx_types, some of which we did not intend to
      support for these sizes.
      
      Change-Id: I44aca4876b261c345623cd04ad6235bca4532701
      90024e44
    • Yue Chen's avatar
      Remove CONFIG_INTERINTRA · 670b660d
      Yue Chen authored
      Change-Id: Icbedc16b01adf9b48f0f357ec89143462865d54e
      670b660d
    • Yue Chen's avatar
      Guard the limits in for-loops for setting inter_tx_size · c5252a69
      Yue Chen authored
      Always at least set the first row and the first col, otherwise will
      have unattended entries if we enable 4x16 or 16x4 txs.
      
      Change-Id: I403ea3694ccf8c7ee1584f55bbc3e9b8355d9272
      c5252a69
  4. 31 Oct, 2017 3 commits
    • Sebastien Alaiwan's avatar
      Remove experimental flag of WARPED_MOTION · 1f56b8e3
      Sebastien Alaiwan authored
      This experiment has been adopted, we can simplify the code
      by dropping the associated preprocessor conditionals.
      
      Change-Id: I54fee8a18afaec5333ae852eb6bd9ac0e2550db1
      1f56b8e3
    • Sebastien Alaiwan's avatar
      Remove experimental flag of MOTION_VAR · 1bc94fcc
      Sebastien Alaiwan authored
      This experiment has been adopted, we can simplify the code
      by dropping the associated preprocessor conditionals.
      
      Change-Id: I2dce80e1e1b2116708b6ba9feeacaacc12af8fc4
      1bc94fcc
    • Debargha Mukherjee's avatar
      Adding a speed feature for tx_size search · 51666866
      Debargha Mukherjee authored
      This patch factors out a function that computes the rd cost for
      a given transform type given the transform partition already
      computed. This is then used to develop a speed feature where the
      transform size search disables trellis optimization but once the
      transform sizes are decided, a final search is conducted with
      optimization turned back on.
      This patch does not change anything in speed 0 yet.
      
      Change-Id: I30acfc5e2dd353d711e5f4260d5b344847b03ade
      51666866
  5. 30 Oct, 2017 5 commits
    • Debargha Mukherjee's avatar
      Remove compound-segment/wedge config flags · 371968cd
      Debargha Mukherjee authored
      Change-Id: I39cfbb135add0553cadf64481b13786831fbdddd
      371968cd
    • Jingning Han's avatar
      Speed up inter frame rate-distortion optimization · cf842ad2
      Jingning Han authored
      The frame marker system supports one to map the reference frame
      index into the natural order. It allows direct checking on the
      efficacy of the reference frames given their relative locations
      with respect to the current coding frame.
      
      This commit uses such property to filter out reference frames
      less likely to contribute coding gains from the rate-distortion
      optimization process. For example, it takes out the check on
      last2 / 3 frames, when their actual location is further away
      from the golden frame.
      
      The AWCY results show 0.6% performance regression. The encoding
      speed gets doubled.
      
      To use the speed up, one needs to turn on frame-marker experiment
      before we turn it on by default, and enable selective_ref_frame
      entry in the speed feature.
      
      Change-Id: Ifb03ed90acd980bbc7ff1c2e17982e21e68d2588
      cf842ad2
    • Sebastien Alaiwan's avatar
      Remove experimental flag of GLOBAL_MOTION · 48795807
      Sebastien Alaiwan authored
      This experiment has been adopted, we can simplify the code
      by dropping the associated preprocessor conditionals.
      
      Change-Id: I9c9d6ef5317798cbf237307a9754fe7e03bdda47
      48795807
    • Luc Trudeau's avatar
      [CFL] Sub8x8 Validation Code Rewrite · c7af36d4
      Luc Trudeau authored
      Sub8x8 Validation code is changed to be more robust. The scope of the
      validation is narrowed to validating that all of the required content in
      the storage buffer was stored between CfL predictions. The early
      termination used in the current mode decision code does not allow to
      validate more than that.
      
      This change does not change encoder output
      
      BUG=aomedia:925
      
      Change-Id: I7f1ed84da5037dcfaaf5da9cf33b4b8d664d2352
      c7af36d4
    • Debargha Mukherjee's avatar
      Remove experimental flag for rect-tx · 11812967
      Debargha Mukherjee authored
      Change-Id: I0cc53a03f07a11a6f7ea0570ff4ee8cf7c18c5aa
      11812967
  6. 27 Oct, 2017 6 commits
    • Joe Young's avatar
      Ext-intra modification/tuning · 3ca43bf0
      Joe Young authored
      For ext-intra direcation intra modes:
      
      1. Use neighbor block modes to modify edge filtering strength
         Coding gain (lowres/midres/hdres):
           (8 keyframes)
           PSNR: -0.19 -0.22 -0.10
           SSIM: -0.29 -0.27 -0.13
      
      2. Use context-based cdf to code angle_delta syntax
           (8 keyframes)
           PSNR: -0.20 -0.24 -0.27
           SSIM: -0.29 -0.33 -0.37
      
      3. Filter corner sample:
           (8 keyframes)
           PSNR: -0.01 -0.02 -0.05
           SSIM: -0.03 -0.04 -0.05
      
      Combined Bd-rate improvement for 8 keyframes
           PSNR: -0.40 -0.47 -0.40
           SSIM: -0.57 -0.60 -0.51
      
      Change-Id: Id47ac17b6bf91cd810b70cacfc5b457341f417f3
      3ca43bf0
    • Hui Su's avatar
      Resolve an issue between palette and delta-q · efb755cd
      Hui Su authored
      Avoid resetting current_q_index.
      
      Change-Id: I76d27605b6a8c0cd6b1032e24ea5ca8561115e5a
      efb755cd
    • Cheng Chen's avatar
      JNT_COMP: 5. Support compound_segment · 3392c436
      Cheng Chen authored
      Let jnt_comp work with compound_segment. When frame distance weights
      are used, compound_segment blending is turned off. When normal compound
      mode is used, compound_segment blending is searched. The rdcost should
      select between {frame distance weights, normal compound,
      compound_segment blending}.
      
      Change-Id: I162d1d204ba6a5976538357d9dbc71fc5b1aa8d5
      3392c436
    • Cheng Chen's avatar
      JNT_COMP: 4. add context and entropy read/write · 0a7f2f51
      Cheng Chen authored
      Change-Id: I0e6f7ab981e31f7120105515f6204568b6dc82d3
      0a7f2f51
    • Cheng Chen's avatar
      JNT_COMP: 3. rd select the best weight · ca6958c6
      Cheng Chen authored
      Select the best compound_idx in rd.
      The rate/cost for compound_idx and their ctx will be in patch 4.
      
      But there's a bug for now if we don't encode one more time using the
      selected compound_idx. It remains a issue to be solved in the future.
      
      Change-Id: I5e1ba51da2b6ab5bacd8aba752dda43bd2257014
      ca6958c6
    • RogerZhou's avatar
      Fixed build conflict (amvr,intrabc). · 10a0380a
      RogerZhou authored
      Change-Id: Ibfeb424bf0ebab7bbeb69f6f6df24a4f4924ec97
      10a0380a
  7. 26 Oct, 2017 1 commit
  8. 25 Oct, 2017 3 commits
    • Cheng Chen's avatar
      JNT_COMP: 2. assign proper weigths in rdopt · efc55fd9
      Cheng Chen authored
      Change-Id: I255be6e0193dd6b91424ce53ed41aeaaeb1c01a7
      efc55fd9
    • Yue Chen's avatar
      Optimizations for filter_intra · 57b8ff68
      Yue Chen authored
      Reduce number of modes from 10 to 6, and disable fi modes in UV.
      To reduce complexity, apply filter directly without subtracting
      the estimated means.
      
      Change-Id: Iaf78d92d31e4a7cc30ea7863b57a9611c5f503e6
      57b8ff68
    • David Michael Barr's avatar
      [CFL] Switch to txfm_rd_in_plane in alpha search · 1f8d0950
      David Michael Barr authored
      This is more precise than the dist functions it replaces.
      
      Results on Subset1 (compared with previous commit with CfL enabled)
        PSNR | PSNR Cb | PSNR Cr | PSNR HVS |   SSIM | MS SSIM | CIEDE 2000
      0.0634 | -0.9188 | -0.9429 |   0.0609 | 0.0722 |  0.0593 |    -0.3226
      
      Change-Id: I955a7d7eceea50482edb40b0d1041b300e3c9042
      1f8d0950
  9. 24 Oct, 2017 3 commits
  10. 23 Oct, 2017 1 commit
    • Yaowu Xu's avatar
      Use type with smaller size · abe5215f
      Yaowu Xu authored
      This is to reduce the size of the huge token buffer by at least more
      than 10%.
      
      BUG=aomedia:940
      
      Change-Id: I2ee7f7f62e8e8fb819cf7fba40a7ca7f860acaa4
      abe5215f
  11. 21 Oct, 2017 2 commits
  12. 20 Oct, 2017 3 commits
  13. 19 Oct, 2017 2 commits
    • Yue Chen's avatar
      Disable residue hash feature on cross-border blocks · 25dc0701
      Yue Chen authored
      Disable this feature unless the entire block is within the frame.
      The reason is, rd decisions in mbmi, e.g. inter_tx_block[][], made
      for blocks partially out of the border can be partly nonsense
      therefore cannot be reused by blocks at other locations.
      
      It caused an infinite loop when encoding a clip with repetitive
      patterns. A cross-border block has an invalid big tx stored
      in inter_tx_block[0][1] and the other block (same residue, within
      frame) reused this mbmi, which makes encoder never reach the
      termination condition when tx blocks are being recursively
      partitioned.
      
      BUG=aomedia:913
      
      Change-Id: Id25a1dbc4a68b5136f6bdf9f6b5811b7ec6920b0
      25dc0701
    • Luc Trudeau's avatar
      [CFL] Fix negative rounding issue in alpha dist · a45b104a
      Luc Trudeau authored
      get_scaled_luma_q0(-alpha_q3, pred_buf_q3[i]) is NOT equivalent to
      -get_scaled_luma_q0(alpha_q3, pred_buf_q3[i]). When the product
      alpha_q3*pred_buf_q3[i] is an exact multiple of 32 (0.5 in Q6), then the right
      shift will round both positive and negative values towards infinity, creating a
      bias. So, e.g., get_scaled_luma_q0(-4, 8) will yield 0, but
      -get_scaled_luma_q0(4, 8) will yield -1.
      
      Results on Subset1 (compared to parent With CfL enabled)
        PSNR | PSNR Cb | PSNR Cr | PSNR HVS |   SSIM | MS SSIM | CIEDE 2000
      0.0000 |  0.0000 |  0.0000 |   0.0000 | 0.0000 |  0.0000 |     0.0000
      
      https://arewecompressedyet.com/?job=cfl-no-round-fix%402017-10-07T11%3A50%3A47.711Z&job=cfl-round-fix%402017-10-07T02%3A15%3A51.359Z
      
      Change-Id: I8a7900c32fbd7213f1ed4e09c3626c063800e186
      a45b104a
  14. 18 Oct, 2017 1 commit