1. 26 Mar, 2013 5 commits
    • Ronald S. Bultje's avatar
      Add col/row-based coefficient scanning patterns for 1D 8x8/16x16 ADSTs. · d9094d8f
      Ronald S. Bultje authored
      These are mostly just for experimental purposes. I saw small gains (in
      the 0.1% range) when playing with this on derf.
      
      Change-Id: Ib21eed477bbb46bddcd73b21c5c708a5b46abedc
      d9094d8f
    • Ronald S. Bultje's avatar
      Redo banding for all transforms. · 3120dbdd
      Ronald S. Bultje authored
      Now that the first AC coefficient in both directions use the same DC
      as their context, there no longer is a purpose in letting both have
      their own band. Merging these two bands allows us to split bands for
      some of the very high-frequency AC bands.
      
      In addition, I'm redoing the banding for the 1D-ADST col/row scans. I
      don't think the old banding made any sense at all (it merged the last
      coefficient of the first row/col in the same band as the first two of
      the second row/col), which was clearly an oversight from the band being
      applied in scan-order (rather than in their actual position). Now,
      coefficients at the same position will be in the same band, regardless
      what scan order is used. I think this makes most sense for the purpose
      of banding, which is basically "predict energy for this coefficient
      depending on the energy of context coefficients" (i.e. pt).
      
      After full re-training, together with previous patch, derf gains about
      1.2-1.3%, and hd/stdhd gain about 0.9-1.0%.
      
      Change-Id: I7a0cc12ba724e88b278034113cb4adaaebf87e0c
      3120dbdd
    • Ronald S. Bultje's avatar
      Use above/left (instead of previous in scan-order) as token context. · 790fb132
      Ronald S. Bultje authored
      Pearson correlation for above or left is significantly higher than for
      previous-in-scan-order (absolute values depend on position in scan, but
      in general, we gain about 0.1-0.2 by using either above or left; using
      both basically just makes this even better). For eob branch skipping,
      we continue to use the previous token in scan order.
      
      This helps about 0.9% on derf after re-training on a limited data set.
      Full re-training and results on larger-resolution clips are pending.
      
      Note that this commit breaks trellis, so we can probably get further
      gains out of it by fixing trellis at some later point.
      
      Change-Id: Iead68e296fc3a105cca746b5e3da9555d6010cfe
      790fb132
    • John Koleszar's avatar
      Add an in-loop deringing experiment · 441e2eab
      John Koleszar authored
      Adds a per-frame, strength adjustable, in loop deringing filter. Uses
      the existing vp9_post_proc_down_and_across 5 tap thresholded blur
      code, with a brute force search for the threshold.
      
      Results almost strictly positive on the YT HD set, either having no
      effect or helping PSNR in the range of 1-3% (overall average 0.8%).
      Results more mixed for the CIF set, (-0.5 min, 1.4 max, 0.1 avg).
      This has an almost strictly negative impact to SSIM, so examining a
      different filter or a more balanced search heuristic is in order.
      
      Other test set results pending.
      
      Change-Id: I5ca6ee8fe292dfa3f2eab7f65332423fa1710b58
      441e2eab
    • Deb Mukherjee's avatar
      Modeling default coef probs with distribution · fd18d5df
      Deb Mukherjee authored
      Replaces the default tables for single coefficient magnitudes with
      those obtained from an appropriate distribution. The EOB node
      is left unchanged. The model is represeted as a 256-size codebook
      where the index corresponds to the probability of the Zero or the
      One node. Two variations are implemented corresponding to whether
      the Zero node or the One-node is used as the peg. The main advantage
      is that the default prob tables will become considerably smaller and
      manageable. Besides there is substantially less risk of over-fitting
      for a training set.
      
      Various distributions are tried and the one that gives the best
      results is the family of Generalized Gaussian distributions with
      shape parameter 0.75. The results are within about 0.2% of fully
      trained tables for the Zero peg variant, and within 0.1% of the
      One peg variant.
      
      The forward updates are optionally (controlled by a macro)
      model-based, i.e. restricted to only convey probabilities from the
      codebook. Backward updates can also be optionally (controlled by
      another macro) model-based, but is turned off by default. Currently
      model-based forward updates work about the same as unconstrained
      updates, but there is a drop in performance with backward-updates
      being model based.
      
      The model based approach also allows the probabilities for the key
      frames to be adjusted from the defaults based on the base_qindex of
      the frame. Currently the adjustment function is a placeholder that
      adjusts the prob of EOB and Zero node from the nominal one at higher
      quality (lower qindex) or lower quality (higher qindex) ends of the
      range. The rest of the probabilities are then derived based on the
      model from the adjusted prob of zero.
      
      Change-Id: Iae050f3cbcc6d8b3f204e8dc395ae47b3b2192c9
      fd18d5df
  2. 22 Mar, 2013 2 commits
    • Paul Wilkins's avatar
      Disable zero bin mode boost. · ec080fa9
      Paul Wilkins authored
      As things stand the zero bin mode boost is hurting somewhat.
      In part this seems to be because the boost applied as is
      interferes with the rd mode selection loop.
      
      Average gains (derf 0.072, yt 0.243, ythd 0.179 std-hd 0.212%)
      
      Change-Id: Icaecea3908d9a7352370e49b8fa822f2c2c49dc1
      ec080fa9
    • Paul Wilkins's avatar
      Minor code clean up · 815734e5
      Paul Wilkins authored
      Change-Id: Ifa864e0acb253b238b03cdeed0fe5d6ee30a45d8
      815734e5
  3. 20 Mar, 2013 1 commit
  4. 18 Mar, 2013 5 commits
    • Paul Wilkins's avatar
      Changes to rd error_per_bit calculation. · d8ffee45
      Paul Wilkins authored
      Specifically changes to retain more precision
      especially at low Q through to the point of use.
      
      Change-Id: Ief5f010f2ca4daaabef49520e7edb46c35daf397
      d8ffee45
    • Ronald S. Bultje's avatar
      Fix ENTROPY_STATS code in vp9_tokenize.c. · b99dce68
      Ronald S. Bultje authored
      Change-Id: I9b4cb1e2ce6c6a99cffd473ff2fa7579bd318fcd
      b99dce68
    • Yunqing Wang's avatar
      Optimize 8x8 idct function · 6344c84c
      Yunqing Wang authored
      Wrote sse2 functions of vp9_short_idct8x8 and vp9_short_idct10_8x8.
      Compared to c version, the sse2 version is 2X faster. The decoder
      test didn't show noticeable gain since 8x8 idct doesn't take much
      of decoding time (less than 1% in my test).
      
      Change-Id: I56313e18cd481700b3b52c4eda5ca204ca6365f3
      6344c84c
    • John Koleszar's avatar
      Replace scaling byte with explicit display size · 8a3f55f2
      John Koleszar authored
      If the intended display size is different than the size the frame is
      coded at, then send that size explicitly in the bitstream. Adds a new
      bit to the frame header to indicate whether the extra size fields
      are present.
      
      Change-Id: I525c66f22d207efaf1e5f903c6a2a91b80245854
      8a3f55f2
    • Paul Wilkins's avatar
      Adapt ARNR filter length and strength. · cdb322dd
      Paul Wilkins authored
      Adjust the filter length and strength for each
      ARF group based on a measure of difficulty (the boost)
      and the active q range.
      
      Remove lower limit on RDMULT value.
      
      Average gains on the different sets in range 0.4%-0.9%.
      However the ARNR changes give a very big boost on a
      few clips.
      
      Eg. Soccer ~5%, in derf set and Cyclist ~ 10% in the std-hd set
      
      Change-Id: I2078d78798e27ad2bcc2b32d703ea37b67412ec4
      cdb322dd
  5. 16 Mar, 2013 3 commits
    • Deb Mukherjee's avatar
      Context-pred fix to not use top/left on edges · b1921b2f
      Deb Mukherjee authored
      This fix resolves some of the mismatch issues being seen
      recently. While this is the right thing to do when tiling
      is used for this experiment, it is not the underlying cause
      of the the mismatches.
      Something else is causing writing outside of the allowable
      frame area in the encoder leading to this mismatch.
      
      Change-Id: If52c6f67555aa18ab8762865384e323b47237277
      b1921b2f
    • John Koleszar's avatar
      Remove some unused rate control variables · b8ac9f2f
      John Koleszar authored
      These variables are unused, and are subject to overflowing, causing
      assertions when built with -ftrapv.
      
      Change-Id: Ia00a3201af309906c05bcd4b23a643925ed6ea86
      b8ac9f2f
    • John Koleszar's avatar
      Fix use of NaN in firstpass · db5f2cb5
      John Koleszar authored
      If the second reference is better than the first in the long term,
      it was possible to try to take the fractional exponent of a
      negative number, giving an undefined result.
      
      Change-Id: I1dd08286747ceae960eb03bb5d98a383cc9d253b
      db5f2cb5
  6. 15 Mar, 2013 1 commit
    • Christian Duvivier's avatar
      Faster vp9_short_fdct16x16. · 4418b790
      Christian Duvivier authored
      Scalar path is about 1.5x faster (3.1% overall encoder speedup).
      SSE2 path is about 7.2x faster (7.8% overall encoder speedup).
      
      Change-Id: I06da5ad0cdae2488431eabf002b0d898d66d8289
      4418b790
  7. 14 Mar, 2013 4 commits
    • Yaowu Xu's avatar
      force lossless coding at very high quality end · 374a1736
      Yaowu Xu authored
      Change-Id: I75fc4eee10bee9efd419d248827290cce8e6d637
      374a1736
    • Yaowu Xu's avatar
      Remove leftover reference to 2nd order dc/ac quant · f4d2ad69
      Yaowu Xu authored
      Change-Id: Ib8dacf1d2797743569771b8f699e40e1aeb085cb
      f4d2ad69
    • John Koleszar's avatar
      Fix pulsing issue with scaling · 9b7be888
      John Koleszar authored
      Updates the YV12_BUFFER_CONFIG structure to be crop-aware. The
      exiting width/height parameters are left unchanged, storing the
      width and height algined to a 16 byte boundary. The cropped
      dimensions are added as new fields.
      
      This fixes a nasty visual pulse when switching between scaled and
      unscaled frame dimensions due to a mismatch between the scaling
      ratio and the 16-byte aligned sizes.
      
      Change-Id: Id4a3f6aea6b9b9ae38bdfa1b87b7eb2cfcdd57b6
      9b7be888
    • John Koleszar's avatar
      Add VP9_GET_REFERENCE control · b3c350a1
      John Koleszar authored
      This is like VP8_COPY_REFERENCE, but returns a pointer to the reference
      frame rather than a copy of it. This is useful when the application
      doesn't know what the size of the reference is, as is the case when
      scaling is in effect.
      
      Change-Id: I63667109f65510364d0e397ebe56217140772085
      b3c350a1
  8. 13 Mar, 2013 2 commits
    • Jingning Han's avatar
      Support +/-2048 motion vector coding · 76c12ab9
      Jingning Han authored
      Enable entropy coding of motion vectors up to +/-2048. Also
      extend the motion search range accordingly.
      
      Change-Id: Iac2bb015e8934521cef83a19edbe967d9f097436
      76c12ab9
    • Yaowu Xu's avatar
      removed reference to "LLM" and "x8" · 00555263
      Yaowu Xu authored
      The commit changed the name of files and function to remove obselete
      reference to LLM and x8.
      
      Change-Id: I973b20fc1a55149ed68b5408b3874768e6f88516
      00555263
  9. 12 Mar, 2013 3 commits
    • John Koleszar's avatar
      disambiguate superframe index in vp9_stop_encode() · 872fc3de
      John Koleszar authored
      If the bool-coded partition naturally ends in a byte that matches the
      superframe index marker, it could lead to a parse error. This commit
      ensures that if such a marker is seen, it is padded out with an
      additional zero byte to disambiguate it.
      
      Change-Id: Id977de05745b6fa9ef08afb71e210a2a3ecca02e
      872fc3de
    • Paul Wilkins's avatar
      Change buffer update rules on ARF overlay. · 8be3056c
      Paul Wilkins authored
      When coding the frame that corresponds to the midpoint frame
      defining an ARF, do not update the last reference frame buffer.
      Previously this buffer was updated meaning that when coding the next
      ARF all the reference buffers were the same (or nearly so).
      Turning the update off means that the frame before is still available
      as an alternative predictor and for use in compound prediction.
      
      Also fixed inconsistency in test for mismatch (patch from JK).
      
      Net average gains (derf 0.049, yt 0.163, yt-hd 0.207, std-hd 0.286)
      
      Change-Id: Ifee21da21ccbb1648ac2eafe890d3ce60562c7bc
      8be3056c
    • Dmitry Kovalev's avatar
      Code cleanup. · 2891d70b
      Dmitry Kovalev authored
      Removing redundant code, introducing new functions for better
      decomposition, adding 'clamp' function to vp9_common.h.
      
      Change-Id: Ic3b8ca13bbc38f60f0c9c43910b5802005e31aaf
      2891d70b
  10. 11 Mar, 2013 2 commits
    • Paul Wilkins's avatar
      Changes to maximum gf/arf interval. · 08d2c382
      Paul Wilkins authored
      This patch puts in an adjustment to the maximum gf/arf
      interval based on the active q range.  It sets a fixed
      baseline maximum of 16 but can drop this down to 12 at
      lower q. This required some re-ordering in the first pass
      code to insure we have a Q range estimate before defining
      the first gf sequence.
      
      The main gains seed are int he STD hd set on 50fps clips
      where previously the interval could rise as high as 25.
      On the std hd clip the gains are around 2.8% with limit set
      to 300 frames.
      
      When combined with the one shot rate control flags we get
      combined of:
      
      derf 1.55% (limit300), yt 7.25%, hd 5.17% std-hd 5.84% (limit300)
      
      Change-Id: Ib380d51354511f2ff0f171a8df4e74291c0421f9
      08d2c382
    • John Koleszar's avatar
      Fix vp9_tree_probs_from_distribution with CONFIG_CODE_NONZEROCOUNT · 9b4095c5
      John Koleszar authored
      The automatic merge result was incomplete.
      
      Change-Id: I8976318bfc346d867660a013a302c80edb25fc29
      9b4095c5
  11. 10 Mar, 2013 1 commit
    • John Koleszar's avatar
      Optimize vp9_tree_probs_from_distribution · bd84685f
      John Koleszar authored
      The previous implementation visited each node in the tree multiple times
      because it used each symbol's encoding to revisit the branches taken and
      increment its count. Instead, we can traverse the tree depth first and
      calculate the probabilities and branch counts as we walk back up. The
      complexity goes from somewhere between O(nlogn) and O(n^2) (depending on
      how balanced the tree is) to O(n).
      
      Only tested one clip (256kbps, CIF), saw 13% decoding perf improvement.
      
      Note that this optimization should port trivially to VP8 as well. In VP8,
      the decoder doesn't use this function, but it does routinely show up
      on the profile for realtime encoding.
      
      Change-Id: I4f2848e4f41dc9a7694f73f3e75034bce08d1b12
      bd84685f
  12. 09 Mar, 2013 1 commit
    • Deb Mukherjee's avatar
      Continued experiment with nonzero count · a28139c8
      Deb Mukherjee authored
      Adds probability updates for extra bits for the nzcs, code for
      getting nzc stats, plus some minor cleanups and fixes.
      
      Change-Id: If2814e7f04fb52f5025ad9f400f3e6c50a00b543
      a28139c8
  13. 08 Mar, 2013 2 commits
    • Jingning Han's avatar
      Extend diff MV limit from +/-256 to +/-1024 · 2a5278bd
      Jingning Han authored
      Increase the motion search range by 4x. Change MV_CLASS tree of the
      entropy coding to allow two additional mv classes to cover the
      extended motion vector limit. The codec determines the effective
      motion search range conditioned on the actual frame dimension.
      
      It provides coding gains:
      
      stdhd 0.39%
      yt    0.56%
      hd    0.47%
      
      Major coding performance gains are packed in several sequences with
      intense motion activities, e.g., ped_1080p gains 7% at high bit-rates,
      and on average 3%.
      
      TODO: Need to further tune the rate control and motion search units.
      
      Change-Id: Ib842540a6796fbee5a797809433ef6a477c6d78d
      2a5278bd
    • Ronald S. Bultje's avatar
      Add support for tx_select in i8x8 encoding in keyframes. · b41dee84
      Ronald S. Bultje authored
      Also enable tx_select for keyframes.
      
      Change-Id: Iadb1231d9fa7af0c8dce3d9b41830b93a302479e
      b41dee84
  14. 07 Mar, 2013 3 commits
    • Dmitry Kovalev's avatar
      Consistent usage of ROUND_POWER_OF_TWO macro. · 3603dfb6
      Dmitry Kovalev authored
      Change-Id: I44660975e9985310d8c654c158ee7a61291b5a08
      3603dfb6
    • Ronald S. Bultje's avatar
      Re-add support for ADST in superblocks. · d3724abe
      Ronald S. Bultje authored
      This also changes the RD search to take account of the correct block
      index when searching (this is required for ADST positioning to work
      correctly in combination with tx_select).
      
      Change-Id: Ie50d05b3a024a64ecd0b376887aa38ac5f7b6af6
      d3724abe
    • Deb Mukherjee's avatar
      Coding con-zero count rather than EOB for coeffs · eb6ef241
      Deb Mukherjee authored
      This patch revamps the entropy coding of coefficients to code first
      a non-zero count per coded block and correspondingly remove the EOB
      token from the token set.
      
      STATUS:
      Main encode/decode code achieving encode/decode sync - done.
      Forward and backward probability updates to the nzcs - done.
      Rd costing updates for nzcs - done.
      Note: The dynamic progrmaming apporach used in trellis quantization
      is not exactly compatible with nzcs. A suboptimal approach has been
      used instead where branch costs are updated to account for changes
      in the nzcs.
      
      TODO:
      Training the default probs/counts for nzcs
      
      Change-Id: I951bc1e22f47885077a7453a09b0493daa77883d
      eb6ef241
  15. 06 Mar, 2013 1 commit
    • Paul Wilkins's avatar
      Added stricter Q control flag. · db6ad013
      Paul Wilkins authored
      Added a variant of the one shot maxQ flag
      for two pass that forces a fixed Q for the
      normal inter frames. Disabled by default.
      Also small adjustment to the Bits per MB
      estimation.
      Change-Id: I87efdfb2d094fe1340ca9ddae37470d7b278c8b8
      db6ad013
  16. 05 Mar, 2013 4 commits
    • Dmitry Kovalev's avatar
      Code cleanup. · 7f99c3c5
      Dmitry Kovalev authored
      Removing redundant 'extern' keywords, fixing formatting and #include order,
      code simplification.
      
      Change-Id: I0e5fdc8009010f3f885f13b5d76859b9da511758
      7f99c3c5
    • John Koleszar's avatar
      Add 'superframe' index · 522d4bf8
      John Koleszar authored
      A 'superframe' is a group of frames that share the same PTS, but have a
      defined decoding order. This commit adds the ability to append an index
      to such a group of frames, allowing for random access to the constituent
      frames. This could be useful for frame-level parallelism or partial
      decoding in a multilayer scenario.
      
      Decoding the stream serially without such an index should work as a
      fallback, and VP9/TestSuperframeIndexIsOptional verifies that.
      
      Change-Id: Idff83b7560e1a7077d8fb067bfbc45b567e78b1c
      522d4bf8
    • Ronald S. Bultje's avatar
      Make superblocks independent of macroblock code and data. · 111ca421
      Ronald S. Bultje authored
      Split macroblock and superblock tokenization and detokenization
      functions and coefficient-related data structs so that the bitstream
      layout and related code of superblock coefficients looks less like it's
      a hack to fit macroblocks in superblocks.
      
      In addition, unify chroma transform size selection from luma transform
      size (i.e. always use the same size, as long as it fits the predictor);
      in practice, this means 32x32 and 64x64 superblocks using the 16x16 luma
      transform will now use the 16x16 (instead of the 8x8) chroma transform,
      and 64x64 superblocks using the 32x32 luma transform will now use the
      32x32 (instead of the 16x16) chroma transform.
      
      Lastly, add a trellis optimize function for 32x32 transform blocks.
      
      HD gains about 0.3%, STDHD about 0.15% and derf about 0.1%. There's
      a few negative points here and there that I might want to analyze
      a little closer.
      
      Change-Id: Ibad7c3ddfe1acfc52771dfc27c03e9783e054430
      111ca421
    • John Koleszar's avatar
      Reinitialize motion search tables on frame size change · daa9b29e
      John Koleszar authored
      Make sure the motion search is done with the offsets calculated from
      the correct stride.
      
      Change-Id: Ifbcc0f742eda3399c255bfcfa1cdee9a4bb4b4e7
      daa9b29e