hashirama/uvg266
1
0
Fork 0
mirror of https://github.com/ultravideo/uvg266.git synced 2024-11-24 18:34:06 +00:00
Code Issues Actions Projects Releases Wiki Activity
2848 commits 3 branches 6 tags 9.2 MiB
Commit graph

500 commits

Author SHA1 Message Date
Laurent Fasnacht 63adb54a3d Add --threads <int> command line parameter 2014-05-14 13:27:09 +02:00
Laurent Fasnacht e772799d5e encoder_state_encode uses now the threadqueue 2014-05-14 13:27:08 +02:00
Laurent Fasnacht baede7f6c4 threadqueue 2014-05-14 13:27:08 +02:00
Laurent Fasnacht bc3ca90bdf Fix tiles when SAO or deblock is enabled. 
Was broken by previous commit.
 2014-05-14 13:27:07 +02:00
Laurent Fasnacht 4815a0604b Entropy coding sync works without parallelism, without SAO and without deblocking 2014-05-14 13:27:06 +02:00
Laurent Fasnacht 2c2a2528f3 Remove openmp stuff 2014-05-14 13:27:06 +02:00
Ari Koivula fb763f7940 Move coefficient generation functions from encoder.c to transform.c. 
- These functions probably should have been there to begin with.
 2014-05-12 11:37:39 +03:00
Ari Koivula a3478ecd20 Move transform skip decision to it's own function. 2014-05-12 11:18:27 +03:00
Ari Koivula d9b890de6e Remove redundant variables. 
- Redefine LCU_WIDTH to be 64. Stuff will break horribly if it's
  anything else anyway.

- Add LCU_WIDTH_C for chroma LCU width. It should be more readable than the
  constant (LCU_WIDTH >> 1).
 2014-05-12 10:58:07 +03:00
Ari Koivula 59e0e98523 Separate luma and chroma coefficient generation variables. 2014-05-12 10:38:24 +03:00
Ari Koivula 0ca65e7606 Move chroma coefficient generation to it's own function. 
- It's time to chop up this monster that is encode_transform_tree.
 2014-05-12 10:24:06 +03:00
Ari Koivula 3c3c9a26c6 Move scan order selection to a function. 2014-05-12 08:47:16 +03:00
Ari Koivula 623d9001a8 Reorder chroma coefficient generation. 2014-05-12 08:47:16 +03:00
Ari Koivula 93141c7d2e Avoid unnecessary copying of predicted pixels when there are no coeffs. 
- These are probably from a time when reconstruction happened in this
  function.
 2014-05-09 16:39:58 +03:00
Ari Koivula 27ab882c25 Clean up coefficient generation. 2014-05-09 16:33:10 +03:00
Ari Koivula ce945ab4ef Handle coefficient initialization better. 
- Coefficients are no longer required to be pre-zeroed. The resulting zeroes
  are copied in even in the case where we already know they are all zeroes.

- Move cbf clearing code to only happen at the leaves of the recursion.
 2014-05-09 16:30:28 +03:00
Laurent Fasnacht b274558139 Refactor and fix entry_points functions. 
Seems to be OK with HM now
 2014-05-09 12:42:37 +02:00
Ari Koivula cb5d7e6541 Fix compilation for VS2010. 2014-05-08 17:28:12 +03:00
Laurent Fasnacht 0452806ec4 Entry points 2014-05-08 15:04:56 +02:00
Laurent Fasnacht da588af2ba Partial support for wavefront 2014-05-08 15:04:55 +02:00
Laurent Fasnacht 4de5660254 Fix missing offset in LCU range computation for wavefronts 2014-05-08 15:04:55 +02:00
Laurent Fasnacht dc34a5eac6 LCU borders 2014-05-08 15:04:54 +02:00
Laurent Fasnacht 24f4a8cad1 Wavefront also needs entrypoints 2014-05-08 15:04:53 +02:00
Laurent Fasnacht d05f8b52aa Rewrite of encoder_state_write_bitstream_leaf: handle slice + tiles + wavefronts correctly 2014-05-08 15:04:53 +02:00
Laurent Fasnacht 27f694e3e8 Some initial code to support wpp and slices 2014-05-08 15:04:52 +02:00
Laurent Fasnacht 163189c3c7 Bitstream for leaves can be computed in parallel 2014-05-08 15:04:51 +02:00
Laurent Fasnacht be9882f5b2 Leaf bitstream write 2014-05-08 15:04:50 +02:00
Laurent Fasnacht ae6a7a9c4b Leaf encoder uses encoder_state->lcu_order 2014-05-08 15:04:49 +02:00
Laurent Fasnacht b740142325 Add is_leaf to encoder_state 2014-05-08 15:04:48 +02:00
Laurent Fasnacht 8451d5b100 Move some init code to encoder_state_new_frame 2014-05-08 15:04:48 +02:00
Laurent Fasnacht 1cb3f14dfe lcu_order_count in (leaves) encoder 2014-05-08 15:04:47 +02:00
Laurent Fasnacht ef6ae3e723 Remove dead code 2014-05-08 15:04:46 +02:00
Ari Koivula 535b42bc9b Fix compilation for VS2010. 2014-05-07 15:26:44 +03:00
Laurent Fasnacht 05eef82896 Remove extra [ from graphviz dump 2014-05-07 13:40:29 +02:00
Laurent Fasnacht 84e5dbee39 Remove quote from graphviz dump 2014-05-07 13:33:02 +02:00
Laurent Fasnacht b48a687d3c Restored parallelism, but it will be done in another way... OpenMP is not very efficient in these kind of dynamic situation 2014-05-07 11:55:56 +02:00
Laurent Fasnacht 0e6f1c99fc Refactor picture to remove hidden dependency between slice and tiles 
picture.type -> encoder_state->global->pictype
picture.slicetype -> encoder_state->global->slicetype
picture.slice_sao_luma_flag -> 1 (was constant)
picture.slice_sao_chroma_flag -> 1 (was constant)

This may be changed later. For now it's better to avoid having slice related stuff in picture.
 2014-05-07 11:55:48 +02:00
Laurent Fasnacht 39d96e0546 Fix bug with cabac stream pointing to bad data 2014-05-07 11:55:41 +02:00
Laurent Fasnacht e144f817ef Works when not using tiles 2014-05-07 11:55:16 +02:00
Laurent Fasnacht 24c2bd70ca Fix small bugs with compilation 2014-05-07 11:54:35 +02:00
Laurent Fasnacht a03f0cba19 encoder_control_input_init near the other encoder_control_* functions 2014-05-07 11:53:21 +02:00
Laurent Fasnacht 1e2671ac30 Renamed encoder_clear_refs to encoder_state_clear_refs 2014-05-07 11:53:12 +02:00
Laurent Fasnacht 831b221cf8 Parsing seems to work now 2014-05-07 11:53:01 +02:00
Laurent Fasnacht 8b5cb62237 Debug code to generate a graph 2014-05-07 11:52:04 +02:00
Laurent Fasnacht cee6bb0e71 Fix iteration on children 2014-05-07 11:49:14 +02:00
Laurent Fasnacht 699669ee35 fixed typo 2014-05-07 11:48:16 +02:00
Laurent Fasnacht 6c6adf18c7 Refactor encoder_state 2014-05-07 11:47:31 +02:00
Laurent Fasnacht a23edd0339 added parent to encoder_state 2014-05-07 11:42:54 +02:00
Laurent Fasnacht 5ce518a47a lcu_at_tile_start and lcu_at_tile_end helper functions 2014-05-07 11:42:30 +02:00
Laurent Fasnacht c2872bd6b0 Slices and WPP in command line and encoder 2014-05-07 11:42:04 +02:00
Laurent Fasnacht 2d6f199246 reorganized encoder_state structure 2014-05-07 11:41:27 +02:00
Laurent Fasnacht f0b076876f Moved all the stream related stuff into substream_write_bitstream 2014-05-07 11:40:20 +02:00
Ari Koivula eaf8835bda Add some comments and const qualifiers. 2014-05-06 19:20:38 +03:00
Ari Koivula 3910b7989a Clear old cbf data before recursion in encode_transform_tree. 
- Because encode_transform_tree also maintains the CBF data and assumes that
  the CBFs are initially zeroed, calling the function more than once would
  result in incorrect CBF data.
 2014-05-06 19:03:29 +03:00
Ari Koivula bdc16d2612 Improve cu_info coded block flag data structure a bit. 
- It works just like the old structure except that the flags are checked with
  bitmasks instead of having the flag value be propagated upwards. There isn't
  really any benefit to this because the flags still have to be propagated to
  parent CUs.

- Wrapped them inside a struct to make copying them easier. (Just need to copy
  the struct instead of making individual copies)
 2014-05-06 18:28:04 +03:00
Ari Koivula d123b98aea Remove unnecessary tertiary expressions from usages of CABAC_BIN. 2014-05-06 17:39:25 +03:00
Marko Viitanen bf2c2a1330 Small changes to fix compiling on VS 
- Added threads.h to VS project
- Included Windows.h in threads.h
 2014-05-05 11:18:43 +03:00
Laurent Fasnacht f3d4e6eb09 Move bitstream write to a separate function, and add assertions about the part which should not write to bitstream. 2014-05-05 09:24:57 +02:00
Laurent Fasnacht 7f6f4fe9c1 Reference count for picture 2014-05-05 08:03:24 +02:00
Laurent Fasnacht 323054d5e2 naming: alloc_yuv_t -> yuv_t_alloc dealloc_yuv_t -> yuv_t_free 2014-05-02 11:45:27 +02:00
Laurent Fasnacht 7d6d1d5536 Remove pic->pred_* 2014-05-02 11:38:07 +02:00
Laurent Fasnacht 92e14cc80d rename picture_init to picture alloc and picture_destroy to picture_free 2014-05-02 10:58:28 +02:00
Laurent Fasnacht b76f7377b6 Always initialize tiles data structures (even with only one tile) 2014-05-02 10:00:22 +02:00
Laurent Fasnacht 161fe38f5e Remove USE_TILES define 2014-05-01 13:58:13 +02:00
Laurent Fasnacht a84fd6486d Add function subencoder_blit_pixels 2014-05-01 11:16:11 +02:00
Laurent Fasnacht b8b28635ff Iterable structure for sub-encoders (more flexibility) 2014-05-01 11:16:10 +02:00
Laurent Fasnacht 212d390003 Cleanup of encoder_state_init and encoder_state_finalize 2014-05-01 11:16:10 +02:00
Laurent Fasnacht 161053f86b Do not allow more tiles than dimension in LCU 2014-05-01 07:11:31 +02:00
Ari Koivula d1fbc6dc80 Fix a small memory leak. 
- Malloced pointer returned by alloc_yuv_t was not being freed in
  substream_encode.

- Remove use of yuv_t from encode_one_frame, as it's not used there anymore.
 2014-04-30 11:15:34 +03:00
Ari Koivula 27b94d4b45 Address gcc -Wtype-limits errors. 
- Fixes warnings in #19 and #16.
 2014-04-29 09:15:52 +03:00
Ari Koivula 4490e8afd6 Remove depth dimension from picture->cu_array. 
- It isn't used for anything anymore.

- It was used in the past to hold information during search, but now that
  information is held in lcu_t structs.
 2014-04-28 10:18:22 +03:00
Laurent Fasnacht 76ec605b72 SAO works with tiles now 2014-04-28 06:29:21 +02:00
Laurent Fasnacht 7719837f17 Simple OpenMP parallelization 2014-04-25 09:11:10 +02:00
Laurent Fasnacht 4e34859e66 Fix compilation error with USE_TILES=1 and -Werror=maybe-uninitialized 2014-04-24 08:41:05 +02:00
Laurent Fasnacht 59392c4a62 Fix compilation issue with USE_TILES=0 2014-04-24 08:38:24 +02:00
Laurent Fasnacht af9a1c0fbb Use same reference images for all subencoders 2014-04-24 08:38:23 +02:00
Laurent Fasnacht 73c574fb45 P-frame: first try... 2014-04-24 08:38:22 +02:00
Laurent Fasnacht 03361dcf2c sao try... still not working 2014-04-24 08:38:22 +02:00
Laurent Fasnacht 3db4c59478 Recontruct full frame from tiles 2014-04-24 08:38:21 +02:00
Laurent Fasnacht 35d5d22ccc Fix tile size not to go outside of the original picture 2014-04-24 08:38:20 +02:00
Laurent Fasnacht b36e154c38 Some cleanup 2014-04-24 08:38:19 +02:00
Laurent Fasnacht 01580a93c3 Encoding with tiles now more or less works with -p 1 --no-sao --no-deblock 2014-04-24 08:38:19 +02:00
Laurent Fasnacht 5fb4d9c36e substream_encode function 2014-04-24 08:38:17 +02:00
Laurent Fasnacht e292b2c274 allocate subencoders 2014-04-24 08:38:17 +02:00
Laurent Fasnacht fba4f5432a Fix debug code 2014-04-24 08:38:16 +02:00
Laurent Fasnacht b255133460 Debug for tiles 2014-04-24 08:38:15 +02:00
Laurent Fasnacht 11629ce811 Use tile scan order in encode_one_frame() 2014-04-24 08:38:14 +02:00
Laurent Fasnacht 0036afa056 Write tiles related information picture parameter set and slice header 2014-04-24 08:38:14 +02:00
Laurent Fasnacht 1e9c894eba Coding tree block raster and tile scanning conversion process, according to ITU-T Rec. H.265 (04/2013) 6.5.1 2014-04-24 08:38:13 +02:00
Laurent Fasnacht 7bd6aa2e9c encoder_control_input_init call moved to encoder_control_init 2014-04-24 08:38:13 +02:00
Laurent Fasnacht ff318ae0e9 Tiles in encoder_control 2014-04-24 08:38:12 +02:00
Laurent Fasnacht 19b1642aa2 Removed all cabac parameters (cabac is part of encoder_state) 2014-04-22 11:46:53 +02:00
Laurent Fasnacht 5fea5875a5 Huge refactoring 
Split some parts of encoder_control into encoder_state
(idea: encoder_control is immutable)

Goal is to allow multiple substreams in the future.
 2014-04-22 10:39:12 +02:00
Ari Koivula 88a67a4e49 Fix faulty assert that stops the program from working with inter frames. 
- The assert would be true after the next if block, but in it's current place
  it's false.
 2014-04-22 10:57:38 +03:00
Ari Koivula 1b437a5989 Address clang-analyzer warnings about garbage values. 
- False alarm, but surprisingly difficult to convince clang of that. It
  doesn't seem to understand bit shifts very well.

- Only assert and changing LCU_WIDTH>>depth to width was necessary to satisfy
  clang.

- Closes #35.
 2014-04-17 18:43:09 +03:00
Laurent Fasnacht 3396264f3c Moved g_cur_lambda_cost into encoder_control.cur_lambda_cost 2014-04-17 12:00:21 +02:00
Laurent Fasnacht 534013be77 Remove g_lambda_cost 2014-04-17 11:49:27 +02:00
Laurent Fasnacht 83360918ba Removed table generation from main code, moved it to tools. 2014-04-17 11:13:15 +02:00
Laurent Fasnacht 4a9c239027 Remove g_bitdepth 2014-04-17 11:13:13 +02:00
Laurent Fasnacht 7a2b883059 Remove encoder_input width, height, height_in_lcu, and width_in_lcu 2014-04-17 11:13:12 +02:00
Laurent Fasnacht d01e3ae67f bitstream is a union, and is statically in encoder_control structure 2014-04-17 11:13:12 +02:00
Laurent Fasnacht 9ac3b7bf2b encoder->in.cur_pic --> cur_pic 2014-04-17 11:13:10 +02:00
Laurent Fasnacht 21d34613c2 Replace encoder->stream by stream 2014-04-17 11:13:09 +02:00
Laurent Fasnacht 2286175378 nal are now written to a bitstream, not a FILE* 2014-04-17 11:13:09 +02:00
Laurent Fasnacht ec9d70f70c Moved scalinglist_process into init_encoder_control 2014-04-16 11:45:51 +02:00
Laurent Fasnacht e06253d437 scalinglist changes missing in previous commit 2014-04-16 11:00:29 +02:00
Laurent Fasnacht 9901c38dd5 scalinglist in independent file 2014-04-16 10:25:16 +02:00
Laurent Fasnacht 9112cbb58c Generate and use static tables 2014-04-16 09:49:09 +02:00
Laurent Fasnacht 960f2cb4b0 g_sig_last_scan -> const uint32_t* 2014-04-15 16:09:52 +02:00
Laurent Fasnacht 763b775d3e encoder_control->cfg is const 2014-04-15 16:09:52 +02:00
Laurent Fasnacht ae2d79c954 Remove encoder_control.cqmfile 2014-04-15 16:09:51 +02:00
Laurent Fasnacht 486768fc79 scalinglist privatization 2014-04-14 13:39:28 +02:00
Laurent Fasnacht 78c579053a encoder_control should be const in nearly all the code 2014-04-14 10:56:06 +02:00
Laurent Fasnacht 418e6eae51 Changed bitstream handling to allow in-memory bitstream. 2014-04-14 08:13:00 +02:00
Marko Viitanen c38ec1aa10 Added commandline option for RDO (--rd) 2014-04-09 12:29:15 +03:00
Ari Koivula 92ac5025f9 Take intra mode based coeff scan mode into account for coeff bit cost. 
- Previously only diagonal scan mode, the most common one, would be used.

- This improved bdrate by 0.1-0.5 % for p0 and 0-0.2 % for p60.
 2014-04-09 10:44:44 +03:00
Ari Koivula c5dfcdf3aa Simplify scan mode selection. 
- The scan mode selection for chroma was a bit complicated so I checked it
  and it was all unnecessary. The mode selection is the same as for luma.
 2014-04-09 10:36:39 +03:00
Ari Koivula 3764688f84 Fix lambda initialization. 
- Lambda was initialized before slice type was set in encoder_control.
 2014-04-08 16:58:36 +03:00
Ari Koivula 6e0bc655e2 Resolve unused variable warning. 
- This unexpectedly changes bitstream, but as that makes no sense, it must be
  because some part of the program uses uninitialized memory.
 2014-04-04 17:28:50 +03:00
Marko Viitanen e15a86268d Clean up tabs and whitespaces 2014-04-04 16:04:44 +03:00
Laurent Fasnacht 816ae13b1d Moved context information inside cabac_data. 
This is required in order to be able to work on parallelism.
 2014-04-04 14:28:50 +02:00
Laurent Fasnacht 8a14bd3b7b Remove cabac global variable 2014-04-04 14:26:40 +02:00
Laurent Fasnacht 946c815932 init_context directly has a QP parameter, instead of passing an encoder_control* 
This makes context less tightly coupled with encoder.
 2014-04-04 14:26:39 +02:00
Marko Viitanen b83559d3f3 Use RDO to check for transform skip mode 2014-04-04 13:09:42 +03:00
Ari Koivula b19e4f3f2d Resolve possible uninitialized variable warnings. 
- Working towards issue #11.

- Neither variable was actually used as uninitialized.
 2014-04-04 13:02:06 +03:00
Ari Koivula d44d1837bb Remove unreferenced parameters. 
- Working towards issue #11.
 2014-04-04 12:56:24 +03:00
Ari Koivula 46d33d3945 Resolve unsigned/signed mismatch warnings. 
- Working towards issue #11.
 2014-04-04 12:56:23 +03:00
Ari Koivula 27a3329dfb Remove unreferenced_parameter macro. 
- It was a silly hack to selectively silence compiler warnings from VS, but
  there is no point as it causes compiler warnings in GCC.
 2014-04-03 15:38:17 +03:00
Marko Viitanen 0da8071300 Changed final cost (and transform skip) error function from SAD to SSD 2014-04-02 14:51:39 +03:00
Marko Viitanen a14fb14e33 Added new commandline parameter --no-transform-skip 2014-04-02 14:49:48 +03:00
Marko Viitanen cfb21c0e4c Implemented transform skipping (for 4x4 blocks) 
transform skip vs. normal transform selection criteria might need more work, currently both are calculated for each 4x4 block and SAD+coeff_SSE is compared.
 2014-04-02 10:54:03 +03:00
Panu Sjövall c8f629495d Remove unnecessary buffer from bitstream. 
- Writing encoded data to file is done in bitstream_put one byte at a time and nal_write only writes the packet headers
 2014-03-25 11:46:56 +02:00
Ari Koivula 953aef0379 Move rest of LCU encoding inside the LCU loop. 
- Move SAO search inside the LCU loop.

- Move CU coding inside the LCU loop.

- Move SAO frame reconstruction loop to sao module.
 2014-03-21 12:41:44 +02:00
Ari Koivula 746eaa3671 Move deblocking code to filter module. 2014-03-21 11:57:12 +02:00
Ari Koivula c42b25054a Modify deblocking to be done per-LCU in the encoding loop. 
- Intra works. There is still something wrong in inter.

- Avoid horizontal deblocking of the rightmost 4 pixels in the LCU.
  This is because vertical deblocking must be done for all pixels
  before horizontal, but vertical deblocking can't be done for those
  pixels before the next LCU is finished.

- Add separate deblocking of the rightmost pixels of the last LCU
  after the LCU edge has been deblocked.

- This is a pretty ugly hack but will have to do for now.
 2014-03-20 18:14:43 +02:00
Ari Koivula 7328fc2897 Add special handling for bottom right LCU pixel. 
I didn't take into account that the reference pixel on the top-left of the
LCU gets over written if we just replace the top reference pixels for
current LCU with the bottom reference pixels after doing the search.
To handle this I copy the pixel that gets overwritten to the vertical
reference pixels.
 2014-03-19 16:01:31 +02:00
Ari Koivula 57ce7e990b Add new reference pixel buffer management to encoding loop. 
- This is necessary because after we add in-loop filters to be done per LCU,
  the reconstruction buffer will have the deblocked pixels. We only need the
  edge-pixels for intra prediction though so we just save those.

- Right now it only copies the pixels and passes them on to search, where
  the copied pixels are asserted to be the same ones we copy from
  reconstruction buffer.

- New yuv_t struct added for arrays of dynamic length. We might want to change
  other buffers to use it or something like it in the future.
 2014-03-19 16:01:30 +02:00
Ari Koivula 34e453c880 Clean up encode_one_frame. 
There is a lot of duplicated code due to handling random access and trailing
pictures separately. I merged the code for these two branches so it would be
easier to modify.
 2014-03-17 18:04:41 +02:00
Ari Koivula b1596eb76c Refactor coding of end_of_slice_segment_flag. 
According to spec the end_of_slice_segment_flag is always coded, but in the
code it looked like it was not coded for the last LCU in picture. This was
due to the end_of_slice_segment_flag being coded inside cabac_flush, like it
is in HM. This is a bit silly so I moved it out of cabac_flush.
 2014-03-17 18:04:41 +02:00
Ari Koivula 6a384112cc Remove old unnecessary functions from search module. 
- As we move to per-LCU encoding only search_lcu will be necessary.

- Function search_best_mode is no longer used.
 2014-03-17 18:04:41 +02:00
Ari Koivula 9563b50a34 Refactor coefficient group scan mapping lists. 
The relation between coefficients positions and coefficient group positions
was a big confusing due to the use of 16x16 diagonal coefficient mappings
also as coefficient group mappings.

- Moved all coefficient group mappings to their own const arrays and added
  a new array the select the correct coefficient group mapping. This removes
  special cases for 8x8 and 32x32 transform sizes.

- Removed all coefficient group mapping initialization from init_sig_last_scan.

- Removed 128x128 and 64x64 from regular coefficient group array as those
  transform sizes don't exist anymore in HEVC.
 2014-03-14 16:56:49 +02:00
Ari Koivula e93ae54e70 Fix bug with selecting wrong coeff scan mode for 4x4 blocks. 2014-03-12 14:56:05 +02:00
Ari Koivula 409b094acf Try and get NxN to work again. 
- Moved NxN search to be done on the same level as other searches, as it's
  really not any different from 2Nx2N.
- Produces working bitstream but reconstruction is different.
 2014-03-11 19:19:20 +02:00
Ari Koivula 470f36ce34 Remove left-over commented out code. 2014-03-07 12:44:45 +02:00
Ari Koivula 5d9aa3409b Merge branch 'new-search-wip' 
Conflicts:
	src/encoder.c
	src/intra.c
	src/intra.h
	src/search.c
 2014-03-06 19:00:53 +02:00
Ari Koivula 47af5207c5 Remove dead code and fix white space. 2014-03-06 18:35:17 +02:00
Marko Viitanen c3e47c6a8d Removed redundant coeff_y/u/v variables from CU 2014-03-05 17:18:03 +02:00
Marko Viitanen b00f4c8dbf Fixed transform offset value and set tr_depth to correct value when depth 0 2014-03-05 17:10:17 +02:00
Ari Koivula 1cf03cfb37 Add --no-rdoq parameter. 2014-03-05 17:02:21 +02:00
Ari Koivula ca51f660d5 Fix fluctuation of coefficients on identical frames when RDOQ is on. 
Moves CABAC context initialization to take place before search. This fixes
an issue with RDOQ returning different coefficients for identical adjacent
frames.

- This actually probably worsens BD-rate a little for all frames except the
  first one because we were using last frames final CABAC context for every
  LCU and now we are using initialized CABAC contexts. The fix is to encode
  the LCU before we start compressing the next LCU so we can update CABAC
  contexts.
 2014-03-05 17:02:13 +02:00