Commit Graph

4371 Commits

Author SHA1 Message Date
Yaowu Xu
b6a41802c4 Simplify select_tx_mode()
The function is called only once, right after all stats counters are
reset to 0. Therefore all the computations have zero effect on return
values. This commmit to removed those effectless code.

Change-Id: I50d27c0802547921fa36c60aa4bd92d76247f595
2014-08-13 11:48:29 -07:00
Jim Bankoski
07c3abe21e Merge "intra blocks disallowed inadvertently" 2014-08-13 08:40:14 -07:00
Paul Wilkins
a3dbc8b304 Merge "Minor Two pass rate control adjustments." 2014-08-13 02:56:19 -07:00
James Zern
a6b7bd6a1c Merge "fixes several -Wunused-function warnings" 2014-08-12 20:15:14 -07:00
Jim Bankoski
f452961765 fixes several -Wunused-function warnings
Change-Id: I4dc2cb255f4fe30998b6ee61184895dee9f5da8e
2014-08-12 16:51:07 -07:00
Jim Bankoski
5c55202c6b intra blocks disallowed inadvertently
At speed 6 the smallest partitioning was 16x16 and biggest
intra block was 8x8, essentially disallowing all intra blocks
which produces ugly artifacts when revealing new video.

Change-Id: I364042d4c64e09be0666ade64aac94d0a1b586cf
2014-08-12 16:22:32 -07:00
Alex Converse
f55f68c13a Fix pickmode for intra DC blocks
Change-Id: I4906043ec1e3577ba37622a235c54adacb66125a
2014-08-12 16:04:14 -07:00
Paul Wilkins
ea55e4d26a Minor Two pass rate control adjustments.
Alter way in which we identify static scenes.
Remove some redundant code.

Change-Id: I883c2ea2e341770648a8663d8881f720ed099b61
2014-08-12 16:27:32 +01:00
James Zern
1d9e2b5003 Merge "remove unused fields from BOOL_CODER / vp9_writer" 2014-08-11 23:08:45 -07:00
James Zern
2ee221a2dc Merge "vp9_write_bit_buffer: fix function name typo" 2014-08-11 18:43:51 -07:00
Jingning Han
245e57c78e Merge "Enable fast forward txfm and quant for rate-distortion search" 2014-08-11 17:56:48 -07:00
Dmitry Kovalev
ccddd5d0f9 Merge "Removing unised OUTPUT_YUV_SRC define." 2014-08-11 11:03:20 -07:00
Jingning Han
5b63c2797a Merge "Integrate fast txfm and quant path into skip_recode system" 2014-08-11 08:53:34 -07:00
Jingning Han
9da4cd94f5 Merge "Extend skip_txfm flag into array to cover YUV planes" 2014-08-11 08:53:25 -07:00
Dmitry Kovalev
55f6ef9e53 Removing unised OUTPUT_YUV_SRC define.
Change-Id: Ia88eebf10c8620dadaa66722543db291e44fe104
2014-08-08 17:14:01 -07:00
Dmitry Kovalev
cd1fbc67f9 Simplifying vp9_set_speed_features() function.
Change-Id: I3e67230690b81ef54ef48ae26107fe7bc880ab8e
2014-08-08 16:29:24 -07:00
Jingning Han
b4b09c9796 Enable fast forward txfm and quant for rate-distortion search
This commit enables encoder to select fast forward transform and
quantization path according to the prediction residual sse/variance,
in the rate-distortion optimization scheme.

Change-Id: Ief9fc3844fd4107166d401970e800c6e5ce2b5fe
2014-08-08 16:16:51 -07:00
Dmitry Kovalev
6e5149b68d Merge "Moving pass from VP9_COMP to VP9EncoderConfig." 2014-08-08 16:07:47 -07:00
Adrian Grange
a334cb0b7d Merge "Common encode/decode function to get reference frame" 2014-08-08 15:56:18 -07:00
Dmitry Kovalev
5a78c40657 Moving framerate from VP9EncoderConfig to VP9_COMP.
Framerate changes over time, VP9EncoderConfig supposed to have static
read-only data.

Change-Id: I36580a7ee71243c1eabb57691089691731d03e67
2014-08-08 15:29:37 -07:00
Dmitry Kovalev
91c2f1e45a Moving pass from VP9_COMP to VP9EncoderConfig.
We had a very complicated way to initialize cpi->pass from
cfg->g_pass:
switch (cfg->g_pass) {
  case VPX_RC_ONE_PASS:
    oxcf->mode = ONE_PASS_GOOD;
    break;
  case VPX_RC_FIRST_PASS:
    oxcf->mode = TWO_PASS_FIRST;
    break;
  case VPX_RC_LAST_PASS:
    oxcf->mode = TWO_PASS_SECOND_BEST;
    break;
}

cpi->pass = get_pass(oxcf->mode).

Now pass is moved to VP9EncoderConfig and initialization is simple:
switch (cfg->g_pass) {
  case VPX_RC_ONE_PASS:
    oxcf->pass = 0;
    break;
  case VPX_RC_FIRST_PASS:
    oxcf->pass = 1;
    break;
  case VPX_RC_LAST_PASS:
    oxcf->pass = 2;
    break;
}

Change-Id: I8f582203a4575f5e39b071598484a8ad2b72e0d9
2014-08-08 14:27:54 -07:00
Scott LaVarnway
7035527c9a Improved vp9_quantize_fp_neon()
Eliminated instructions by using better neon instructions
and rearranging the loop.

On a Nexus 7, vpxenc (in realtime mode, speed -12)
reported a performance improvement of ~1.0%.

Change-Id: I6b1700e79318f647ea67ef25e954c308932950ec
2014-08-08 13:57:25 -07:00
Dmitry Kovalev
2fe6fa72fc Merge "Cleaning up vp9_encodeframe.c." 2014-08-08 13:55:34 -07:00
Adrian Grange
1ebf52df2c Common encode/decode function to get reference frame
Replaced encoder and decoder functions to get a pointer
to a reference frame with a common function, vp9_get_ref_frame,
and simplified it.

Change-Id: Icb206fcce8caace3bfd1db3dbfa318dde79043ee
2014-08-08 11:37:11 -07:00
Adrian Grange
75b42a4977 Remove coding_use_prev_mi member from VP9_COMMON
This was shadowing the use of error_resilient_mode, but with
the opposite sense.

Change-Id: Ie4d30263a304fe4b3e94f0c7741db6888cc6afd8
2014-08-08 09:40:38 -07:00
Alex Converse
bac9beff72 Merge "Fix active_map speed 6." 2014-08-08 09:30:47 -07:00
Alex Converse
6acc5c630e Merge "Cleanup SEG_LVL_SKIP handling in encode_superblock." 2014-08-08 09:30:44 -07:00
levytamar82
69a5f5ecf7 Fix bug 807
in the sub_pixel_*variance* function the dst is aligned to 16 bytes and not
to 32 bytes - now load unaligned data

Change-Id: I2e0b9745543697efc56fefa32857ea10117af135
2014-08-07 18:51:02 -07:00
Alex Converse
2a5c46d8f5 Fix active_map speed 6.
Fix the interaction between active map and reuse_inter_pred_sby. The
reuse_inter_pred_sby feature expects inter predictors to already be
built, but blocks with active map on skip this step.

Change-Id: Ibb2bf0d228f678935d82a0ede9cb0919ab7c8878
2014-08-07 15:57:58 -07:00
Alex Converse
e874aea74c Cleanup SEG_LVL_SKIP handling in encode_superblock.
Change-Id: Ib7497ba08696765cbc1b2cc4218d37f4298f278c
2014-08-07 15:57:58 -07:00
Dmitry Kovalev
b539705916 Cleaning up vp9_encodeframe.c.
Change-Id: Ia3001ae5c44faee3978fc3eb7a027cd9712a0373
2014-08-07 14:55:54 -07:00
levytamar82
af10457e02 Fix bug 806
in the function sad32x32x4d and sad64x64x4d the source is aligned to 16 bytes
and not to 32 bytes - the load is now unaligned.

Change-Id: I922fdba56d0936b5cf72e4503519f185645a168c
2014-08-07 14:13:30 -07:00
Dmitry Kovalev
65234504b9 Merge "Removing direct references to VP9_COMP." 2014-08-07 14:12:32 -07:00
James Zern
79716fb668 Merge "vp9_bitstream: make some function params const" 2014-08-07 12:34:11 -07:00
Deb Mukherjee
a468170804 Merge "Changes hdr for profiles > 1 for intraonly frames" 2014-08-07 11:15:38 -07:00
James Zern
cffe2df033 vp9_bitstream: make some function params const
Change-Id: Ic49e290c7c88c55e730a0b2b0eb16ee3d0f120e7
2014-08-07 10:56:06 -07:00
Deb Mukherjee
09bf1d61ca Changes hdr for profiles > 1 for intraonly frames
Specifies the bit-depth, color sampling and colorspace
for intra only frames for profiles > 0

Also adds checks to ensure that profile 1 and 3 are
exclusively used for non 420 streams.

Change-Id: Icfb15fa1acccbce8f757c78fa8a2f60591360745
2014-08-07 09:47:14 -07:00
Jim Bankoski
8e5fa0aa05 Merge "vp9_rd.c: cleanup ugly expression" 2014-08-07 06:24:17 -07:00
Jim Bankoski
4e7cd38f9e vp9_rd.c: cleanup ugly expression
Change-Id: I066763c2e3ae69ab772bb986da97af9b504f0329
2014-08-06 16:34:47 -07:00
Jingning Han
8684c23260 Integrate fast txfm and quant path into skip_recode system
This commit integrates the fast transform and quantization process
into skip_recode scheme in the rate-distortion optimization loop.
Previously the fast transform and quantization process was only
enabled for non-RD coding flow.

Change-Id: Ib7db4d39b7033f1495c75897271f769799198ba8
2014-08-06 16:11:22 -07:00
Minghai Shang
390ad737b8 [spatial svc]Add is_spatial_svc() helper function.
Change-Id: Ice5376100d8e27cbdaddfd3cd06898cedd2720fe
2014-08-06 15:20:51 -07:00
Adrian Grange
7ec06cddc6 Merge "Create function vp9_preserve_existing_gf" 2014-08-06 14:50:51 -07:00
Dmitry Kovalev
09b3d04aac Removing direct references to VP9_COMP.
Change-Id: Ic37624d807884e71f08b50fd04892f03f2708ba7
2014-08-06 12:59:02 -07:00
James Zern
f80245bdea Merge "vp9_subexp: remove vp9_compute_update_table()" 2014-08-06 12:15:36 -07:00
Alex Converse
a6aaab3a9f Merge "Use INTER_ALL for VAR based partitions for screencast material." 2014-08-06 10:31:03 -07:00
Alex Converse
b26d9eab21 Merge "Add a codec control option to tune for screen content." 2014-08-06 10:30:40 -07:00
Pengchong Jin
92cdb96e5f Merge "Directly split the block in partition search" 2014-08-06 08:38:56 -07:00
James Zern
8c3886df78 remove unused fields from BOOL_CODER / vp9_writer
measure_cost, bit_counter, value

Change-Id: I8f93444ddddbee4ece7db6076a922ad96786044c
2014-08-05 21:58:03 -07:00
James Zern
c5a7eefdca vp9_write_bit_buffer: fix function name typo
vp9_rb_bytes_written -> vp9_wb_bytes_written
+ move limits.h from the header to the source file where it's needed

Change-Id: Ifcdc856b4d4dcc2fff555ef11f86c86a0d83dab3
2014-08-05 21:57:54 -07:00
Pengchong Jin
74593c1eae Directly split the block in partition search
This patch allows the encoder to directly split the block
in partition search, therefore skip searching NONE. It
computes a score which measures whether 16x16 motion vectors
from the first pass in the current block are consistent with
each others. If they are inconsistent and we have enough Q
to encode, split the block directly, and skip searching NONE.

This feature is under flag CONFIG_FP_MB_STATS. In speed 2,
it further gives a speedup of 3-8% on sample yt clips as
compared to the previous version under the same flag. Overall,
the features under the flag will give 7-15% on typical yt
clips at up to 6000kbps data rate. The speedup at very high
data rate is not significant.

For hard stdhd clips:
park_joy_1080p @ 15000kbps:       504541ms -> 506293ms (-0.35%)
pedestrian_area_1080p @ 2000kbps: 326610ms -> 290090ms (+11.2%)

The compression performance using the features under the flag:
derf: -0.068%
yt:   -0.189%
hd:   -0.318%
stdhd:-0.183%

To use the feature, set CONFIG_FP_MB_STATS and turn on
cpi->use_fp_mb_stats.

Change-Id: Iad58a2966515c8861aa9eb211565b1864048d47f
2014-08-05 16:13:40 -07:00
Jingning Han
1a8d45f309 Extend skip_txfm flag into array to cover YUV planes
Change-Id: Ieae182d72d625d0d3fd4ed7c7d24cb521a0f21b0
2014-08-05 15:42:12 -07:00
Adrian Grange
0984121f49 Create function vp9_preserve_existing_gf
This code was being called from two places and
difficult to parse. I rationalized it in to a
function to improve readability.

Change-Id: I154b8fe0b84e6c01e69601e78e67bd47c954d8b6
2014-08-05 15:25:21 -07:00
Alex Converse
2be9ea610f Use INTER_ALL for VAR based partitions for screencast material.
This offers 25% more compression on my HD screencast testset.

Change-Id: I85eaef95fd8f2e03e326443e9514482b2ee35cef
2014-08-05 15:23:50 -07:00
Alex Converse
572de92991 Add a codec control option to tune for screen content.
Change-Id: I12b0b0a06750cb9a5f1bdc3368deb44137acc9a3
2014-08-05 15:23:45 -07:00
James Zern
82e5665f74 vp9_subexp: remove vp9_compute_update_table()
makes update_bits[] a constant table

Change-Id: If105bf0522487f54a31e12a7cc73aa24671b19d4
2014-08-05 14:34:15 -07:00
Minghai Shang
854acfbbea Merge "[spatial svc]Enabl golden frame for base layer and fix wrong ref_frame_flag for upper layers in first frame" 2014-08-05 10:40:04 -07:00
Pengchong Jin
5971e8985b Merge "Store first pass motion vector directions" 2014-08-04 17:35:42 -07:00
Pengchong Jin
233e0ccc73 Store first pass motion vector directions
Re-organize the one-byte structure for 16x16 first pass
block. Add bits to indicate motion vector directions.

Change-Id: Id10754ba343dfc712c7fed5bcc85c67fa0bbcb89
2014-08-04 16:17:47 -07:00
Jingning Han
ef1023b786 Merge "Cosmetic indent fix in vp9_rd_pick_inter_mode_sb" 2014-08-04 15:39:51 -07:00
Minghai Shang
5de4368b43 [spatial svc]Enabl golden frame for base layer and fix wrong ref_frame_flag for upper layers in first frame
Change-Id: I1fec9c084b499b2f445b79726066d01b136b59fa
2014-08-04 15:06:21 -07:00
Jingning Han
395c934720 Cosmetic indent fix in vp9_rd_pick_inter_mode_sb
Change-Id: I303213fbb0ad5fbf28f0ca17271624b75e9d6319
2014-08-04 15:00:47 -07:00
Jim Bankoski
63c1c3ee64 energy -> int to avoid unsigned / signed mismatch
Change-Id: Idd1327852f0df0eab0ea3b33959f2b8292b77301
2014-08-04 12:07:26 -07:00
Jim Bankoski
967c7c0828 vp9_firstpass.c: fix unused static function
Change-Id: Ia989a01d79f1f02defbeccb1b8eee23fdf5e764e
2014-08-04 12:07:20 -07:00
Jim Bankoski
64126c3aa4 set but unused var in vp9_rdopt
Change-Id: Ie813af24d23c7d72b38749b3382b315daf516be5
2014-08-04 08:56:05 -07:00
Jim Bankoski
7f63dabfe9 break at the end of clauses with assert(0) to avoid gcc warning
Change-Id: I1b3c5337f018dde27dc819ab18bd081d169a91e8
2014-08-04 08:52:53 -07:00
Jim Bankoski
3cf5908e24 uint8_t segment and skip to avoid signed / unsigned warnings
Change-Id: I2e2765b851fb0a1b15351c2aa0e079197cbee373
2014-08-04 08:52:40 -07:00
Jim Bankoski
c3591463f4 avoid signed mismatch warning in gcc
Change-Id: I9d995178fd764be252e571fa23901da0f991bbb8
2014-08-04 08:52:40 -07:00
Frank Galligan
5f8fa13258 Merge "Added vp9_sad8x8_neon()" 2014-08-01 14:11:38 -07:00
Scott LaVarnway
98165ec074 Neon version of vp9_sub_pixel_variance8x8(),
vp9_variance8x8(), and vp9_get8x8var().

On a Nexus 7, vpxenc (in realtime mode, speed -12)
reported a performance improvement of ~1.2%.

Change-Id: I8a66ac2a0f550b407caa27816833bdc563395102
2014-08-01 11:35:55 -07:00
Frank Galligan
5487b6067c Merge "Neon version of vp9_sub_pixel_variance32x32()," 2014-08-01 09:46:37 -07:00
Scott LaVarnway
545be78136 Added vp9_sad8x8_neon()
Change-Id: I3be8911121ef9a5f39f6c1a2e28f9e00972e0624
2014-08-01 06:36:18 -07:00
Jingning Han
f94227c81d Merge "Skip calling vp9_block_energy when aq-mode is off" 2014-07-31 14:38:20 -07:00
Yunqing Wang
d5b8a1b324 Merge "Code cleanup in rdopt.c" 2014-07-31 12:20:06 -07:00
Jingning Han
1c3a80b9a1 Skip calling vp9_block_energy when aq-mode is off
The mb_energy value is used by aq-mode. Turn off computing its
value when aq-mode is off.

Change-Id: I26c239f124eca45a5ee58b90d19eae00d9a7cda5
2014-07-31 11:51:59 -07:00
Frank Galligan
e589d8ea5c Merge "Neon version of vp9_subtract_block()" 2014-07-31 10:24:26 -07:00
Yunqing Wang
678d747259 Code cleanup in rdopt.c
Moved encode_breakout code out of handle_inter_mode().

Change-Id: Icd661136b05fdf163768c406f91e0c98a8df89eb
2014-07-31 09:55:40 -07:00
Scott LaVarnway
6f4b8dcdc2 Neon version of vp9_subtract_block()
On a Nexus 7, vpxenc (in realtime mode, speed -12)
reported a performance improvement of ~3.2%

Change-Id: I8862497264142171b7efc32df1a67714a23539f4
2014-07-31 09:28:06 -07:00
Jingning Han
a6a348b85e Merge "Refactor rd_pick_parition interface" 2014-07-31 09:25:34 -07:00
Scott LaVarnway
d39448e2d4 Neon version of vp9_sub_pixel_variance32x32(),
vp9_variance32x32(), and vp9_get32x32var().

Change-Id: I8137e2540e50984744da59ae3a41e94f8af4a548
2014-07-31 08:00:36 -07:00
Jingning Han
a3b062c56f Merge "Chessboard pattern partition search" 2014-07-30 14:34:42 -07:00
Pengchong Jin
7f29d22e51 Merge "Early termination after partition NONE is done in RD." 2014-07-30 13:35:02 -07:00
Pengchong Jin
49866baae6 Early termination after partition NONE is done in RD.
This patch allows the encoder to skip the search for partition
SPLIT, HORZ, VERT after the search for partition NONE is done
in RD optimization. It uses the first pass block-wise statistics
to make the decision. If all 16x16 blocks in the current partition
have zero motions and small residues from the frist pass statistics,
and it has small difference variance, further partition search is
skipped.

For speed 2 setting, experiments on general youtube clips show that
the speedup varies from 1% - 10%, 5% on average. On the performance
side in PSNR, derf 0.004%, yt -0.059%, hd -0.106%, stdhd 0.032%.

For hard stdhd clips:
park_joy_1080p, 502952 ms -> 503307 ms (-0.07%)
pedestrian_area_1080p, 227049 ms -> 220531 ms (+3%)

This feature is under the compilation flag CONFIG_FP_MB_STATS and
it is off in current setting.

Change-Id: I554537e9242178263b65ebe14a04f9c221b58bae
2014-07-30 11:54:49 -07:00
Frank Galligan
81c2db591f Merge "Neon version of vp9_quantize_fp()" 2014-07-30 11:12:20 -07:00
Jingning Han
d82ff94284 Refactor rd_pick_parition interface
Remove the variable that indicates the relative block index. This
is explicitly covered by the use of pc_tree.

Change-Id: Ib13142582fff926c85e375bde656aa050add8350
2014-07-30 10:53:57 -07:00
Jingning Han
ca2dcb7fed Chessboard pattern partition search
This commit enables a chessboard pattern constrained partition
search for 720p and above resolutions. The scheme applies stricter
partition search to alternative blocks based on its above/left
neighboring blocks' partition range, as well as that of the
collocated blocks in the previous frame. It is currently turned
on at 16x16 block size level. The chessboard pattern is flipped
per coding frame.

The speed 3 runtime is reduced:
park_joy_1080p, 652832 ms -> 607738 ms (7% speed-up)
pedestrian_area_1080p, 215998 ms -> 200589 ms (8% speed-up)

The compression performance is changed:
hd     -0.223%
stdhd  -0.295%

Change-Id: I2d4d123ae89f7171562f618febb4d81789575b19
2014-07-30 10:32:41 -07:00
Jingning Han
e9935a4ca0 Merge "Clean up max/min allowed block size in rd_pick_partition" 2014-07-30 10:05:26 -07:00
Jingning Han
22cf82a14c Merge "Use frame index directly in get_chessboard_index" 2014-07-30 10:05:03 -07:00
Jim Bankoski
e71adcd834 Merge "clear up cfg unused warning in vp9_pick_inter_mode" 2014-07-30 09:56:33 -07:00
Scott LaVarnway
d4a37db5b8 Neon version of vp9_quantize_fp()
On a Nexus 7, vpxenc (in realtime mode, speed -12)
reported a performance improvement of ~12.4%

Change-Id: Id29d215acf58bb108489e218a259adf74b4768d7
2014-07-30 09:33:46 -07:00
Jim Bankoski
6647ca795d clear up cfg unused warning in vp9_pick_inter_mode
Change-Id: Iefcf0a25aaf5e44e8e791839aa82d876555025e0
2014-07-30 08:55:22 -07:00
Scott LaVarnway
521cf7e879 Neon version of vp9_sub_pixel_variance16x16(),
vp9_variance16x16(), and vp9_get16x16var().

On a Nexus 7, vpxenc (in realtime mode, speed -12)
reported a performance improvement of ~16.7%.

Change-Id: Ib163aa99f56e680194aabe00dacdd7f0899a4ecb
2014-07-30 08:17:32 -07:00
Pengchong Jin
838b53b9fb Merge "Remove the redundant index computation in the first pass" 2014-07-29 14:27:24 -07:00
Jingning Han
6646ea73e2 Clean up max/min allowed block size in rd_pick_partition
This commit replace the repetitive retrieve of max and min allowed
partition from speed_feature with local variables max_size and
min_size.

Change-Id: Ib06f11f16615e4876e4dd5fb6a968c6bf5f7b216
2014-07-29 11:03:52 -07:00
Jingning Han
c36f78b054 Use frame index directly in get_chessboard_index
The get_chessboard_index() used to call the entire VP9_COMMON
struct pointer to retrieve the chessboard pattern index. This cl
makes it call the frame index directly.

Change-Id: I3cad9d209ea2e77a358085a04fe1ff0ddec5ba03
2014-07-29 10:55:56 -07:00
Scott LaVarnway
d19d222db6 Added vp9_fdct8x8_neon(), vp9_fdct8x8_1_neon()
On a Nexus 7, vpxenc (in realtime mode, speed -12)
reported a performance improvement of ~3.7%.

Change-Id: I428c72c40df82c6d537955e320a8debf99343004
2014-07-29 08:56:05 -07:00
Pengchong Jin
6491065775 Remove the redundant index computation in the first pass
Remove the redundant index computation when store the first
pass block-wise statistics. Currently, a single byte is
allocated for a 16x16 blocks, and all the frame statistics
saved during the first pass will be kept in memory for use
in the second pass. For a 1920x1080 300-frame clip, it will
take about 2.3 MB memory. This feature is off in current
setting.

Change-Id: I135a95b348ec093d54c6a07e1e8237626909e3bd
2014-07-28 18:31:36 -07:00
levytamar82
4ba92dc5ab Fix bug 805
Remove all the redundant dct functions (dct4x4, dct8x8)
in avx2 except dct32x32 those functions were copied originally from dct_sse2

Change-Id: I742576fbf5175f3ac09f2076976a9247b259323e
2014-07-28 15:46:01 -07:00
Pengchong Jin
c580428928 Merge "Store block-wise statistics obtained in the first pass" 2014-07-28 14:49:05 -07:00
Pengchong Jin
bae652245d Store block-wise statistics obtained in the first pass
Change-Id: I9956db2ba2f7d28f484daaf5022d8d1ef5db473c
2014-07-28 09:12:40 -07:00
Jingning Han
ac1f06188d Merge "Fix rd_pick_partition search loop for 4x4 blocks" 2014-07-25 15:57:35 -07:00
Jingning Han
3d5f17311c Merge "Remove unnecessary conditional assignment" 2014-07-25 15:56:31 -07:00
Minghai Shang
8433c8f92d Merge "[spatial svc]Fix reference issues" 2014-07-25 13:24:27 -07:00
Alex Converse
f5827aee38 Merge "Refactor inter/intra_suberblock_yrd." 2014-07-25 10:51:48 -07:00
Jingning Han
84af0486f9 Fix rd_pick_partition search loop for 4x4 blocks
The partition search for 4x4 blocks takes unnecessary steps to
reconstruct pixels and an extra partition type update. This commit
removes such operations. No visible compression/speed difference.
Thanks to Yue (yuec@) for finding this issue.

Change-Id: I3f83824aa3fd3717d63be0b280fa57258939a70a
2014-07-25 07:17:58 -07:00
Jingning Han
7112d70f24 Remove unnecessary conditional assignment
The assignment of the variable mode_excluded in
vp9_rd_pick_inter_mode_sub8x8 takes redundant conditional jump.
This commit removes it.

Change-Id: Ie195fbe6e54ec2ade7093d562c456a2e93143704
2014-07-24 15:34:11 -07:00
Tim Kopp
9d337d34f2 s/CONFIG_DENOISING/CONFIG_VP9_TEMPORAL_DENOISING
This should prevent confusion with the VP8 CONFIG_TEMPORAL_DENOISING and other
flags.

Change-Id: I1fe4e2977895b7966841d861ab74317ad875b6c8
2014-07-24 13:43:52 -07:00
Alex Converse
6eae35c07f Refactor inter/intra_suberblock_yrd.
Move txfm_rd_in_plane into choose_tx_size_from_rd and cleanup callers.

Change-Id: I1df2d7dc984802bd5e204cbe881ada0d75fbb3f7
2014-07-24 11:21:51 -07:00
Minghai Shang
929001bf22 [spatial svc]Fix reference issues
1. Remove last reference flag for first frame upper layers in one pass mode.
2. Disable refresh golden frame flag for key frames.

Change-Id: I44ac1bd2c795169e4fbfdd078ea79a1d33a204d6
2014-07-23 16:54:14 -07:00
Jingning Han
374c885919 Merge "Remove redundant argument entry in handle_inter_mode" 2014-07-23 15:07:01 -07:00
Jingning Han
787e8240d5 Merge "Use the chessboard pattern pred search in newmv mode" 2014-07-23 15:06:52 -07:00
Yaowu Xu
5dcb2e3237 Merge "Moved call to vp9_clear_system_state() to a proper location" 2014-07-23 12:46:11 -07:00
Jingning Han
e945c56d4a Remove redundant argument entry in handle_inter_mode
The value of mode_excluded has been properly set in
vp9_rd_pick_inter_mode_sb(). It is redundant to send it in
handle_inter_mode() and re-set the value again.

Change-Id: I408d4731f2f42e0bcf3ae62e85757717bb410471
2014-07-23 12:04:45 -07:00
Jingning Han
4f2f86725b Use the chessboard pattern pred search in newmv mode
This commit extends the chessboard pattern prediction filter search.
If the above and left blocks have the same prediction filter type,
the encoder will skip the prediction filter type search and use the
reference one.

The overall chessboard pattern prediction filter type search reduces
speed 3 runtime for hard clips. Experiments on park joy at 1080p
and 15000 kbps show that the runtime goes from 723265 ms to 65832 ms,
i.e., about 10% speed-up. Compression performance wise, it affects
the coding quality by

Change-Id: I880975497c7ad166532e9eea9bf46684d77ff327
derf:    -0.326%
yt:      -0.257%
hd:      -0.241%
stdhd:   -0.417%
2014-07-23 11:59:52 -07:00
Jingning Han
66d5757695 Merge "Remove redundant num_refs definition" 2014-07-23 10:34:54 -07:00
Jingning Han
353819103e Remove redundant num_refs definition
Use is_comp_pred to replace the use case of num_refs.

Change-Id: I4d0c1e14d5f728428a2ae3d293cd2b4a8b2f31d8
2014-07-23 09:29:51 -07:00
Jingning Han
0e5edf4eae Merge "Enable chessboard inter prediction filter type search" 2014-07-23 09:12:56 -07:00
Jingning Han
54ad09586c Enable chessboard inter prediction filter type search
This commit enables a chessboard pattern prediction filter type
search scheme for rate-distortion optimization speed-up. For the
inferred motion vector modes, the encoder can re-use its above/left
neighbor blocks' prediction filter type and skip a full test on
all possible filter types. Such operation is turned on/off
alternatively in a chessboard manner.

It is turned on in speed 3. For test clip pedestrian 1080p, the
runtime is reduced from 231500 ms -> 221700 ms. The compression
performance is changed:
derf:  -0.147%
yt:    -0.134%
hd:    -0.079%
stdhd: -0.220%

Change-Id: I1912f278e7576c2dc632688e3ad7a257410c605a
2014-07-22 16:49:03 -07:00
Adrian Grange
1f3c43e602 Merge "Fix get_frame_type function" 2014-07-22 15:17:27 -07:00
Tim Kopp
75441e1e08 Merge "VP9 denoiser bugfix in debugging code." 2014-07-22 14:48:42 -07:00
Jingning Han
f0f428e9ba Merge "USE local best_filter variable in handle_inter_mode" 2014-07-22 14:21:34 -07:00
Tim Kopp
1fe18acb92 VP9 denoiser bugfix in debugging code.
When OUTPUT_YUV_DENOISED is enabled the encoder outputs the uncompressed,
denoised video to a separate file. Moved the point at which the file is
written to in order to avoid an extra blank frame at the beginning of the video.

Change-Id: I805f6a912b18b3d9cae59b13c5b8108279439ce3
2014-07-22 14:10:16 -07:00
James Zern
dcb6aa8e41 Merge "vp9_bitstream.c: cosmetics" 2014-07-22 13:17:40 -07:00
Adrian Grange
caad1686d4 Fix get_frame_type function
Fixed the function get_frame_type to return the correct
frame type for golden and last frames.

Change-Id: I8edddd9aa26cbe7a1de8ff211389410b22b1bd14
2014-07-22 12:12:16 -07:00
James Zern
de4db2dc4f vp9_bitstream.c: cosmetics
fix indent, spelling and drop some vertical whitespace

Change-Id: I722671381a374a24763b07a02805ab1d149ab3f4
2014-07-22 11:38:53 -07:00
Jingning Han
5de6114e8f USE local best_filter variable in handle_inter_mode
This should be a local variable. Move the definition from
vp9_rd_pick_inter_mode_sb to handle_inter_mode.

Change-Id: I14f4168bb1c896ed04e8f6d4cd89fbf4c9839944
2014-07-22 11:35:59 -07:00
Minghai Shang
24c9d6ad43 [spatial svc]Use #if instead of #ifdef on macro CONFIG_SPATIAL_SVC
Change-Id: Ifc94377a0d05d66e3d21b007893a985b66db6082
2014-07-22 11:11:55 -07:00
Jingning Han
97ccebac8f Merge "Turn on adaptive pred filter scheme for sub8x8 below 720p" 2014-07-22 09:11:52 -07:00
Jingning Han
ffd948bbd5 Turn on adaptive pred filter scheme for sub8x8 below 720p
For sequences of resolution below 720p, the encoder will check
intra prediction modes and inter prediction modes from LAST_FRAME.
This commit turns on adaptive prediction filter scheme for sub8x8
blocks, where inter prediction modes are enabled. For the test
sequence bus at CIF, the speed 2 runtime goes down from 17879 ms
to 16783 ms, i.e., 6% speed up. The compression performance of
derf set is down by -0.128%.

Change-Id: I01d5321a5ceab4e0666ac5be56c52d896c7a8d45
2014-07-21 16:22:56 -07:00
Alex Converse
5926e7c0e8 Remove unfinished VP9 alpha channel.
Change-Id: Ic5d3a3a0dac10b49495771886a31e793bb78b5ca
2014-07-21 15:55:50 -07:00
Yaowu Xu
bcaf1d69ec Moved call to vp9_clear_system_state() to a proper location
The commit moved a call to vp9_clear_system_state() to a correct
location, i.e. prior function calls using floating point numbers.
This was to fix a mismatch mmx code and sse2 version, where a
floating point number used in adjust_frame_rate(cpi) gets NAN due
to mmx registers being in wrong state.

Change-Id: I40e0a6de98812000ccee6a729badb630604fd7e6
2014-07-21 15:55:12 -07:00
Yunqing Wang
765485cab2 Add -DNDEBUG when config option debug is disabled
For gcc, when libvpx config option debug is disabled, added the
flag -DNDEBUG to disable the assertions in libvpx for some speedup.

Change-Id: Ifcb7b9e8ef5cbe5d07a24407b53b9a2923f596ee
2014-07-21 09:20:03 -07:00
Tim Kopp
c66f612c4b VP9 denoiser fix: ref frames now updated properly
The ALT_REF_FRAME is now updated in the case of a KEY_FRAME in the VP9 denoiser.

Change-Id: Idf9a9772706f50e774fb240afcc01db38841043c
2014-07-18 15:26:19 -07:00
Deb Mukherjee
727f384085 Merge "Separates profile 2 into 2 profiles 2 and 3" 2014-07-18 03:23:51 -07:00
Deb Mukherjee
c447a50aea Separates profile 2 into 2 profiles 2 and 3
Separates HBD profile int two profiles (2 and 3) consistent with the
highbitdepth branch. This patch is ported from the original highbitdepth
branch patch: https://gerrit.chromium.org/gerrit/#/c/70460/

Two of the invalid file tests needed to be updated.

Change-Id: I6a4acd2f7a60b1fb4cbcc8e0dad4eab4248431e3
2014-07-17 20:51:59 -07:00
Pengchong Jin
ac638125ea Merge "Fixed a bug of setting wrong first pass mb stats pointer" 2014-07-17 14:24:52 -07:00
Adrian Grange
8cb8aef7c7 Merge "Modified frame buffer handling" 2014-07-17 12:15:16 -07:00
Pengchong Jin
e358ab5fc9 Fixed a bug of setting wrong first pass mb stats pointer
The bug sets the wrong pointer to the first pass mb stats
if the encoder does the re-coding in the second pass.

Change-Id: I8a11f45dd7dceb38de814adec24cecccae370d00
2014-07-17 12:04:15 -07:00
Scott LaVarnway
ba0652e83a Merge "Added vp9_sad64x64_neon(), vp9_sad32x32_neon()" 2014-07-17 11:42:16 -07:00
Adrian Grange
f68aaa38d6 Modified frame buffer handling
This patch is the first step toward simplifying the
frame buffer handling.

The final goal is to have a common frame buffer handling
framework for both encoder and decoder that incorporates
the existing ability to use externally allocated memory.

Change-Id: I2c378a4f54a39908915f46c4260e17a080db7ff1
2014-07-17 11:06:35 -07:00
Paul Wilkins
93960c869e Merge "Changes to rd balance and multi-arf bug fix." 2014-07-17 07:01:31 -07:00
Yaowu Xu
42a68e6701 Merge "make default_interp_filter choice a speed feature" 2014-07-16 19:12:22 -07:00
Guillaume Martres
f744f19cff Merge "vp9_ratectrl.c: refactor get_active_quality usage" 2014-07-16 14:29:32 -07:00
Yaowu Xu
51c60a891e make default_interp_filter choice a speed feature
This commit changed the hard-coded DEFAULT_INTERP_FILTER to a speed
feature with the same default value: SWITCHABLE.

Change-Id: I7f54f40f1bd3f5277841d04b85db7a84e47313f1
2014-07-16 14:28:51 -07:00
Scott LaVarnway
696fa52eaa Added vp9_sad64x64_neon(), vp9_sad32x32_neon()
and vp9_sad16x16_neon()

On a Nexus 7, vpxenc (in realtime mode, speed -6)
reported a performance improvement of ~17%.

Change-Id: I91e070cde2973451083d3f3d63b49b7886de9a85
2014-07-16 12:54:46 -07:00
Tim Kopp
ca752e3320 Merge "VP9 Denoiser denoises after mode/bsize search" 2014-07-16 08:22:14 -07:00
Paul Wilkins
b691230dea Changes to rd balance and multi-arf bug fix.
2 pass only change to calculation of rd mult based on Q.
Make a small adjustment based on frame type and also
replace adjustment based on iifactor with an one based
on the ambient GF/ARF boost level.

Also fix multi arf bug / issue.

Overall these change give an slight improvement in ssim
but hurt psnr a little.

Change-Id: I5e1751e3ff5390a26f543d7855059e6fbcce105e
2014-07-16 13:58:47 +01:00
Yaowu Xu
faa686bb1b Added a rt speed 12
We target this speed to achieve similar encoding speed and better
compression than vp8 rt mode with cpu-used at -12.

Change-Id: Ic1bb4371c81a17ea80e83459c1cbf4c09a3498e8
2014-07-15 16:46:22 -07:00
Yaowu Xu
257f16cc54 Merge "Make non-rd pick_mode work with Golden/Altref" 2014-07-15 12:01:31 -07:00
Tim Kopp
03819ed9ab VP9 Denoiser denoises after mode/bsize search
In vp8, statistics are collected about the different modes as they are searched.
This process is more complicated due to the variable block size. Fields were
added to the PICM_MODE_CONTEXT struct to hold this information for each point in
the search. The information is then taken from the appropriate part of the tree
during denoising.

Change-Id: I89261ab77ad637821287ae157dfdf694702b8e77
2014-07-15 08:43:43 -07:00
Pengchong Jin
f349b071c6 Rewrite functions related to first pass block stats
Change-Id: I28679f88e2911b06eef5cbc83ecb62b8c69e4c53
2014-07-14 17:45:27 -07:00
Deb Mukherjee
1f6aaeddc5 Merge "Some extra bit probability cleanups" 2014-07-14 17:26:54 -07:00
Jingning Han
2806ea91f5 Merge "Fix a potential invalid memory access in non-RD coding flow" 2014-07-14 17:25:43 -07:00
Yaowu Xu
f1885bc0ca Make non-rd pick_mode work with Golden/Altref
This is to fix a reported issue #825:
https://code.google.com/p/webm/issues/detail?id=825

Change-Id: I196535aee81a8967551c058849d7f9c6874cb730
2014-07-14 17:16:24 -07:00
Minghai Shang
e899859c48 [spatial svc]Implement alt reference frames
All changes are for spatial svc only.
1. Enable encoding hidden frames in each layer and use alt reference idex to reference the hidden frame in each layer
2. Use golden reference idx for spatial reference
3. For those layers that don't have hidden frames (caused by lack of frame buffers), reference a hidden frame in lower layers
4. Add "auto-alt-refs" in svc options
Change-Id: Idf27d1fd2fb5f3ffd9e86d2119235e3dad36c178
2014-07-14 11:24:17 -07:00
Jingning Han
6ce515b9ff Merge "Fix chrome valgrind warning due to the use of mismatched bsize" 2014-07-13 11:07:44 -07:00
Jingning Han
b957439c87 Fix a potential invalid memory access in non-RD coding flow
This commit fixes a potential out-of-boundary memory access due to
the use of reuse_inter_pred_sby in the non-RD coding flow. It
resolves the corresponding asan error.

Change-Id: Iff605f5921230966990013541cd855d698810922
2014-07-11 15:50:43 -07:00
Jingning Han
3cddd81c6d Fix chrome valgrind warning due to the use of mismatched bsize
This commit fixes a mismatched use case of block size in non-RD
intra prediction check. The residual SSE and variance should be
calculated per transform block size, instead of operating block
size, which caused chrome valgrind warning on conditional jump
based on uninitialized value (webm issue 823). This commit
resolves this issue.

Change-Id: I595c06599c7e0fd0e4a08736519ba68fc14bc79a
2014-07-11 15:49:22 -07:00
Yunqing Wang
6298e232e1 Merge "Code refactoring: use defined inline functions" 2014-07-11 13:46:46 -07:00
Deb Mukherjee
6957e7a077 Some extra bit probability cleanups
Refactoring to remove some duplication of probability
tables between tokenization and detokenization.

Change-Id: I2fc6a6497f9c0410021a9b41f828bc58a864e466
2014-07-11 11:39:18 -07:00
Yaowu Xu
84744a497a Merge "Remove an unused parameter in vp9_init_search_range()" 2014-07-11 11:13:22 -07:00
Adrian Grange
af4f390fff Merge "Re-factor and simplify arnr filter." 2014-07-11 10:52:09 -07:00
Yaowu Xu
6673d2f309 Remove an unused parameter in vp9_init_search_range()
Change-Id: I3d9130e726a1299fd258f6dfe93315e2d12f76da
2014-07-11 10:32:39 -07:00
Yunqing Wang
1b5e9871f7 Code refactoring: use defined inline functions
Changed to use defined inline functions consistently through
the code.

Change-Id: I7644d24fa7a837378564a6e0790416d3725dd200
2014-07-11 10:30:25 -07:00
Paul Wilkins
e3e6e06155 Re-factor and simplify arnr filter.
Use a weaker filter for second level arf frames.
Average gain across all sets and metrics ~0.3%

Remove code for arnr_type which is no longer
supported in VP9 which always uses a centered blur.

Re-factor and some cleanup.

Change-Id: Ieb4b8940e99e4e02b3fcc9fca6f2d4109e6ed639
2014-07-11 17:45:40 +01:00
Yaowu Xu
3265585326 Merge "Minor cleanup" 2014-07-10 16:39:48 -07:00
Yunqing Wang
75cd57503d Refactor vp9_diamond_search_sad function
Currently, vp9_diamond_search_sadx4() is only called when sse3 is
enabled, which is improper since sse2 optimization of sdx4df
functions are available. Changed to always use
vp9_diamond_search_sadx4().

Change-Id: I4b95d6b7a3c6c645783c373f0ba8d645ece24717
2014-07-10 09:19:03 -07:00
Yunqing Wang
30117a576d Refactor refining_search_sad code
There are sse2 optimization of sdx4df functions. Instead of calling
vp9_refining_search_sadx4 only when sse3 is enabled, call it always.

Change-Id: I24f93818f7d4209d1425039e0eb099ff9ff08fe9
2014-07-09 16:50:11 -07:00
Yaowu Xu
87cf002e9d Minor cleanup
Change-Id: I3a3ceeeed489f8b1ccd7199ff97f3fb991bbf5a4
2014-07-09 15:42:10 -07:00
Yunqing Wang
a581da218e Remove repetitive code in mcomp.c
Deleted vp9_find_best_sub_pixel_comp_tree(), and combined it in
vp9_find_best_sub_pixel_tree().

Change-Id: Ifb25763c8b19822df5537cc1daa76ce88dc3b056
2014-07-09 14:50:50 -07:00
Yunqing Wang
a51e389b42 Merge "Adjust full-pixel search method in real-time mode" 2014-07-09 13:46:42 -07:00
Yaowu Xu
0e99f3a387 Merge "Combined non-rd motion searchs into a single function" 2014-07-09 13:02:25 -07:00
Yunqing Wang
9bd3be69a4 Adjust full-pixel search method in real-time mode
Use FAST_HEX in speed 5 and 6, which covers more points than
FAST_DIAMOND and improves motion search quality.

At speed 6, RTC set borg tests showed slight quality gain (psnr
gain: 0.143%, ssim gain: 0.226%). No noticeable encoding speed
change.

Change-Id: Ifa62875d9a52ee382ec494f271382bb77d8c67bf
2014-07-09 12:56:25 -07:00
Yaowu Xu
c788bceb55 Combined non-rd motion searchs into a single function
This commit combined the full pel and sub pel motion search into a
single function to avoid code duplication. The commit does not change
encoder outputs.

Change-Id: Ibe18342c4f64073bef20f9cf6c6ca0a20d01bf0d
2014-07-09 12:07:52 -07:00
Jingning Han
f6bf614b2f Merge "Re-design quantization process for 32x32 transform block" 2014-07-09 11:55:26 -07:00
Jingning Han
9ad1b9fc67 Re-design quantization process for 32x32 transform block
This commit enables a new quantization process for 32x32 2D-DCT
transform coefficient blocks. It improves the compression
performance of speed 5 by 1.4%. The overall compression gains of
speed 5 due to the new quantization scheme is 4.7%. It also includes
the SSSE3 implementation of the 32x32 quantization process.

Change-Id: I0855b124fd6462418683f783f5bcb44255c9993b
2014-07-08 16:55:28 -07:00
Tim Kopp
3c86228cd3 Vp9 denoiser MC bugfix
In the previous version, only certain buffers in the macroblockd were saved and
the restored. In this version, all of the buffers are saved and restored. The
code was then rolled into a loop for readability.

Also contains a tiny fix for when the -DOUTPUT_YUV_DENOISED flag is used.

Change-Id: Id925ef8b3fa122ae88acfa1d9a1e4df45df83518
2014-07-08 15:13:13 -07:00
Johann
fe4b663559 Merge "Use the VP9 version of extend_borders" 2014-07-07 15:20:37 -07:00
Jingning Han
ffd2213660 Merge "Tune SSSE3 implementation of fast path quantization" 2014-07-07 12:07:09 -07:00
Jingning Han
6214038be9 Merge "Remove an empty line" 2014-07-07 11:42:48 -07:00
Jingning Han
00fc0e3ff5 Tune SSSE3 implementation of fast path quantization
This commit further simplifies the SSSE3 implementation of the fast
path quantization process.

Change-Id: I5be3286ec0f1bd81d1cf5be3168fece6384fb9ca
2014-07-07 11:06:53 -07:00
Jingning Han
3316918b3b Remove an empty line
Change-Id: Id6eedc502c86433df1456dd994aee6bc9a1359a2
2014-07-07 10:28:05 -07:00
Alex Converse
f60a1178c6 Cleanup motion search speed features.
* Replace max_step_search_steps with constant MAX_MVSEARCH_STEPS
* Fold (reduce_first_step_size + speed > 5) into reduce_first_step_size
  replacing uses of reduce_first_step_size that don't add the speed
  check with zero.

Change-Id: Iae46395dbf3eaca138bf4d18b838a9e364b5a198
2014-07-07 10:08:45 -07:00
Alex Converse
f0e0d01e94 Merge "Allow lossless skipping in RD mode decision." 2014-07-07 09:58:43 -07:00
Guillaume Martres
44666b7ce3 vp9_ratectrl.c: refactor get_active_quality usage
Change-Id: I53db06acf5bc434f9584136b848322f5870300b3
2014-07-06 18:49:46 +02:00
Paul Wilkins
14f570f6c3 Merge "Multi-arf: Add code to turn it on and off." 2014-07-03 02:16:49 -07:00
Alex Converse
49f00ba77e Merge "Cleanup vp9_rd." 2014-07-02 21:33:54 -07:00
Alex Converse
5fa37fb526 Merge "Split vp9_rdopt into vp9_rdopt and vp9_rd." 2014-07-02 21:33:44 -07:00
Dmitry Kovalev
362b53ecf7 Merge "Cleaning up and simplifying read_frame_stats()." 2014-07-02 15:56:50 -07:00
Alex Converse
15123db753 Cleanup vp9_rd.
Change-Id: I39a37335ba5b3a969d328afb1f425ddb2cf7ddda
2014-07-02 15:54:36 -07:00
Alex Converse
03c276ea17 Split vp9_rdopt into vp9_rdopt and vp9_rd.
vp9_rdopt is for making rd optimal mode decisions. vp9_rd is for all
other rd related routines. Anything used outside of making an rd optimal
decision belongs in rd.

Change-Id: I772a3073f7588bdf139f551fb9810b6864d8e64b
2014-07-02 15:33:33 -07:00
Dmitry Kovalev
4635a2ba11 Cleaning up and simplifying read_frame_stats().
Change-Id: I262ecac02d376de83097bb40f744f5584e987603
2014-07-02 13:52:50 -07:00
Yunqing Wang
ee5d0335ca Merge "Fix rd threshold overflow issue" 2014-07-02 13:06:41 -07:00
Tim Kopp
6c0a2692e4 Merge "VP9 denoiser implemented FILTER_BLOCK case" 2014-07-02 12:48:28 -07:00
Tim Kopp
3302147f93 Merge "VP9 denoising enabled by noise_sensitivity param" 2014-07-02 12:45:34 -07:00
Yunqing Wang
3bc1193201 Fix rd threshold overflow issue
Moved the threshold adjustment before reference flag checking,
which could set the threshold to INT_MAX for disabled reference
frame, and cause overflow if the adjustment is done after that.

Change-Id: I85e94f8726d5e3ae93f65965aa978721dddc9957
2014-07-02 12:16:27 -07:00
Tim Kopp
1e511a539c Merge "Replaced loops with vpx_memcpy()" 2014-07-02 11:35:32 -07:00
Tim Kopp
03a3ba4a0d VP9 denoiser implemented FILTER_BLOCK case
Renamed updating_running_avg() to filter(). Extended function with the rest of
the filter procedure. Made all of the empirically-determined constants used in
VP8 into functions so they can be tweaked more easily.

Change-Id: I41730c8c92370c76885950a43742347477ca4e7e
2014-07-02 11:23:29 -07:00
Tim Kopp
9c9922df97 VP9 denoising enabled by noise_sensitivity param
As in VP8.

Currently, this parameter is set with the VP8E_SET_NOISE_SENSITIVITY flag.
The flag was not renamed so that we don't break the interface for webrtc. This
should probably be changed at some point in the future.

Change-Id: Ic73fcb0dde9d1d019e9d042050b617333ac65472
2014-07-02 11:20:35 -07:00
Tim Kopp
49741fee9f Replaced loops with vpx_memcpy()
Change-Id: Icbe05657f0e92c3838e6a5a975f4f82d21328a2e
2014-07-02 10:36:25 -07:00
Yaowu Xu
6c417077be Merge "Added a speed feature controlling a motion search parameter" 2014-07-02 10:31:51 -07:00
Paul Wilkins
8830772370 Multi-arf: Add code to turn it on and off.
Add test code to turn multi-arf on and off depending
on group length and zero motion.

Changes to active max group length for mult-arf.

Fund second arf only from normal frame bits.

Change-Id: I920287fac1c886428c15a39f731a25d07c2b796c
2014-07-02 17:53:23 +01:00
Paul Wilkins
579c7bcca5 Merge "Adapt strength of AQ2." 2014-07-02 09:49:34 -07:00
Yaowu Xu
92a6db7928 Added a speed feature controlling a motion search parameter
This commit added a speed feature to control the step_param used in
full pixel motion search. The intention is to reduced the search
steps for high speed real time coding.

Change-Id: I21d2f0105c2b647783a6688615da7fcf2b6d670b
2014-07-02 09:30:43 -07:00
Pengchong Jin
2c04e85d06 Merge "Store/read 16x16 block statistics obtained from the first pass" 2014-07-02 08:36:22 -07:00
Paul Wilkins
adf4293e4e Adapt strength of AQ2.
Adapt the use of segmentation in AQ mode 2 based on
the ambient kf/arf/gf Q.

Disable segmentation where the rate per SB is very
low and overheads are likely to outweigh the benefits.

This patch reduces the -ve average metrics impact
of AQ mode 2 while allowing stronger 3 segment AQ
in some cases. Average improvement ~0.5-1.0%.

Change-Id: I5892dfcc7507c5cc6444531cc7fe17554cf8d0c7
2014-07-02 16:34:26 +01:00
Tim Kopp
08cb2b0211 Merge "VP9 denoiser used s/int/enum where appropriate" 2014-07-01 23:03:24 -07:00
Tim Kopp
73799aa3f7 Merge "Denoised output is now grayscale" 2014-07-01 23:03:07 -07:00
Yaowu Xu
82fd084b35 Merge "Re-design quantization process" 2014-07-01 19:04:01 -07:00
Jingning Han
9ac2f66320 Re-design quantization process
This commit re-designs the quantization process for transform
coefficient blocks of size 4x4 to 16x16. It improves compression
performance for speed 7 by 3.85%. The SSSE3 version for the
new quantization process is included.

The average runtime of the 8x8 block quantization is reduced
from 285 cycles -> 255 cycles, i.e., over 10% faster.

Change-Id: I61278aa02efc70599b962d3314671db5b0446a50
2014-07-01 17:00:07 -07:00
Alex Converse
0256a75950 Allow lossless skipping in RD mode decision.
Change-Id: I2fc4ecfc2dd3ff1dd241a68c9ed4c280291b41f2
2014-07-01 16:57:21 -07:00
Pengchong Jin
aaabbd67b2 Store/read 16x16 block statistics obtained from the first pass
Add a conditional compile flag for this feature. Also add a
switch to enable the encoder to use these statistics in the
second pass. Currently, the switch is turned off.

Change-Id: Ia1c858c35ec90e36f19f5cffe156b97ddaa04922
2014-07-01 16:47:17 -07:00
Yunqing Wang
f31ff029df Elevate NEWMV mode checking threshold in real time
The current threshold is knid of low, and in many cases NEWMV
mode is checked but not picked as the best mode. This patch
added a speed feature to increase NEWMV threshold, so that
less partition mode checking goes to check NEWMV. This feature
is enabled for speed 6 and 7.

Rtc set borg tests showed:
1. Speed 6, overall psnr: -0.088%, ssim: -1.339%;
   Average speedup on rtc set is 11.1%.
2. Speed 7, overall psnr: -0.505%, ssim: -2.320%
   Average speedup on rtc set is 12.9%.

Change-Id: I953b849eeb6e0d5a1f13eacba30c14204472c5be
2014-07-01 14:50:39 -07:00
Tim Kopp
1a66dab93a VP9 denoiser used s/int/enum where appropriate
Change-Id: Id52a7869fd1f31bb060de170e3295da7435adb9e
2014-07-01 14:07:40 -07:00
Tim Kopp
2f71de77f0 Denoised output is now grayscale
Grayscale is conditionally compiled.

Change-Id: I482ab237560d0bae8d397fd9999e78d38104f2a1
2014-07-01 14:07:40 -07:00
Dmitry Kovalev
19cbf54143 Merge "Fix visual studio build issue" 2014-07-01 10:30:39 -07:00
Yunqing Wang
9ba1d60bd1 Fix visual studio build issue
Fixed the signed/unsigned mismatch.

Change-Id: Id83d603b8f1745b71f4cf695a0751e55518b1316
2014-07-01 08:58:05 -07:00
Alex Converse
6c54dbcb69 Merge "BITSTREAM: Handle transform size and motion vectors more logically for non-420." 2014-06-30 17:44:01 -07:00
Yunqing Wang
4bb4c2291b Merge "Encode_breakout code refactoring" 2014-06-30 16:05:25 -07:00
Yaowu Xu
370618ffb4 Merge "change to not force interp_type as SWITCHABLE" 2014-06-30 15:44:08 -07:00
Yaowu Xu
186bd4eb52 change to not force interp_type as SWITCHABLE
Encoder still uses SWITCHABLE as default via DEFAULT_INTERP_FILTER,
but does not override the default if it is not SWITCHABLE.

Change-Id: I3c0f6653bd228381a623a026c66599b0a87d01d5
2014-06-30 12:48:21 -07:00
Jingning Han
6643b8868d Merge "Remove unused set_mode_info function" 2014-06-30 12:20:30 -07:00
Yunqing Wang
3779ccaf98 Encode_breakout code refactoring
Moved the encode_breakout_test out of vp9_pick_inter_mode().

Change-Id: I6966d0293ae5210a5a28b0e8debacb24d1c0d2d4
2014-06-30 11:22:32 -07:00
Jingning Han
30ab37019c Remove unused set_mode_info function
When the frame is intra coded only, the encoder takes the RD
coding flow. Hence the function set_mode_info is not practically
in use. This commit removes it and the associated conditional
branches.

Change-Id: I1e42659ceb55b771ba712d1cdecacb446aa6460d
2014-06-30 10:59:04 -07:00
Yunqing Wang
dee5782f93 Enable encode breakout in real time
For real time speed 7, once encode breakout is on(i.e. encoding
setting --static-thresh=1), a proper encode breakout threshold
is set to speed up the encoder.

Set --static-thresh=1, RTC set borg test showed a slight overall
psnr loss of 0.162%, but ssim gain of 0.287%. The average speedup
on RTC set is 6%, and for some clips, the speedup can be 10+%.

Change-Id: Id522d9ce779ff7c699936d13d0c47083de4afb85
2014-06-30 10:41:12 -07:00
Yunqing Wang
9d41313e4b Decide the partitioning threshold from the variance histogram
Before encoding a frame, calculate and store each 16x16 block's
variance of source difference between last and current frame.
Find partitioning threshold T for the frame from its variance
histogram, and then use T to make partition decisions.

Comparing with fixed 16x16 partitioning, rtc set test showed an
overall psnr gain of 3.242%, and ssim gain of 3.751%. The best
psnr gain is 8.653%.

The overall encoding speed didn't change much. It got faster for
some clips(for example, 12% speedup for vidyo1), and a little
slower for others.

Also, a minor modification was made in datarate unit test.

Change-Id: Ie290743aa3814e83607b93831b667a2a49d0932c
2014-06-30 09:36:23 -07:00
Tim Kopp
04d9720c63 Merge "Implemented motion compensation for VP9 denoiser" 2014-06-30 08:29:32 -07:00
Jim Bankoski
a13bf65315 remove unused parms from rd_pick_inter_mode_sb_seg_skip
Change-Id: I7f989d197444d166133ad91eb23ac1033109f58d
2014-06-29 09:23:21 -07:00
Yaowu Xu
0e79906dc0 Merge "Allow encoder to set lpf level to 0" 2014-06-27 16:47:16 -07:00
Yaowu Xu
303aa7e42a Merge "Added a new speed 7 in rt mode" 2014-06-27 16:47:06 -07:00
Tim Kopp
a5f49183da Merge "fix: Only do spatial SVC when there are > 1 layers" 2014-06-27 15:42:14 -07:00
Tim Kopp
b0959b8195 Merge "VP9 denoiser: implemented update_frame_stats()" 2014-06-27 15:41:51 -07:00
Yaowu Xu
d0cb273e04 Allow encoder to set lpf level to 0
As a way to speed-up rtc encoding at speed 7.

Change-Id: Ie36a010392cf7b741dc130df21a4e733622a75b7
2014-06-27 15:23:41 -07:00
Yaowu Xu
3f92b7b994 Added a new speed 7 in rt mode
To experiment with different speed/quality compromises.

Change-Id: Ia9d4b85243554d620498a327da37c356e752b07f
2014-06-27 13:29:09 -07:00
Alex Converse
3cac9f0a04 Merge "Use UV prediction when deciding to skip in for lossless." 2014-06-27 10:06:54 -07:00
Tim Kopp
2826c1d259 Implemented motion compensation for VP9 denoiser
Change-Id: Iee21eb0ecc5a1fe2c56fb3df0cee0ead6d139ed1
2014-06-27 08:56:09 -07:00
Tim Kopp
0299a60334 fix: Only do spatial SVC when there are > 1 layers
Bug introduced in I930dced169c9d53f8044d2754a04332138347409. If
svc.number_temporal_layers == 1 and svc.number_spatial_layers == 1, the system
attempt to do spatial SVC. It no longer does that.

Change-Id: Ie6b130a72b1eea40c547c9a64447e40695f811c5
2014-06-27 08:56:09 -07:00
Tim Kopp
52462bf7a8 VP9 denoiser: implemented update_frame_stats()
Also added reset_frame_stats()

Change-Id: I8e6ca00dbd5fa85cd39485d81c9343c0ff207d6c
2014-06-27 08:56:09 -07:00
Paul Wilkins
1d5223c627 Multi-arf: Change ref buffer for primary arf.
For the primary arf in a group, if multiple arfs
are enabled and we were using arfs in the previous
group, then allow the second arf from the previous
group to be used as an additional reference.

Change-Id: Iaf41706a52f54ef21548026851cd77100d6aebda
2014-06-27 12:12:21 +01:00
Jingning Han
5a3e3c6d3f Adaptive txfm size selection depending on residual sse/variance
This commit enables an adaptive transform size selection method
for speed -6. It uses largest transform size when the sse is more
than 4 times of variance, i.e., most energy is compacted in the
DC coefficient. Otherwise, use the default TX_8X8. It improves
the compression efficiency for rtc set of speed -6 by 0.8%, no
speed change observed.

Change-Id: Ie6ed1e728ff7bf88ebe940a60811361cdd19969c
2014-06-26 16:00:42 -07:00
Pengchong Jin
73eeb3beff Merge "Skip the partition search for the frame with no motion" 2014-06-26 14:36:10 -07:00
Alex Converse
aed5271876 Use UV prediction when deciding to skip in for lossless.
Change-Id: Ic149749157d762039446d14472d40d9211c6451a
2014-06-26 14:34:56 -07:00
Pengchong Jin
1286126073 Skip the partition search for the frame with no motion
This patch allows the encoder to skip the partition search for the
frame if it is an inter frame and only zero motion vectors have
been detected in the first pass. The partition size is directly
assigned according to the difference variance.

Borg tests show overall little performance changes in term of PSNR
(derf -0.027%, yt 0.152%, hd 0.078%, stdhd 0%). The worst case of
PSNR loss is -0.514% from yt. The best PSNR gain is 4.293% from yt.
The second pass encoding speedup for slideshow clips is 15%-40%.

Change-Id: I881f347d286553ee5594a9ea09ba1a61ac684045
2014-06-26 12:10:34 -07:00
Jingning Han
e15f6bc19c Merge "Add const mark to const values in non-RD coding mode" 2014-06-26 11:00:34 -07:00
Jingning Han
56afb9c41a Merge "Enable real-time version reference motion vector search" 2014-06-26 11:00:25 -07:00
Jingning Han
46ea9ec719 Enable real-time version reference motion vector search
This commit enables a fast reference motion vector search scheme.
It checks the nearest top and left neighboring blocks to decide the
most probable predicted motion vector. If it finds the two have
the same motion vectors, it then skip finding exterior range for
the second most probable motion vector, and correspondingly skips
the check for NEARMV.

The runtime of speed -5 goes down
pedestrian at 1080p 29377 ms -> 27783 ms
vidyo at 720p       11830 ms -> 10990 ms
i.e., 6%-8% speed-up.

For rtc set, the compression performance
goes down by about -1.3% for both speed -5 and -6.

Change-Id: I2a7794fa99734f739f8b30519ad4dfd511ab91a5
2014-06-26 09:49:13 -07:00
Jingning Han
99e25ec469 Add const mark to const values in non-RD coding mode
Change-Id: I65209fd1e06fc06833f6647cb028b414391a7017
2014-06-26 09:42:03 -07:00
Paul Wilkins
46218c9cb9 Merge "Fix quality regression for multi arf off case." 2014-06-26 09:41:40 -07:00
Jingning Han
e84e868570 Merge "Make non-RD intra mode search txfm size dependent" 2014-06-26 09:10:07 -07:00
Paul Wilkins
1c27e1f127 Fix quality regression for multi arf off case.
Bug introduced during multiple iterations on: I3831*

gf_group->arf_update_idx[] cannot currently be used
to select the arf buffer index if buffer flipping on overlays
is enabled (still currently the case when multi arf OFF).

Change-Id: I4ce9ea08f1dd03ac3ad8b3e27375a91ee1d964dc
2014-06-26 09:59:53 +01:00
Paul Wilkins
601be5a29e Merge "Dual arf: Name changes." 2014-06-26 01:55:00 -07:00
Jingning Han
2aa50eafb2 Make non-RD intra mode search txfm size dependent
This commit fixes the potential issue in the non-RD mode decision
flow that only checks part of the block to estimate the cost. It
was due to the use of fixed transform size, in replacing the
largest transform block size. This commit enables per transform
block cost estimation of the intra prediction mode in the non-RD
mode decision.

Change-Id: I14ff92065e193e3e731c2bbf7ec89db676f1e132
2014-06-25 18:52:18 -07:00
Alex Converse
bd1fc3402c Merge "Allow lossless breakout in non-rd mode decision." 2014-06-25 10:51:57 -07:00
Minghai Shang
e319e4bfa6 Merge "[spatial svc]Don't skip motion search in first pass encoding" 2014-06-25 10:31:21 -07:00
Minghai Shang
0a103ae999 Merge "[spatial svc]Implement lag in frames for spatial svc" 2014-06-25 10:31:05 -07:00
Jingning Han
35bd31cd0a Merge "Replace cpi->common with preset variable cm" 2014-06-25 08:57:18 -07:00
Jingning Han
9f3f5c8bc4 Merge "Add vp9_ prefix to mv_pred and setup_pred_block functions" 2014-06-25 08:57:08 -07:00
Yunqing Wang
bccc785f63 Merge "Reuse inter prediction result in real-time speed 6" 2014-06-25 08:18:33 -07:00
Paul Wilkins
9f76c1ec50 Dual arf: Name changes.
Cosmetic patch only in response to comments on
previous patches suggesting a couple of name changes
for consistency and clarity.

Change-Id: Ida3a359b0d5755345660d304a7697a3a3686b2a3
2014-06-25 10:37:02 +01:00
Paul Wilkins
b8c382f8e7 Merge "Dual ARF changes: Buffer index selection." 2014-06-25 02:35:56 -07:00
Paul Wilkins
0f446165bc Merge "Adjust arf Q limits with multi-arf." 2014-06-25 02:35:45 -07:00
Alex Converse
1409d1e1ff Allow lossless breakout in non-rd mode decision.
This is very helpful for large moving windows in screencasts.

Change-Id: I91b5f9acb133281ee85ccd8f843e6bae5cadefca
2014-06-24 16:44:35 -07:00
Jingning Han
9e55834426 Replace cpi->common with preset variable cm
This commit replaces a few use cases of cpi->common with preset
variable cm, to avoid unnecessary pointer fetch in the non-RD
coding mode.

Change-Id: I4038f1c1a47373b8fd7bc5d69af61346103702f6
2014-06-24 16:07:17 -07:00
Jingning Han
85cfae818b Add vp9_ prefix to mv_pred and setup_pred_block functions
Make these two functions accessible by both RD and non-RD coding
modes.

Change-Id: Iecb39dbf3d65436286ea3c7ffaa9920d0b3aff85
2014-06-24 16:06:21 -07:00
Minghai Shang
6bebe65118 [spatial svc]Don't skip motion search in first pass encoding
Change-Id: Ia6bcdaf5a5b80e68176f60d8d00e9b5cf3f9bfe3
2014-06-24 14:29:13 -07:00
Minghai Shang
277338f748 [spatial svc]Implement lag in frames for spatial svc
Change-Id: I930dced169c9d53f8044d2754a04332138347409
2014-06-24 14:01:17 -07:00
Yunqing Wang
0aae100076 Reuse inter prediction result in real-time speed 6
In real-time speed 6, no partition search is done. The inter
prediction results got from picking mode can be reused in the
following encoding process. A speed feature reuse_inter_pred_sby
is added to only enable the resue in speed 6.

This patch doesn't change encoding result. RTC set tests showed
that the encoding speed gain is 2% - 5%.

Change-Id: I3884780f64ef95dd8be10562926542528713b92c
2014-06-24 12:46:33 -07:00
Johann
58ac00e9ab Use the VP9 version of extend_borders
Change-Id: Ie16f12b4763a45465e130fb39cbb727c08529ac8
2014-06-24 12:27:08 -07:00
Adrian Grange
8357292a5a Fix test on maximum downscaling limits
There is a normative scaling range of (x1/2, x16)
for VP9. This patch fixes the maximum downscaling
tests that are applied in the convolve function.

The code used a maximum downscaling limit of x1/5
for historic reasons related to the scalable
coding work. Since the downsampling in this
application is non-normative it will revert to
using a separate non-normative scaler.

Change-Id: Ide80ed712cee82fe5cb3c55076ac428295a6019f
2014-06-24 10:26:09 -07:00
Tim Kopp
4efcf83833 Merge "Fixed VP9 denoiser COPY_BLOCK case" 2014-06-24 09:48:11 -07:00
Paul Wilkins
60244ec1f4 Dual ARF changes: Buffer index selection.
Add indirection to the section of buffer indices.
This is to help simplify things in the future if we
have other codec features that switch indices.

Limit the max GF interval for static sections to fit
the gf_group structures.

Change-Id: I38310daaf23fd906004c0e8ee3e99e15570f84cb
2014-06-24 16:30:44 +01:00
Paul Wilkins
11b34f1e19 Adjust arf Q limits with multi-arf.
Adjust enforced minimum arf Q deltas for non primary arfs
in the middle of an arf/gf group.

Change-Id: Ie8034ffb3ac00f887d74ae1586d4cac91d6cace2
2014-06-24 16:29:24 +01:00
Paul Wilkins
9aca602e07 Further dual arf changes: multi_arf_allowed.
Add multi_arf_allowed flag.
Re-initialize buffer indices every kf.
Add some const indicators.

Change-Id: If86c39153517c427182691d2d4d4b7e90594be71
2014-06-24 13:19:17 +01:00
Paul Wilkins
8160a26fa0 Fix some bugs in multi-arf
Fix some bugs relating to the use of buffers
in the overlay frames.

Fix bug where a mid sequence overlay was
propagating large partition and transform sizes into
the subsequent frame because of :-
  sf->last_partitioning_redo_frequency  > 1 and
  sf->tx_size_search_method == USE_LARGESTALL

Change-Id: Ibf9ef39a5a5150f8cbdd2c9275abb0316c67873a
2014-06-24 13:07:48 +01:00
Paul Wilkins
2611022504 Clean out old CONFIG_MULTIPLE_ARF code.
Remove the old experimental multi arf code that was under
the flag CONFIG_MULTIPLE_ARF.

Change-Id: Ib24865abc11691d6ac8cb0434ada1da674368a61
2014-06-24 13:00:19 +01:00
Paul Wilkins
2e430cba61 Experiment for mid group second arf.
This patch implements a mechanism for inserting a second
arf at the mid position of arf groups.

It is currently disabled by default using the flag multi_arf_enabled.

Results are currently down somewhat in initial testing if
multi-arf is enabled. Most of the loss is attributable to the
fact that code to preserve the previous golden frame
(in the arf buffer) in cases where we are coding an overlay
frame, is currently disabled in the multi-arf case.

Change-Id: I1d777318ca09f147db2e8c86d7315fe86168c865
2014-06-24 12:59:14 +01:00
Alex Converse
2518e33bec Merge "Switch active map implementation to segment based." 2014-06-23 18:25:51 -07:00
Alex Converse
20adfc5350 Merge "Fork vp9_rd_pick_inter_mode_sb_seg_skip" 2014-06-23 18:25:46 -07:00
Adrian Grange
8c1f071f1e Allocate buffers based on correct chroma format
The encoder currently allocates frame buffers before
it establishes what the chroma sub-sampling factor is,
always allocating based on the 4:4:4 format.

This patch detects the chroma format as early as
possible allowing the encoder to allocate buffers of
the correct size.

Future patches will change the encoder to allocate
frame buffers on demand to further reduce the memory
profile of the encoder and rationalize the buffer
management in the encoder and decoder.

Change-Id: Ifd41dd96e67d0011719ba40fada0bae74f3a0d57
2014-06-23 11:45:13 -07:00
Alex Converse
6118dcfe40 Merge "Actually skip blocks in skip segments in non-rd encoder." 2014-06-23 10:23:20 -07:00
Jingning Han
961bafc366 Merge "Remove unused vp9_init_quant_tables function" 2014-06-23 09:37:30 -07:00
Alex Converse
aeacaac574 Switch active map implementation to segment based.
Change-Id: Ibb841a1fa4d08d164cf5461246ec290f582b1f80
2014-06-20 13:13:23 -07:00
Alex Converse
e8a4edf49e Fork vp9_rd_pick_inter_mode_sb_seg_skip
Change-Id: I549868725b789f0f4f89828005a65972c20df888
2014-06-20 13:13:18 -07:00
Alex Converse
173a86b2a2 Actually skip blocks in skip segments in non-rd encoder.
Copy split from macroblock to pick mode context so it doesn't get lost.

Change-Id: Ie37aa12558dbe65c4f8076cf808250fffb7f27a8
2014-06-20 11:49:02 -07:00
Tim Kopp
6d2ebfabb1 Merge "VP9 denoiser bugfixes" 2014-06-20 10:37:46 -07:00
Jingning Han
48b8ce21f0 Merge "Allow key frame more flexibility in mode search" 2014-06-20 09:38:02 -07:00
Tim Kopp
b79d5b62bd Fixed VP9 denoiser COPY_BLOCK case
Now copies the src to the correct location in the running average buffer.

Change-Id: I9c83c96dc7a97f42c8df16ab4a9f18b733181f34
2014-06-20 07:18:42 -07:00
Tim Kopp
31c03b31fe VP9 denoiser bugfixes
s/stdint.h/vpx\/vpx_int.h

Added missing 'break;'s

Also included other minor changes, mostly cosmetic.

Change-Id: I852bba3e85e794f1d4af854c45c16a23a787e6a3
2014-06-20 07:18:42 -07:00
Tim Kopp
7820abe53c Merge "Added CFLAG for outputting vp9 denoised signal" 2014-06-20 07:16:08 -07:00
Jingning Han
c99a8fd7c8 Allow key frame more flexibility in mode search
This commit allows the key frame to search through more prediction
modes and more flexible block sizes. No speed change observed. The
coding performance for rtc set is improved by 1.7% for speed -5 and
3.0% for speed -6.

Change-Id: Ifd1bc28558017851b210b4004f2d80838938bcc5
2014-06-19 14:47:12 -07:00
Tim Kopp
ab9755f3af Merge "Fixes in VP9 alloc, free, and COPY_FRAME case" 2014-06-19 12:43:00 -07:00
Jingning Han
b202e475e9 Merge "Separate rate-distortion modeling for DC and AC coefficients" 2014-06-19 11:47:55 -07:00
Tim Kopp
26955b2b6a Merge "Improved vp9 denoiser running avg update." 2014-06-19 11:22:50 -07:00
Tim Kopp
40d8a20106 Merge "Implemented COPY_BLOCK case for vp9 denoiser" 2014-06-19 11:02:46 -07:00
Yunqing Wang
8297b4d7cd Merge "Modify non-rd intra mode checking" 2014-06-19 09:30:04 -07:00
Tim Kopp
c9c4e13d09 Added CFLAG for outputting vp9 denoised signal
Change-Id: Iab9b4e11cad927f3282e486c203564e1a658f377
2014-06-19 08:41:36 -07:00
Tim Kopp
b56f3af7db Fixes in VP9 alloc, free, and COPY_FRAME case
Change-Id: I1216f17e2206ef521fe219b6d72d8e41d1ba1147
2014-06-19 08:41:36 -07:00
Tim Kopp
0fec8f9712 Improved vp9 denoiser running avg update.
Change-Id: Ie0aa41fb7957755544321897b3bb2dd92f392027
2014-06-19 08:41:36 -07:00
Tim Kopp
ff38807165 Implemented COPY_BLOCK case for vp9 denoiser
Change-Id: Ie89ad1e3aebbd474e1a0db69c1961b4d1ddcd33e
2014-06-19 08:41:36 -07:00
Tim Kopp
2614e56c58 Changed buf_2ds in vp9 denoiser to YV12 buffers
Changed alloc, free, and running average code as necessary.

Change-Id: Ifc4d9ccca462164214019963b3768a457791b9c1
2014-06-18 14:18:09 -07:00
Tim Kopp
a4b7a713a4 Update running avg for VP9 denoiser
Change-Id: I9577d648542064052795bf5770428fbd5c276b7b
2014-06-18 14:18:09 -07:00
Tim Kopp
2a72067301 Implemented vp9_denoiser_{alloc,free}()
Change-Id: I79eba79f7c52eec19ef2356278597e06620d5e27
2014-06-18 14:18:09 -07:00
Alex Converse
7557a65d16 BITSTREAM: Handle transform size and motion vectors more logically for non-420.
This breaks the profile 1 bitstream.

Don't force non420 uv transform size to 1/4 y size. In the 4:2:0 case the
chroma corresponding to a luma block is 1/4 its size. In the 4:4:4 case
chroma and luma planes are the same size. Disallowing larger transforms
can result in a loss of compression efficiency and is inconsistent.

For sub-8x8 blocks only average corresponding motion vectors.

4:2:0 and profile 0 behavior remains unchanged.

Change-Id: I560ae07183012c6734dd1860ea54ed6f62f3cae8
2014-06-18 13:07:51 -07:00
Jingning Han
3b9c19aaa7 Remove unused vp9_init_quant_tables function
This function is not effectively used, hence removed.

Change-Id: I2e8e48fa07c7518931690f3b04bae920cb360e49
2014-06-18 11:51:41 -07:00
Yunqing Wang
55834d42cc Modify non-rd intra mode checking
Speed 6 uses small tx size, namely 8x8. max_intra_bsize needs to
be modified accordingly to ensure valid intra mode checking.
Borg test on RTC set showed an overall PSNR gain of 0.335% in speed
-6.

This also changes speed -5 encoding by allowing DC_PRED checking
for block32x32. Borg test on RTC set showed a slight PSNR gain of
0.145%, and no noticeable speed change.

Change-Id: I1502978d8fbe265b3bb235db0f9c35ba0703cd45
2014-06-18 11:38:44 -07:00
Jingning Han
7c45dc98a8 Separate rate-distortion modeling for DC and AC coefficients
This is the first step to rework the rate-distortion modeling used
in rtc coding mode. The overall goal is to make the modeling
customized for the statistics encountered in the rtc coding.

This commit makes encoder to perform rate-distortion modeling for
DC and AC coefficients separately. No speed changes observed.
The coding performance for pedestrian_area_1080p is largely
improved:

speed -5, from 79558 b/f, 37.871 dB -> 79598 b/f, 38.600 dB
speed -6, from 79515 b/f, 37.822 dB -> 79544 b/f, 38.130 dB

Overall performance for rtc set at speed -6 is improved by 0.67%.

Change-Id: I9153444567e5f75ccdcaac043c2365992c005c0c
2014-06-18 10:50:38 -07:00
Dmitry Kovalev
bf46feb379 Merge "Moving RD-opt related code from vp9_encoder.h to vp9_rdopt.h." 2014-06-17 14:20:17 -07:00
Pengchong Jin
bed7cf2eeb Merge "skip the un-necessary motion search in the first pass" 2014-06-17 12:08:42 -07:00
Jingning Han
6cfb854eef Merge "Fix C versions of DC calculation functions" 2014-06-16 18:33:21 -07:00
Pengchong Jin
cdc954fdc8 skip the un-necessary motion search in the first pass
This patch allows the VP9 encoder to skip the un-necessary
motion search in the first pass. It computes the motion error
of 0,0 motion using the last source frame as the reference,
and skips the further motion search if this error is small.

Borg test shows overall the patch gives PSNR gain (derf -0.001%,
yt 0.341%, hd 0.282%). Individual clips may have PSNR gain or
loss. The best PSNR performance is 7.347% and the worst is -0.662%.
The first pass encoding speedup for slideshow clips is over 30%.

Change-Id: I4cac4dbd911f277ee858e161f3ca652c771344fe
2014-06-16 10:16:27 -07:00
unknown
45648532bc vp9_pickmode.c: fix vs12 compiler warnings
Change-Id: I5042b76a7050c121bf960ecb20c79d35adcc4cd5
2014-06-15 12:47:48 -07:00
Jingning Han
6b0bc34b62 Fix C versions of DC calculation functions
This commit fixes the scaling factors used in the C versions of the
DC calculation functions.

Change-Id: Iab41108c2bb93c2f2e78667214f3a772a2b707b5
2014-06-13 16:09:40 -07:00
Yunqing Wang
feaae409c8 Merge "Revert "skip un-neccessary motion search in the first pass"" 2014-06-13 15:21:26 -07:00
Dmitry Kovalev
3f8508eb61 Moving RD-opt related code from vp9_encoder.h to vp9_rdopt.h.
Change-Id: I8fab776c8801e19d3f5027ed55a6aa69eee951de
2014-06-13 12:34:40 -07:00
Dmitry Kovalev
bcfbd2f948 Replacing RC_MODE with vpx_rc_mode.
Both enums are identical.
Change-Id: I06653f9c90a2d3a2dd5c741e75b17ee7d066a56f
2014-06-13 12:22:35 -07:00
Paul Wilkins
af8d4054d6 Revert "skip un-neccessary motion search in the first pass"
This patch appears to have introduced non-determinism and/or
mismatch from debug vs release.

This reverts commit 5daef90efc.

Change-Id: I80081e55cfeaaa821b510b58a4e6e6328003c7da
2014-06-13 18:53:36 +01:00
Tim Kopp
123cd3a52c Merge "Added skeleton for VP9 denoiser" 2014-06-13 09:44:39 -07:00
Paul Wilkins
3082565b8d Merge "Cleaning up accumulate_frame_motion_stats()." 2014-06-13 02:27:03 -07:00
Dmitry Kovalev
7336903545 Merge "Adding MV_SPEED_FEATURES struct." 2014-06-12 17:15:33 -07:00
Tim Kopp
ab8bfb077b Added skeleton for VP9 denoiser
Change-Id: Iccf6ede4c4f85646b0f8daec47050ce93e267c90
2014-06-12 15:12:22 -07:00
Dmitry Kovalev
48f0935b81 Merge "Removing unused ssim_weighted_pred_err field from FIRSTPASS_STATS." 2014-06-12 14:16:18 -07:00
Dmitry Kovalev
4ff1a614f1 Adding MV_SPEED_FEATURES struct.
Moving all motion vector related speed parameters from SPEED_FEATURES to
MV_SPEED_FEATURES.

Change-Id: I3e9af0039c7162f8671878c5920bce3cb256a84e
2014-06-12 14:15:27 -07:00
Dmitry Kovalev
c90cd4d572 Merge "Moving full_pixel_search() to vp9_mcomp.c." 2014-06-12 14:12:45 -07:00
Dmitry Kovalev
ab449cd9ba Merge "Adding is_altref_enabled() function." 2014-06-12 13:24:42 -07:00
Dmitry Kovalev
f80a346e0e Merge "Replacing txfm_size with tx_size." 2014-06-12 13:07:11 -07:00
Dmitry Kovalev
442cbf565d Moving full_pixel_search() to vp9_mcomp.c.
Change-Id: I12389f801ebd3bd2ae3bf31e125433bfb429ee65
2014-06-12 13:06:37 -07:00
Dmitry Kovalev
86583b2bec Adding is_altref_enabled() function.
Change-Id: I54cdb4ce11590511e6f86bc2fd55771f1c18a20a
2014-06-12 12:13:20 -07:00
Jingning Han
d5ae43318e Merge "Fast computation path for forward transform and quantization" 2014-06-12 11:59:52 -07:00
Dmitry Kovalev
4345d12d28 Replacing txfm_size with tx_size.
Change-Id: Ifa6374e9db5919322733b656e0865f5f19ee6f2c
2014-06-12 11:57:26 -07:00
Dmitry Kovalev
eaeda536a4 Removing unused ssim_weighted_pred_err field from FIRSTPASS_STATS.
Change-Id: Ia8c7e3905ac21732cb6b8099eaf8df72c7e36b73
2014-06-12 11:28:54 -07:00
Jingning Han
ccba289f8d Fast computation path for forward transform and quantization
This commit enables a fast path computational flow for forward
transformation. It checks the sse and variance of prediction
residuals and decides if the quantized coefficients are all
zero, dc only, or more. It then selects the corresponding coding
path in the forward transformation and quantization stage.

It is currently enabled in rtc coding mode. Will do it for rd
coding mode next.

In speed -6, the runtime for pedestrian_area 1080p at 1000 kbps
goes down from 14234 ms to 13704 ms, i.e., about 4% speed-up.
Overall coding performance for rtc set is changed by -0.18%.

Change-Id: I0452da1786d59bc8bcbe0a35fdae9f623d1d44e1
2014-06-12 11:10:54 -07:00
Alex Converse
893433be31 Merge "Fix SEG_LVL_SKIP in non-RD inter mode selection." 2014-06-12 10:38:06 -07:00
Alex Converse
130d9ade25 Merge "Fix SEG_LVL_SKIP in RD inter mode selection." 2014-06-12 10:37:20 -07:00
Yunqing Wang
f9d1e66f6a Merge "skip un-neccessary motion search in the first pass" 2014-06-12 09:43:47 -07:00
Pengchong Jin
5daef90efc skip un-neccessary motion search in the first pass
This patch allows the encoder to skip the
un-neccessary motion search in the first pass. It
calculates the error of the zero motion vector using
the last source frame as reference and skips the
further motion search in the first pass if the error
is small.

The encoding speedup of the first pass for slideshow
videos is over 30%. Borg test shows the overall PSNR
performance remain approximately the same (derf -0.009,
hd 0.387, yt 0.021, stdhd 0.065). Individual clips may
have either PSNR gain or loss. The worst PSNR perfomance
is from yt set, with a PSNR loss of -1.1.

Change-Id: I08b2ab110b695e4689573b2567fa531b6457616e
2014-06-12 08:55:52 -07:00
Alex Converse
6c3f311ba2 Fix SEG_LVL_SKIP in non-RD inter mode selection.
Add a set_mode_info_seg_skip function that fills the requisite mode info.

Change-Id: I460b1b6845d720d9b09ed5b64df0ea0aac443f62
2014-06-11 17:53:26 -07:00
Alex Converse
b0a8057f67 Fix SEG_LVL_SKIP in RD inter mode selection.
* Only use ZEROMV, disalowing the intra modes that were previously
  tested.
* Score rate and distortion as zero.

Change-Id: Ifcf99e272095725f11da1dcd26bd0f850683e680
2014-06-11 17:52:15 -07:00
Dmitry Kovalev
e6fadb5ba8 Merge "Cleaning up vp9_variance_mmx.c." 2014-06-10 17:27:12 -07:00
Dmitry Kovalev
bc93f425d0 Removing two unused TX_SIZE_SEARCH_METHOD members.
Change-Id: I33a38bb9f46e7ef509bbbf0cfd7bc3ea5072d022
2014-06-10 11:08:30 -07:00
Dmitry Kovalev
22368479c0 Merge "Removing chessboard_index from SPEED_FEATURES." 2014-06-10 10:53:53 -07:00
Dmitry Kovalev
9636601146 Merge "Removing unused motion_vector_context enum from vp9_encodeframe.c" 2014-06-10 10:53:25 -07:00
Yunqing Wang
70eb862fd3 Merge "Use small transform size in non-rd real-time mode" 2014-06-09 13:07:24 -07:00
Dmitry Kovalev
e0c6507229 Merge "Removing unused tt_activity_measure()." 2014-06-09 10:45:56 -07:00
Yunqing Wang
b04d766800 Use small transform size in non-rd real-time mode
In non-rd real-time mode, choosing smaller transform size in
encoding gives better video quality and good speed gain than
choosing larger transform size. This patch set tx size search
method to ALLOW_8X8, which is better than using 4x4 or other
larger sizes.

Borg tests on rtc set at speed 6 showed significant gain on quality.
PSNR gain: 11.034% and SSIM gain: 15.466%.

The speed gain is 5% - 12% for <720p clips, and 2% - 7% for
720p clips.

Change-Id: If4dc74ed2df359346b059f47fb73b4a0193ec548
2014-06-09 08:26:50 -07:00
Adrian Grange
61c4295af8 Merge "Fix internal stats printing" 2014-06-09 07:13:20 -07:00
Adrian Grange
b447b9d978 Merge "Revert "Removing this_frame_stats member from TWO_PASS struct."" 2014-06-06 14:03:52 -07:00
Adrian Grange
a4f747921a Revert "Removing this_frame_stats member from TWO_PASS struct."
Use of stack frame variable "fps" beyond the lifetime of the function.

fps is sent as a paremeter to output_stats and stored in the
packet holding this encoded frame. This has scope beyond the
lifetime of the calling function.

This reverts commit 3f95a230c7

Change-Id: Icd8e14b3d7dd733590ada12e619b9dce95b6b0f5
2014-06-06 12:51:56 -07:00
Dmitry Kovalev
5f72de91a8 Merge "Adding encode_tiles() function." 2014-06-06 10:03:18 -07:00
Dmitry Kovalev
923c30a174 Removing chessboard_index from SPEED_FEATURES.
This is not a speed feature, adding inline function instead.

Change-Id: Ia48c41802eec9e92cf990339d724097279695c9a
2014-06-05 18:17:54 -07:00
Dmitry Kovalev
31403fd7d7 Adding encode_tiles() function.
Change-Id: Ib8187c8f2556e1e9268b0683cd2b6ff3489f0205
2014-06-05 18:03:40 -07:00
Deb Mukherjee
e219622b80 Fixes qindex for first frame in 1-pass cq/q modes
Produces sane qindex for the first frame in 1-pass constant and
constrained qualirty modes.

Change-Id: Ib2a5091df15a23489e9bb5534a2019cf2689755e
2014-06-05 12:29:44 -07:00