1 2021-09-27 v1.11.0 "Smew Duck"
2 This maintenance release adds support for VBR mode in VP9 rate control
3 interface, new codec controls to get quantization parameters and loop filter
4 levels, and includes several improvements to NEON and numerous bug fixes.
7 New codec control is added to get quantization parameters and loop filter
10 VBR mode is supported in VP9 rate control library.
13 Numerous improvements for Neon optimizations.
14 Code clean-up and refactoring.
15 Calculation of rd multiplier is changed with BDRATE gains.
18 Fix to overflow on duration.
19 Fix to several instances of -Wunused-but-set-variable.
20 Fix to avoid chroma resampling for 420mpeg2 input.
21 Fix to overflow in calc_iframe_target_size.
22 Fix to disallow skipping transform and quantization.
23 Fix some -Wsign-compare warnings in simple_encode.
24 Fix input file path in simple_encode_test.
25 Fix valid range for under/over_shoot pct.
27 2021-03-09 v1.10.0 "Ruddy Duck"
28 This maintenance release adds support for darwin20 and new codec controls, as
29 well as numerous bug fixes.
32 New codec control is added to disable loopfilter for VP9.
34 New encoder control is added to disable feature to increase Q on overshoot
37 Configure support for darwin20 is added.
39 New codec control is added for VP9 rate control. The control ID of this
40 interface is VP9E_SET_EXTERNAL_RATE_CONTROL. To make VP9 use a customized
41 external rate control model, users will have to implement each callback
42 function in vpx_rc_funcs_t and register them using libvpx API
43 vpx_codec_control_() with the control ID.
46 Use -std=gnu++11 instead of -std=c++11 for c++ files.
49 Override assembler with --as option of configure for MSVS.
50 Fix several compilation issues with gcc 4.8.5.
51 Fix to resetting rate control for temporal layers.
52 Fix to the rate control stats of SVC example encoder when number of spatial
54 Fix to reusing motion vectors from the base spatial layer in SVC.
55 2 pass related flags removed from SVC example encoder.
57 2020-07-29 v1.9.0 "Quacking Duck"
58 This release adds support for NV12, a separate library for rate control, as
59 well as incremental improvements.
62 NV12 support is added to this release.
63 A new interface is added for VP9 rate control. The new library libvp9rc.a
64 must be linked by applications.
65 Googletest is updated to v1.10.0.
66 simple_encode.cc is compiled into a new library libsimple_encode.a with
70 Various changes to improve VP9 SVC, rate control, quality and speed to real
74 Fix key frame update refresh simulcast flexible svc.
75 Fix to disable_16x16part speed feature for real time encoding.
76 Fix some signed integer overflows for VP9 rate control.
77 Fix initialization of delta_q_uv.
78 Fix condition in regulate_q for cyclic refresh.
79 Various fixes to dynamic resizing for VP9 SVC.
81 2019-12-09 v1.8.2 "Pekin Duck"
82 This release collects incremental improvements to many aspects of the library.
85 ARCH_* defines have been removed in favor of VPX_ARCH_*.
87 2019-07-15 v1.8.1 "Orpington Duck"
88 This release collects incremental improvements to many aspects of the library.
91 VP8E_SET_CPUUSED now accepts values up to 9 for vp9.
92 VPX_CTRL_VP9E_SET_MAX_INTER_BITRATE_PCT had a spelling fix (was VP8E).
93 The --sdk-path option has been removed. If you were using it to build for
94 Android please read build/make/Android.mk for alternatives.
95 All PPC optimizations have been disabled:
96 https://bugs.chromium.org/p/webm/issues/detail?id=1522.
99 Various changes to improve encoder rate control, quality and speed
100 for practically every use case.
103 vp9-rtc: Fix color artifacts for speed >= 8.
105 2019-01-31 v1.8.0 "Northern Shoveler Duck"
106 This release focused on encoding performance for realtime and VOD use cases.
109 This adds and improves several vp9 controls. Most are related to SVC:
110 VP9E_SET_SVC_FRAME_DROP_LAYER:
111 - Frame dropping in SVC.
112 VP9E_SET_SVC_INTER_LAYER_PRED:
113 - Inter-layer prediction in SVC.
114 VP9E_SET_SVC_GF_TEMPORAL_REF:
115 - Enable long term temporal reference in SVC.
116 VP9E_SET_SVC_REF_FRAME_CONFIG/VP9E_GET_SVC_REF_FRAME_CONFIG:
117 - Extend and improve this control for better flexibility in setting SVC
119 VP9E_SET_POSTENCODE_DROP:
120 - Allow for post-encode frame dropping (applies to non-SVC too).
121 VP9E_SET_SVC_SPATIAL_LAYER_SYNC:
122 - Enable spatial layer sync frames.
123 VP9E_SET_SVC_LAYER_ID:
124 - Extend api to specify temporal id for each spatial layers.
126 - Extend Region of Interest functionality to VP9.
129 2 pass vp9 encoding has improved substantially. When using --auto-alt-ref=6,
130 we see approximately 8% for VBR and 10% for CQ. When using --auto-alt-ref=1,
131 the gains are approximately 4% for VBR and 5% for CQ.
133 For real-time encoding, speed 7 has improved by ~5-10%. Encodes targeted at
134 screen sharing have improved when the content changes significantly (slide
135 sharing) or scrolls. There is a new speed 9 setting for mobile devices which
136 is about 10-20% faster than speed 8.
140 VP9 partition issue for 1080p.
141 VP9 rate control improvments.
142 Postprocessing Multi Frame Quality Enhancement (MFQE) issue.
143 VP8 multithread decoder issues.
144 A variety of fuzzing issues.
146 2018-01-04 v1.7.0 "Mandarin Duck"
147 This release focused on high bit depth performance (10/12 bit) and vp9
148 encoding improvements.
151 This release is ABI incompatible due to new vp9 encoder features.
153 Frame parallel decoding for vp9 has been removed.
156 vp9 encoding supports additional threads with --row-mt. This can be greater
157 than the number of tiles.
159 Two new vp9 encoder options have been added:
163 Additional tooling for respecting the vp9 "level" profiles has been added.
166 A variety of fuzzing issues.
167 vp8 threading fix for ARM.
168 Codec control VP9_SET_SKIP_LOOP_FILTER fixed.
169 Reject invalid multi resolution configurations.
171 2017-01-09 v1.6.1 "Long Tailed Duck"
172 This release improves upon the VP9 encoder and speeds up the encoding and
176 This release is ABI compatible with 1.6.0.
179 Faster VP9 encoding and decoding.
180 High bit depth builds now provide similar speed for 8 bit encode and decode
181 for x86 targets. Other platforms and higher bit depth improvements are in
185 A variety of fuzzing issues.
187 2016-07-20 v1.6.0 "Khaki Campbell Duck"
188 This release improves upon the VP9 encoder and speeds up the encoding and
192 This release is ABI incompatible with 1.5.0 due to a new 'color_range' enum
193 in vpx_image and some minor changes to the VP8_COMP structure.
195 The default key frame interval for VP9 has changed from 128 to 9999.
198 A core focus has been performance for low end Intel processors. SSSE3
199 instructions such as 'pshufb' have been avoided and instructions have been
200 reordered to better accommodate the more constrained pipelines.
202 As a result, devices based on Celeron processors have seen substantial
203 decoding improvements. From Indian Runner Duck to Javan Whistling Duck,
204 decoding speed improved between 10 and 30%. Between Javan Whistling Duck
205 and Khaki Campbell Duck, it improved another 10 to 15%.
207 While Celeron benefited most, Core-i5 also improved 5% and 10% between the
210 Realtime performance for WebRTC for both speed and quality has received a
214 A number of fuzzing issues, found variously by Mozilla, Chromium and others,
215 have been fixed and we strongly recommend updating.
217 2015-11-09 v1.5.0 "Javan Whistling Duck"
218 This release improves upon the VP9 encoder and speeds up the encoding and
222 This release is ABI incompatible with 1.4.0. It drops deprecated VP8
223 controls and adds a variety of VP9 controls for testing.
225 The vpxenc utility now prefers VP9 by default.
228 Faster VP9 encoding and decoding
229 Smaller library size by combining functions used by VP8 and VP9
232 A variety of fuzzing issues
234 2015-04-03 v1.4.0 "Indian Runner Duck"
235 This release includes significant improvements to the VP9 codec.
238 This release is ABI incompatible with 1.3.0. It drops the compatibility
239 layer, requiring VPX_IMG_FMT_* instead of IMG_FMT_*, and adds several codec
243 Faster VP9 encoding and decoding
244 Multithreaded VP9 decoding (tile and frame-based)
245 Multithreaded VP9 encoding - on by default
246 YUV 4:2:2 and 4:4:4 support in VP9
247 10 and 12bit support in VP9
248 64bit ARM support by replacing ARM assembly with intrinsics
251 Fixes a VP9 bitstream issue in Profile 1. This only affected non-YUV 4:2:0
255 Frame Parallel decoding fails for segmented and non-420 files.
257 2013-11-15 v1.3.0 "Forest"
258 This release introduces the VP9 codec in a backward-compatible way.
259 All existing users of VP8 can continue to use the library without
260 modification. However, some VP8 options do not map to VP9 in the same manner.
262 The VP9 encoder in this release is not feature complete. Users interested in
263 the encoder are advised to use the git master branch and discuss issues on
264 libvpx mailing lists.
267 This release is ABI and API compatible with Duclair (v1.0.0). Users
268 of older releases should refer to the Upgrading notes in this document
272 Get rid of bashisms in the main build scripts
273 Added usage info on command line options
274 Add lossless compression mode
276 Add additional Mac OS X targets: 10.7, 10.8 and 10.9 (darwin11-13)
277 Add option to disable documentation
278 configure: add --enable-external-build support
279 make: support V=1 as short form of verbose=yes
280 configure: support mingw-w64
281 configure: support hardfloat armv7 CHOSTS
282 configure: add support for android x86
283 Add estimated completion time to vpxenc
284 Don't exit on decode errors in vpxenc
285 vpxenc: support scaling prior to encoding
286 vpxdec: support scaling output
287 vpxenc: improve progress indicators with --skip
288 msvs: Don't link to winmm.lib
289 Add a new script for producing vcxproj files
290 Produce Visual Studio 10 and 11 project files
291 Produce Windows Phone project files
292 msvs-build: use msbuild for vs >= 2005
293 configure: default configure log to config.log
294 Add encoding option --static-thresh
297 Miscellaneous speed optimizations for VP8 and VP9.
300 In general, quality is consistent with the Eider release.
303 This release represents approximately a year of engineering effort,
304 and contains multiple bug fixes. Please refer to git history for details.
308 This release acts as a checkpoint for a large amount of internal refactoring
309 and testing. It also contains a number of small bugfixes, so all users are
310 encouraged to upgrade.
313 This release is ABI and API compatible with Duclair (v1.0.0). Users
314 of older releases should refer to the Upgrading notes in this
315 document for that release.
318 VP8 optimizations for MIPS dspr2
319 vpxenc: add -quiet option
322 Encoder and decoder speed is consistent with the Eider release.
325 In general, quality is consistent with the Eider release.
327 Minor tweaks to ARNR filtering
328 Minor improvements to real time encoding with multiple temporal layers
331 Fixes multithreaded encoder race condition in loopfilter
332 Fixes multi-resolution threaded encoding
333 Fix potential encoder dead-lock after picture resize
336 2012-05-09 v1.1.0 "Eider"
337 This introduces a number of enhancements, mostly focused on real-time
338 encoding. In addition, it fixes a decoder bug (first introduced in
339 Duclair) so all users of that release are encouraged to upgrade.
342 This release is ABI and API compatible with Duclair (v1.0.0). Users
343 of older releases should refer to the Upgrading notes in this
344 document for that release.
346 This release introduces a new temporal denoiser, controlled by the
347 VP8E_SET_NOISE_SENSITIVITY control. The temporal denoiser does not
348 currently take a strength parameter, so the control is effectively
349 a boolean - zero (off) or non-zero (on). For compatibility with
350 existing applications, the values accepted are the same as those
351 for the spatial denoiser (0-6). The temporal denoiser is enabled
352 by default, and the older spatial denoiser may be restored by
353 configuring with --disable-temporal-denoising. The temporal denoiser
354 is more computationally intensive than the spatial one.
356 This release removes support for a legacy, decode only API that was
357 supported, but deprecated, at the initial release of libvpx
358 (v0.9.0). This is not expected to have any impact. If you are
359 impacted, you can apply a reversion to commit 2bf8fb58 locally.
360 Please update to the latest libvpx API if you are affected.
363 Adds a motion compensated temporal denoiser to the encoder, which
364 gives higher quality than the older spatial denoiser. (See above
365 for notes on upgrading).
367 In addition, support for new compilers and platforms were added,
369 improved support for XCode
370 Android x86 NDK build
374 Changing resolution with vpx_codec_enc_config_set() is now
375 supported. Previously, reinitializing the codec was required to
376 change the input resolution.
378 The vpxenc application has initial support for producing multiple
379 encodes from the same input in one call. Resizing is not yet
380 supported, but varying other codec parameters is. Use -- to
381 delineate output streams. Options persist from one stream to the
384 Also, the vpxenc application will now use a keyframe interval of
385 5 seconds by default. Use the --kf-max-dist option to override.
388 Decoder performance improved 2.5% versus Duclair. Encoder speed is
389 consistent with Duclair for most material. Two pass encoding of
390 slideshow-like material will see significant improvements.
392 Large realtime encoding speed gains at a small quality expense are
393 possible by configuring the on-the-fly bitpacking experiment with
394 --enable-onthefly-bitpacking. Realtime encoder can be up to 13%
395 faster (ARM) depending on the number of threads and bitrate
396 settings. This technique sees constant gain over the 5-16 speed
397 range. For VC style input the loss seen is up to 0.2dB. See commit
398 52cf4dca for further details.
401 On the whole, quality is consistent with the Duclair release. Some
404 Reduced blockiness in easy sections by applying a penalty to
407 Improved quality of static sections (like slideshows) with
410 Improved keyframe sizing with multiple temporal layers
413 Corrected alt-ref contribution to frame rate for visible updates
414 to the alt-ref buffer. This affected applications making manual
415 usage of the frame reference flags, or temporal layers.
417 Additional constraints were added to disable multi-frame quality
418 enhancement (MFQE) in sections of the frame where there is motion.
421 Fixed corruption issues when vpx_codec_enc_config_set() was called
422 with spatial resampling enabled.
424 Fixed a decoder error introduced in Duclair where the segmentation
425 map was not being reinitialized on keyframes (#378)
428 2012-01-27 v1.0.0 "Duclair"
429 Our fourth named release, focused on performance and features related to
430 real-time encoding. It also fixes a decoder crash bug introduced in
431 v0.9.7, so all users of that release are encouraged to upgrade.
434 This release is ABI incompatible with prior releases of libvpx, so the
435 "major" version number has been bumped to 1. You must recompile your
436 applications against the latest version of the libvpx headers. The
437 API remains compatible, and this should not require code changes in most
441 This release introduces several substantial new features to the encoder,
442 of particular interest to real time streaming applications.
444 Temporal scalability allows the encoder to produce a stream that can
445 be decimated to different frame rates, with independent rate targeting
448 Multiframe quality enhancement postprocessing can make visual quality
449 more consistent in the presence of frames that are substantially
450 different quality than the surrounding frames, as in the temporal
451 scalability case and in some forced keyframe scenarios.
453 Multiple-resolution encoding support allows the encoding of the
454 same content at different resolutions faster than encoding them
458 Optimization targets for this release included the decoder and the real-
459 time modes of the encoder. Decoder speed on x86 has improved 10.5% with
460 this release. Encoder improvements followed a curve where speeds 1-3
461 improved 4.0%-1.5%, speeds 4-8 improved <1%, and speeds 9-16 improved
462 1.5% to 10.5%, respectively. "Best" mode speed is consistent with the
466 Encoder quality in the single stream case is consistent with the Cayuga
470 This release fixes an OOB read decoder crash bug present in v0.9.7
471 related to the clamping of motion vectors in SPLITMV blocks. This
472 behavior could be triggered by corrupt input or by starting
473 decoding from a P-frame.
476 2011-08-15 v0.9.7-p1 "Cayuga" patch 1
477 This is an incremental bugfix release against Cayuga. All users of that
478 release are strongly encouraged to upgrade.
480 - Fix potential OOB reads (cdae03a)
482 An unbounded out of bounds read was discovered when the
483 decoder was requested to perform error concealment (new in
484 Cayuga) given a frame with corrupt partition sizes.
486 A bounded out of bounds read was discovered affecting all
487 versions of libvpx. Given an multipartition input frame that
488 is truncated between the mode/mv partition and the first
489 residiual paritition (in the block of partition offsets), up
490 to 3 extra bytes could have been read from the source buffer.
491 The code will not take any action regardless of the contents
492 of these undefined bytes, as the truncated buffer is detected
493 immediately following the read based on the calculated
494 starting position of the coefficient partition.
496 - Fix potential error concealment crash when the very first frame
497 is missing or corrupt (a609be5)
499 - Fix significant artifacts in error concealment (a4c2211, 99d870a)
501 - Revert 1-pass CBR rate control changes (e961317)
502 Further testing showed this change produced undesirable visual
503 artifacts, rolling back for now.
506 2011-08-02 v0.9.7 "Cayuga"
507 Our third named release, focused on a faster, higher quality, encoder.
510 This release is backwards compatible with Aylesbury (v0.9.5) and
511 Bali (v0.9.6). Users of older releases should refer to the Upgrading
512 notes in this document for that release.
515 Stereo 3D format support for vpxenc
516 Runtime detection of available processor cores.
517 Allow specifying --end-usage by enum name
518 vpxdec: test for frame corruption
519 vpxenc: add quantizer histogram display
520 vpxenc: add rate histogram display
521 Set VPX_FRAME_IS_DROPPABLE
522 update configure for ios sdk 4.3
523 Avoid text relocations in ARM vp8 decoder
524 Generate a vpx.pc file for pkg-config.
525 New ways of passing encoded data between encoder and decoder.
528 This release includes across-the-board speed improvements to the
529 encoder. On x86, these measure at approximately 11.5% in Best mode,
530 21.5% in Good mode (speed 0), and 22.5% in Realtime mode (speed 6).
531 On ARM Cortex A9 with Neon extensions, real-time encoding of video
532 telephony content is 35% faster than Bali on single core and 48%
533 faster on multi-core. On the NVidia Tegra2 platform, real time
534 encoding is 40% faster than Bali.
536 Decoder speed was not a priority for this release, but improved
537 approximately 8.4% on x86.
539 Reduce motion vector search on alt-ref frame.
540 Encoder loopfilter running in its own thread
541 Reworked loopfilter to precalculate more parameters
542 SSE2/SSSE3 optimizations for build_predictors_mbuv{,_s}().
543 Make hor UV predict ~2x faster (73 vs 132 cycles) using SSSE3.
544 Removed redundant checks
545 Reduced structure sizes
546 utilize preload in ARMv6 MC/LPF/Copy routines
547 ARM optimized quantization, dfct, variance, subtract
548 Increase chrow row alignment to 16 bytes.
549 disable trellis optimization for first pass
550 Write SSSE3 sub-pixel filter function
551 Improve SSE2 half-pixel filter funtions
552 Add vp8_sub_pixel_variance16x8_ssse3 function
553 Reduce unnecessary distortion computation
554 Use diamond search to replace full search
555 Preload reference area in sub-pixel motion search (real-time mode)
558 This release focused primarily on one-pass use cases, including
559 video conferencing. Low latency data rate control was significantly
560 improved, improving streamability over bandwidth constrained links.
561 Added support for error concealment, allowing frames to maintain
562 visual quality in the presence of substantial packet loss.
564 Add rc_max_intra_bitrate_pct control
565 Limit size of initial keyframe in one-pass.
566 Improve framerate adaptation
567 Improved 1-pass CBR rate control
568 Improved KF insertion after fades to still.
569 Improved key frame detection.
570 Improved activity masking (lower PSNR impact for same SSIM boost)
571 Improved interaction between GF and ARFs
572 Adding error-concealment to the decoder.
573 Adding support for independent partitions
574 Adjusted rate-distortion constants
578 Removed firstpass motion map
579 Fix parallel make install
580 Fix multithreaded encoding for 1 MB wide frame
581 Fixed iwalsh_neon build problems with RVDS4.1
582 Fix semaphore emulation, spin-wait intrinsics on Windows
583 Fix build with xcode4 and simplify GLOBAL.
584 Mark ARM asm objects as allowing a non-executable stack.
585 Fix vpxenc encoding incorrect webm file header on big endian
588 2011-03-07 v0.9.6 "Bali"
589 Our second named release, focused on a faster, higher quality, encoder.
592 This release is backwards compatible with Aylesbury (v0.9.5). Users
593 of older releases should refer to the Upgrading notes in this
594 document for that release.
597 vpxenc --psnr shows a summary when encode completes
598 --tune=ssim option to enable activity masking
599 improved postproc visualizations for development
600 updated support for Apple iOS to SDK 4.2
601 query decoder to determine which reference frames were updated
602 implemented error tracking in the decoder
603 fix pipe support on windows
606 Primary focus was on good quality mode, speed 0. Average improvement
607 on x86 about 40%, up to 100% on user-generated content at that speed.
608 Best quality mode speed improved 35%, and realtime speed 10-20%. This
609 release also saw significant improvement in realtime encoding speed
612 Improved encoder threading
613 Dont pick encoder filter level when loopfilter is disabled.
614 Avoid double copying of key frames into alt and golden buffer
616 x86 sse2 temporal filter
617 SSSE3 version of fast quantizer
618 vp8_rd_pick_best_mbsegmentation code restructure
619 Adjusted breakout RD for SPLITMV
620 Changed segmentation check order
621 Improved rd_pick_intra4x4block
622 Adds armv6 optimized variance calculation
623 ARMv6 optimized sad16x16
624 ARMv6 optimized half pixel variance calculations
625 Full search SAD function optimization in SSE4.1
626 Improve MV prediction accuracy to achieve performance gain
627 Improve MV prediction in vp8_pick_inter_mode() for speed>3
630 Best quality mode improved PSNR 6.3%, and SSIM 6.1%. This release
631 also includes support for "activity masking," which greatly improves
632 SSIM at the expense of PSNR. For now, this feature is available with
633 the --tune=ssim option. Further experimentation in this area
634 is ongoing. This release also introduces a new rate control mode
635 called "CQ," which changes the allocation of bits within a clip to
636 the sections where they will have the most visual impact.
638 Tuning for the more exact quantizer.
639 Relax rate control for last few frames
641 Limit key frame quantizer for forced key frames.
643 Add simple version of activity masking.
644 make rdmult adaptive for intra in quantizer RDO
645 cap the best quantizer for 2nd order DC
646 change the threshold of DC check for encode breakout
649 Fix crash on Sparc Solaris.
650 Fix counter of fixed keyframe distance
651 ARNR filter pointer update bug fix
652 Fixed use of motion percentage in KF/GF group calc
653 Changed condition for using RD in Intra Mode
654 Fix encoder real-time only configuration.
655 Fix ARM encoder crash with multiple token partitions
656 Fixed bug first cluster timecode of webm file is wrong.
657 Fixed various encoder bugs with odd-sized images
658 vp8e_get_preview fixed when spatial resampling enabled
659 quantizer: fix assertion in fast quantizer path
660 Allocate source buffers to be multiples of 16
661 Fix for manual Golden frame frequency
662 Fix drastic undershoot in long form content
665 2010-10-28 v0.9.5 "Aylesbury"
666 Our first named release, focused on a faster decoder, and a better encoder.
669 This release incorporates backwards-incompatible changes to the
670 ivfenc and ivfdec tools. These tools are now called vpxenc and vpxdec.
673 * the -q (quiet) option has been removed, and replaced with
674 -v (verbose). the output is quiet by default. Use -v to see
675 the version number of the binary.
677 * The default behavior is now to write output to a single file
678 instead of individual frames. The -y option has been removed.
679 Y4M output is the default.
681 * For raw I420/YV12 output instead of Y4M, the --i420 or --yv12
682 options must be specified.
684 $ ivfdec -o OUTPUT INPUT
685 $ vpxdec --i420 -o OUTPUT INPUT
687 * If an output file is not specified, the default is to write
688 Y4M to stdout. This makes piping more natural.
690 $ ivfdec -y -o - INPUT | ...
693 * The output file has additional flexibility for formatting the
694 filename. It supports escape characters for constructing a
695 filename from the width, height, and sequence number. This
696 replaces the -p option. To get the equivalent:
698 $ ivfdec -p frame INPUT
699 $ vpxdec --i420 -o frame-%wx%h-%4.i420 INPUT
702 * The output file must be specified with -o, rather than as the
705 $ ivfenc <options> INPUT OUTPUT
706 $ vpxenc <options> -o OUTPUT INPUT
708 * The output defaults to webm. To get IVF output, use the --ivf
711 $ ivfenc <options> INPUT OUTPUT.ivf
712 $ vpxenc <options> -o OUTPUT.ivf --ivf INPUT
716 ivfenc and ivfdec have been renamed to vpxenc, vpxdec.
717 vpxdec supports .webm input
718 vpxdec writes .y4m by default
719 vpxenc writes .webm output by default
720 vpxenc --psnr now shows the average/overall PSNR at the end
721 ARM platforms now support runtime cpu detection
722 vpxdec visualizations added for motion vectors, block modes, references
723 vpxdec now silent by default
724 vpxdec --progress shows frame-by-frame timing information
725 vpxenc supports the distinction between --fps and --timebase
726 NASM is now a supported assembler
727 configure: enable PIC for shared libs by default
728 configure: add --enable-small
729 configure: support for ppc32-linux-gcc
730 configure: support for sparc-solaris-gcc
733 Improve handling of invalid frames
734 Fix valgrind errors in the NEON loop filters.
735 Fix loopfilter delta zero transitions
736 Fix valgrind errors in vp8_sixtap_predict8x4_armv6().
737 Build fixes for darwin-icc
740 20-40% (average 28%) improvement in libvpx decoder speed,
742 Rewrite vp8_short_walsh4x4_sse2()
743 Optimizations on the loopfilters.
744 Miscellaneous improvements for Atom
745 Add 4-tap version of 2nd-pass ARMv6 MC filter.
746 Improved multithread utilization
747 Better instruction choices on x86
748 reorder data to use wider instructions
749 Update NEON wide idcts
750 Make block access to frame buffer sequential
751 Improved subset block search
752 Bilinear subpixel optimizations for ssse3.
753 Decrease memory footprint
755 Encoder speed improvements (percentage gain not measured):
756 Skip unnecessary search of identical frames
757 Add SSE2 subtract functions
758 Improve bounds checking in vp8_diamond_search_sadx4()
759 Added vp8_fast_quantize_b_sse2
762 Over 7% overall PSNR improvement (6.3% SSIM) in "best" quality
763 encoding mode, and up to 60% improvement on very noisy, still
764 or slow moving source video
766 Motion compensated temporal filter for Alt-Ref Noise Reduction
767 Improved use of trellis quantization on 2nd order Y blocks
768 Tune effect of motion on KF/GF boost in two pass
769 Allow coefficient optimization for good quality speed 0.
770 Improved control of active min quantizer for two pass.
771 Enable ARFs for non-lagged compress
775 Disable frame dropping by default
776 Improved multithreaded performance
777 Improved Force Key Frame Behaviour
778 Increased rate control buffer level precision
779 Fix bug in 1st pass motion compensation
780 ivfenc: correct fixed kf interval, --disable-kf
782 Changed above and left context data layout
783 Rework idct calling structure.
784 Removed unnecessary MB_MODE_INFO copies
785 x86: SSSE3 sixtap prediction
786 Reworked IDCT to include reconstruction (add) step
787 Swap alt/gold/new/last frame buffer ptrs instead of copying.
788 Improve SSE2 loopfilter functions
789 Change bitreader to use a larger window.
790 Avoid loopfilter reinitialization when possible
792 Normalize quantizer's zero bin and rounding factors
793 Add trellis quantization.
794 Make the quantizer exact.
795 Updates to ARNR filtering algorithm
796 Fix breakout thresh computation for golden & AltRef frames
797 Redo the forward 4x4 dct
798 Improve the accuracy of forward walsh-hadamard transform
799 Further adjustment of RD behaviour with Q and Zbin.
801 Allow linking of libs built with MinGW to MSVC
802 Fix target auto-detection on mingw32
803 Allow --cpu= to work for x86.
804 configure: pass original arguments through to make dist
805 Fix builds without runtime CPU detection
806 msvs: fix install of codec sources
807 msvs: Change devenv.com command line for better msys support
808 msvs: Add vs9 targets.
809 Add x86_64-linux-icc target
811 Potential crashes on older MinGW builds
812 Fix two-pass framrate for Y4M input.
813 Fixed simple loop filter, other crashes on ARM v6
814 arm: fix missing dependency with --enable-shared
815 configure: support directories containing .o
816 Replace pinsrw (SSE) with MMX instructions
817 apple: include proper mach primatives
818 Fixed rate control bug with long key frame interval.
819 Fix DSO link errors on x86-64 when not using a version script
820 Fixed buffer selection for UV in AltRef filtering
825 * ivfenc/ivfdec now support YUV4MPEG2 input and pipe I/O
826 * Speed optimizations
829 * Prevent out-of-bounds accesses on invalid data
830 - Build system updates:
831 * Detect toolchain to be used automatically for native builds
832 * Support building shared libraries
833 * Better autotools emulation (--prefix, --libdir, DESTDIR)
835 * http://webmproject.blogspot.com/2010/06/changes-to-webm-open-source-license.html
839 - Initial open source release. Welcome to WebM and VP8!