1 2022-06-17 v1.12.0 "Torrent Duck"
2 This release adds optimizations for Loongarch, adds support for vp8 in the
3 real-time rate control library, upgrades GoogleTest to v1.11.0, updates
4 libwebm to libwebm-1.0.0.28-20-g206d268, and includes numerous bug fixes.
7 This release is ABI compatible with the previous release.
9 vp8 support in the real-time rate control library.
10 New codec control VP8E_SET_RTC_EXTERNAL_RATECTRL is added.
12 Configure support for darwin21 is added.
14 GoogleTest is upgraded to v1.11.0.
16 libwebm is updated to libwebm-1.0.0.28-20-g206d268.
18 Allow SimpleEncode environment to take target level as input to match
19 the level conformance in vp9.
22 Numerous improvements on checking memory allocations.
23 Optimizations for Loongarch.
27 Fix to a crash related to {vp8/vp9}_set_roi_map.
28 Fix to compiling failure with -Wformat-nonliteral.
29 Fix to integer overflow with vp9 with high resolution content.
30 Fix to AddNoiseTest failure with ARMv7.
31 Fix to libvpx Null-dereference READ in vp8.
33 2021-09-27 v1.11.0 "Smew Duck"
34 This maintenance release adds support for VBR mode in VP9 rate control
35 interface, new codec controls to get quantization parameters and loop filter
36 levels, and includes several improvements to NEON and numerous bug fixes.
39 New codec control is added to get quantization parameters and loop filter
42 VBR mode is supported in VP9 rate control library.
45 Numerous improvements for Neon optimizations.
46 Code clean-up and refactoring.
47 Calculation of rd multiplier is changed with BDRATE gains.
50 Fix to overflow on duration.
51 Fix to several instances of -Wunused-but-set-variable.
52 Fix to avoid chroma resampling for 420mpeg2 input.
53 Fix to overflow in calc_iframe_target_size.
54 Fix to disallow skipping transform and quantization.
55 Fix some -Wsign-compare warnings in simple_encode.
56 Fix input file path in simple_encode_test.
57 Fix valid range for under/over_shoot pct.
59 2021-03-09 v1.10.0 "Ruddy Duck"
60 This maintenance release adds support for darwin20 and new codec controls, as
61 well as numerous bug fixes.
64 New codec control is added to disable loopfilter for VP9.
66 New encoder control is added to disable feature to increase Q on overshoot
69 Configure support for darwin20 is added.
71 New codec control is added for VP9 rate control. The control ID of this
72 interface is VP9E_SET_EXTERNAL_RATE_CONTROL. To make VP9 use a customized
73 external rate control model, users will have to implement each callback
74 function in vpx_rc_funcs_t and register them using libvpx API
75 vpx_codec_control_() with the control ID.
78 Use -std=gnu++11 instead of -std=c++11 for c++ files.
81 Override assembler with --as option of configure for MSVS.
82 Fix several compilation issues with gcc 4.8.5.
83 Fix to resetting rate control for temporal layers.
84 Fix to the rate control stats of SVC example encoder when number of spatial
86 Fix to reusing motion vectors from the base spatial layer in SVC.
87 2 pass related flags removed from SVC example encoder.
89 2020-07-29 v1.9.0 "Quacking Duck"
90 This release adds support for NV12, a separate library for rate control, as
91 well as incremental improvements.
94 NV12 support is added to this release.
95 A new interface is added for VP9 rate control. The new library libvp9rc.a
96 must be linked by applications.
97 Googletest is updated to v1.10.0.
98 simple_encode.cc is compiled into a new library libsimple_encode.a with
102 Various changes to improve VP9 SVC, rate control, quality and speed to real
106 Fix key frame update refresh simulcast flexible svc.
107 Fix to disable_16x16part speed feature for real time encoding.
108 Fix some signed integer overflows for VP9 rate control.
109 Fix initialization of delta_q_uv.
110 Fix condition in regulate_q for cyclic refresh.
111 Various fixes to dynamic resizing for VP9 SVC.
113 2019-12-09 v1.8.2 "Pekin Duck"
114 This release collects incremental improvements to many aspects of the library.
117 ARCH_* defines have been removed in favor of VPX_ARCH_*.
119 2019-07-15 v1.8.1 "Orpington Duck"
120 This release collects incremental improvements to many aspects of the library.
123 VP8E_SET_CPUUSED now accepts values up to 9 for vp9.
124 VPX_CTRL_VP9E_SET_MAX_INTER_BITRATE_PCT had a spelling fix (was VP8E).
125 The --sdk-path option has been removed. If you were using it to build for
126 Android please read build/make/Android.mk for alternatives.
127 All PPC optimizations have been disabled:
128 https://bugs.chromium.org/p/webm/issues/detail?id=1522.
131 Various changes to improve encoder rate control, quality and speed
132 for practically every use case.
135 vp9-rtc: Fix color artifacts for speed >= 8.
137 2019-01-31 v1.8.0 "Northern Shoveler Duck"
138 This release focused on encoding performance for realtime and VOD use cases.
141 This adds and improves several vp9 controls. Most are related to SVC:
142 VP9E_SET_SVC_FRAME_DROP_LAYER:
143 - Frame dropping in SVC.
144 VP9E_SET_SVC_INTER_LAYER_PRED:
145 - Inter-layer prediction in SVC.
146 VP9E_SET_SVC_GF_TEMPORAL_REF:
147 - Enable long term temporal reference in SVC.
148 VP9E_SET_SVC_REF_FRAME_CONFIG/VP9E_GET_SVC_REF_FRAME_CONFIG:
149 - Extend and improve this control for better flexibility in setting SVC
151 VP9E_SET_POSTENCODE_DROP:
152 - Allow for post-encode frame dropping (applies to non-SVC too).
153 VP9E_SET_SVC_SPATIAL_LAYER_SYNC:
154 - Enable spatial layer sync frames.
155 VP9E_SET_SVC_LAYER_ID:
156 - Extend api to specify temporal id for each spatial layers.
158 - Extend Region of Interest functionality to VP9.
161 2 pass vp9 encoding has improved substantially. When using --auto-alt-ref=6,
162 we see approximately 8% for VBR and 10% for CQ. When using --auto-alt-ref=1,
163 the gains are approximately 4% for VBR and 5% for CQ.
165 For real-time encoding, speed 7 has improved by ~5-10%. Encodes targeted at
166 screen sharing have improved when the content changes significantly (slide
167 sharing) or scrolls. There is a new speed 9 setting for mobile devices which
168 is about 10-20% faster than speed 8.
172 VP9 partition issue for 1080p.
173 VP9 rate control improvments.
174 Postprocessing Multi Frame Quality Enhancement (MFQE) issue.
175 VP8 multithread decoder issues.
176 A variety of fuzzing issues.
178 2018-01-04 v1.7.0 "Mandarin Duck"
179 This release focused on high bit depth performance (10/12 bit) and vp9
180 encoding improvements.
183 This release is ABI incompatible due to new vp9 encoder features.
185 Frame parallel decoding for vp9 has been removed.
188 vp9 encoding supports additional threads with --row-mt. This can be greater
189 than the number of tiles.
191 Two new vp9 encoder options have been added:
195 Additional tooling for respecting the vp9 "level" profiles has been added.
198 A variety of fuzzing issues.
199 vp8 threading fix for ARM.
200 Codec control VP9_SET_SKIP_LOOP_FILTER fixed.
201 Reject invalid multi resolution configurations.
203 2017-01-09 v1.6.1 "Long Tailed Duck"
204 This release improves upon the VP9 encoder and speeds up the encoding and
208 This release is ABI compatible with 1.6.0.
211 Faster VP9 encoding and decoding.
212 High bit depth builds now provide similar speed for 8 bit encode and decode
213 for x86 targets. Other platforms and higher bit depth improvements are in
217 A variety of fuzzing issues.
219 2016-07-20 v1.6.0 "Khaki Campbell Duck"
220 This release improves upon the VP9 encoder and speeds up the encoding and
224 This release is ABI incompatible with 1.5.0 due to a new 'color_range' enum
225 in vpx_image and some minor changes to the VP8_COMP structure.
227 The default key frame interval for VP9 has changed from 128 to 9999.
230 A core focus has been performance for low end Intel processors. SSSE3
231 instructions such as 'pshufb' have been avoided and instructions have been
232 reordered to better accommodate the more constrained pipelines.
234 As a result, devices based on Celeron processors have seen substantial
235 decoding improvements. From Indian Runner Duck to Javan Whistling Duck,
236 decoding speed improved between 10 and 30%. Between Javan Whistling Duck
237 and Khaki Campbell Duck, it improved another 10 to 15%.
239 While Celeron benefited most, Core-i5 also improved 5% and 10% between the
242 Realtime performance for WebRTC for both speed and quality has received a
246 A number of fuzzing issues, found variously by Mozilla, Chromium and others,
247 have been fixed and we strongly recommend updating.
249 2015-11-09 v1.5.0 "Javan Whistling Duck"
250 This release improves upon the VP9 encoder and speeds up the encoding and
254 This release is ABI incompatible with 1.4.0. It drops deprecated VP8
255 controls and adds a variety of VP9 controls for testing.
257 The vpxenc utility now prefers VP9 by default.
260 Faster VP9 encoding and decoding
261 Smaller library size by combining functions used by VP8 and VP9
264 A variety of fuzzing issues
266 2015-04-03 v1.4.0 "Indian Runner Duck"
267 This release includes significant improvements to the VP9 codec.
270 This release is ABI incompatible with 1.3.0. It drops the compatibility
271 layer, requiring VPX_IMG_FMT_* instead of IMG_FMT_*, and adds several codec
275 Faster VP9 encoding and decoding
276 Multithreaded VP9 decoding (tile and frame-based)
277 Multithreaded VP9 encoding - on by default
278 YUV 4:2:2 and 4:4:4 support in VP9
279 10 and 12bit support in VP9
280 64bit ARM support by replacing ARM assembly with intrinsics
283 Fixes a VP9 bitstream issue in Profile 1. This only affected non-YUV 4:2:0
287 Frame Parallel decoding fails for segmented and non-420 files.
289 2013-11-15 v1.3.0 "Forest"
290 This release introduces the VP9 codec in a backward-compatible way.
291 All existing users of VP8 can continue to use the library without
292 modification. However, some VP8 options do not map to VP9 in the same manner.
294 The VP9 encoder in this release is not feature complete. Users interested in
295 the encoder are advised to use the git master branch and discuss issues on
296 libvpx mailing lists.
299 This release is ABI and API compatible with Duclair (v1.0.0). Users
300 of older releases should refer to the Upgrading notes in this document
304 Get rid of bashisms in the main build scripts
305 Added usage info on command line options
306 Add lossless compression mode
308 Add additional Mac OS X targets: 10.7, 10.8 and 10.9 (darwin11-13)
309 Add option to disable documentation
310 configure: add --enable-external-build support
311 make: support V=1 as short form of verbose=yes
312 configure: support mingw-w64
313 configure: support hardfloat armv7 CHOSTS
314 configure: add support for android x86
315 Add estimated completion time to vpxenc
316 Don't exit on decode errors in vpxenc
317 vpxenc: support scaling prior to encoding
318 vpxdec: support scaling output
319 vpxenc: improve progress indicators with --skip
320 msvs: Don't link to winmm.lib
321 Add a new script for producing vcxproj files
322 Produce Visual Studio 10 and 11 project files
323 Produce Windows Phone project files
324 msvs-build: use msbuild for vs >= 2005
325 configure: default configure log to config.log
326 Add encoding option --static-thresh
329 Miscellaneous speed optimizations for VP8 and VP9.
332 In general, quality is consistent with the Eider release.
335 This release represents approximately a year of engineering effort,
336 and contains multiple bug fixes. Please refer to git history for details.
340 This release acts as a checkpoint for a large amount of internal refactoring
341 and testing. It also contains a number of small bugfixes, so all users are
342 encouraged to upgrade.
345 This release is ABI and API compatible with Duclair (v1.0.0). Users
346 of older releases should refer to the Upgrading notes in this
347 document for that release.
350 VP8 optimizations for MIPS dspr2
351 vpxenc: add -quiet option
354 Encoder and decoder speed is consistent with the Eider release.
357 In general, quality is consistent with the Eider release.
359 Minor tweaks to ARNR filtering
360 Minor improvements to real time encoding with multiple temporal layers
363 Fixes multithreaded encoder race condition in loopfilter
364 Fixes multi-resolution threaded encoding
365 Fix potential encoder dead-lock after picture resize
368 2012-05-09 v1.1.0 "Eider"
369 This introduces a number of enhancements, mostly focused on real-time
370 encoding. In addition, it fixes a decoder bug (first introduced in
371 Duclair) so all users of that release are encouraged to upgrade.
374 This release is ABI and API compatible with Duclair (v1.0.0). Users
375 of older releases should refer to the Upgrading notes in this
376 document for that release.
378 This release introduces a new temporal denoiser, controlled by the
379 VP8E_SET_NOISE_SENSITIVITY control. The temporal denoiser does not
380 currently take a strength parameter, so the control is effectively
381 a boolean - zero (off) or non-zero (on). For compatibility with
382 existing applications, the values accepted are the same as those
383 for the spatial denoiser (0-6). The temporal denoiser is enabled
384 by default, and the older spatial denoiser may be restored by
385 configuring with --disable-temporal-denoising. The temporal denoiser
386 is more computationally intensive than the spatial one.
388 This release removes support for a legacy, decode only API that was
389 supported, but deprecated, at the initial release of libvpx
390 (v0.9.0). This is not expected to have any impact. If you are
391 impacted, you can apply a reversion to commit 2bf8fb58 locally.
392 Please update to the latest libvpx API if you are affected.
395 Adds a motion compensated temporal denoiser to the encoder, which
396 gives higher quality than the older spatial denoiser. (See above
397 for notes on upgrading).
399 In addition, support for new compilers and platforms were added,
401 improved support for XCode
402 Android x86 NDK build
406 Changing resolution with vpx_codec_enc_config_set() is now
407 supported. Previously, reinitializing the codec was required to
408 change the input resolution.
410 The vpxenc application has initial support for producing multiple
411 encodes from the same input in one call. Resizing is not yet
412 supported, but varying other codec parameters is. Use -- to
413 delineate output streams. Options persist from one stream to the
416 Also, the vpxenc application will now use a keyframe interval of
417 5 seconds by default. Use the --kf-max-dist option to override.
420 Decoder performance improved 2.5% versus Duclair. Encoder speed is
421 consistent with Duclair for most material. Two pass encoding of
422 slideshow-like material will see significant improvements.
424 Large realtime encoding speed gains at a small quality expense are
425 possible by configuring the on-the-fly bitpacking experiment with
426 --enable-onthefly-bitpacking. Realtime encoder can be up to 13%
427 faster (ARM) depending on the number of threads and bitrate
428 settings. This technique sees constant gain over the 5-16 speed
429 range. For VC style input the loss seen is up to 0.2dB. See commit
430 52cf4dca for further details.
433 On the whole, quality is consistent with the Duclair release. Some
436 Reduced blockiness in easy sections by applying a penalty to
439 Improved quality of static sections (like slideshows) with
442 Improved keyframe sizing with multiple temporal layers
445 Corrected alt-ref contribution to frame rate for visible updates
446 to the alt-ref buffer. This affected applications making manual
447 usage of the frame reference flags, or temporal layers.
449 Additional constraints were added to disable multi-frame quality
450 enhancement (MFQE) in sections of the frame where there is motion.
453 Fixed corruption issues when vpx_codec_enc_config_set() was called
454 with spatial resampling enabled.
456 Fixed a decoder error introduced in Duclair where the segmentation
457 map was not being reinitialized on keyframes (#378)
460 2012-01-27 v1.0.0 "Duclair"
461 Our fourth named release, focused on performance and features related to
462 real-time encoding. It also fixes a decoder crash bug introduced in
463 v0.9.7, so all users of that release are encouraged to upgrade.
466 This release is ABI incompatible with prior releases of libvpx, so the
467 "major" version number has been bumped to 1. You must recompile your
468 applications against the latest version of the libvpx headers. The
469 API remains compatible, and this should not require code changes in most
473 This release introduces several substantial new features to the encoder,
474 of particular interest to real time streaming applications.
476 Temporal scalability allows the encoder to produce a stream that can
477 be decimated to different frame rates, with independent rate targeting
480 Multiframe quality enhancement postprocessing can make visual quality
481 more consistent in the presence of frames that are substantially
482 different quality than the surrounding frames, as in the temporal
483 scalability case and in some forced keyframe scenarios.
485 Multiple-resolution encoding support allows the encoding of the
486 same content at different resolutions faster than encoding them
490 Optimization targets for this release included the decoder and the real-
491 time modes of the encoder. Decoder speed on x86 has improved 10.5% with
492 this release. Encoder improvements followed a curve where speeds 1-3
493 improved 4.0%-1.5%, speeds 4-8 improved <1%, and speeds 9-16 improved
494 1.5% to 10.5%, respectively. "Best" mode speed is consistent with the
498 Encoder quality in the single stream case is consistent with the Cayuga
502 This release fixes an OOB read decoder crash bug present in v0.9.7
503 related to the clamping of motion vectors in SPLITMV blocks. This
504 behavior could be triggered by corrupt input or by starting
505 decoding from a P-frame.
508 2011-08-15 v0.9.7-p1 "Cayuga" patch 1
509 This is an incremental bugfix release against Cayuga. All users of that
510 release are strongly encouraged to upgrade.
512 - Fix potential OOB reads (cdae03a)
514 An unbounded out of bounds read was discovered when the
515 decoder was requested to perform error concealment (new in
516 Cayuga) given a frame with corrupt partition sizes.
518 A bounded out of bounds read was discovered affecting all
519 versions of libvpx. Given an multipartition input frame that
520 is truncated between the mode/mv partition and the first
521 residiual paritition (in the block of partition offsets), up
522 to 3 extra bytes could have been read from the source buffer.
523 The code will not take any action regardless of the contents
524 of these undefined bytes, as the truncated buffer is detected
525 immediately following the read based on the calculated
526 starting position of the coefficient partition.
528 - Fix potential error concealment crash when the very first frame
529 is missing or corrupt (a609be5)
531 - Fix significant artifacts in error concealment (a4c2211, 99d870a)
533 - Revert 1-pass CBR rate control changes (e961317)
534 Further testing showed this change produced undesirable visual
535 artifacts, rolling back for now.
538 2011-08-02 v0.9.7 "Cayuga"
539 Our third named release, focused on a faster, higher quality, encoder.
542 This release is backwards compatible with Aylesbury (v0.9.5) and
543 Bali (v0.9.6). Users of older releases should refer to the Upgrading
544 notes in this document for that release.
547 Stereo 3D format support for vpxenc
548 Runtime detection of available processor cores.
549 Allow specifying --end-usage by enum name
550 vpxdec: test for frame corruption
551 vpxenc: add quantizer histogram display
552 vpxenc: add rate histogram display
553 Set VPX_FRAME_IS_DROPPABLE
554 update configure for ios sdk 4.3
555 Avoid text relocations in ARM vp8 decoder
556 Generate a vpx.pc file for pkg-config.
557 New ways of passing encoded data between encoder and decoder.
560 This release includes across-the-board speed improvements to the
561 encoder. On x86, these measure at approximately 11.5% in Best mode,
562 21.5% in Good mode (speed 0), and 22.5% in Realtime mode (speed 6).
563 On ARM Cortex A9 with Neon extensions, real-time encoding of video
564 telephony content is 35% faster than Bali on single core and 48%
565 faster on multi-core. On the NVidia Tegra2 platform, real time
566 encoding is 40% faster than Bali.
568 Decoder speed was not a priority for this release, but improved
569 approximately 8.4% on x86.
571 Reduce motion vector search on alt-ref frame.
572 Encoder loopfilter running in its own thread
573 Reworked loopfilter to precalculate more parameters
574 SSE2/SSSE3 optimizations for build_predictors_mbuv{,_s}().
575 Make hor UV predict ~2x faster (73 vs 132 cycles) using SSSE3.
576 Removed redundant checks
577 Reduced structure sizes
578 utilize preload in ARMv6 MC/LPF/Copy routines
579 ARM optimized quantization, dfct, variance, subtract
580 Increase chrow row alignment to 16 bytes.
581 disable trellis optimization for first pass
582 Write SSSE3 sub-pixel filter function
583 Improve SSE2 half-pixel filter funtions
584 Add vp8_sub_pixel_variance16x8_ssse3 function
585 Reduce unnecessary distortion computation
586 Use diamond search to replace full search
587 Preload reference area in sub-pixel motion search (real-time mode)
590 This release focused primarily on one-pass use cases, including
591 video conferencing. Low latency data rate control was significantly
592 improved, improving streamability over bandwidth constrained links.
593 Added support for error concealment, allowing frames to maintain
594 visual quality in the presence of substantial packet loss.
596 Add rc_max_intra_bitrate_pct control
597 Limit size of initial keyframe in one-pass.
598 Improve framerate adaptation
599 Improved 1-pass CBR rate control
600 Improved KF insertion after fades to still.
601 Improved key frame detection.
602 Improved activity masking (lower PSNR impact for same SSIM boost)
603 Improved interaction between GF and ARFs
604 Adding error-concealment to the decoder.
605 Adding support for independent partitions
606 Adjusted rate-distortion constants
610 Removed firstpass motion map
611 Fix parallel make install
612 Fix multithreaded encoding for 1 MB wide frame
613 Fixed iwalsh_neon build problems with RVDS4.1
614 Fix semaphore emulation, spin-wait intrinsics on Windows
615 Fix build with xcode4 and simplify GLOBAL.
616 Mark ARM asm objects as allowing a non-executable stack.
617 Fix vpxenc encoding incorrect webm file header on big endian
620 2011-03-07 v0.9.6 "Bali"
621 Our second named release, focused on a faster, higher quality, encoder.
624 This release is backwards compatible with Aylesbury (v0.9.5). Users
625 of older releases should refer to the Upgrading notes in this
626 document for that release.
629 vpxenc --psnr shows a summary when encode completes
630 --tune=ssim option to enable activity masking
631 improved postproc visualizations for development
632 updated support for Apple iOS to SDK 4.2
633 query decoder to determine which reference frames were updated
634 implemented error tracking in the decoder
635 fix pipe support on windows
638 Primary focus was on good quality mode, speed 0. Average improvement
639 on x86 about 40%, up to 100% on user-generated content at that speed.
640 Best quality mode speed improved 35%, and realtime speed 10-20%. This
641 release also saw significant improvement in realtime encoding speed
644 Improved encoder threading
645 Dont pick encoder filter level when loopfilter is disabled.
646 Avoid double copying of key frames into alt and golden buffer
648 x86 sse2 temporal filter
649 SSSE3 version of fast quantizer
650 vp8_rd_pick_best_mbsegmentation code restructure
651 Adjusted breakout RD for SPLITMV
652 Changed segmentation check order
653 Improved rd_pick_intra4x4block
654 Adds armv6 optimized variance calculation
655 ARMv6 optimized sad16x16
656 ARMv6 optimized half pixel variance calculations
657 Full search SAD function optimization in SSE4.1
658 Improve MV prediction accuracy to achieve performance gain
659 Improve MV prediction in vp8_pick_inter_mode() for speed>3
662 Best quality mode improved PSNR 6.3%, and SSIM 6.1%. This release
663 also includes support for "activity masking," which greatly improves
664 SSIM at the expense of PSNR. For now, this feature is available with
665 the --tune=ssim option. Further experimentation in this area
666 is ongoing. This release also introduces a new rate control mode
667 called "CQ," which changes the allocation of bits within a clip to
668 the sections where they will have the most visual impact.
670 Tuning for the more exact quantizer.
671 Relax rate control for last few frames
673 Limit key frame quantizer for forced key frames.
675 Add simple version of activity masking.
676 make rdmult adaptive for intra in quantizer RDO
677 cap the best quantizer for 2nd order DC
678 change the threshold of DC check for encode breakout
681 Fix crash on Sparc Solaris.
682 Fix counter of fixed keyframe distance
683 ARNR filter pointer update bug fix
684 Fixed use of motion percentage in KF/GF group calc
685 Changed condition for using RD in Intra Mode
686 Fix encoder real-time only configuration.
687 Fix ARM encoder crash with multiple token partitions
688 Fixed bug first cluster timecode of webm file is wrong.
689 Fixed various encoder bugs with odd-sized images
690 vp8e_get_preview fixed when spatial resampling enabled
691 quantizer: fix assertion in fast quantizer path
692 Allocate source buffers to be multiples of 16
693 Fix for manual Golden frame frequency
694 Fix drastic undershoot in long form content
697 2010-10-28 v0.9.5 "Aylesbury"
698 Our first named release, focused on a faster decoder, and a better encoder.
701 This release incorporates backwards-incompatible changes to the
702 ivfenc and ivfdec tools. These tools are now called vpxenc and vpxdec.
705 * the -q (quiet) option has been removed, and replaced with
706 -v (verbose). the output is quiet by default. Use -v to see
707 the version number of the binary.
709 * The default behavior is now to write output to a single file
710 instead of individual frames. The -y option has been removed.
711 Y4M output is the default.
713 * For raw I420/YV12 output instead of Y4M, the --i420 or --yv12
714 options must be specified.
716 $ ivfdec -o OUTPUT INPUT
717 $ vpxdec --i420 -o OUTPUT INPUT
719 * If an output file is not specified, the default is to write
720 Y4M to stdout. This makes piping more natural.
722 $ ivfdec -y -o - INPUT | ...
725 * The output file has additional flexibility for formatting the
726 filename. It supports escape characters for constructing a
727 filename from the width, height, and sequence number. This
728 replaces the -p option. To get the equivalent:
730 $ ivfdec -p frame INPUT
731 $ vpxdec --i420 -o frame-%wx%h-%4.i420 INPUT
734 * The output file must be specified with -o, rather than as the
737 $ ivfenc <options> INPUT OUTPUT
738 $ vpxenc <options> -o OUTPUT INPUT
740 * The output defaults to webm. To get IVF output, use the --ivf
743 $ ivfenc <options> INPUT OUTPUT.ivf
744 $ vpxenc <options> -o OUTPUT.ivf --ivf INPUT
748 ivfenc and ivfdec have been renamed to vpxenc, vpxdec.
749 vpxdec supports .webm input
750 vpxdec writes .y4m by default
751 vpxenc writes .webm output by default
752 vpxenc --psnr now shows the average/overall PSNR at the end
753 ARM platforms now support runtime cpu detection
754 vpxdec visualizations added for motion vectors, block modes, references
755 vpxdec now silent by default
756 vpxdec --progress shows frame-by-frame timing information
757 vpxenc supports the distinction between --fps and --timebase
758 NASM is now a supported assembler
759 configure: enable PIC for shared libs by default
760 configure: add --enable-small
761 configure: support for ppc32-linux-gcc
762 configure: support for sparc-solaris-gcc
765 Improve handling of invalid frames
766 Fix valgrind errors in the NEON loop filters.
767 Fix loopfilter delta zero transitions
768 Fix valgrind errors in vp8_sixtap_predict8x4_armv6().
769 Build fixes for darwin-icc
772 20-40% (average 28%) improvement in libvpx decoder speed,
774 Rewrite vp8_short_walsh4x4_sse2()
775 Optimizations on the loopfilters.
776 Miscellaneous improvements for Atom
777 Add 4-tap version of 2nd-pass ARMv6 MC filter.
778 Improved multithread utilization
779 Better instruction choices on x86
780 reorder data to use wider instructions
781 Update NEON wide idcts
782 Make block access to frame buffer sequential
783 Improved subset block search
784 Bilinear subpixel optimizations for ssse3.
785 Decrease memory footprint
787 Encoder speed improvements (percentage gain not measured):
788 Skip unnecessary search of identical frames
789 Add SSE2 subtract functions
790 Improve bounds checking in vp8_diamond_search_sadx4()
791 Added vp8_fast_quantize_b_sse2
794 Over 7% overall PSNR improvement (6.3% SSIM) in "best" quality
795 encoding mode, and up to 60% improvement on very noisy, still
796 or slow moving source video
798 Motion compensated temporal filter for Alt-Ref Noise Reduction
799 Improved use of trellis quantization on 2nd order Y blocks
800 Tune effect of motion on KF/GF boost in two pass
801 Allow coefficient optimization for good quality speed 0.
802 Improved control of active min quantizer for two pass.
803 Enable ARFs for non-lagged compress
807 Disable frame dropping by default
808 Improved multithreaded performance
809 Improved Force Key Frame Behaviour
810 Increased rate control buffer level precision
811 Fix bug in 1st pass motion compensation
812 ivfenc: correct fixed kf interval, --disable-kf
814 Changed above and left context data layout
815 Rework idct calling structure.
816 Removed unnecessary MB_MODE_INFO copies
817 x86: SSSE3 sixtap prediction
818 Reworked IDCT to include reconstruction (add) step
819 Swap alt/gold/new/last frame buffer ptrs instead of copying.
820 Improve SSE2 loopfilter functions
821 Change bitreader to use a larger window.
822 Avoid loopfilter reinitialization when possible
824 Normalize quantizer's zero bin and rounding factors
825 Add trellis quantization.
826 Make the quantizer exact.
827 Updates to ARNR filtering algorithm
828 Fix breakout thresh computation for golden & AltRef frames
829 Redo the forward 4x4 dct
830 Improve the accuracy of forward walsh-hadamard transform
831 Further adjustment of RD behaviour with Q and Zbin.
833 Allow linking of libs built with MinGW to MSVC
834 Fix target auto-detection on mingw32
835 Allow --cpu= to work for x86.
836 configure: pass original arguments through to make dist
837 Fix builds without runtime CPU detection
838 msvs: fix install of codec sources
839 msvs: Change devenv.com command line for better msys support
840 msvs: Add vs9 targets.
841 Add x86_64-linux-icc target
843 Potential crashes on older MinGW builds
844 Fix two-pass framrate for Y4M input.
845 Fixed simple loop filter, other crashes on ARM v6
846 arm: fix missing dependency with --enable-shared
847 configure: support directories containing .o
848 Replace pinsrw (SSE) with MMX instructions
849 apple: include proper mach primatives
850 Fixed rate control bug with long key frame interval.
851 Fix DSO link errors on x86-64 when not using a version script
852 Fixed buffer selection for UV in AltRef filtering
857 * ivfenc/ivfdec now support YUV4MPEG2 input and pipe I/O
858 * Speed optimizations
861 * Prevent out-of-bounds accesses on invalid data
862 - Build system updates:
863 * Detect toolchain to be used automatically for native builds
864 * Support building shared libraries
865 * Better autotools emulation (--prefix, --libdir, DESTDIR)
867 * http://webmproject.blogspot.com/2010/06/changes-to-webm-open-source-license.html
871 - Initial open source release. Welcome to WebM and VP8!