Merge branch '0.3.0'

* 0.3.0: (65 commits) Update ChangeLog Cosmetic fixes misc style fix add missing YUVA->ARGB automatic conversion in WebPEncode() Container spec: Clarify frame disposal container doc: add a note about the 'ANMF' payload Container spec: clarify the background color field container doc: move RIFF description to own section libwebp/mux: fix double free use WebPDataCopy() instead of re-coding it. demux: keep a frame tail pointer; used in AddFrame add doc precision about WebPParseHeaders() return codes gif2webp: Bgcolor fix for a special case fix bad saturation order in QuantizeBlock vwebp/animation: fix background dispose Makefile.vc: fix dynamic builds update ChangeLog examples: don't use C99 %zu update ChangeLog update NEWS ... Conflicts: src/webp/format_constants.h Change-Id: Ie659644d3ea5592cde64ec3af90a00cd17838247
2024-11-20 04:18:26 +01:00 · 2013-04-01 19:11:52 -07:00 · 2013-04-01 19:11:52 -07:00 · df4a406d8d
commit df4a406d8d
parent e9a7990bc5 1e0d4b8c8b
59 changed files with 1492 additions and 583 deletions
--- a/.mailmap
+++ b/.mailmap
@ -1,2 +1,6 @@
+<johann.koenig@duck.com> <johannkoenig@google.com>
 Mikołaj Zalewski <mikolajz@google.com>
 Pascal Massimino <pascal.massimino@gmail.com>
+Vikas Arora <vikasa@google.com>
+<vikasa@google.com> <vikasa@gmail.com>
+<vikasa@google.com> <vikaas.arora@gmail.com>
--- a/3
+++ b/3
@ -1,9 +1,10 @@
 Contributors:
 - James Zern (jzern at google dot com)
 - Jan Engelhardt (jengelh at medozas dot de)
- Johann (johannkoenig at google dot com)
+- Johann (johann dot koenig at duck dot com)
 - Jyrki Alakuijala (jyrki at google dot com)
 - Lou Quillio (louquillio at google dot com)
+- Mans Rullgard (mans at mansr dot com)
 - Martin Olsson (mnemo at minimum dot se)
 - Mikołaj Zalewski (mikolajz at google dot com)
 - Noel Chromium (noel at chromium dot org)
--- a/383
+++ b/383
@ -1,3 +1,333 @@
+d52b405 Cosmetic fixes
+6cb4a61 misc style fix
+68111ab add missing YUVA->ARGB automatic conversion in WebPEncode()
+403bfe8 Container spec: Clarify frame disposal
+3e7a13a Merge "Container spec: clarify the background color field" into 0.3.0
+14af774 container doc: add a note about the 'ANMF' payload
+cc635ef Container spec: clarify the background color field
+e3e3394 container doc: move RIFF description to own section
+4299f39 libwebp/mux: fix double free
+33f9a69 Merge "demux: keep a frame tail pointer; used in AddFrame" into 0.3.0
+a2a7b95 use WebPDataCopy() instead of re-coding it.
+6f18f12 demux: keep a frame tail pointer; used in AddFrame
+e5af49e add doc precision about WebPParseHeaders() return codes
+db46daa Merge "Makefile.vc: fix dynamic builds" into 0.3.0
+53c77af Merge "gif2webp: Bgcolor fix for a special case" into 0.3.0
+a5ebd14 gif2webp: Bgcolor fix for a special case
+6378f23 Merge "vwebp/animation: fix background dispose" into 0.3.0
+3c8eb9a fix bad saturation order in QuantizeBlock
+04c7a2e vwebp/animation: fix background dispose
+81a5069 Makefile.vc: fix dynamic builds
+5f25c39 update ChangeLog
+14d42af examples: don't use C99 %zu
+5ccf1fe update ChangeLog
+2560c24 update NEWS
+f43bafc Merge changes Iecccb09c,If5ee9fd2,I3e181ce4 into 0.3.0
+a788644 dwebp: warn when decoding animated webp's
+302efcd Decode: return more meaningful error for animation
+ad45273 WebPBitstreamFeatures: add has_animation field
+783dfa4 disable FRGM decoding for good in libwebpmux
+4b956be Update ChangeLog
+ad8b86d update NEWS
+3e084f6 Merge "demux cosmetics: comments/rename internal function" into 0.3.0
+d3f8c62 Merge "move WebPFeatureFlags declaration" into 0.3.0
+7386fe5 Merge "libwebp{demux,mux}: install mux_types.h" into 0.3.0
+d6cd4e9 Merge "bump decode abi" into 0.3.0
+17f8da5 bump decode abi
+97684ae Merge "add doc precision about WebPDemuxPartial()" into 0.3.0
+f933fd2 move WebPFeatureFlags declaration
+289bc47 libwebp{demux,mux}: install mux_types.h
+224e8d4 add doc precision about WebPDemuxPartial()
+4c18e80 demux cosmetics: comments/rename internal function
+7cfd1bf update AUTHORS
+401f7b8 Merge "speed-up lossless (~3%) with ad-hoc histogram cost evaluation" into 0.3.0
+1fc8ffc Merge "makefile.unix: dist related changes" into 0.3.0
+8a89c6e Merge changes I466c377f,Ib761ebd3,I694857fc into 0.3.0
+f4ffb2d speed-up lossless (~3%) with ad-hoc histogram cost evaluation
+723847d gif2webp: only write error messages to stderr
+701b9e2 makefile.unix: dist related changes
+bb85b43 Merge "update NEWS" into 0.3.0
+59423a2 gif2webp: fix crash on open failure with libgif5
+9acb17d gif2webp: silence a unused param warning
+7d9fdc2 Merge "README updates" into 0.3.0
+5621934 Merge "build: fix install race on shared headers" into 0.3.0
+70809d8 Merge "bump version to 0.3.0" into 0.3.0
+d851cd1 demux: make the parse a bit more strict
+28bb410 update NEWS
+cef9388 bump version to 0.3.0
+9048494 build: fix install race on shared headers
+1e67e8e README updates
+42b611a Merge "configure: drop experimental from mux/demux" into 0.3.0
+096a8e3 Merge "vwebp: add color profile support" into 0.3.0
+ddfee5d vwebp: add color profile support
+0d6927d Merge "Mark fragment options as experimental in webpmux" into 0.3.0
+5dbd403 Mark fragment options as experimental in webpmux
+a0a6648 configure: drop experimental from mux/demux
+ee65bad Merge "add support for BITS > 32" into 0.3.0
+744930d add support for BITS > 32
+7dd288f cwebp: fix build
+19a8dd0 Merge "Makefile.vc: add vwebp.exe target" into 0.3.0
+50eedda Merge "examples: normalize icc related program arguments" into 0.3.0
+757f637 Merge "Makefile.vc: add libwebpdecoder target" into 0.3.0
+b65c4b7 Makefile.vc: add libwebpdecoder target
+f8db7b4 Merge "vwebp: replace doubles w/floats where appropriate" into 0.3.0
+d99aa56 Makefile.vc: add vwebp.exe target
+013023e vwebp: replace doubles w/floats where appropriate
+9b3db89 README.mux: add version reference
+7b6a26c Merge "cwebp: output metadata statistics" into 0.3.0
+d8dc72a examples: normalize icc related program arguments
+7bfc905 Merge "make alpha unfilter work in-place" into 0.3.0
+0037b2d Merge "add LUT-free reference code for YUV->RGB conversion." into 0.3.0
+166bf74 Merge "demux: disable fragment parsing" into 0.3.0
+126974b add LUT-free reference code for YUV->RGB conversion.
+0aef3eb make alpha unfilter work in-place
+14ef500 Merge "Remove 'status: experimental' from container spec" into 0.3.0
+d40c98e Merge "webpmux binary: tiny style fix" into 0.3.0
+0bc4268 cwebp: output metadata statistics
+bc03980 Merge "autoconf: normalize experimental define" into 0.3.0
+d1e21b1 Remove 'status: experimental' from container spec
+7681bb9 webpmux binary: tiny style fix
+a3dd3d0 avoid installing example_util.h
+252320e demux: disable fragment parsing
+537bde0 autoconf: normalize experimental define
+5e338e0 Merge changes I33e8a613,I8e8a7b44 into 0.3.0
+d9d0ea1 Merge changes If21e3ec7,I991fc30b into 0.3.0
+627f5ca automake: add reference to libwebp for mux/demux
+eef73d0 don't consolidate proba stats too often
+05ec4cc libwebp{,decoder}.pc: add pthread flags
+1bfcf5b add libwebpmux.pc
+26ca843 add libwebpdemux.pc
+69e2590 Merge "Tune Lossless compression for lower qualities."
+0478b5d Tune Lossless compression for lower qualities.
+39f7586 add a mention of parallel alpha encoding in the NEWS
+5a21d96 Merge "1.5x-2x faster encoding for method 3 and up"
+9bfbdd1 1.5x-2x faster encoding for method 3 and up
+27dc741 Correct frame options order in README.mux
+be2fd17 Mux: fix a scenario with bad ANMF/FRGM size
+19eb012 Merge "Demux: Add option to get frame count using GetI()"
+7368b8c Merge "WebPGetFeatures() out of if condition for clarity."
+f604c9a Merge "fix windows build"
+153f94e fix windows build
+847b492 Merge "vwebp: use magenta for 'i'nfo display"
+25ea46b Merge "vwebp: add keyboard shortcuts to help output"
+bea7cca vwebp: use magenta for 'i'nfo display
+8fab161 webpmux: correct -frame param order in help output
+03cc23d vwebp: add keyboard shortcuts to help output
+068eba8 Demux: Add option to get frame count using GetI()
+988b8f5 WebPGetFeatures() out of if condition for clarity.
+6933d91 Merge "gif2webp: Be lenient about background color index."
+4d0f7c5 Merge "WebPGetFeatures() behavior change:"
+fdeeb01 gif2webp: Be lenient about background color index.
+ad25032 Merge "multi-threaded alpha encoding for lossy"
+4e32d3e Merge "fix compilation of token.c"
+f817930 multi-threaded alpha encoding for lossy
+8805035 fix compilation of token.c
+fc81621 code using the actual values for num_parts_, not the ones from config
+7265535 Merge "move the config check from .c to .h"
+dd9e76f move the config check from .c to .h
+956b217 WebPGetFeatures() behavior change:
+df02e4c WebPDemuxGetI behavior change:
+633c004 Merge "rebalance method tools (-m) for methods [0..4]"
+58ca6f6 rebalance method tools (-m) for methods [0..4]
+7648c3c Merge "describe rd-opt levels introduce VP8RDLevel enum"
+67fb100 Merge "autoconf: enable silent-rules by default"
+a5042a3 GetVersion() methods for mux and demux
+5189957 describe rd-opt levels introduce VP8RDLevel enum
+4e094ac autoconf: enable silent-rules by default
+b7eaa85 inline VP8LFastLog2() and VP8LFastSLog2 for small values
+5cf7792 split quant_levels.c into decoder and encoder version
+e5d3ffe Merge "Update code example in README.mux"
+ac5a915 Update code example in README.mux
+38a91e9 Add example code snippet for demux API
+5f557f3 README.mux: add info about Demux API and vwebp
+c0ba090 backward_references: avoid signed integer overflow
+943386d disable SSE2 for now
+9479fb7 lossless encoding speedup
+ec2030a merge two lines together
+b67956c Merge "Remove ReadOneBit() and ReadSymbolUnsafe()"
+1667bde Remove ReadOneBit() and ReadSymbolUnsafe()
+3151669 wicdec + dwebp cosmetics: normalize formatting
+92668da change default filtering parameters:   * type is now 'strong'   * strength is now '60'
+b7490f8 introduce WEBP_REFERENCE_IMPLEMENTATION compile option
+3383885 faster decoding (3%-6%)
+5c3e381 Merge "add a -jpeg_like option"
+c231104 remove unused declaration of VP8Zigzag
+3615295 Merge "wicdec: add alpha support for paletted formats"
+c9f1649 wicdec: add alpha support for paletted formats
+1262f81 Merge "wicdec: silence some warnings"
+e7ea61e wicdec: silence some warnings
+23c0f35 fix missing intptr_t->int cast for MSVC
+e895059 add a -jpeg_like option
+1f803f6 Merge "Tune alpha quality mapping to more reasonable values."
+1267d49 Tune alpha quality mapping to more reasonable values.
+043076e Merge "speed-up lossless in BackwardTrace"
+f3a44dc remove one malloc from TraceBackwards()
+0fc1a3a speed-up lossless in BackwardTrace
+7c732e5 cwebp: centralize WebPCleanupTransparentArea()
+7381254 Merge "wicdec: add ICC profile extraction"
+e83ff7d wicdec: add ICC profile extraction
+146c6e3 Merge "cosmetics: pngdec: normalize default label location"
+a8f549d Merge "manpages: italicize option parameters"
+e118db8 Merge "encode.h: note the need to free() WebPMemoryWriter"
+1dfee6d cosmetics: pngdec: normalize default label location
+14c3820 manpages: italicize option parameters
+7defbfa encode.h: note the need to free() WebPMemoryWriter
+88d382a cwebp: cleanup after memory_writer
+12d6cec fix extra space in dwebp.1 man
+b01681a Fix for demuxer frame iteration:
+56c12aa Demuxer creation fix:
+66c810b add a -yuv option to dwebp (very similar to -pgm)
+841a3ba Merge "Remove -Wshadow warnings."
+8fd0252 Merge "upsampling_neon.c: fix build"
+6efed26 Remove -Wshadow warnings.
+60904aa Merge "allow WebPINewRGB/YUVA to be passed a NULL output buffer."
+b7adf37 allow WebPINewRGB/YUVA to be passed a NULL output buffer.
+27f8f74 upsampling_neon.c: fix build
+06b9cdf gitignore: add IOS related directories
+f112221 Merge "Fix more comments for iobuild.sh"
+fe4d25d Fix more comments for iobuild.sh
+1de3e25 Merge "NEON optimised yuv to rgb conversion"
+090b708 NEON optimised yuv to rgb conversion
+daa0647 Merge "Add ios build script for building iOS library."
+79fe39e Add ios build script for building iOS library.
+126c035 remove some more -Wshadow warnings
+522e9d6 Merge "cwebp: enable '-metadata'"
+76ec5fa cwebp: enable '-metadata'
+aeb91a9 Merge "cosmetics: break a few long lines"
+be7c96b cosmetics: break a few long lines
+cff8ddb Merge "add libwebpdecoder.pc"
+93148ab Merge "libwebp.pc.in: detab"
+6477f95 Merge "Makefile.vc: normalize path separator"
+bed1ed7 add libwebpdecoder.pc
+46168b2 libwebp.pc.in: detab
+a941a34 Fixed few nits in the build files.
+dd7a49b Makefile.vc: normalize path separator
+9161be8 Merge "cwebp: extract WIC decoding to its own module"
+08e7c58 Merge "Provide an option to build decoder library."
+0aeba52 Provide an option to build decoder library.
+757ebcb catch malloc(0)/calloc(0) with an assert
+152ec3d Merge "handle malloc(0) and calloc(0) uniformly on all platforms"
+a452a55 cwebp: extract WIC decoding to its own module
+2b252a5 Merge "Provide option to swap bytes for 16 bit colormodes"
+94a48b4 Provide option to swap bytes for 16 bit colormodes
+42f8f93 handle malloc(0) and calloc(0) uniformly on all platforms
+8b2152c Merge "add an extra assert to check memory bounds"
+0d19fbf remove some -Wshadow warnings
+cd22f65 add an extra assert to check memory bounds
+8189fed Merge "Add details and reference about the YUV->RGB conversion"
+1d2702b Merge "Formatting fixes in lossless bitstream spec"
+8425aae Formatting fixes in lossless bitstream spec
+a556cb1 Add details and reference about the YUV->RGB conversion
+d8f21e0 add link to SSIM description on Wikipedia
+18e9167 Merge "WebP-lossless spec clarifications:"
+98e25b9 Merge "cwebp: add -metadata option"
+f01c2a5 WebP-lossless spec clarifications:
+f4a9797 Merge "Disto4x4 and Disto16x16 in NEON"
+47b7b0b Disto4x4 and Disto16x16 in NEON
+7eaee9f cwebp: add -metadata option
+36c52c2 tiffdec: use toff_t for exif ifd offset
+7c8111e Merge "cwebp/tiffdec: add TIFF metadata extraction"
+e6409ad Remove redundant include from dsp/lossless code.
+1ab5b3a Merge "configure: fix --with-gifincludedir"
+03c749e configure: fix --with-gifincludedir
+8b65063 multiple libgif versions support for gif2webp
+476e293 gif2webp: Use DGifOpenFileName()
+b50f277 tiffdec: correct format string
+2b9048e Merge "tiffdec: check error returns for width/height"
+a1b5a9a Merge "cwebp/tiff: use the first image directory"
+079423f tiffdec: check error returns for width/height
+d62824a Merge "cwebp/jpegdec: add JPEG metadata extraction"
+03afaca Merge "cwebp: add PNG metadata extraction"
+2c72496 cwebp/jpegdec: add JPEG metadata extraction
+dba64d9 cwebp: add PNG metadata extraction
+1f075f8 Lossless spec corrections/rewording/clarifications
+2914ecf cwebp/tiffdec: add TIFF metadata extraction
+d82a3e3 More corrections/clarifications in lossless spec:
+bd00255 cwebp/tiff: use the first image directory
+df7aa07 Merge "Cleanup around jpegdec"
+0f57dcc decoding speed-up (~1%)
+bcec339 Lossless bitstream clarification:
+6bf2087 add examples/metadata.c
+207f89c Merge "configure: add libwebpdemux status to summary"
+1bd287a Cleanup around jpegdec
+9145567 Merge "cosmetics: use '== 0' in size checks"
+d6b88b7 cosmetics: use '== 0' in size checks
+d3dace2 cosmetics: jpegdec
+2f69af7 configure: add libwebpdemux status to summary
+1c1c564 cwebp: extract tiff decoding to its own module
+6a871d6 cwebp: extract jpeg decoding to its own module
+2ee228f cwebp: extract png decoding to its own module
+4679db0 Merge "cwebp: add metadata framework"
+63aba3a cwebp: add metadata framework
+931bd51 lossless bitstream: block size bits correction
+e4fc4c1 lossless bitstream: block size bits correction
+d65ec67 fix build, move token.c to src/enc/
+657f5c9 move token buffer to its own file (token.c)
+c34a375 introduce GetLargeValue() to slim-fast GetCoeffs().
+d5838cd faster non-transposing SSE2 4x4 FTransform
+f76191f speed up GetResidualCost()
+ba2aa0f Add support for BITS=24 case
+2e7f6e8 makefile.unix: Dependency on libraries
+dca8421 Merge "Separate out mux and demux code and libraries:"
+23782f9 Separate out mux and demux code and libraries:
+bd56a01 configure: add summary output
+90e5e31 dwebp manual: point to webpmux, gif2webp.
+540790c gif2webp.c: add a note about prerequisites
+d1edf69 cwebp man page: meaning of '-q' for lossy/lossless
+79efa1d Add man page for gif2webp utility
+2243e40 Merge "gif2webp build support with autoconf tools"
+c40efca gif2webp build support with autoconf tools
+6523e2d WebP Container:
+4da788d Merge "simplify the fwd transform"
+42c3b55 simplify the fwd transform
+41a6ced user GLfloat instead of float
+b542611 fix indentation
+68f282f * handle offset in anim viewer 'vwebp' * fix gif2webp to handle disposal method and odd offset correctly
+118cb31 Merge "add SSE2 version of Sum of Square error for 16x16, 16x8 and 8x8 case"
+8a7c3cc Merge "Change the order of -frame argument to be more natural"
+99e0a70 Merge "Simplify the texture evaluation Disto4x4()"
+0f923c3 make the bundling work in a tmp buffer
+e5c3b3f Simplify the texture evaluation Disto4x4()
+4860008 Change the order of -frame argument to be more natural
+35bfd4c add SSE2 version of Sum of Square error for 16x16, 16x8 and 8x8 case
+a7305c2 Clarification for unknown chunks
+4c4398e Refine WebP Container Spec wrt unknown chunks.
+2ca642e Rectify WebPMuxGetFeatures:
+7caab1d Some cosmetic/comment fixes.
+60b2651 Merge "Write a GIF to WebP converter based on libgif."
+c7127a4 Merge "Add NEON version of FTransformWHT"
+11b2721 Write a GIF to WebP converter based on libgif.
+e9a15a3 ExUtilWriteFile() to write memory segment to file
+74356eb Add a simple cleanup step in mux assembly:
+51bb1e5 mux.h: correct WebPDemuxSelectFragment() prototype
+22a0fd9 Add NEON version of FTransformWHT
+fa30c86 Update mux code to match the spec wrt animation
+d9c5fbe by-pass Analysis pass in case segments=1
+d2ad445 Merge changes Ibeccffc3,Id1585b16
+5c8be25 Merge "Chunk fourCCs for XMP/EXIF"
+a00a3da Use 'frgm' instead of 'tile' in webpmux parameters
+81b8a74 Design change in ANMF and FRGM chunks:
+f903cba Chunk fourCCs for XMP/EXIF
+812933d Tune performance of HistogramCombine
+52ad197 Animation specification in container spec
+001b930 Image fragment specification in container spec
+391f9db Ordering of description of bits in container spec
+d573577 Metadata specification in container spec
+1c4609b Merge commit 'v0.2.1'
+0ca584c Merge "Color profile specification in container spec"
+e8b41ad add NEON asm version for WHT inverse transform
+af6f0db Color profile specification in container spec
+a61a824 Merge "Add NULL check in chunk APIs"
+0e8b7ee fix WebPPictureView() unassigned strides
+75e5f17 ARM/NEON: 30% encoding speed-up
+02b4356 Add NULL check in chunk APIs
+a077072 mux struct naming
+6c66dde Merge "Tune Lossless encoder"
+ab5ea21 Tune Lossless encoder
+74fefc8 Update ChangeLog (v0.2.1, origin/0.2.0)
+92f8059 Rename some chunks:
+3bb4bbe Merge "Mux API change:"
+d0c79f0 Mux API change:
 abc0604 Merge "update NEWS" into 0.2.0
 57cf313 update NEWS
 25f585c bump version to 0.2.1
@ -15,6 +345,55 @@ b30add2 EncodeImageInternal: fix uninitialized free
 e5970bd Make *InitSSE2() functions be empty on non-SSE2 platform
 ef5cc47 make *InitSSE2() functions be empty on non-SSE2 platform
 c4ea259 make VP8DspInitNEON() public
+8344ead Merge "libwebp: validate chunk size in ParseOptionalChunks"
+4828bb9 Merge "cwebp (windows): fix alpha image import on XP"
+3076333 libwebp: validate chunk size in ParseOptionalChunks
+7048189 AccumulateLSIM: fix double -> float warnings
+eda8ee4 cwebp (windows): fix alpha image import on XP
+c6e9865 Merge "add EXPERIMENTAL code for YUV-JPEG colorspace"
+f0360b4 add EXPERIMENTAL code for YUV-JPEG colorspace
+f86e6ab add LSIM metric to WebPPictureDistortion()
+c3aa215 Speed up HistogramCombine for lower qualities.
+1765cb1 Merge "autoconf/libwebp: enable dll builds for mingw"
+a13562e autoconf/libwebp: enable dll builds for mingw
+9f469b5 typo: no_fancy -> no_fancy_upsampling
+1a27f2f Merge "fix double to float conversion warning"
+cf1e90d Merge "cwebp: fix jpg encodes on XP"
+f2b5d19 [cd]webp: always output windows errors
+e855208 fix double to float conversion warning
+ecd66f7 cwebp: fix jpg encodes on XP
+7b3eb37 Tune lossless compression to get better gains.
+ce8bff4 Merge "VP8LAllocateHistogramSet: fix overflow in size calculation"
+ab5b67a Merge "EncodeImageInternal: fix uninitialized free"
+7fee5d1 Merge "GetHistoBits: fix integer overflow"
+a6ae04d VP8LAllocateHistogramSet: fix overflow in size calculation
+80237c4 GetHistoBits: fix integer overflow
+8a99723 EncodeImageInternal: fix uninitialized free
+0b9e682 minor cosmetics
+a792b91 fix the -g/O3 discrepancy for 32bit compile
+73ba435 Merge "detect and merge similar segments"
+fee6627 detect and merge similar segments
+0c44f41 src/webp/*.h: don't forward declare enums in C++
+d7a5ac8 vwebp: use demux interface
+931e0ea Merge "replace 'typedef struct {} X;" by "typedef struct X X; struct X {};""
+8f216f7 remove cases of equal comparison for qsort()
+28d25c8 replace 'typedef struct {} X;" by "typedef struct X X; struct X {};"
+2afee60 speed up for ARM using 8bit for boolean decoder
+5725cab new segmentation algorithm
+2cf1f81 Merge "fix the BITS=8 case"
+12f78ae fix the BITS=8 case
+6920c71 fix MSVC warnings regarding implicit uint64 to uint32 conversions
+f6c096a webpmux binary: Rename 'xmp' option to 'meta'
+ddfe871 webpmux help correction
+b7c5544 Merge "Make *InitSSE2() functions be empty on non-SSE2 platform"
+1c04a0d Common APIs for chunks metadata and color profile.
+2a3117a Merge "Create WebPMuxFrameInfo struct for Mux APIs"
+5c3a723 Make *InitSSE2() functions be empty on non-SSE2 platform
+7c6e60f make *InitSSE2() functions be empty on non-SSE2 platform
+c7eb457 make VP8DspInitNEON() public
+ab3234a Create WebPMuxFrameInfo struct for Mux APIs
+e3990fd Alignment fixes
+e55fbd6 Merge branch '0.2.0'
 4238bc0 Update ChangeLog (v0.2.0)
 c655380 dec/io.c: cosmetics
 fe1958f RGBA4444: harmonize lossless/lossy alpha values
@ -135,7 +514,7 @@ f0b5def bump versions
 05108f6 Merge "More spec/code matching in mux:"
 6808e69 More spec/code matching in mux:
 bd2b46f Merge "doc/webp-container-spec: light cosmetics"
-20ead32 doc/webp-container-spec: light cosmetics (full)
+20ead32 doc/webp-container-spec: light cosmetics
 1d40a8b configure: add pthread detection
 b5e9067 fix some int <-> size_t mix for buffer sizes
 e41a759 build: remove libwebpmux from default targets/config
@ -258,7 +637,7 @@ f8f9408 libwebp: add WebPDemux stub functions
 fb47bb5 Merge "NumNamedElements() should take an enum param."
 7c68980 Fix asserts in Palette and BackwardReference code.
 fbdcb7e NumNamedElements() should take an enum param.
-fb4943b modify WebPParseHeaders to allow reuse by GetFeatures (old-decode-alph-3)
+fb4943b modify WebPParseHeaders to allow reuse by GetFeatures
 3697b5c write an ad-hoc EncodeImageInternal variant
 eaee9e7 Bug-Fix: Decode small (less than 32 bytes) images.
 0bceae4 Merge "cwebp: fix alpha reporting in stats output"
--- a/Makefile.vc
+++ b/Makefile.vc
@ -1,6 +1,7 @@
 #
 # Stem for static libs and DLLs
 #
+LIBWEBPDECODER_BASENAME = libwebpdecoder
 LIBWEBP_BASENAME = libwebp
 LIBWEBPMUX_BASENAME = libwebpmux
 LIBWEBPDEMUX_BASENAME = libwebpdemux
@ -75,6 +76,7 @@ STATICLIBBUILD = TRUE
 CC             = $(CCDEBUG)
 RTLIB          = $(RTLIBD)
 STATICLIBBUILD = TRUE
+LIBWEBPDECODER_BASENAME = $(LIBWEBPDECODER_BASENAME)_debug
 LIBWEBP_BASENAME = $(LIBWEBP_BASENAME)_debug
 LIBWEBPMUX_BASENAME = $(LIBWEBPMUX_BASENAME)_debug
 LIBWEBPDEMUX_BASENAME = $(LIBWEBPDEMUX_BASENAME)_debug
@ -85,6 +87,7 @@ DLLBUILD  = TRUE
 CC        = $(CCDEBUG)
 RTLIB     = $(RTLIBD)
 DLLBUILD  = TRUE
+LIBWEBPDECODER_BASENAME = $(LIBWEBPDECODER_BASENAME)_debug
 LIBWEBP_BASENAME = $(LIBWEBP_BASENAME)_debug
 LIBWEBPMUX_BASENAME = $(LIBWEBPMUX_BASENAME)_debug
 LIBWEBPDEMUX_BASENAME = $(LIBWEBPDEMUX_BASENAME)_debug
@ -93,17 +96,19 @@ LIBWEBPDEMUX_BASENAME = $(LIBWEBPDEMUX_BASENAME)_debug
 !IF "$(STATICLIBBUILD)" == "TRUE"
 CC     = $(CC) $(RTLIB)
 CFGSET = TRUE
+LIBWEBPDECODER = $(DIRLIB)\$(LIBWEBPDECODER_BASENAME).lib
 LIBWEBP = $(DIRLIB)\$(LIBWEBP_BASENAME).lib
 LIBWEBPMUX = $(DIRLIB)\$(LIBWEBPMUX_BASENAME).lib
 LIBWEBPDEMUX = $(DIRLIB)\$(LIBWEBPDEMUX_BASENAME).lib
 !ELSE IF "$(DLLBUILD)" == "TRUE"
 DLLC   = webp_dll.c
 DLLINC = webp_dll.h
+DLL_OBJS = $(DIROBJ)\$(DLLC:.c=.obj)
 CC     = $(CC) /I$(DIROBJ) /FI$(DLLINC) $(RTLIB) /DWEBP_DLL
+LIBWEBPDECODER = $(DIRLIB)\$(LIBWEBPDECODER_BASENAME)_dll.lib
 LIBWEBP = $(DIRLIB)\$(LIBWEBP_BASENAME)_dll.lib
 LIBWEBPMUX = $(DIRLIB)\$(LIBWEBPMUX_BASENAME)_dll.lib
 LIBWEBPDEMUX = $(DIRLIB)\$(LIBWEBPMDEMUX_BASENAME)_dll.lib
-LIBWEBP_OBJS = $(DIROBJ)\$(DLLC:.c=.obj)
 LIBWEBP_PDBNAME = $(DIROBJ)\$(LIBWEBP_BASENAME)_dll.pdb
 CFGSET = TRUE
 !ENDIF
@ -160,17 +165,22 @@ DEC_OBJS = \
 DEMUX_OBJS = \
    $(DIROBJ)\demux\demux.obj \

-DSP_OBJS = \
+DSP_DEC_OBJS = \
    $(DIROBJ)\dsp\cpu.obj \
    $(DIROBJ)\dsp\dec.obj \
+    $(DIROBJ)\dsp\dec_neon.obj \
    $(DIROBJ)\dsp\dec_sse2.obj \
-    $(DIROBJ)\dsp\enc.obj \
-    $(DIROBJ)\dsp\enc_sse2.obj \
    $(DIROBJ)\dsp\lossless.obj \
    $(DIROBJ)\dsp\upsampling.obj \
+    $(DIROBJ)\dsp\upsampling_neon.obj \
    $(DIROBJ)\dsp\upsampling_sse2.obj \
    $(DIROBJ)\dsp\yuv.obj \

+DSP_ENC_OBJS = \
+    $(DIROBJ)\dsp\enc.obj \
+    $(DIROBJ)\dsp\enc_neon.obj \
+    $(DIROBJ)\dsp\enc_sse2.obj \
+
 EX_FORMAT_DEC_OBJS = \
    $(DIROBJ)\examples\jpegdec.obj \
    $(DIROBJ)\examples\metadata.obj \
@ -205,29 +215,35 @@ MUX_OBJS = \
    $(DIROBJ)\mux\muxinternal.obj \
    $(DIROBJ)\mux\muxread.obj \

-UTILS_OBJS = \
+UTILS_DEC_OBJS = \
    $(DIROBJ)\utils\bit_reader.obj \
-    $(DIROBJ)\utils\bit_writer.obj \
    $(DIROBJ)\utils\color_cache.obj \
    $(DIROBJ)\utils\filters.obj \
    $(DIROBJ)\utils\huffman.obj \
-    $(DIROBJ)\utils\huffman_encode.obj \
-    $(DIROBJ)\utils\quant_levels.obj \
    $(DIROBJ)\utils\quant_levels_dec.obj \
    $(DIROBJ)\utils\rescaler.obj \
    $(DIROBJ)\utils\thread.obj \
    $(DIROBJ)\utils\utils.obj \

-LIBWEBP_OBJS = $(DEC_OBJS) $(DSP_OBJS) $(ENC_OBJS) $(UTILS_OBJS) $(LIBWEBP_OBJS)
+UTILS_ENC_OBJS = \
+    $(DIROBJ)\utils\bit_writer.obj \
+    $(DIROBJ)\utils\huffman_encode.obj \
+    $(DIROBJ)\utils\quant_levels.obj \
+
+LIBWEBPDECODER_OBJS = $(DEC_OBJS) $(DSP_DEC_OBJS) $(UTILS_DEC_OBJS)
+LIBWEBP_OBJS = $(LIBWEBPDECODER_OBJS) $(ENC_OBJS) $(DSP_ENC_OBJS) \
+               $(UTILS_ENC_OBJS) $(DLL_OBJS)
 LIBWEBPMUX_OBJS = $(MUX_OBJS) $(LIBWEBPMUX_OBJS)
 LIBWEBPDEMUX_OBJS = $(DEMUX_OBJS) $(LIBWEBPDEMUX_OBJS)

-OUT_LIBS = $(LIBWEBP)
+OUT_LIBS = $(LIBWEBPDECODER) $(LIBWEBP)
 OUT_EXAMPLES = $(DIRBIN)\cwebp.exe $(DIRBIN)\dwebp.exe

 all: $(OUT_LIBS) $(OUT_EXAMPLES)
 $(DIRBIN)\cwebp.exe: $(DIROBJ)\examples\cwebp.obj $(EX_FORMAT_DEC_OBJS)
 $(DIRBIN)\dwebp.exe: $(DIROBJ)\examples\dwebp.obj
+$(DIRBIN)\vwebp.exe: $(DIROBJ)\examples\vwebp.obj
+$(DIRBIN)\vwebp.exe: $(EX_UTIL_OBJS) $(LIBWEBPDEMUX) $(LIBWEBP)
 $(DIRBIN)\webpmux.exe: $(DIROBJ)\examples\webpmux.obj $(LIBWEBPMUX)
 $(DIRBIN)\webpmux.exe: $(EX_UTIL_OBJS) $(LIBWEBP)
 $(OUT_EXAMPLES): $(EX_UTIL_OBJS) $(LIBWEBP)
@ -237,6 +253,7 @@ experimental:
 	    CFG=$(CFG) \
 	    CFLAGS="$(CFLAGS) /DWEBP_EXPERIMENTAL_FEATURES" /$(MAKEFLAGS)

+$(LIBWEBPDECODER): $(LIBWEBPDECODER_OBJS)
 $(LIBWEBP): $(LIBWEBP_OBJS)
 $(LIBWEBPMUX): $(LIBWEBPMUX_OBJS)
 $(LIBWEBPDEMUX): $(LIBWEBPDEMUX_OBJS)
@ -244,8 +261,8 @@ $(LIBWEBPDEMUX): $(LIBWEBPDEMUX_OBJS)
 $(LIBWEBP_OBJS) $(LIBWEBPMUX_OBJS) $(LIBWEBPDEMUX_OBJS): $(OUTPUT_DIRS)

 !IF "$(DLLBUILD)" == "TRUE"
-$(LIBWEBP_OBJS) $(LIBWEBPMUX_OBJS) $(LIBWEBPDEMUX_OBJS):
-	$(DIROBJ)\$(DLLINC) $(DIROBJ)\$(DLLC)
+$(LIBWEBP_OBJS) $(LIBWEBPMUX_OBJS) $(LIBWEBPDEMUX_OBJS): \
+    $(DIROBJ)\$(DLLINC) $(DIROBJ)\$(DLLC)

 {$(DIROBJ)}.c{$(DIROBJ)}.obj:
 	$(CC) $(CFLAGS) /Fd$(LIBWEBP_PDBNAME) /Fo$@  $<
@ -253,14 +270,14 @@ $(LIBWEBP_OBJS) $(LIBWEBPMUX_OBJS) $(LIBWEBPDEMUX_OBJS):
 $(LIBWEBPMUX): $(LIBWEBP)
 $(LIBWEBPDEMUX): $(LIBWEBP)

-$(LIBWEBP) $(LIBWEBPMUX) $(LIBWEBPDEMUX):
+$(LIBWEBPDECODER) $(LIBWEBP) $(LIBWEBPMUX) $(LIBWEBPDEMUX):
 	$(LNKDLL) /out:$(DIRBIN)\$(@B:_dll=.dll) /implib:$@ $(LFLAGS) $**
 	-xcopy $(DIROBJ)\*.pdb $(DIRLIB) /y

 clean::
 	@-erase /s $(DIROBJ)\$(DLLC) $(DIROBJ)\$(DLLINC) 2> NUL
 !ELSE
-$(LIBWEBP) $(LIBWEBPMUX) $(LIBWEBPDEMUX):
+$(LIBWEBPDECODER) $(LIBWEBP) $(LIBWEBPMUX) $(LIBWEBPDEMUX):
 	$(LNKLIB) /out:$@ $(LFLAGS) $**
 	-xcopy $(DIROBJ)\*.pdb $(DIRLIB) /y
 !ENDIF
--- a/10
+++ b/10
@ -1,9 +1,17 @@
- next version:
+- 3/20/13: version 0.3.0
+  This is a binary compatible release.
  * WebPINewRGB/WebPINewYUVA accept being passed a NULL output buffer
    and will perform auto-allocation.
  * default filter option is now '-strong -f 60'
  * encoding speed-up for lossy methods 3 to 6
  * alpha encoding can be done in parallel to lossy using 'cwebp -mt ...'
+  * color profile, metadata (XMP/EXIF) and animation support finalized in the
+    container.
+  * various NEON assembly additions
+  Tool updates / additions:
+    * gif2webp added
+    * vwebp given color profile & animation support
+    * cwebp can preserve color profile / metadata with '-metadata'

 - 10/30/12: version 0.2.1
  * Various security related fixes
--- a/62
+++ b/62
@ -4,7 +4,7 @@
          \__\__/\____/\_____/__/ ____  ___
                / _/ /    \    \ /  _ \/ _/
               /  \_/   / /   \ \   __/  \__
-               \____/____/\_____/_____/____/v0.2.1
+               \____/____/\_____/_____/____/v0.3.0

 Description:
 ============
@ -49,7 +49,7 @@ will build the binaries examples/cwebp and examples/dwebp, along
 with the static library src/libwebp.a. No system-wide installation
 is supplied, as this is a simple alternative to the full installation
 system based on the autoconf tools (see below).
-Please refer to the makefile.unix for additional details and customizations.
+Please refer to makefile.unix for additional details and customizations.

 Using autoconf tools:
 ---------------------
@ -71,10 +71,11 @@ should be all you need to have the following files

 installed.

-Note: The encoding and decoding libraries are compiled separately
-(as src/dec/libwebpdecode.* and src/dec/libwebpencode.*). They
-can be installed independently using a minor modification in the
-corresponding Makefile.am configure files (see comments there).
+Note: A decode-only library, libwebpdecoder, is available using the
+'--enable-libwebpdecoder' flag. The encode library is built separately and can
+be installed independently using a minor modification in the corresponding
+Makefile.am configure files (see comments there). See './configure --help' for
+more options.

 SWIG bindings:
 --------------
@ -176,7 +177,7 @@ options:

  -metadata <string> ..... comma separated list of metadata to
                           copy from the input to the output if present.
-                           Valid values: all, none (default), exif, iccp, xmp
+                           Valid values: all, none (default), exif, icc, xmp

  -short ................. condense printed message
  -quiet ................. don't print anything.
@ -263,12 +264,51 @@ Visualization tool:

 There's a little self-serve visualization tool called 'vwebp' under the
 examples/ directory. It uses OpenGL to open a simple drawing window and show
-a decoded WebP file. It's not yet integrated in the automake or makefile.unix
-build system, but you can try to manually compile it using the recommendations
-at the top of the source file.
+a decoded WebP file. It's not yet integrated in the automake build system, but
+you can try to manually compile it using the recommendations below.

-Usage: 'vwebp my_picture.webp'
+Usage: vwebp in_file [options]

+Decodes the WebP image file and visualize it using OpenGL
+Options are:
+  -version  .... print version number and exit.
+  -noicc ....... don't use the icc profile if present.
+  -nofancy ..... don't use the fancy YUV420 upscaler.
+  -nofilter .... disable in-loop filtering.
+  -mt .......... use multi-threading.
+  -info ........ print info.
+  -h     ....... this help message.
+
+Keyboard shortcuts:
+  'c' ................ toggle use of color profile.
+  'i' ................ overlay file information.
+  'q' / 'Q' / ESC .... quit.
+
+Building:
+---------
+
+Prerequisites:
+1) OpenGL & OpenGL Utility Toolkit (GLUT)
+  Linux:
+    $ sudo apt-get install freeglut3-dev mesa-common-dev
+  Mac + XCode:
+    - These libraries should be available in the OpenGL / GLUT frameworks.
+  Windows:
+    http://freeglut.sourceforge.net/index.php#download
+
+2) (Optional) qcms (Quick Color Management System)
+  i. Download qcms from Mozilla / Chromium:
+    http://hg.mozilla.org/mozilla-central/file/0e7639e3bdfb/gfx/qcms
+    http://src.chromium.org/viewvc/chrome/trunk/src/third_party/qcms
+  ii. Build and archive the source files as libqcms.a / qcms.lib
+  iii. Update makefile.unix / Makefile.vc
+    a) Define WEBP_HAVE_QCMS
+    b) Update include / library paths to reference the qcms directory.
+
+Build using makefile.unix / Makefile.vc:
+$ make -f makefile.unix examples/vwebp
+> nmake /f Makefile.vc CFG=release-static \
+    ../obj/x64/release-static/bin/vwebp.exe

 Encoding API:
 =============
--- a/README.mux
+++ b/README.mux
@ -1,7 +1,7 @@
          __   __  ____  ____  ____  __ __  _     __ __
         /  \\/  \/  _ \/  _ \/  _ \/  \  \/ \___/_ / _\
         \       /   __/  _  \   __/      /  /  (_/  /__
-          \__\__/\_____/_____/__/  \__//_/\_____/__/___/
+          \__\__/\_____/_____/__/  \__//_/\_____/__/___/v0.1.0


 Description:
@ -9,9 +9,9 @@ Description:

 WebPMux: set of two libraries 'Mux' and 'Demux' for creation, extraction and
 manipulation of an extended format WebP file, which can have features like
-color profile, metadata, animation and fragmented images. Reference
-command-line tools 'webpmux' and 'vwebp' as well as the WebP container
-specification 'doc/webp-container-spec.txt' are also provided in this package.
+color profile, metadata and animation. Reference command-line tools 'webpmux'
+and 'vwebp' as well as the WebP container specification
+'doc/webp-container-spec.txt' are also provided in this package.

 WebP Mux tool:
 ==============
@ -26,7 +26,6 @@ A list of options is available using the -help command line flag:
 Usage: webpmux -get GET_OPTIONS INPUT -o OUTPUT
       webpmux -set SET_OPTIONS INPUT -o OUTPUT
       webpmux -strip STRIP_OPTIONS INPUT -o OUTPUT
-       webpmux -frgm FRAGMENT_OPTIONS [-frgm...] -o OUTPUT
       webpmux -frame FRAME_OPTIONS [-frame...] [-loop LOOP_COUNT]
               [-bgcolor BACKGROUND_COLOR] -o OUTPUT
       webpmux -info INPUT
@ -38,7 +37,6 @@ GET_OPTIONS:
   icc       Get ICC profile.
   exif      Get EXIF metadata.
   xmp       Get XMP metadata.
-   frgm n    Get nth fragment.
   frame n   Get nth frame.

 SET_OPTIONS:
@ -47,7 +45,7 @@ SET_OPTIONS:
   exif file.exif    Set EXIF metadata.
   xmp  file.xmp     Set XMP metadata.
   where:    'file.icc' contains the ICC profile to be set,
-             'file.exif' contains the EXIF metadata to be set and
+             'file.exif' contains the EXIF metadata to be set
             'file.xmp' contains the XMP metadata to be set

 STRIP_OPTIONS:
@ -56,12 +54,6 @@ STRIP_OPTIONS:
   exif      Strip EXIF metadata.
   xmp       Strip XMP metadata.

-FRAGMENT_OPTIONS(i):
- Create fragmented image.
-   file_i +xi+yi
-   where:    'file_i' is the i'th fragment (WebP format),
-             'xi','yi' specify the image offset for this fragment.
-
 FRAME_OPTIONS(i):
 Create animation.
   file_i +di+xi+yi+mi
@ -86,29 +78,18 @@ INPUT & OUTPUT are in WebP format.
 Note: The nature of EXIF, XMP and ICC data is not checked and is assumed to be
 valid.

-WebP viewer tool:
-================
+Visualization tool:
+===================

 The examples/ directory also contains a tool (vwebp) for viewing WebP files.
-It decodes the image and visualizes it using OpenGL.
-
-A list of options is available using the -h command line flag:
-
-> vwebp -h
-Decodes the WebP image file and visualize it using OpenGL
-Options are:
-  -version  .... print version number and exit.
-  -nofancy ..... don't use the fancy YUV420 upscaler.
-  -nofilter .... disable in-loop filtering.
-  -mt .......... use multi-threading
-  -info ........ print info.
-  -h     ....... this help message.
+It decodes the image and visualizes it using OpenGL. See the libwebp README
+for details on building and running this program.

 Mux API:
 ========
 The Mux API contains methods for adding data to and reading data from WebP
-files. This API currently supports XMP/EXIF metadata, ICC profile, animation
-and fragmented images. Other features may be added in subsequent releases.
+files. This API currently supports XMP/EXIF metadata, ICC profile and animation.
+Other features may be added in subsequent releases.

 Example#1 (pseudo code): Creating a WebPMux object with image data, color
 profile and XMP metadata.
@ -145,11 +126,11 @@ For a detailed Mux API reference, please refer to the header file
 (src/webp/mux.h).

 Demux API:
-=========
+==========
 The Demux API enables extraction of images and extended format data from
 WebP files. This API currently supports reading of XMP/EXIF metadata, ICC
-profile, animation and fragmented images. Other features may be added in
-subsequent releases.
+profile and animated images. Other features may be added in subsequent
+releases.

 Code Example: Demuxing WebP data to extract all the frames, ICC profile
 and EXIF/XMP metadata.
--- a/configure.ac
+++ b/configure.ac
@ -1,4 +1,4 @@
-AC_INIT([libwebp], [0.2.1],
+AC_INIT([libwebp], [0.3.0],
        [http://code.google.com/p/webp/issues],,
        [http://developers.google.com/speed/webp])
 AC_CANONICAL_TARGET
@ -176,7 +176,7 @@ CPPFLAGS=$SAVED_CPPFLAGS
 LIBS=$SAVED_LIBS

 if test "$gif_support" = "yes" -a \
-        "$enable_experimental_libwebpmux" = "yes"; then
+        "$enable_libwebpmux" = "yes"; then
  build_gif2webp=yes
 fi
 AM_CONDITIONAL([BUILD_GIF2WEBP], [test "${build_gif2webp}" = "yes"])
@ -244,27 +244,27 @@ AC_MSG_CHECKING(if --enable-experimental option is specified)
 AC_ARG_ENABLE([experimental], AS_HELP_STRING([--enable-experimental],
                                             [Activate experimental features]))
 if test "$enable_experimental" = "yes"; then
-        AC_DEFINE(EXPERIMENTAL,,[Enable experimental code])
-        USE_EXPERIMENTAL_CODE="-DWEBP_EXPERIMENTAL_FEATURES"
+  AC_DEFINE(WEBP_EXPERIMENTAL_FEATURES, [1], [Enable experimental code])
+  USE_EXPERIMENTAL_CODE="-DWEBP_EXPERIMENTAL_FEATURES"
 fi
 AC_MSG_RESULT(${enable_experimental-no})
 AC_SUBST(USE_EXPERIMENTAL_CODE)

 dnl === Check whether libwebpmux should be built
 AC_MSG_CHECKING(whether libwebpmux is to be built)
-AC_ARG_ENABLE([experimental-libwebpmux],
-              AS_HELP_STRING([--enable-experimental-libwebpmux],
+AC_ARG_ENABLE([libwebpmux],
+              AS_HELP_STRING([--enable-libwebpmux],
                             [Build libwebpmux @<:@default=no@:>@]))
-AC_MSG_RESULT(${enable_experimental_libwebpmux-no})
-AM_CONDITIONAL([WANT_MUX], [test "$enable_experimental_libwebpmux" = "yes"])
+AC_MSG_RESULT(${enable_libwebpmux-no})
+AM_CONDITIONAL([WANT_MUX], [test "$enable_libwebpmux" = "yes"])

 dnl === Check whether libwebpdemux should be built
 AC_MSG_CHECKING(whether libwebpdemux is to be built)
-AC_ARG_ENABLE([experimental-libwebpdemux],
-              AS_HELP_STRING([--enable-experimental-libwebpdemux],
+AC_ARG_ENABLE([libwebpdemux],
+              AS_HELP_STRING([--enable-libwebpdemux],
                             [Build libwebpdemux @<:@default=no@:>@]))
-AC_MSG_RESULT(${enable_experimental_libwebpdemux-no})
-AM_CONDITIONAL([WANT_DEMUX], [test "$enable_experimental_libwebpdemux" = "yes"])
+AC_MSG_RESULT(${enable_libwebpdemux-no})
+AM_CONDITIONAL([WANT_DEMUX], [test "$enable_libwebpdemux" = "yes"])

 dnl === Check whether decoder library should be built.
 AC_MSG_CHECKING(whether decoder library is to be built)
@ -283,7 +283,8 @@ AC_CONFIG_FILES([Makefile src/Makefile man/Makefile \
                 src/enc/Makefile src/dsp/Makefile \
                 src/demux/Makefile src/mux/Makefile \
                 src/utils/Makefile \
-                 src/libwebp.pc src/libwebpdecoder.pc])
+                 src/libwebp.pc src/libwebpdecoder.pc \
+                 src/demux/libwebpdemux.pc src/mux/libwebpmux.pc])


 AC_OUTPUT
@ -297,8 +298,8 @@ Static libraries: ${enable_static}
 Threaded decode: ${enable_threading-no}
 libwebp: yes
 libwebpdecoder: ${enable_libwebpdecoder-no}
-libwebpdemux: ${enable_experimental_libwebpdemux-no}
-libwebpmux: ${enable_experimental_libwebpmux-no}
+libwebpdemux: ${enable_libwebpdemux-no}
+libwebpmux: ${enable_libwebpmux-no}

 Tools:
 cwebp: yes
@ -314,5 +315,5 @@ dwebp: yes
  PNG  : ${png_support-no}
  WIC  : ${wic_support-no}
 gif2webp: ${build_gif2webp-no}
-webpmux: ${enable_experimental_libwebpmux-no}
+webpmux: ${enable_libwebpmux-no}
 ])
--- a/doc/webp-container-spec.txt
+++ b/doc/webp-container-spec.txt
@ -60,9 +60,9 @@ The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
 document are to be interpreted as described in [RFC 2119][].

 **Note:** Out of the features mentioned above, lossy compression, lossless
-compression and transparency are finalized and are to be considered stable.
-On the other hand, metadata, color profile, animation and image fragmentation
-are experimental as of now, and are open to discussion, feedback and comments.
+compression, transparency, metadata, color profile and animation are finalized
+and are to be considered stable. On the other hand, image fragmentation is
+experimental as of now, and is open to discussion, feedback and comments.
 The same is indicated using annotation "_status: experimental_" in the relevant
 sections of this document.

@ -93,11 +93,21 @@ _uint32_

 : A 32-bit, little-endian, unsigned integer.

+_FourCC_
+
+: A _FourCC_ (four-character code) is a _uint32_ created by concatenating four
+  ASCII characters in little-endian order.
+
 _1-based_

 : An unsigned integer field storing values offset by `-1`. e.g., Such a field
 would store value _25_ as _24_.

+RIFF file format
+----------------
+The WebP file format is based on the RIFF (resource interchange file format)
+document format.
+
 The basic element of a RIFF file is a _chunk_. It consists of:

     0                   1                   2                   3
@ -112,12 +122,12 @@ The basic element of a RIFF file is a _chunk_. It consists of:

 Chunk FourCC: 32 bits

-: ASCII four character code or _chunk tag_ used for chunk identification.
+: ASCII four-character code used for chunk identification.

 Chunk Size: 32 bits (_uint32_)

-: The size of the chunk (_ckSize_) not including this field, the chunk
-  identifier and padding.
+: The size of the chunk not including this field, the chunk identifier or
+  padding.

 Chunk Payload: _Chunk Size_ bytes

@ -126,43 +136,13 @@ Chunk Payload: _Chunk Size_ bytes

 _ChunkHeader('ABCD')_

-: This is used to describe the fourcc and size header of individual
-  chunks, where 'ABCD' is the fourcc for the chunk. This element's
+: This is used to describe the _FourCC_ and _Chunk Size_ header of individual
+  chunks, where 'ABCD' is the FourCC for the chunk. This element's
  size is 8 bytes.

-: Note that, in this specification, all chunk tag characters are in
-  file order, not in byte order of a uint32 of any particular
-  architecture.
-
-_list of chunks_
-
-: A concatenation of multiple chunks.
-
-: We will refer to the first chunk as having _position_ 0, the second
-  as position 1, etc. By _chunk with index 0 among "ABCD"_ we mean
-  the first chunk among the chunks of type "ABCD" in the list, the
-  _chunk with index 1 among "ABCD"_ is the second such chunk, etc.
-
-A WebP file MUST begin with a single chunk with a tag 'RIFF'. All
-other defined chunks are contained within this chunk. The file SHOULD
-NOT contain anything after it.
-
-The maximum size of RIFF's _ckSize_ is 2^32 minus 10 bytes. The size
-of the whole file is at most 4GiB minus 2 bytes.
-
-**Note:** some RIFF libraries are said to have bugs when handling files
-larger than 1GiB or 2GiB. If you are using an existing library, check
-that it handles large files correctly.
-
-The first four bytes of the RIFF chunk contents (i.e., bytes 8-11 of the file)
-MUST be the ASCII string "WEBP". They are followed by a list of chunks. As the
-size of any chunk is even, the size of the RIFF chunk is also even.  The
-contents of the chunks in that list will be described in the following sections.
-
-**Note:** RIFF has a convention that all-uppercase chunks are standard
-chunks that apply to any RIFF file format, while chunks specific to a
-file format are all lowercase. WebP does not follow this convention.
-
+**Note:** RIFF has a convention that all-uppercase chunk FourCCs are standard
+chunks that apply to any RIFF file format, while FourCCs specific to a file
+format are all lowercase. WebP does not follow this convention.

 WebP file header
 ----------------
@ -183,12 +163,20 @@ WebP file header

 File Size: 32 bits (_uint32_)

-: The size of the file in bytes starting at offset 8.
+: The size of the file in bytes starting at offset 8. The maximum value of
+this field is 2^32 minus 10 bytes and thus the size of the whole file is at
+most 4GiB minus 2 bytes.

 'WEBP': 32 bits

 : The ASCII characters 'W' 'E' 'B' 'P'.

+A WebP file MUST begin with a RIFF header with the FourCC 'WEBP'. The file size
+in the header is the total size of the chunks that follow plus `4` bytes for
+the 'WEBP' FourCC. The file SHOULD NOT contain anything after it. As the size
+of any chunk is even, the size given by the RIFF header is also even. The
+contents of individual chunks will be described in the following sections.
+
 Simple file format (lossy)
 --------------------------

@ -274,15 +262,15 @@ An extended format file consists of:

  * A 'VP8X' chunk with information about features used in the file.

-  * An optional 'ICCP' chunk with color profile. _\[status: experimental\]_
+  * An optional 'ICCP' chunk with color profile.

-  * An optional 'ANIM' chunk with animation control data. _\[status: experimental\]_
+  * An optional 'ANIM' chunk with animation control data.

  * Image data.

-  * An optional 'EXIF' chunk with EXIF metadata. _\[status: experimental\]_
+  * An optional 'EXIF' chunk with EXIF metadata.

-  * An optional 'XMP ' chunk with XMP metadata. _\[status: experimental\]_
+  * An optional 'XMP ' chunk with XMP metadata.

  * An optional list of [unknown chunks](#unknown-chunks). _\[status: experimental\]_

@ -325,7 +313,7 @@ Reserved (Rsv): 2 bits

 : SHOULD be `0`.

-ICC profile (I): 1 bit _\[status: experimental\]_
+ICC profile (I): 1 bit

 : Set if the file contains an ICC profile.

@ -334,15 +322,15 @@ Alpha (L): 1 bit
 : Set if any of the frames of the image contain transparency information
 ("alpha").

-EXIF metadata (E): 1 bit _\[status: experimental\]_
+EXIF metadata (E): 1 bit

 : Set if the file contains EXIF metadata.

-XMP metadata (X): 1 bit _\[status: experimental\]_
+XMP metadata (X): 1 bit

 : Set if the file contains XMP metadata.

-Animation (A): 1 bit _\[status: experimental\]_
+Animation (A): 1 bit

 : Set if this is an animated image. Data in 'ANIM' and 'ANMF' chunks should be
 used to control the animation.
@ -371,11 +359,12 @@ Future specifications MAY add more fields.

 ### Chunks

-#### Animation _\[status: experimental\]_
+#### Animation

 An animation is controlled by ANIM and ANMF chunks.

 ANIM Chunk:
+{:#anim_chunk}

 For an animated image, this chunk contains the _global parameters_ of the
 animation.
@ -392,10 +381,14 @@ animation.

 Background Color: 32 bits (_uint32_)

-: The background color of the canvas in \[Blue, Green, Red, Alpha\] byte order.
-The background color is the color used for those pixels of the canvas that are
-not covered by a frame. Background color is also used when disposal method is
-`1`.
+: The default background color of the canvas in \[Blue, Green, Red, Alpha\]
+byte order. This color is used to fill the unused space on the canvas around the
+frames, as well as the transparent pixels of the first frame. Background color
+is also used when disposal method is `1`.
+
+**Note**: Viewers that have a preferred background against which to present the
+images (web browsers, for example) should ignore this value and use their
+preferred background color instead.

 Loop Count: 16 bits (_uint16_)

@ -457,14 +450,45 @@ Reserved: 7 bits

 Disposal method (D): 1 bit

-: Indicates how the area used by this frame is to be treated before rendering
-the next frame on canvas:
+: Indicates how _the current frame_ is to be treated after it has been displayed
+(before rendering the next frame) on the canvas:

-  * `0`: Do not dispose. Keep the area used by this frame as it is and render
-    the next frame on top of it.
+  * `0`: Do not dispose. Leave the canvas as is.

-  * `1`: Dispose to background color (also part of this chunk). Restore the
-    area used by this frame to background color before rendering the next frame.
+  * `1`: Dispose to background color. Fill the _rectangle_ on the canvas covered
+    by the _current frame_ with background color specified in the
+    [ANIM chunk](#anim_chunk).
+
+After disposing the current frame, render the next frame on the canvas using
+[alpha-blending](#alpha-blending). If the next frame does not have an alpha
+channel, assume alpha value of 255, effectively replacing the rectangle.
+
+**Notes**:
+
+  * The frame disposal only applies to the _frame rectangle_, that is, the
+    rectangle defined by _Frame X_, _Frame Y_, _frame width_ and _frame height_.
+    It may or may not cover the whole canvas.
+
+{:#alpha-blending}
+  * **Alpha-blending**:
+
+    Given that each of the R, G, B and A channels is 8-bit, and the RGB
+    channels are _not premultiplied_ by alpha, the formula for blending
+    'dst' onto 'src' is:
+
+~~~~~
+    blend.A = src.A + dst.A * (1 - src.A / 255)
+    if blend.A = 0 then
+      blend.RGB = 0
+    else
+      blend.RGB = (src.RGB * src.A +
+                   dst.RGB * dst.A * (1 - src.A / 255)) / blend.A
+~~~~~
+
+  * Alpha-blending SHOULD be done in linear color space, by taking into account
+    the [color profile](#color-profile) of the image. If the color profile is
+    not present, sRGB is to be assumed. (Note that sRGB also needs to be
+    linearized due to a gamma of ~2.2).

 Frame Data: _Chunk Size_ - `16` bytes

@ -478,6 +502,9 @@ Frame Data: _Chunk Size_ - `16` bytes

  * An optional list of [unknown chunks](#unknown-chunks).

+**Note**: The 'ANMF' payload, _Frame Data_ above, consists of individual
+_padded_ chunks as described by the [RIFF file format](#riff-file-format).
+
 #### Fragments _\[status: experimental\]_

 For images that are represented by fragments, this chunk contains data for
@ -644,7 +671,7 @@ The formats of VP8 and VP8L chunks are as described in sections
 [Simple file format (lossy)](#simple-file-format-lossy)
 and [Simple file format (lossless)](#simple-file-format-lossless) respectively.

-#### Color profile _\[status: experimental\]_
+#### Color profile

     0                   1                   2                   3
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
@ -666,7 +693,7 @@ See the [ICC Specification][iccspec] for details.

 If this chunk is not present, sRGB SHOULD be assumed.

-#### Metadata _\[status: experimental\]_
+#### Metadata

 Metadata can be stored in 'EXIF' or 'XMP ' chunks.

@ -727,14 +754,14 @@ A file MAY contain unknown chunks:
 Readers SHOULD ignore these chunks. Writers SHOULD preserve them in their
 original order (unless they specifically intend to modify these chunks).

-### Assembling the Canvas from fragments/frames _\[status: experimental\]_
+### Assembling the Canvas from fragments/frames

 Here we provide an overview of how a reader should assemble a canvas in case
 of a fragmented-image and in case of an animated image. The notation
 _VP8X.field_ means the field in the 'VP8X' chunk with the same description.

 Displaying a _fragmented image_ canvas MUST be equivalent to the following
-pseudocode:
+pseudocode: _\[status: experimental\]_

 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 assert VP8X.flags.hasFragments
--- a/examples/Makefile.am
+++ b/examples/Makefile.am
@ -11,9 +11,7 @@ endif

 noinst_LTLIBRARIES = libexampleutil.la

-libexampleutil_la_SOURCES = example_util.c
-libexampleutilinclude_HEADERS = example_util.h
-libexampleutilincludedir =
+libexampleutil_la_SOURCES = example_util.c example_util.h

 dwebp_SOURCES = dwebp.c stopwatch.h
 dwebp_CPPFLAGS  = $(AM_CPPFLAGS) $(USE_EXPERIMENTAL_CODE)
--- a/examples/cwebp.c
+++ b/examples/cwebp.c
@ -366,14 +366,30 @@ static int DumpPicture(const WebPPicture* const picture, const char* PGM_name) {

 enum {
  METADATA_EXIF = (1 << 0),
-  METADATA_ICCP = (1 << 1),
+  METADATA_ICC  = (1 << 1),
  METADATA_XMP  = (1 << 2),
-  METADATA_ALL  = METADATA_EXIF | METADATA_ICCP | METADATA_XMP
+  METADATA_ALL  = METADATA_EXIF | METADATA_ICC | METADATA_XMP
 };

 static const int kChunkHeaderSize = 8;
 static const int kTagSize = 4;

+static void PrintMetadataInfo(const Metadata* const metadata,
+                              int metadata_written) {
+  if (metadata == NULL || metadata_written == 0) return;
+
+  fprintf(stderr, "Metadata:\n");
+  if (metadata_written & METADATA_ICC) {
+    fprintf(stderr, "  * ICC profile:  %6d bytes\n", (int)metadata->iccp.size);
+  }
+  if (metadata_written & METADATA_EXIF) {
+    fprintf(stderr, "  * EXIF data:    %6d bytes\n", (int)metadata->exif.size);
+  }
+  if (metadata_written & METADATA_XMP) {
+    fprintf(stderr, "  * XMP data:     %6d bytes\n", (int)metadata->xmp.size);
+  }
+}
+
 // Outputs, in little endian, 'num' bytes from 'val' to 'out'.
 static int WriteLE(FILE* const out, uint32_t val, int num) {
  uint8_t buf[4];
@ -424,7 +440,8 @@ static int WriteWebPWithMetadata(FILE* const out,
                                 const WebPPicture* const picture,
                                 const WebPMemoryWriter* const memory_writer,
                                 const Metadata* const metadata,
-                                 int keep_metadata) {
+                                 int keep_metadata,
+                                 int* const metadata_written) {
  const char kVP8XHeader[] = "VP8X\x0a\x00\x00\x00";
  const int kAlphaFlag = 0x10;
  const int kEXIFFlag  = 0x08;
@ -439,13 +456,16 @@ static int WriteWebPWithMetadata(FILE* const out,
                                            !!(keep_metadata & METADATA_EXIF),
                                            kEXIFFlag, &flags, &metadata_size);
  const int write_iccp = UpdateFlagsAndSize(&metadata->iccp,
-                                            !!(keep_metadata & METADATA_ICCP),
+                                            !!(keep_metadata & METADATA_ICC),
                                            kICCPFlag, &flags, &metadata_size);
  const int write_xmp  = UpdateFlagsAndSize(&metadata->xmp,
                                            !!(keep_metadata & METADATA_XMP),
                                            kXMPFlag, &flags, &metadata_size);
  uint8_t* webp = memory_writer->mem;
  size_t webp_size = memory_writer->size;
+
+  *metadata_written = 0;
+
  if (webp_size < kMinSize) return 0;
  if (webp_size - kChunkHeaderSize + metadata_size > kMaxChunkPayload) {
    fprintf(stderr, "Error! Addition of metadata would exceed "
@ -482,11 +502,20 @@ static int WriteWebPWithMetadata(FILE* const out,
      ok = ok && WriteLE24(out, picture->width - 1);
      ok = ok && WriteLE24(out, picture->height - 1);
    }
-    if (write_iccp) ok = ok && WriteMetadataChunk(out, "ICCP", &metadata->iccp);
+    if (write_iccp) {
+      ok = ok && WriteMetadataChunk(out, "ICCP", &metadata->iccp);
+      *metadata_written |= METADATA_ICC;
+    }
    // Image
    ok = ok && (fwrite(webp, webp_size, 1, out) == 1);
-    if (write_exif) ok = ok && WriteMetadataChunk(out, "EXIF", &metadata->exif);
-    if (write_xmp)  ok = ok && WriteMetadataChunk(out, "XMP ", &metadata->xmp);
+    if (write_exif) {
+      ok = ok && WriteMetadataChunk(out, "EXIF", &metadata->exif);
+      *metadata_written |= METADATA_EXIF;
+    }
+    if (write_xmp) {
+      ok = ok && WriteMetadataChunk(out, "XMP ", &metadata->xmp);
+      *metadata_written |= METADATA_XMP;
+    }
    return ok;
  } else {
    // No metadata, just write the original image file.
@ -575,7 +604,7 @@ static void HelpLong(void) {
  printf("                           ");
  printf("copy from the input to the output if present.\n");
  printf("                           "
-         "Valid values: all, none (default), exif, iccp, xmp\n");
+         "Valid values: all, none (default), exif, icc, xmp\n");

  printf("\n");
  printf("  -short ................. condense printed message\n");
@ -631,6 +660,7 @@ int main(int argc, const char *argv[]) {
  int resize_w = 0, resize_h = 0;
  int show_progress = 0;
  int keep_metadata = 0;
+  int metadata_written = 0;
  WebPPicture picture;
  int print_distortion = -1;        // -1=off, 0=PSNR, 1=SSIM, 2=LSIM
  WebPPicture original_picture;    // when PSNR or SSIM is requested
@ -812,7 +842,7 @@ int main(int argc, const char *argv[]) {
        { "all",  METADATA_ALL },
        { "none", 0 },
        { "exif", METADATA_EXIF },
-        { "iccp", METADATA_ICCP },
+        { "icc",  METADATA_ICC },
        { "xmp",  METADATA_XMP },
      };
      const size_t kNumTokens = sizeof(kTokens) / sizeof(kTokens[0]);
@ -844,7 +874,7 @@ int main(int argc, const char *argv[]) {
        start = token + 1;
      }
 #ifdef HAVE_WINCODEC_H
-      if (keep_metadata != 0 && keep_metadata != METADATA_ICCP) {
+      if (keep_metadata != 0 && keep_metadata != METADATA_ICC) {
        // TODO(jzern): remove when -metadata is supported on all platforms.
        fprintf(stderr, "Warning: only ICC profile extraction is currently"
                        " supported on this platform!\n");
@ -978,7 +1008,7 @@ int main(int argc, const char *argv[]) {

  if (keep_metadata != 0 && out != NULL) {
    if (!WriteWebPWithMetadata(out, &picture, &memory_writer,
-                               &metadata, keep_metadata)) {
+                               &metadata, keep_metadata, &metadata_written)) {
      fprintf(stderr, "Error writing WebP file with metadata!\n");
      goto Error;
    }
@ -990,6 +1020,9 @@ int main(int argc, const char *argv[]) {
    } else {
      PrintExtraInfoLossy(&picture, short_output, config.low_memory, in_file);
    }
+    if (!short_output) {
+      PrintMetadataInfo(&metadata, metadata_written);
+    }
  }
  if (!quiet && !short_output && print_distortion >= 0) {  // print distortion
    static const char* distortion_names[] = { "PSNR", "SSIM", "LSIM" };
--- a/examples/dwebp.c
+++ b/examples/dwebp.c
@ -460,6 +460,13 @@ int main(int argc, const char *argv[]) {
      goto end;
    }

+    if (bitstream->has_animation) {
+      fprintf(stderr,
+              "Error! Decoding of an animated WebP file is not supported.\n"
+              "       Use webpmux to extract the individual frames or\n"
+              "       vwebp to view this image.\n");
+    }
+
    switch (format) {
      case PNG:
 #ifdef HAVE_WINCODEC_H
--- a/examples/example_util.c
+++ b/examples/example_util.c
@ -44,8 +44,8 @@ int ExUtilReadFile(const char* const file_name,
  fclose(in);

  if (!ok) {
-    fprintf(stderr, "Could not read %zu bytes of data from file %s\n",
-            file_size, file_name);
+    fprintf(stderr, "Could not read %d bytes of data from file %s\n",
+            (int)file_size, file_name);
    free(file_data);
    return 0;
  }
--- a/examples/gif2webp.c
+++ b/examples/gif2webp.c
@ -117,36 +117,42 @@ static int ReadSubImage(GifFileType* gif, WebPPicture* pic, WebPPicture* view) {
  return ok;
 }

-static int GetColorFromIndex(const ColorMapObject* const color_map, GifWord idx,
-                             uint32_t* const argb) {
-  assert(color_map != NULL && color_map->Colors != NULL);
-  if (idx >= color_map->ColorCount) {
-    return 0;  // Invalid index.
+static int GetBackgroundColor(const ColorMapObject* const color_map,
+                              GifWord bgcolor_idx, uint32_t* const bgcolor) {
+  if (transparent_index != -1 && bgcolor_idx == transparent_index) {
+    *bgcolor = TRANSPARENT_COLOR;  // Special case.
+    return 1;
+  } else if (color_map == NULL || color_map->Colors == NULL
+             || bgcolor_idx >= color_map->ColorCount) {
+    return 0;  // Invalid color map or index.
  } else {
-    const GifColorType color = color_map->Colors[idx];
-    *argb = (0xff        << 24)
-          | (color.Red   << 16)
-          | (color.Green <<  8)
-          | (color.Blue  <<  0);
+    const GifColorType color = color_map->Colors[bgcolor_idx];
+    *bgcolor = (0xff        << 24)
+             | (color.Red   << 16)
+             | (color.Green <<  8)
+             | (color.Blue  <<  0);
    return 1;
  }
 }

-static void DisplayGifError(const GifFileType* const gif) {
+static void DisplayGifError(const GifFileType* const gif, int gif_error) {
  // GIFLIB_MAJOR is only defined in libgif >= 4.2.0.
  // libgif 4.2.0 has retired PrintGifError() and added GifErrorString().
 #if defined(GIFLIB_MAJOR) && defined(GIFLIB_MINOR) && \
        ((GIFLIB_MAJOR == 4 && GIFLIB_MINOR >= 2) || GIFLIB_MAJOR > 4)
 #if GIFLIB_MAJOR >= 5
    // Static string actually, hence the const char* cast.
-    const char* error_str = (const char*)GifErrorString(gif->Error);
+    const char* error_str = (const char*)GifErrorString(
+        (gif == NULL) ? gif_error : gif->Error);
 #else
    const char* error_str = (const char*)GifErrorString();
+    (void)gif;
 #endif
    if (error_str == NULL) error_str = "Unknown error";
-    fprintf(stderr, "GIFLib Error: %s\n", error_str);
+    fprintf(stderr, "GIFLib Error %d: %s\n", gif_error, error_str);
 #else
-    fprintf(stderr, "GIFLib Error: ");
+    (void)gif;
+    fprintf(stderr, "GIFLib Error %d: ", gif_error);
    PrintGifError();
    fprintf(stderr, "\n");
 #endif
@ -184,7 +190,7 @@ static void Help(void) {

 int main(int argc, const char *argv[]) {
  int verbose = 0;
-  int gif_error = -1;
+  int gif_error = GIF_ERROR;
  WebPMuxError err = WEBP_MUX_OK;
  int ok = 0;
  const char *in_file = NULL, *out_file = NULL;
@ -196,6 +202,7 @@ int main(int argc, const char *argv[]) {
  WebPMuxFrameInfo frame;
  WebPMuxAnimParams anim = { WHITE_COLOR, 0 };

+  int is_first_frame = 1;
  int done;
  int c;
  int quiet = 0;
@ -280,14 +287,6 @@ int main(int argc, const char *argv[]) {
  picture.custom_ptr = &memory;
  if (!WebPPictureAlloc(&picture)) goto End;

-  if (gif->SColorMap != NULL &&
-      !GetColorFromIndex(gif->SColorMap, gif->SBackGroundColor,
-                         &anim.bgcolor)) {
-    fprintf(stderr, "GIF decode warning: invalid background color index. "
-            "Assuming white background.\n");
-  }
-  ClearPicture(&picture, anim.bgcolor);
-
  mux = WebPMuxNew();
  if (mux == NULL) {
    fprintf(stderr, "ERROR: could not create a mux object.\n");
@ -340,11 +339,11 @@ int main(int argc, const char *argv[]) {
          goto End;
        }
        if (verbose) {
-          fprintf(stderr, "Added frame %dx%d (offset:%d,%d duration:%d) ",
-                  view.width, view.height, frame.x_offset, frame.y_offset,
-                  frame.duration);
-          fprintf(stderr, "dispose:%d transparent index:%d\n",
-                  frame.dispose_method, transparent_index);
+          printf("Added frame %dx%d (offset:%d,%d duration:%d) ",
+                 view.width, view.height, frame.x_offset, frame.y_offset,
+                 frame.duration);
+          printf("dispose:%d transparent index:%d\n",
+                 frame.dispose_method, transparent_index);
        }
        WebPDataClear(&frame.bitstream);
        break;
@ -376,6 +375,15 @@ int main(int argc, const char *argv[]) {
                                 : WEBP_MUX_DISPOSE_NONE;
            }
            transparent_index = (flags & GIF_TRANSPARENT_MASK) ? data[4] : -1;
+            if (is_first_frame) {
+              if (!GetBackgroundColor(gif->SColorMap, gif->SBackGroundColor,
+                                      &anim.bgcolor)) {
+                fprintf(stderr, "GIF decode warning: invalid background color "
+                        "index. Assuming white background.\n");
+              }
+              ClearPicture(&picture, anim.bgcolor);
+              is_first_frame = 0;
+            }
            break;
          }
          case PLAINTEXT_EXT_FUNC_CODE: {
@ -389,7 +397,7 @@ int main(int argc, const char *argv[]) {
              if (data == NULL) goto End;  // Loop count sub-block missing.
              if (data[0] != 3 && data[1] != 1) break;   // wrong size/marker
              anim.loop_count = data[2] | (data[3] << 8);
-              if (verbose) fprintf(stderr, "Loop count: %d\n", anim.loop_count);
+              if (verbose) printf("Loop count: %d\n", anim.loop_count);
            } else if (!memcmp(data + 1, "XMP dataXMP", 11)) {
              // Read XMP metadata.
              WebPData xmp;
@ -398,7 +406,7 @@ int main(int argc, const char *argv[]) {
              xmp.bytes = (uint8_t*)data;
              xmp.size = data[0] + 1;
              WebPMuxSetChunk(mux, "XMP ", &xmp, 1);
-              if (verbose) fprintf(stderr, "XMP size: %zu\n", xmp.size);
+              if (verbose) printf("XMP size: %d\n", (int)xmp.size);
            } else if (!memcmp(data + 1, "ICCRGBG1012", 11)) {
              // Read ICC profile.
              WebPData icc;
@ -407,7 +415,7 @@ int main(int argc, const char *argv[]) {
              icc.bytes = (uint8_t*)data;
              icc.size = data[0] + 1;
              WebPMuxSetChunk(mux, "ICCP", &icc, 1);
-              if (verbose) fprintf(stderr, "ICC size: %zu\n", icc.size);
+              if (verbose) printf("ICC size: %d\n", (int)icc.size);
            }
            break;
          }
@ -451,12 +459,18 @@ int main(int argc, const char *argv[]) {
      fprintf(stderr, "Error writing output file: %s\n", out_file);
      goto End;
    }
-    if (!quiet) fprintf(stderr, "Saved output file: %s\n", out_file);
+    if (!quiet) {
+      printf("Saved output file: %s\n", out_file);
+    }
+  } else {
+    if (!quiet) {
+      printf("Nothing written; use -o flag to save the result.\n");
+    }
  }

  // All OK.
  ok = 1;
-  gif_error = 0;
+  gif_error = GIF_OK;

 End:
  WebPDataClear(&webp_data);
@ -464,8 +478,8 @@ int main(int argc, const char *argv[]) {
  WebPPictureFree(&picture);
  if (out != NULL && out_file != NULL) fclose(out);

-  if (gif_error != 0) {
-    DisplayGifError(gif);
+  if (gif_error != GIF_OK) {
+    DisplayGifError(gif, gif_error);
  }
  if (gif != NULL) {
    DGifCloseFile(gif);
--- a/examples/jpegdec.c
+++ b/examples/jpegdec.c
@ -79,9 +79,9 @@ static int StoreICCP(j_decompress_ptr dinfo, MetadataPayload* const iccp) {
      ICCPSegment* segment;

      if (segment_size == 0 || count == 0 || seq == 0) {
-        fprintf(stderr, "[ICCP] size (%zu) / count (%d) / sequence number (%d)"
+        fprintf(stderr, "[ICCP] size (%d) / count (%d) / sequence number (%d)"
                        " cannot be 0!\n",
-                segment_size, seq, count);
+                (int)segment_size, seq, count);
        return 0;
      }

--- a/examples/vwebp.c
+++ b/examples/vwebp.c
@ -8,7 +8,7 @@
 //  Simple WebP file viewer.
 //
 // Compiling on linux:
-//   sudo apt-get install libglut3-dev mesa-common-dev
+//   sudo apt-get install freeglut3-dev mesa-common-dev
 //   gcc -o vwebp vwebp.c -O3 -lwebp -lwebpmux -lglut -lGL -lpthread -lm
 // Compiling on Mac + XCode:
 //   gcc -o vwebp vwebp.c -lwebp -lwebpmux -framework GLUT -framework OpenGL
@ -19,9 +19,6 @@
 #include <stdlib.h>
 #include <string.h>

-#include "webp/decode.h"
-#include "webp/demux.h"
-
 #ifdef __APPLE__
 #include <GLUT/glut.h>
 #else
@ -31,6 +28,13 @@
 #endif
 #endif

+#ifdef WEBP_HAVE_QCMS
+#include <qcms.h>
+#endif
+
+#include "webp/decode.h"
+#include "webp/demux.h"
+
 #include "./example_util.h"

 #ifdef _MSC_VER
@ -42,9 +46,11 @@ static void Help(void);
 // Unfortunate global variables. Gathered into a struct for comfort.
 static struct {
  int has_animation;
+  int has_color_profile;
  int done;
  int decoding_error;
  int print_info;
+  int use_color_profile;

  int canvas_width, canvas_height;
  int loop_count;
@ -56,6 +62,12 @@ static struct {
  const WebPDecBuffer* pic;
  WebPDemuxer* dmux;
  WebPIterator frameiter;
+  struct {
+    int width, height;
+    int x_offset, y_offset;
+    enum WebPMuxAnimDispose dispose_method;
+  } prev_frame;
+  WebPChunkIterator iccp;
 } kParams;

 static void ClearPreviousPic(void) {
@ -67,10 +79,128 @@ static void ClearParams(void) {
  ClearPreviousPic();
  WebPDataClear(&kParams.data);
  WebPDemuxReleaseIterator(&kParams.frameiter);
+  WebPDemuxReleaseChunkIterator(&kParams.iccp);
  WebPDemuxDelete(kParams.dmux);
  kParams.dmux = NULL;
 }

+// -----------------------------------------------------------------------------
+// Color profile handling
+static int ApplyColorProfile(const WebPData* const profile,
+                             WebPDecBuffer* const rgba) {
+#ifdef WEBP_HAVE_QCMS
+  int i, ok = 0;
+  uint8_t* line;
+  uint8_t major_revision;
+  qcms_profile* input_profile = NULL;
+  qcms_profile* output_profile = NULL;
+  qcms_transform* transform = NULL;
+  const qcms_data_type input_type = QCMS_DATA_RGBA_8;
+  const qcms_data_type output_type = QCMS_DATA_RGBA_8;
+  const qcms_intent intent = QCMS_INTENT_DEFAULT;
+
+  if (profile == NULL || rgba == NULL) return 0;
+  if (profile->bytes == NULL || profile->size < 10) return 1;
+  major_revision = profile->bytes[8];
+
+  qcms_enable_iccv4();
+  input_profile = qcms_profile_from_memory(profile->bytes, profile->size);
+  // qcms_profile_is_bogus() is broken with ICCv4.
+  if (input_profile == NULL ||
+      (major_revision < 4 && qcms_profile_is_bogus(input_profile))) {
+    fprintf(stderr, "Color profile is bogus!\n");
+    goto Error;
+  }
+
+  output_profile = qcms_profile_sRGB();
+  if (output_profile == NULL) {
+    fprintf(stderr, "Error creating output color profile!\n");
+    goto Error;
+  }
+
+  qcms_profile_precache_output_transform(output_profile);
+  transform = qcms_transform_create(input_profile, input_type,
+                                    output_profile, output_type,
+                                    intent);
+  if (transform == NULL) {
+    fprintf(stderr, "Error creating color transform!\n");
+    goto Error;
+  }
+
+  line = rgba->u.RGBA.rgba;
+  for (i = 0; i < rgba->height; ++i, line += rgba->u.RGBA.stride) {
+    qcms_transform_data(transform, line, line, rgba->width);
+  }
+  ok = 1;
+
+ Error:
+  if (input_profile != NULL) qcms_profile_release(input_profile);
+  if (output_profile != NULL) qcms_profile_release(output_profile);
+  if (transform != NULL) qcms_transform_release(transform);
+  return ok;
+#else
+  (void)profile;
+  (void)rgba;
+  return 1;
+#endif  // WEBP_HAVE_QCMS
+}
+
+//------------------------------------------------------------------------------
+// File decoding
+
+static int Decode(void) {   // Fills kParams.frameiter
+  const WebPIterator* const iter = &kParams.frameiter;
+  WebPDecoderConfig* const config = kParams.config;
+  WebPDecBuffer* const output_buffer = &config->output;
+  int ok = 0;
+
+  ClearPreviousPic();
+  output_buffer->colorspace = MODE_RGBA;
+  ok = (WebPDecode(iter->fragment.bytes, iter->fragment.size,
+                   config) == VP8_STATUS_OK);
+  if (!ok) {
+    fprintf(stderr, "Decoding of frame #%d failed!\n", iter->frame_num);
+  } else {
+    kParams.pic = output_buffer;
+    if (kParams.use_color_profile) {
+      ok = ApplyColorProfile(&kParams.iccp.chunk, output_buffer);
+      if (!ok) {
+        fprintf(stderr, "Applying color profile to frame #%d failed!\n",
+                iter->frame_num);
+      }
+    }
+  }
+  return ok;
+}
+
+static void decode_callback(int what) {
+  if (what == 0 && !kParams.done) {
+    int duration = 0;
+    if (kParams.dmux != NULL) {
+      WebPIterator* const iter = &kParams.frameiter;
+      if (!WebPDemuxNextFrame(iter)) {
+        WebPDemuxReleaseIterator(iter);
+        if (WebPDemuxGetFrame(kParams.dmux, 1, iter)) {
+          --kParams.loop_count;
+          kParams.done = (kParams.loop_count == 0);
+        } else {
+          kParams.decoding_error = 1;
+          kParams.done = 1;
+          return;
+        }
+      }
+      duration = iter->duration;
+    }
+    if (!Decode()) {
+      kParams.decoding_error = 1;
+      kParams.done = 1;
+    } else {
+      glutPostRedisplay();
+      glutTimerFunc(duration, decode_callback, what);
+    }
+  }
+}
+
 //------------------------------------------------------------------------------
 // Callbacks

@ -84,6 +214,24 @@ static void HandleKey(unsigned char key, int pos_x, int pos_y) {
    ClearParams();
    exit(0);
 #endif
+  } else if (key == 'c') {
+    if (kParams.has_color_profile && !kParams.decoding_error) {
+      kParams.use_color_profile = 1 - kParams.use_color_profile;
+
+      if (kParams.has_animation) {
+        // Restart the completed animation to pickup the color profile change.
+        if (kParams.done && kParams.loop_count == 0) {
+          kParams.loop_count =
+              (int)WebPDemuxGetI(kParams.dmux, WEBP_FF_LOOP_COUNT) + 1;
+          kParams.done = 0;
+          // Start the decode loop immediately.
+          glutTimerFunc(0, decode_callback, 0);
+        }
+      } else {
+        Decode();
+        glutPostRedisplay();
+      }
+    }
  } else if (key == 'i') {
    kParams.print_info = 1 - kParams.print_info;
    glutPostRedisplay();
@ -141,24 +289,46 @@ static void HandleDisplay(void) {
  glPixelZoom(1, -1);
  xoff = (GLfloat)(2. * iter->x_offset / kParams.canvas_width);
  yoff = (GLfloat)(2. * iter->y_offset / kParams.canvas_height);
-  glRasterPos2f(-1. + xoff, 1. - yoff);
+  glRasterPos2f(-1.f + xoff, 1.f - yoff);
  glPixelStorei(GL_UNPACK_ALIGNMENT, 1);
  glPixelStorei(GL_UNPACK_ROW_LENGTH, pic->u.RGBA.stride / 4);
-  if (iter->dispose_method == WEBP_MUX_DISPOSE_BACKGROUND) {
+
+  if (kParams.prev_frame.dispose_method == WEBP_MUX_DISPOSE_BACKGROUND) {
+    // TODO(later): these offsets and those above should factor in window size.
+    //              they will be incorrect if the window is resized.
+    // glScissor() takes window coordinates (0,0 at bottom left).
+    const int window_x = kParams.prev_frame.x_offset;
+    const int window_y = kParams.canvas_height -
+                         kParams.prev_frame.y_offset -
+                         kParams.prev_frame.height;
+    glEnable(GL_SCISSOR_TEST);
+    // Only updated the requested area, not the whole canvas.
+    glScissor(window_x, window_y,
+              kParams.prev_frame.width, kParams.prev_frame.height);
+
    glClear(GL_COLOR_BUFFER_BIT);  // use clear color
+    DrawCheckerBoard();
+
+    glDisable(GL_SCISSOR_TEST);
  }
+  kParams.prev_frame.width = iter->width;
+  kParams.prev_frame.height = iter->height;
+  kParams.prev_frame.x_offset = iter->x_offset;
+  kParams.prev_frame.y_offset = iter->y_offset;
+  kParams.prev_frame.dispose_method = iter->dispose_method;
+
  glDrawPixels(pic->width, pic->height,
               GL_RGBA, GL_UNSIGNED_BYTE,
               (GLvoid*)pic->u.RGBA.rgba);
  if (kParams.print_info) {
    char tmp[32];

-    glColor4f(0.90, 0.0, 0.90, 1.0);
+    glColor4f(0.90f, 0.0f, 0.90f, 1.0f);
    glRasterPos2f(-0.95f, 0.90f);
    PrintString(kParams.file_name);

    snprintf(tmp, sizeof(tmp), "Dimension:%d x %d", pic->width, pic->height);
-    glColor4f(0.90, 0.0, 0.90, 1.0);
+    glColor4f(0.90f, 0.0f, 0.90f, 1.0f);
    glRasterPos2f(-0.95f, 0.80f);
    PrintString(tmp);
    if (iter->x_offset != 0 || iter->y_offset != 0) {
@ -192,55 +362,6 @@ static void StartDisplay(void) {
  DrawCheckerBoard();
 }

-//------------------------------------------------------------------------------
-// File decoding
-
-static int Decode(void) {   // Fills kParams.frameiter
-  const WebPIterator* const iter = &kParams.frameiter;
-  WebPDecoderConfig* const config = kParams.config;
-  WebPDecBuffer* const output_buffer = &config->output;
-  int ok = 0;
-
-  ClearPreviousPic();
-  output_buffer->colorspace = MODE_RGBA;
-  ok = (WebPDecode(iter->fragment.bytes, iter->fragment.size,
-                   config) == VP8_STATUS_OK);
-  if (!ok) {
-    fprintf(stderr, "Decoding of frame #%d failed!\n", iter->frame_num);
-  } else {
-    kParams.pic = output_buffer;
-  }
-  return ok;
-}
-
-static void decode_callback(int what) {
-  if (what == 0 && !kParams.done) {
-    int duration = 0;
-    if (kParams.dmux != NULL) {
-      WebPIterator* const iter = &kParams.frameiter;
-      if (!WebPDemuxNextFrame(iter)) {
-        WebPDemuxReleaseIterator(iter);
-        if (WebPDemuxGetFrame(kParams.dmux, 1, iter)) {
-          --kParams.loop_count;
-          kParams.done = (kParams.loop_count == 0);
-        } else {
-          kParams.decoding_error = 1;
-          kParams.done = 1;
-          return;
-        }
-      }
-      duration = iter->duration;
-    }
-    if (!Decode()) {
-      kParams.decoding_error = 1;
-      kParams.done = 1;
-    } else {
-      glutPostRedisplay();
-      glutTimerFunc(duration, decode_callback, what);
-    }
-  }
-}
-
 //------------------------------------------------------------------------------
 // Main

@ -249,6 +370,7 @@ static void Help(void) {
         "Decodes the WebP image file and visualize it using OpenGL\n"
         "Options are:\n"
         "  -version  .... print version number and exit.\n"
+         "  -noicc ....... don't use the icc profile if present.\n"
         "  -nofancy ..... don't use the fancy YUV420 upscaler.\n"
         "  -nofilter .... disable in-loop filtering.\n"
         "  -mt .......... use multi-threading.\n"
@ -256,6 +378,7 @@ static void Help(void) {
         "  -h     ....... this help message.\n"
         "\n"
         "Keyboard shortcuts:\n"
+         "  'c' ................ toggle use of color profile.\n"
         "  'i' ................ overlay file information.\n"
         "  'q' / 'Q' / ESC .... quit.\n"
        );
@ -270,11 +393,14 @@ int main(int argc, char *argv[]) {
    return -1;
  }
  kParams.config = &config;
+  kParams.use_color_profile = 1;

  for (c = 1; c < argc; ++c) {
    if (!strcmp(argv[c], "-h") || !strcmp(argv[c], "-help")) {
      Help();
      return 0;
+    } else if (!strcmp(argv[c], "-noicc")) {
+      kParams.use_color_profile = 0;
    } else if (!strcmp(argv[c], "-nofancy")) {
      config.options.no_fancy_upsampling = 1;
    } else if (!strcmp(argv[c], "-nofilter")) {
@ -327,6 +453,25 @@ int main(int argc, char *argv[]) {
    printf("Canvas: %d x %d\n", kParams.canvas_width, kParams.canvas_height);
  }

+  kParams.prev_frame.width = kParams.canvas_width;
+  kParams.prev_frame.height = kParams.canvas_height;
+  kParams.prev_frame.x_offset = kParams.prev_frame.y_offset = 0;
+  kParams.prev_frame.dispose_method = WEBP_MUX_DISPOSE_BACKGROUND;
+
+  memset(&kParams.iccp, 0, sizeof(kParams.iccp));
+  kParams.has_color_profile =
+      !!(WebPDemuxGetI(kParams.dmux, WEBP_FF_FORMAT_FLAGS) & ICCP_FLAG);
+  if (kParams.has_color_profile) {
+#ifdef WEBP_HAVE_QCMS
+    if (!WebPDemuxGetChunk(kParams.dmux, "ICCP", 1, &kParams.iccp)) goto Error;
+    printf("VP8X: Found color profile\n");
+#else
+    fprintf(stderr, "Warning: color profile present, but qcms is unavailable!\n"
+            "Build libqcms from Mozilla or Chromium and define WEBP_HAVE_QCMS "
+            "before building.\n");
+#endif
+  }
+
  if (!WebPDemuxGetFrame(kParams.dmux, 1, &kParams.frameiter)) goto Error;

  kParams.has_animation = (kParams.frameiter.num_frames > 1);
--- a/examples/webpmux.c
+++ b/examples/webpmux.c
@ -17,12 +17,6 @@
 /*  Usage examples:

  Create container WebP file:
-    webpmux -frgm fragment_1.webp +0+0 \
-            -frgm fragment_2.webp +960+0 \
-            -frgm fragment_3.webp +0+576 \
-            -frgm fragment_4.webp +960+576 \
-            -o out_fragment_container.webp
-
    webpmux -frame anim_1.webp +100+10+10   \
            -frame anim_2.webp +100+25+25+1 \
            -frame anim_3.webp +100+50+50+1 \
@ -52,6 +46,10 @@
    webpmux -version
 */

+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
+
 #include <assert.h>
 #include <stdio.h>
 #include <stdlib.h>
@ -185,6 +183,9 @@ static WebPMuxError DisplayInfo(const WebPMux* mux) {
  uint32_t flag;

  WebPMuxError err = WebPMuxGetFeatures(mux, &flag);
+#ifndef WEBP_EXPERIMENTAL_FEATURES
+  if (flag & FRAGMENTS_FLAG) err = WEBP_MUX_INVALID_ARGUMENT;
+#endif
  RETURN_IF_ERROR("Failed to retrieve features\n");

  if (flag == 0) {
@ -196,7 +197,7 @@ static WebPMuxError DisplayInfo(const WebPMux* mux) {
  printf("Features present:");
  if (flag & ANIMATION_FLAG) printf(" animation");
  if (flag & FRAGMENTS_FLAG) printf(" image fragments");
-  if (flag & ICCP_FLAG)      printf(" icc profile");
+  if (flag & ICCP_FLAG)      printf(" ICC profile");
  if (flag & EXIF_FLAG)      printf(" EXIF metadata");
  if (flag & XMP_FLAG)       printf(" XMP metadata");
  if (flag & ALPHA_FLAG)     printf(" transparency");
@ -231,7 +232,7 @@ static WebPMuxError DisplayInfo(const WebPMux* mux) {
        RETURN_IF_ERROR3("Failed to retrieve %s#%d\n", type_str, i);
        printf("%3d: %8d %8d ", i, frame.x_offset, frame.y_offset);
        if (is_anim) printf("%8d %7d ", frame.duration, frame.dispose_method);
-        printf("%10zu\n", frame.bitstream.size);
+        printf("%10d\n", (int)frame.bitstream.size);
        WebPDataClear(&frame.bitstream);
      }
    }
@ -241,28 +242,28 @@ static WebPMuxError DisplayInfo(const WebPMux* mux) {
    WebPData icc_profile;
    err = WebPMuxGetChunk(mux, "ICCP", &icc_profile);
    RETURN_IF_ERROR("Failed to retrieve the ICC profile\n");
-    printf("Size of the ICC profile data: %zu\n", icc_profile.size);
+    printf("Size of the ICC profile data: %d\n", (int)icc_profile.size);
  }

  if (flag & EXIF_FLAG) {
    WebPData exif;
    err = WebPMuxGetChunk(mux, "EXIF", &exif);
    RETURN_IF_ERROR("Failed to retrieve the EXIF metadata\n");
-    printf("Size of the EXIF metadata: %zu\n", exif.size);
+    printf("Size of the EXIF metadata: %d\n", (int)exif.size);
  }

  if (flag & XMP_FLAG) {
    WebPData xmp;
    err = WebPMuxGetChunk(mux, "XMP ", &xmp);
    RETURN_IF_ERROR("Failed to retrieve the XMP metadata\n");
-    printf("Size of the XMP metadata: %zu\n", xmp.size);
+    printf("Size of the XMP metadata: %d\n", (int)xmp.size);
  }

  if ((flag & ALPHA_FLAG) && !(flag & (ANIMATION_FLAG | FRAGMENTS_FLAG))) {
    WebPMuxFrameInfo image;
    err = WebPMuxGetFrame(mux, 1, &image);
    RETURN_IF_ERROR("Failed to retrieve the image\n");
-    printf("Size of the image (with alpha): %zu\n", image.bitstream.size);
+    printf("Size of the image (with alpha): %d\n", (int)image.bitstream.size);
  }

  return WEBP_MUX_OK;
@ -272,7 +273,9 @@ static void PrintHelp(void) {
  printf("Usage: webpmux -get GET_OPTIONS INPUT -o OUTPUT\n");
  printf("       webpmux -set SET_OPTIONS INPUT -o OUTPUT\n");
  printf("       webpmux -strip STRIP_OPTIONS INPUT -o OUTPUT\n");
+#ifdef WEBP_EXPERIMENTAL_FEATURES
  printf("       webpmux -frgm FRAGMENT_OPTIONS [-frgm...] -o OUTPUT\n");
+#endif
  printf("       webpmux -frame FRAME_OPTIONS [-frame...] [-loop LOOP_COUNT]"
         "\n");
  printf("               [-bgcolor BACKGROUND_COLOR] -o OUTPUT\n");
@ -286,7 +289,9 @@ static void PrintHelp(void) {
  printf("   icc       Get ICC profile.\n");
  printf("   exif      Get EXIF metadata.\n");
  printf("   xmp       Get XMP metadata.\n");
+#ifdef WEBP_EXPERIMENTAL_FEATURES
  printf("   frgm n    Get nth fragment.\n");
+#endif
  printf("   frame n   Get nth frame.\n");

  printf("\n");
@ -306,6 +311,7 @@ static void PrintHelp(void) {
  printf("   exif      Strip EXIF metadata.\n");
  printf("   xmp       Strip XMP metadata.\n");

+#ifdef WEBP_EXPERIMENTAL_FEATURES
  printf("\n");
  printf("FRAGMENT_OPTIONS(i):\n");
  printf(" Create fragmented image.\n");
@ -313,6 +319,7 @@ static void PrintHelp(void) {
  printf("   where:    'file_i' is the i'th fragment (WebP format),\n");
  printf("             'xi','yi' specify the image offset for this fragment."
         "\n");
+#endif

  printf("\n");
  printf("FRAME_OPTIONS(i):\n");
@ -375,7 +382,8 @@ static int WriteData(const char* filename, const WebPData* const webpdata) {
  if (fwrite(webpdata->bytes, webpdata->size, 1, fout) != 1) {
    fprintf(stderr, "Error writing file %s!\n", filename);
  } else {
-    fprintf(stderr, "Saved file %s (%zu bytes)\n", filename, webpdata->size);
+    fprintf(stderr, "Saved file %s (%d bytes)\n",
+            filename, (int)webpdata->size);
    ok = 1;
  }
  if (fout != stdout) fclose(fout);
@ -596,6 +604,7 @@ static int ParseCommandLine(int argc, const char* argv[],
        arg->params_ = argv[i + 1];
        ++feature_arg_index;
        i += 2;
+#ifdef WEBP_EXPERIMENTAL_FEATURES
      } else if (!strcmp(argv[i], "-frgm")) {
        CHECK_NUM_ARGS_LESS(3, ErrParse);
        if (ACTION_IS_NIL || config->action_type_ == ACTION_SET) {
@ -612,6 +621,7 @@ static int ParseCommandLine(int argc, const char* argv[],
        arg->params_ = argv[i + 2];
        ++feature_arg_index;
        i += 3;
+#endif
      } else if (!strcmp(argv[i], "-o")) {
        CHECK_NUM_ARGS_LESS(2, ErrParse);
        config->output_ = argv[i + 1];
@ -660,8 +670,12 @@ static int ParseCommandLine(int argc, const char* argv[],
        } else {
          ++i;
        }
+#ifdef WEBP_EXPERIMENTAL_FEATURES
      } else if ((!strcmp(argv[i], "frame") ||
                  !strcmp(argv[i], "frgm")) &&
+#else
+      } else if (!strcmp(argv[i], "frame") &&
+#endif
                  (config->action_type_ == ACTION_GET)) {
        CHECK_NUM_ARGS_LESS(2, ErrParse);
        feature->type_ = (!strcmp(argv[i], "frame")) ? FEATURE_ANMF :
@ -763,12 +777,12 @@ static int InitializeConfig(int argc, const char* argv[],
 // Processing.

 static int GetFrameFragment(const WebPMux* mux,
-                            const WebPMuxConfig* config, int isFrame) {
+                            const WebPMuxConfig* config, int is_frame) {
  WebPMuxError err = WEBP_MUX_OK;
  WebPMux* mux_single = NULL;
  long num = 0;
  int ok = 1;
-  const WebPChunkId id = isFrame ? WEBP_CHUNK_ANMF : WEBP_CHUNK_FRGM;
+  const WebPChunkId id = is_frame ? WEBP_CHUNK_ANMF : WEBP_CHUNK_FRGM;
  WebPMuxFrameInfo info;
  WebPDataInit(&info.bitstream);

--- a/makefile.unix
+++ b/makefile.unix
@ -168,7 +168,10 @@ LIBWEBPDEMUX_OBJS = $(DEMUX_OBJS)

 HDRS_INSTALLED = \
    src/webp/decode.h \
+    src/webp/demux.h \
    src/webp/encode.h \
+    src/webp/mux.h \
+    src/webp/mux_types.h \
    src/webp/types.h \

 HDRS = \
@ -191,22 +194,21 @@ HDRS = \
    src/utils/quant_levels_dec.h \
    src/utils/rescaler.h \
    src/utils/thread.h \
-    src/webp/demux.h \
    src/webp/format_constants.h \
-    src/webp/mux.h \
-    src/webp/mux_types.h \
    $(HDRS_INSTALLED) \

 OUT_LIBS = examples/libexample_util.a src/libwebpdecoder.a src/libwebp.a
 OUT_EXAMPLES = examples/cwebp examples/dwebp
+EXTRA_EXAMPLES = examples/gif2webp examples/vwebp examples/webpmux

 OUTPUT = $(OUT_LIBS) $(OUT_EXAMPLES)
 ifeq ($(MAKECMDGOALS),clean)
-  OUTPUT += examples/gif2webp examples/vwebp examples/webpmux
+  OUTPUT += $(EXTRA_EXAMPLES)
  OUTPUT += src/demux/libwebpdemux.a src/mux/libwebpmux.a
 endif

-all: ex
+ex: $(OUT_EXAMPLES)
+all: ex $(EXTRA_EXAMPLES)

 $(EX_FORMAT_DEC_OBJS): %.o: %.h

@ -222,8 +224,6 @@ src/demux/libwebpdemux.a: $(LIBWEBPDEMUX_OBJS)
 %.a:
 	$(AR) $(ARFLAGS) $@ $^

-ex: $(OUT_EXAMPLES)
-
 examples/cwebp: examples/cwebp.o $(EX_FORMAT_DEC_OBJS)
 examples/dwebp: examples/dwebp.o
 examples/gif2webp: examples/gif2webp.o
@ -242,18 +242,21 @@ examples/vwebp: EXTRA_LIBS += $(GL_LIBS)
 examples/webpmux: examples/libexample_util.a src/mux/libwebpmux.a
 examples/webpmux: src/libwebpdecoder.a

-$(OUT_EXAMPLES) examples/gif2webp examples/vwebp examples/webpmux:
+$(OUT_EXAMPLES) $(EXTRA_EXAMPLES):
 	$(CC) -o $@ $^ $(LDFLAGS)

 dist: DESTDIR := dist
+dist: OUT_EXAMPLES += $(EXTRA_EXAMPLES)
 dist: all
 	$(INSTALL) -m755 -d $(DESTDIR)/include/webp \
 	           $(DESTDIR)/doc $(DESTDIR)/lib
 	$(INSTALL) -m755 -s $(OUT_EXAMPLES) $(DESTDIR)
 	$(INSTALL) -m644 $(HDRS_INSTALLED) $(DESTDIR)/include/webp
 	$(INSTALL) -m644 src/libwebp.a $(DESTDIR)/lib
+	$(INSTALL) -m644 src/demux/libwebpdemux.a $(DESTDIR)/lib
+	$(INSTALL) -m644 src/mux/libwebpmux.a $(DESTDIR)/lib
 	umask 022; \
-	for m in man/[cd]webp.1; do \
+	for m in man/[cd]webp.1 man/gif2webp.1 man/webpmux.1; do \
 	  basenam=$$(basename $$m .1); \
 	  $(GROFF) -t -e -man -T utf8 $$m \
 	    | $(COL) -bx >$(DESTDIR)/doc/$${basenam}.txt; \
--- a/man/cwebp.1
+++ b/man/cwebp.1
@ -1,5 +1,5 @@
 .\"                                      Hey, EMACS: -*- nroff -*-
-.TH CWEBP 1 "March 8, 2013"
+.TH CWEBP 1 "March 13, 2013"
 .SH NAME
 cwebp \- compress an image file to a WebP file
 .SH SYNOPSIS
@ -200,7 +200,7 @@ Specify the hint about input image type. Possible values are:
 .BI \-metadata " string
 A comma separated list of metadata to copy from the input to the output if
 present.
-Valid values: \fBall\fP, \fBnone\fP, \fBexif\fP, \fBiccp\fP, \fBxmp\fP.
+Valid values: \fBall\fP, \fBnone\fP, \fBexif\fP, \fBicc\fP, \fBxmp\fP.
 The default is \fBnone\fP.

 Note: each input format may not support all combinations.
--- a/man/webpmux.1
+++ b/man/webpmux.1
@ -1,5 +1,5 @@
 .\"                                      Hey, EMACS: -*- nroff -*-
-.TH WEBPMUX 1 "February 26, 2013"
+.TH WEBPMUX 1 "March 16, 2013"
 .SH NAME
 webpmux \- command line tool to create WebP Mux/container file.
 .SH SYNOPSIS
@ -21,11 +21,6 @@ webpmux \- command line tool to create WebP Mux/container file.
 .B \-o
 .I OUTPUT
 .br
-.B webpmux \-frgm
-.I FRAGMENT_OPTIONS
-.B [ \-frgm ... ] \-o
-.I OUTPUT
-.br
 .B webpmux \-frame
 .I FRAME_OPTIONS
 .B [ \-frame ... ] [ \-loop
@ -64,9 +59,6 @@ Get EXIF metadata.
 .B xmp
 Get XMP metadata.
 .TP
-.BI frgm " n
-Get nth fragment.
-.TP
 .BI frame " n
 Get nth frame.

@ -98,12 +90,6 @@ Strip EXIF metadata.
 .B xmp
 Strip XMP metadata.

-.SS FRAGMENT_OPTIONS (\-frgm)
-.TP
-.I file_i +xi+yi
-Where: 'file_i' is the i'th fragment (WebP format) and 'xi','yi' specify the
-image offset for this fragment.
-
 .SS FRAME_OPTIONS (\-frame)
 .TP
 .I file_i +di[+xi+yi[+mi]]
@ -163,18 +149,13 @@ webpmux \-get exif exif_container.webp \-o image_metadata.exif
 .br
 webpmux \-strip exif exif_container.webp \-o without_exif.webp
 .br
-webpmux \-frame anim_1.webp +0+0+0 \-frame anim_2.webp +50+50+0 \-loop 10
+webpmux \-frame anim_1.webp +100 \-frame anim_2.webp +100+50+50 \-loop 10
 .br
 .RS 8
 \-bgcolor 255,255,255,255 \-o anim_container.webp
 .RE
 .br
 webpmux \-get frame 2 anim_container.webp \-o frame_2.webp
-.br
-webpmux \-tile tile_1.webp +0+0 \-tile tile_2.webp +960+0 \-tile tile_3.webp
-+0+576 \-tile tile_4.webp +960+576 \-o tile_container.webp
-.br
-webpmux \-get tile 2 tile_container.webp \-o tile_2.webp

 .SH AUTHORS
 \fBwebpmux\fP is written by the WebP team.
--- a/src/Makefile.am
+++ b/src/Makefile.am
@ -1,4 +1,6 @@
-SUBDIRS = dec enc dsp utils
+# The mux and demux libraries depend on libwebp, thus the '.' to force the
+# build order so it's available to them.
+SUBDIRS = dec enc dsp utils .
 if WANT_MUX
  SUBDIRS += mux
 endif
@ -13,11 +15,14 @@ if BUILD_LIBWEBPDECODER
  lib_LTLIBRARIES += libwebpdecoder.la
 endif

+common_HEADERS =
+common_HEADERS += webp/decode.h
+common_HEADERS += webp/types.h
+commondir = $(includedir)/webp
+
 libwebp_la_SOURCES =
 libwebpinclude_HEADERS =
-libwebpinclude_HEADERS += webp/decode.h
 libwebpinclude_HEADERS += webp/encode.h
-libwebpinclude_HEADERS += webp/types.h
 noinst_HEADERS =
 noinst_HEADERS += webp/format_constants.h

@ -31,15 +36,12 @@ libwebp_la_LIBADD += utils/libwebputils.la
 # other than the ones listed on the command line, i.e., after linking, it will
 # not have unresolved symbols. Some platforms (Windows among them) require all
 # symbols in shared libraries to be resolved at library creation.
-libwebp_la_LDFLAGS = -no-undefined -version-info 4:1:0
+libwebp_la_LDFLAGS = -no-undefined -version-info 4:2:0
 libwebpincludedir = $(includedir)/webp
 pkgconfig_DATA = libwebp.pc

 if BUILD_LIBWEBPDECODER
  libwebpdecoder_la_SOURCES =
-  libwebpdecoderinclude_HEADERS =
-  libwebpdecoderinclude_HEADERS += webp/decode.h
-  libwebpdecoderinclude_HEADERS += webp/types.h

  libwebpdecoder_la_LIBADD =
  libwebpdecoder_la_LIBADD += dec/libwebpdecode.la
@ -47,7 +49,6 @@ if BUILD_LIBWEBPDECODER
  libwebpdecoder_la_LIBADD += utils/libwebputilsdecode.la

  libwebpdecoder_la_LDFLAGS = -no-undefined -version-info 0:0:0
-  libwebpdecoderincludedir = $(includedir)/webp
  pkgconfig_DATA += libwebpdecoder.pc
 endif

--- a/src/dec/alpha.c
+++ b/src/dec/alpha.c
@ -44,7 +44,6 @@ static int DecodeAlpha(const uint8_t* data, size_t data_size,
                       int width, int height, int stride, uint8_t* output) {
  uint8_t* decoded_data = NULL;
  const size_t decoded_size = height * width;
-  uint8_t* unfiltered_data = NULL;
  WEBP_FILTER_TYPE filter;
  int pre_processing;
  int rsrv;
@ -83,29 +82,19 @@ static int DecodeAlpha(const uint8_t* data, size_t data_size,
  }

  if (ok) {
-    WebPFilterFunc unfilter_func = WebPUnfilters[filter];
+    WebPUnfilterFunc unfilter_func = WebPUnfilters[filter];
    if (unfilter_func != NULL) {
-      unfiltered_data = (uint8_t*)malloc(decoded_size);
-      if (unfiltered_data == NULL) {
-        ok = 0;
-        goto Error;
-      }
      // TODO(vikas): Implement on-the-fly decoding & filter mechanism to decode
      // and apply filter per image-row.
-      unfilter_func(decoded_data, width, height, 1, width, unfiltered_data);
-      // Construct raw_data (height x stride) from alpha data (height x width).
-      CopyPlane(unfiltered_data, width, output, stride, width, height);
-      free(unfiltered_data);
-    } else {
-      // Construct raw_data (height x stride) from alpha data (height x width).
-      CopyPlane(decoded_data, width, output, stride, width, height);
+      unfilter_func(width, height, width, decoded_data);
    }
+    // Construct raw_data (height x stride) from alpha data (height x width).
+    CopyPlane(decoded_data, width, output, stride, width, height);
    if (pre_processing == ALPHA_PREPROCESSED_LEVELS) {
      ok = DequantizeLevels(decoded_data, width, height);
    }
  }

- Error:
  if (method != ALPHA_NO_COMPRESSION) {
    free(decoded_data);
  }
--- a/src/dec/vp8i.h
+++ b/src/dec/vp8i.h
@ -27,8 +27,8 @@ extern "C" {

 // version numbers
 #define DEC_MAJ_VERSION 0
-#define DEC_MIN_VERSION 2
-#define DEC_REV_VERSION 1
+#define DEC_MIN_VERSION 3
+#define DEC_REV_VERSION 0

 #define ONLY_KEYFRAME_CODE      // to remove any code related to P-Frames

--- a/src/dec/webp.c
+++ b/src/dec/webp.c
@ -14,7 +14,7 @@
 #include "./vp8i.h"
 #include "./vp8li.h"
 #include "./webpi.h"
-#include "../webp/format_constants.h"
+#include "../webp/mux_types.h"  // ALPHA_FLAG

 #if defined(__cplusplus) || defined(c_plusplus)
 extern "C" {
@ -276,6 +276,7 @@ static VP8StatusCode ParseHeadersInternal(const uint8_t* data,
                                          int* const width,
                                          int* const height,
                                          int* const has_alpha,
+                                          int* const has_animation,
                                          WebPHeaderStructure* const headers) {
  int found_riff = 0;
  int found_vp8x = 0;
@ -309,6 +310,7 @@ static VP8StatusCode ParseHeadersInternal(const uint8_t* data,
      return VP8_STATUS_BITSTREAM_ERROR;
    }
    if (has_alpha != NULL) *has_alpha = !!(flags & ALPHA_FLAG);
+    if (has_animation != NULL) *has_animation = !!(flags & ANIMATION_FLAG);
    if (found_vp8x && headers == NULL) {
      return VP8_STATUS_OK;  // Return features from VP8X header.
    }
@ -370,10 +372,19 @@ static VP8StatusCode ParseHeadersInternal(const uint8_t* data,
 }

 VP8StatusCode WebPParseHeaders(WebPHeaderStructure* const headers) {
+  VP8StatusCode status;
+  int has_animation = 0;
  assert(headers != NULL);
  // fill out headers, ignore width/height/has_alpha.
-  return ParseHeadersInternal(headers->data, headers->data_size,
-                              NULL, NULL, NULL, headers);
+  status = ParseHeadersInternal(headers->data, headers->data_size,
+                                NULL, NULL, NULL, &has_animation, headers);
+  if (status == VP8_STATUS_OK || status == VP8_STATUS_NOT_ENOUGH_DATA) {
+    // TODO(jzern): full support of animation frames will require API additions.
+    if (has_animation) {
+      status = VP8_STATUS_UNSUPPORTED_FEATURE;
+    }
+  }
+  return status;
 }

 //------------------------------------------------------------------------------
@ -625,10 +636,11 @@ static VP8StatusCode GetFeatures(const uint8_t* const data, size_t data_size,
  }
  DefaultFeatures(features);

-  // Only parse enough of the data to retrieve width/height/has_alpha.
+  // Only parse enough of the data to retrieve the features.
  return ParseHeadersInternal(data, data_size,
                              &features->width, &features->height,
-                              &features->has_alpha, NULL);
+                              &features->has_alpha, &features->has_animation,
+                              NULL);
 }

 //------------------------------------------------------------------------------
--- a/src/dec/webpi.h
+++ b/src/dec/webpi.h
@ -61,10 +61,10 @@ typedef struct {
 } WebPHeaderStructure;

 // Skips over all valid chunks prior to the first VP8/VP8L frame header.
-// Returns VP8_STATUS_OK on success,
-//         VP8_STATUS_BITSTREAM_ERROR if an invalid header/chunk is found, and
-//         VP8_STATUS_NOT_ENOUGH_DATA if case of insufficient data.
-// In 'headers', compressed_size, offset, alpha_data, alpha_size and lossless
+// Returns: VP8_STATUS_OK, VP8_STATUS_BITSTREAM_ERROR (invalid header/chunk),
+// VP8_STATUS_NOT_ENOUGH_DATA (partial input) or VP8_STATUS_UNSUPPORTED_FEATURE
+// in the case of non-decodable features (animation for instance).
+// In 'headers', compressed_size, offset, alpha_data, alpha_size, and lossless
 // fields are updated appropriately upon success.
 VP8StatusCode WebPParseHeaders(WebPHeaderStructure* const headers);

--- a/src/demux/Makefile.am
+++ b/src/demux/Makefile.am
@ -6,7 +6,10 @@ libwebpdemux_la_SOURCES += demux.c

 libwebpdemuxinclude_HEADERS =
 libwebpdemuxinclude_HEADERS += ../webp/demux.h
+libwebpdemuxinclude_HEADERS += ../webp/mux_types.h
 libwebpdemuxinclude_HEADERS += ../webp/types.h

-libwebpdemux_la_LDFLAGS = -version-info 0:0:0
+libwebpdemux_la_LIBADD = ../libwebp.la
+libwebpdemux_la_LDFLAGS = -no-undefined -version-info 0:0:0
 libwebpdemuxincludedir = $(includedir)/webp
+pkgconfig_DATA = libwebpdemux.pc
--- a/src/demux/demux.c
+++ b/src/demux/demux.c
@ -8,6 +8,10 @@
 //  WebP container demux.
 //

+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
+
 #include <assert.h>
 #include <stdlib.h>
 #include <string.h>
@ -65,6 +69,7 @@ struct WebPDemuxer {
  uint32_t bgcolor_;
  int num_frames_;
  Frame* frames_;
+  Frame** frames_tail_;
  Chunk* chunks_;  // non-image chunks
 };

@ -179,15 +184,12 @@ static void AddChunk(WebPDemuxer* const dmux, Chunk* const chunk) {
 // Add a frame to the end of the list, ensuring the last frame is complete.
 // Returns true on success, false otherwise.
 static int AddFrame(WebPDemuxer* const dmux, Frame* const frame) {
-  const Frame* last_frame = NULL;
-  Frame** f = &dmux->frames_;
-  while (*f != NULL) {
-    last_frame = *f;
-    f = &(*f)->next_;
-  }
+  const Frame* const last_frame = *dmux->frames_tail_;
  if (last_frame != NULL && !last_frame->complete_) return 0;
-  *f = frame;
+
+  *dmux->frames_tail_ = frame;
  frame->next_ = NULL;
+  dmux->frames_tail_ = &frame->next_;
  return 1;
 }

@ -231,8 +233,10 @@ static ParseStatus StoreFrame(int frame_num, uint32_t min_size,
          goto Done;
        }
        break;
-      case MKFOURCC('V', 'P', '8', ' '):
      case MKFOURCC('V', 'P', '8', 'L'):
+        if (alpha_chunks > 0) return PARSE_ERROR;  // VP8L has its own alpha
+        // fall through
+      case MKFOURCC('V', 'P', '8', ' '):
        if (image_chunks == 0) {
          // Extract the bitstream features, tolerating failures when the data
          // is incomplete.
@ -295,7 +299,7 @@ static ParseStatus NewFrame(const MemBuffer* const mem,

 // Parse a 'ANMF' chunk and any image bearing chunks that immediately follow.
 // 'frame_chunk_size' is the previously validated, padded chunk size.
-static ParseStatus ParseFrame(
+static ParseStatus ParseAnimationFrame(
    WebPDemuxer* const dmux, uint32_t frame_chunk_size) {
  const int has_frames = !!(dmux->feature_flags_ & ANIMATION_FLAG);
  const uint32_t anmf_payload_size = frame_chunk_size - ANMF_CHUNK_SIZE;
@ -316,8 +320,8 @@ static ParseStatus ParseFrame(
    return PARSE_ERROR;
  }

-  // Store a frame only if the animation flag is set and all data for this frame
-  // is available.
+  // Store a frame only if the animation flag is set there is some data for
+  // this frame is available.
  status = StoreFrame(dmux->num_frames_ + 1, anmf_payload_size, mem, frame,
                      NULL);
  if (status != PARSE_ERROR && has_frames && frame->frame_num_ > 0) {
@ -333,6 +337,7 @@ static ParseStatus ParseFrame(
  return status;
 }

+#ifdef WEBP_EXPERIMENTAL_FEATURES
 // Parse a 'FRGM' chunk and any image bearing chunks that immediately follow.
 // 'fragment_chunk_size' is the previously validated, padded chunk size.
 static ParseStatus ParseFragment(WebPDemuxer* const dmux,
@ -347,12 +352,12 @@ static ParseStatus ParseFragment(WebPDemuxer* const dmux,
      NewFrame(mem, FRGM_CHUNK_SIZE, fragment_chunk_size, &frame);
  if (status != PARSE_OK) return status;

-  frame->is_fragment_  = 1;
+  frame->is_fragment_ = 1;
  frame->x_offset_ = 2 * ReadLE24s(mem);
  frame->y_offset_ = 2 * ReadLE24s(mem);

-  // Store a fragment only if the fragments flag is set and all data for this
-  // fragment is available.
+  // Store a fragment only if the fragments flag is set there is some data for
+  // this fragment is available.
  status = StoreFrame(frame_num, frgm_payload_size, mem, frame, NULL);
  if (status != PARSE_ERROR && has_fragments && frame->frame_num_ > 0) {
    added_fragment = AddFrame(dmux, frame);
@ -366,8 +371,9 @@ static ParseStatus ParseFragment(WebPDemuxer* const dmux,
  if (!added_fragment) free(frame);
  return status;
 }
+#endif  // WEBP_EXPERIMENTAL_FEATURES

-// General chunk storage starting with the header at 'start_offset' allowing
+// General chunk storage, starting with the header at 'start_offset', allowing
 // the user to request the payload via a fourcc string. 'size' includes the
 // header and the unpadded payload size.
 // Returns true on success, false otherwise.
@ -424,7 +430,7 @@ static ParseStatus ParseSingleImage(WebPDemuxer* const dmux) {
  frame = (Frame*)calloc(1, sizeof(*frame));
  if (frame == NULL) return PARSE_ERROR;

-  // For the single image case, we allow parsing of a partial frame. But we need
+  // For the single image case we allow parsing of a partial frame, but we need
  // at least CHUNK_HEADER_SIZE for parsing.
  status = StoreFrame(1, CHUNK_HEADER_SIZE, &dmux->mem_, frame,
                      &has_vp8l_alpha);
@ -500,6 +506,9 @@ static ParseStatus ParseVP8X(WebPDemuxer* const dmux) {
      case MKFOURCC('A', 'L', 'P', 'H'):
      case MKFOURCC('V', 'P', '8', ' '):
      case MKFOURCC('V', 'P', '8', 'L'): {
+        // check that this isn't an animation (all frames should be in an ANMF).
+        if (anim_chunks > 0) return PARSE_ERROR;
+
        Rewind(mem, CHUNK_HEADER_SIZE);
        status = ParseSingleImage(dmux);
        break;
@ -521,13 +530,16 @@ static ParseStatus ParseVP8X(WebPDemuxer* const dmux) {
        break;
      }
      case MKFOURCC('A', 'N', 'M', 'F'): {
-        status = ParseFrame(dmux, chunk_size_padded);
+        if (anim_chunks == 0) return PARSE_ERROR;  // 'ANIM' precedes frames.
+        status = ParseAnimationFrame(dmux, chunk_size_padded);
        break;
      }
+#ifdef WEBP_EXPERIMENTAL_FEATURES
      case MKFOURCC('F', 'R', 'G', 'M'): {
        status = ParseFragment(dmux, chunk_size_padded);
        break;
      }
+#endif
      case MKFOURCC('I', 'C', 'C', 'P'): {
        store_chunk = !!(dmux->feature_flags_ & ICCP_FLAG);
        goto Skip;
@ -615,6 +627,9 @@ static int IsValidExtendedFormat(const WebPDemuxer* const dmux) {

        if (f->width_ <= 0 || f->height_ <= 0) return 0;
      } else {
+        // There shouldn't be a partial frame in a complete file.
+        if (dmux->state_ == WEBP_DEMUX_DONE) return 0;
+
        // Ensure alpha precedes image bitstream.
        if (alpha->size_ > 0 && image->size_ > 0 &&
            alpha->offset_ > image->offset_) {
@ -643,6 +658,7 @@ static void InitDemux(WebPDemuxer* const dmux, const MemBuffer* const mem) {
  dmux->bgcolor_ = 0xFFFFFFFF;  // White background by default.
  dmux->canvas_width_ = -1;
  dmux->canvas_height_ = -1;
+  dmux->frames_tail_ = &dmux->frames_;
  dmux->mem_ = *mem;
 }

--- a/src/demux/libwebpdemux.pc.in
+++ b/src/demux/libwebpdemux.pc.in
@ -0,0 +1,11 @@
+prefix=@prefix@
+exec_prefix=@exec_prefix@
+libdir=@libdir@
+includedir=@includedir@
+
+Name: libwebpdemux
+Description: Library for parsing the WebP graphics format container
+Version: @PACKAGE_VERSION@
+Requires: libwebp >= 0.2.0
+Cflags: -I${includedir}
+Libs: -L${libdir} -lwebpdemux
--- a/src/dsp/Makefile.am
+++ b/src/dsp/Makefile.am
@ -5,7 +5,10 @@ if BUILD_LIBWEBPDECODER
  noinst_LTLIBRARIES += libwebpdspdecode.la
 endif

-COMMON_SOURCES  =
+common_HEADERS = ../webp/types.h
+commondir = $(includedir)/webp
+
+COMMON_SOURCES =
 COMMON_SOURCES += cpu.c
 COMMON_SOURCES += dec.c
 COMMON_SOURCES += dec_neon.c
@ -19,13 +22,12 @@ COMMON_SOURCES += upsampling_sse2.c
 COMMON_SOURCES += yuv.c
 COMMON_SOURCES += yuv.h

-ENC_SOURCES  =
+ENC_SOURCES =
 ENC_SOURCES += enc.c
 ENC_SOURCES += enc_neon.c
 ENC_SOURCES += enc_sse2.c

 libwebpdsp_la_SOURCES = $(COMMON_SOURCES) $(ENC_SOURCES)
-libwebpdspinclude_HEADERS = ../webp/types.h

 noinst_HEADERS =
 noinst_HEADERS += ../dec/decode_vp8.h
@ -33,13 +35,10 @@ noinst_HEADERS += ../webp/decode.h

 libwebpdsp_la_LDFLAGS = -lm
 libwebpdsp_la_CPPFLAGS = $(USE_EXPERIMENTAL_CODE) $(USE_SWAP_16BIT_CSP)
-libwebpdspincludedir = $(includedir)/webp

 if BUILD_LIBWEBPDECODER
  libwebpdspdecode_la_SOURCES = $(COMMON_SOURCES)
-  libwebpdspdecodeinclude_HEADERS = $(libwebpdspinclude_HEADERS)

  libwebpdspdecode_la_LDFLAGS = $(libwebpdsp_la_LDFLAGS)
  libwebpdspdecode_la_CPPFLAGS = $(libwebpdsp_la_CPPFLAGS)
-  libwebpdspdecodeincludedir = $(libwebpdspincludedir)
 endif
--- a/src/dsp/enc.c
+++ b/src/dsp/enc.c
@ -631,13 +631,13 @@ static int QuantizeBlock(int16_t in[16], int16_t out[16],
  for (; n < 16; ++n) {
    const int j = kZigzag[n];
    const int sign = (in[j] < 0);
-    int coeff = (sign ? -in[j] : in[j]) + mtx->sharpen_[j];
-    if (coeff > 2047) coeff = 2047;
+    const int coeff = (sign ? -in[j] : in[j]) + mtx->sharpen_[j];
    if (coeff > mtx->zthresh_[j]) {
      const int Q = mtx->q_[j];
      const int iQ = mtx->iq_[j];
      const int B = mtx->bias_[j];
      out[n] = QUANTDIV(coeff, iQ, B);
+      if (out[n] > MAX_LEVEL) out[n] = MAX_LEVEL;
      if (sign) out[n] = -out[n];
      in[j] = out[n] * Q;
      if (out[n]) last = n;
--- a/src/dsp/enc_sse2.c
+++ b/src/dsp/enc_sse2.c
@ -776,7 +776,7 @@ static int Disto16x16SSE2(const uint8_t* const a, const uint8_t* const b,
 // Simple quantization
 static int QuantizeBlockSSE2(int16_t in[16], int16_t out[16],
                             int n, const VP8Matrix* const mtx) {
-  const __m128i max_coeff_2047 = _mm_set1_epi16(2047);
+  const __m128i max_coeff_2047 = _mm_set1_epi16(MAX_LEVEL);
  const __m128i zero = _mm_setzero_si128();
  __m128i coeff0, coeff8;
  __m128i out0, out8;
@ -812,10 +812,6 @@ static int QuantizeBlockSSE2(int16_t in[16], int16_t out[16],
  coeff0 = _mm_add_epi16(coeff0, sharpen0);
  coeff8 = _mm_add_epi16(coeff8, sharpen8);

-  // if (coeff > 2047) coeff = 2047
-  coeff0 = _mm_min_epi16(coeff0, max_coeff_2047);
-  coeff8 = _mm_min_epi16(coeff8, max_coeff_2047);
-
  // out = (coeff * iQ + B) >> QFIX;
  {
    // doing calculations with 32b precision (QFIX=17)
@ -843,9 +839,14 @@ static int QuantizeBlockSSE2(int16_t in[16], int16_t out[16],
    out_04 = _mm_srai_epi32(out_04, QFIX);
    out_08 = _mm_srai_epi32(out_08, QFIX);
    out_12 = _mm_srai_epi32(out_12, QFIX);
+
    // pack result as 16b
    out0 = _mm_packs_epi32(out_00, out_04);
    out8 = _mm_packs_epi32(out_08, out_12);
+
+    // if (coeff > 2047) coeff = 2047
+    out0 = _mm_min_epi16(out0, max_coeff_2047);
+    out8 = _mm_min_epi16(out8, max_coeff_2047);
  }

  // get sign back (if (sign[j]) out_n = -out_n)
--- a/src/dsp/yuv.c
+++ b/src/dsp/yuv.c
@ -15,7 +15,7 @@
 extern "C" {
 #endif

-enum { YUV_HALF = 1 << (YUV_FIX - 1) };
+#ifdef WEBP_YUV_USE_TABLE

 int16_t VP8kVToR[256], VP8kUToB[256];
 int32_t VP8kVToG[256], VP8kUToG[256];
@ -62,6 +62,12 @@ void VP8YUVInit(void) {
  done = 1;
 }

+#else
+
+void VP8YUVInit(void) {}
+
+#endif  // WEBP_YUV_USE_TABLE
+
 #if defined(__cplusplus) || defined(c_plusplus)
 }    // extern "C"
 #endif
--- a/src/dsp/yuv.h
+++ b/src/dsp/yuv.h
@ -19,8 +19,9 @@
 //   G = 1.164 * (Y-16) - 0.813 * (V-128) - 0.391 * (U-128)
 //   B = 1.164 * (Y-16)                   + 2.018 * (U-128)
 // where Y is in the [16,235] range, and U/V in the [16,240] range.
-// But the common term 1.164 * (Y-16) can be handled as an offset in the
-// VP8kClip[] table. So the formulae should be read as:
+// In the table-lookup version (WEBP_YUV_USE_TABLE), the common factor
+// "1.164 * (Y-16)" can be handled as an offset in the VP8kClip[] table.
+// So in this case the formulae should be read as:
 //   R = 1.164 * [Y + 1.371 * (V-128)                  ] - 18.624
 //   G = 1.164 * [Y - 0.698 * (V-128) - 0.336 * (U-128)] - 18.624
 //   B = 1.164 * [Y                   + 1.733 * (U-128)] - 18.624
@ -33,6 +34,9 @@

 #include "../dec/decode_vp8.h"

+// Define the following to use the LUT-based code:
+#define WEBP_YUV_USE_TABLE
+
 #if defined(WEBP_EXPERIMENTAL_FEATURES)
 // Do NOT activate this feature for real compression. This is only experimental!
 // This flag is for comparison purpose against JPEG's "YUVj" natural colorspace.
@ -51,9 +55,14 @@ extern "C" {
 #endif

 enum { YUV_FIX = 16,                // fixed-point precision
+       YUV_HALF = 1 << (YUV_FIX - 1),
+       YUV_MASK = (256 << YUV_FIX) - 1,
       YUV_RANGE_MIN = -227,        // min value of r/g/b output
       YUV_RANGE_MAX = 256 + 226    // max value of r/g/b output
 };
+
+#ifdef WEBP_YUV_USE_TABLE
+
 extern int16_t VP8kVToR[256], VP8kUToB[256];
 extern int32_t VP8kVToG[256], VP8kUToG[256];
 extern uint8_t VP8kClip[YUV_RANGE_MAX - YUV_RANGE_MIN];
@ -69,6 +78,16 @@ static WEBP_INLINE void VP8YuvToRgb(uint8_t y, uint8_t u, uint8_t v,
  rgb[2] = VP8kClip[y + b_off - YUV_RANGE_MIN];
 }

+static WEBP_INLINE void VP8YuvToBgr(uint8_t y, uint8_t u, uint8_t v,
+                                    uint8_t* const bgr) {
+  const int r_off = VP8kVToR[v];
+  const int g_off = (VP8kVToG[v] + VP8kUToG[u]) >> YUV_FIX;
+  const int b_off = VP8kUToB[u];
+  bgr[0] = VP8kClip[y + b_off - YUV_RANGE_MIN];
+  bgr[1] = VP8kClip[y + g_off - YUV_RANGE_MIN];
+  bgr[2] = VP8kClip[y + r_off - YUV_RANGE_MIN];
+}
+
 static WEBP_INLINE void VP8YuvToRgb565(uint8_t y, uint8_t u, uint8_t v,
                                       uint8_t* const rgb) {
  const int r_off = VP8kVToR[v];
@ -87,12 +106,6 @@ static WEBP_INLINE void VP8YuvToRgb565(uint8_t y, uint8_t u, uint8_t v,
 #endif
 }

-static WEBP_INLINE void VP8YuvToArgb(uint8_t y, uint8_t u, uint8_t v,
-                                     uint8_t* const argb) {
-  argb[0] = 0xff;
-  VP8YuvToRgb(y, u, v, argb + 1);
-}
-
 static WEBP_INLINE void VP8YuvToRgba4444(uint8_t y, uint8_t u, uint8_t v,
                                         uint8_t* const argb) {
  const int r_off = VP8kVToR[v];
@ -110,14 +123,92 @@ static WEBP_INLINE void VP8YuvToRgba4444(uint8_t y, uint8_t u, uint8_t v,
 #endif
 }

+#else   // Table-free version (slower on x86)
+
+// These constants are 16b fixed-point version of ITU-R BT.601 constants
+#define kYScale 76309      // 1.164 = 255 / 219
+#define kVToR   104597     // 1.596 = 255 / 112 * 0.701
+#define kUToG   25674      // 0.391 = 255 / 112 * 0.886 * 0.114 / 0.587
+#define kVToG   53278      // 0.813 = 255 / 112 * 0.701 * 0.299 / 0.587
+#define kUToB   132201     // 2.018 = 255 / 112 * 0.886
+#define kRCst (-kYScale * 16 - kVToR * 128 + YUV_HALF)
+#define kGCst (-kYScale * 16 + kUToG * 128 + kVToG * 128 + YUV_HALF)
+#define kBCst (-kYScale * 16 - kUToB * 128 + YUV_HALF)
+
+static WEBP_INLINE uint8_t VP8Clip8(int v) {
+  return ((v & ~YUV_MASK) == 0) ? (uint8_t)(v >> YUV_FIX)
+                                : (v < 0) ? 0u : 255u;
+}
+
+static WEBP_INLINE uint8_t VP8ClipN(int v, int N) {  // clip to N bits
+  return ((v & ~YUV_MASK) == 0) ? (uint8_t)(v >> (YUV_FIX + (8 - N)))
+                                : (v < 0) ? 0u : (255u >> (8 - N));
+}
+
+static WEBP_INLINE int VP8YUVToR(int y, int v) {
+  return kYScale * y + kVToR * v + kRCst;
+}
+
+static WEBP_INLINE int VP8YUVToG(int y, int u, int v) {
+  return kYScale * y - kUToG * u - kVToG * v + kGCst;
+}
+
+static WEBP_INLINE int VP8YUVToB(int y, int u) {
+  return kYScale * y  + kUToB * u + kBCst;
+}
+
+static WEBP_INLINE void VP8YuvToRgb(uint8_t y, uint8_t u, uint8_t v,
+                                    uint8_t* const rgb) {
+  rgb[0] = VP8Clip8(VP8YUVToR(y, v));
+  rgb[1] = VP8Clip8(VP8YUVToG(y, u, v));
+  rgb[2] = VP8Clip8(VP8YUVToB(y, u));
+}
+
 static WEBP_INLINE void VP8YuvToBgr(uint8_t y, uint8_t u, uint8_t v,
                                    uint8_t* const bgr) {
-  const int r_off = VP8kVToR[v];
-  const int g_off = (VP8kVToG[v] + VP8kUToG[u]) >> YUV_FIX;
-  const int b_off = VP8kUToB[u];
-  bgr[0] = VP8kClip[y + b_off - YUV_RANGE_MIN];
-  bgr[1] = VP8kClip[y + g_off - YUV_RANGE_MIN];
-  bgr[2] = VP8kClip[y + r_off - YUV_RANGE_MIN];
+  bgr[0] = VP8Clip8(VP8YUVToB(y, u));
+  bgr[1] = VP8Clip8(VP8YUVToG(y, u, v));
+  bgr[2] = VP8Clip8(VP8YUVToR(y, v));
+}
+
+static WEBP_INLINE void VP8YuvToRgb565(uint8_t y, uint8_t u, uint8_t v,
+                                       uint8_t* const rgb) {
+  const int r = VP8Clip8(VP8YUVToR(y, u));
+  const int g = VP8ClipN(VP8YUVToG(y, u, v), 6);
+  const int b = VP8ClipN(VP8YUVToB(y, v), 5);
+  const uint8_t rg = (r & 0xf8) | (g >> 3);
+  const uint8_t gb = (g << 5) | b;
+#ifdef WEBP_SWAP_16BIT_CSP
+  rgb[0] = gb;
+  rgb[1] = rg;
+#else
+  rgb[0] = rg;
+  rgb[1] = gb;
+#endif
+}
+
+static WEBP_INLINE void VP8YuvToRgba4444(uint8_t y, uint8_t u, uint8_t v,
+                                         uint8_t* const argb) {
+  const int r = VP8Clip8(VP8YUVToR(y, u));
+  const int g = VP8ClipN(VP8YUVToG(y, u, v), 4);
+  const int b = VP8Clip8(VP8YUVToB(y, v));
+  const uint8_t rg = (r & 0xf0) | g;
+  const uint8_t ba = b | 0x0f;   // overwrite the lower 4 bits
+#ifdef WEBP_SWAP_16BIT_CSP
+  argb[0] = ba;
+  argb[1] = rg;
+#else
+  argb[0] = rg;
+  argb[1] = ba;
+#endif
+}
+
+#endif  // WEBP_YUV_USE_TABLE
+
+static WEBP_INLINE void VP8YuvToArgb(uint8_t y, uint8_t u, uint8_t v,
+                                     uint8_t* const argb) {
+  argb[0] = 0xff;
+  VP8YuvToRgb(y, u, v, argb + 1);
 }

 static WEBP_INLINE void VP8YuvToBgra(uint8_t y, uint8_t u, uint8_t v,
--- a/src/enc/alpha.c
+++ b/src/enc/alpha.c
@ -127,8 +127,8 @@ static int EncodeAlphaInternal(const uint8_t* const data, int width, int height,
  VP8BitWriterAppend(bw, &header, ALPHA_HEADER_LEN);

  filter_func = WebPFilters[filter];
-  if (filter_func) {
-    filter_func(data, width, height, 1, width, tmp_alpha);
+  if (filter_func != NULL) {
+    filter_func(data, width, height, width, tmp_alpha);
    alpha_src = tmp_alpha;
  }  else {
    alpha_src = data;
--- a/src/enc/cost.c
+++ b/src/enc/cost.c
@ -75,7 +75,7 @@ const uint16_t VP8LevelCodes[MAX_VARIABLE_LEVEL][2] = {

 // fixed costs for coding levels, deduce from the coding tree.
 // This is only the part that doesn't depend on the probability state.
-const uint16_t VP8LevelFixedCosts[2048] = {
+const uint16_t VP8LevelFixedCosts[MAX_LEVEL + 1] = {
     0,  256,  256,  256,  256,  432,  618,  630,
   731,  640,  640,  828,  901,  948, 1021, 1101,
  1174, 1221, 1294, 1042, 1085, 1115, 1158, 1202,
--- a/src/enc/cost.h
+++ b/src/enc/cost.h
@ -18,7 +18,8 @@
 extern "C" {
 #endif

-extern const uint16_t VP8LevelFixedCosts[2048];   // approximate cost per level
+// approximate cost per level:
+extern const uint16_t VP8LevelFixedCosts[MAX_LEVEL + 1];
 extern const uint16_t VP8EntropyCost[256];        // 8bit fixed-point log(p)

 // Cost of coding one event with probability 'proba'.
--- a/src/enc/frame.c
+++ b/src/enc/frame.c
@ -901,15 +901,21 @@ int VP8EncLoop(VP8Encoder* const enc) {
 // Single pass using Token Buffer.

 #if !defined(DISABLE_TOKEN_BUFFER)
+
+#define MIN_COUNT 96   // minimum number of macroblocks before updating stats
+
 int VP8EncTokenLoop(VP8Encoder* const enc) {
  int ok;
-  // refresh the proba 8 times per pass
-  const int max_count = (enc->mb_w_ * enc->mb_h_) >> 3;
-  int cnt = max_count;
+  // Roughly refresh the proba height times per pass
+  int max_count = (enc->mb_w_ * enc->mb_h_) >> 3;
+  int cnt;
  VP8EncIterator it;
  VP8Proba* const proba = &enc->proba_;
  const VP8RDLevel rd_opt = enc->rd_opt_level_;

+  if (max_count < MIN_COUNT) max_count = MIN_COUNT;
+  cnt = max_count;
+
  assert(enc->num_parts_ == 1);
  assert(enc->use_tokens_);
  assert(proba->use_skip_proba_ == 0);
--- a/src/enc/histogram.c
+++ b/src/enc/histogram.c
@ -98,8 +98,6 @@ void VP8LHistogramAddSinglePixOrCopy(VP8LHistogram* const histo,
  }
 }

-
-
 static double BitsEntropy(const int* const array, int n) {
  double retval = 0.;
  int sum = 0;
@ -149,25 +147,6 @@ static double BitsEntropy(const int* const array, int n) {
  }
 }

-double VP8LHistogramEstimateBitsBulk(const VP8LHistogram* const p) {
-  double retval = BitsEntropy(&p->literal_[0], VP8LHistogramNumCodes(p))
-                + BitsEntropy(&p->red_[0], 256)
-                + BitsEntropy(&p->blue_[0], 256)
-                + BitsEntropy(&p->alpha_[0], 256)
-                + BitsEntropy(&p->distance_[0], NUM_DISTANCE_CODES);
-  // Compute the extra bits cost.
-  int i;
-  for (i = 2; i < NUM_LENGTH_CODES - 2; ++i) {
-    retval +=
-        (i >> 1) * p->literal_[256 + i + 2];
-  }
-  for (i = 2; i < NUM_DISTANCE_CODES - 2; ++i) {
-    retval += (i >> 1) * p->distance_[i + 2];
-  }
-  return retval;
-}
-
-
 // Returns the cost encode the rle-encoded entropy code.
 // The constants in this function are experimental.
 static double HuffmanCost(const int* const population, int length) {
@ -207,19 +186,150 @@ static double HuffmanCost(const int* const population, int length) {
  return retval;
 }

-// Estimates the Huffman dictionary + other block overhead size.
-static double HistogramEstimateBitsHeader(const VP8LHistogram* const p) {
-  return HuffmanCost(&p->alpha_[0], 256) +
-         HuffmanCost(&p->red_[0], 256) +
-         HuffmanCost(&p->literal_[0], VP8LHistogramNumCodes(p)) +
-         HuffmanCost(&p->blue_[0], 256) +
-         HuffmanCost(&p->distance_[0], NUM_DISTANCE_CODES);
+static double PopulationCost(const int* const population, int length) {
+  return BitsEntropy(population, length) + HuffmanCost(population, length);
 }

-double VP8LHistogramEstimateBits(const VP8LHistogram* const p) {
-  return HistogramEstimateBitsHeader(p) + VP8LHistogramEstimateBitsBulk(p);
+static double ExtraCost(const int* const population, int length) {
+  int i;
+  double cost = 0.;
+  for (i = 2; i < length - 2; ++i) cost += (i >> 1) * population[i + 2];
+  return cost;
 }

+// Estimates the Entropy + Huffman + other block overhead size cost.
+double VP8LHistogramEstimateBits(const VP8LHistogram* const p) {
+  return PopulationCost(p->literal_, VP8LHistogramNumCodes(p))
+       + PopulationCost(p->red_, 256)
+       + PopulationCost(p->blue_, 256)
+       + PopulationCost(p->alpha_, 256)
+       + PopulationCost(p->distance_, NUM_DISTANCE_CODES)
+       + ExtraCost(p->literal_ + 256, NUM_LENGTH_CODES)
+       + ExtraCost(p->distance_, NUM_DISTANCE_CODES);
+}
+
+double VP8LHistogramEstimateBitsBulk(const VP8LHistogram* const p) {
+  return BitsEntropy(p->literal_, VP8LHistogramNumCodes(p))
+       + BitsEntropy(p->red_, 256)
+       + BitsEntropy(p->blue_, 256)
+       + BitsEntropy(p->alpha_, 256)
+       + BitsEntropy(p->distance_, NUM_DISTANCE_CODES)
+       + ExtraCost(p->literal_ + 256, NUM_LENGTH_CODES)
+       + ExtraCost(p->distance_, NUM_DISTANCE_CODES);
+}
+
+// -----------------------------------------------------------------------------
+// Various histogram combine/cost-eval functions
+
+// Adds 'in' histogram to 'out'
+static void HistogramAdd(const VP8LHistogram* const in,
+                         VP8LHistogram* const out) {
+  int i;
+  for (i = 0; i < PIX_OR_COPY_CODES_MAX; ++i) {
+    out->literal_[i] += in->literal_[i];
+  }
+  for (i = 0; i < NUM_DISTANCE_CODES; ++i) {
+    out->distance_[i] += in->distance_[i];
+  }
+  for (i = 0; i < 256; ++i) {
+    out->red_[i] += in->red_[i];
+    out->blue_[i] += in->blue_[i];
+    out->alpha_[i] += in->alpha_[i];
+  }
+}
+
+// Performs out = a + b, computing the cost C(a+b) - C(a) - C(b) while comparing
+// to the threshold value 'cost_threshold'. The score returned is
+//  Score = C(a+b) - C(a) - C(b), where C(a) + C(b) is known and fixed.
+// Since the previous score passed is 'cost_threshold', we only need to compare
+// the partial cost against 'cost_threshold + C(a) + C(b)' to possibly bail-out
+// early.
+static double HistogramAddEval(const VP8LHistogram* const a,
+                               const VP8LHistogram* const b,
+                               VP8LHistogram* const out,
+                               double cost_threshold) {
+  double cost = 0;
+  const double sum_cost = a->bit_cost_ + b->bit_cost_;
+  int i;
+
+  cost_threshold += sum_cost;
+
+  // palette_code_bits_ is part of the cost evaluation for literal_.
+  // TODO(skal): remove/simplify this palette_code_bits_?
+  out->palette_code_bits_ =
+      (a->palette_code_bits_ > b->palette_code_bits_) ? a->palette_code_bits_ :
+                                                        b->palette_code_bits_;
+  for (i = 0; i < PIX_OR_COPY_CODES_MAX; ++i) {
+    out->literal_[i] = a->literal_[i] + b->literal_[i];
+  }
+  cost += PopulationCost(out->literal_, VP8LHistogramNumCodes(out));
+  cost += ExtraCost(out->literal_ + 256, NUM_LENGTH_CODES);
+  if (cost > cost_threshold) return cost;
+
+  for (i = 0; i < 256; ++i) out->red_[i] = a->red_[i] + b->red_[i];
+  cost += PopulationCost(out->red_, 256);
+  if (cost > cost_threshold) return cost;
+
+  for (i = 0; i < 256; ++i) out->blue_[i] = a->blue_[i] + b->blue_[i];
+  cost += PopulationCost(out->blue_, 256);
+  if (cost > cost_threshold) return cost;
+
+  for (i = 0; i < NUM_DISTANCE_CODES; ++i) {
+    out->distance_[i] = a->distance_[i] + b->distance_[i];
+  }
+  cost += PopulationCost(out->distance_, NUM_DISTANCE_CODES);
+  cost += ExtraCost(out->distance_, NUM_DISTANCE_CODES);
+  if (cost > cost_threshold) return cost;
+
+  for (i = 0; i < 256; ++i) out->alpha_[i] = a->alpha_[i] + b->alpha_[i];
+  cost += PopulationCost(out->alpha_, 256);
+
+  out->bit_cost_ = cost;
+  return cost - sum_cost;
+}
+
+// Same as HistogramAddEval(), except that the resulting histogram
+// is not stored. Only the cost C(a+b) - C(a) is evaluated. We omit
+// the term C(b) which is constant over all the evaluations.
+static double HistogramAddThresh(const VP8LHistogram* const a,
+                                 const VP8LHistogram* const b,
+                                 double cost_threshold) {
+  int tmp[PIX_OR_COPY_CODES_MAX];  // <= max storage we'll need
+  int i;
+  double cost = -a->bit_cost_;
+
+  for (i = 0; i < PIX_OR_COPY_CODES_MAX; ++i) {
+    tmp[i] = a->literal_[i] + b->literal_[i];
+  }
+  // note that the tests are ordered so that the usually largest
+  // cost shares come first.
+  cost += PopulationCost(tmp, VP8LHistogramNumCodes(a));
+  cost += ExtraCost(tmp + 256, NUM_LENGTH_CODES);
+  if (cost > cost_threshold) return cost;
+
+  for (i = 0; i < 256; ++i) tmp[i] = a->red_[i] + b->red_[i];
+  cost += PopulationCost(tmp, 256);
+  if (cost > cost_threshold) return cost;
+
+  for (i = 0; i < 256; ++i) tmp[i] = a->blue_[i] + b->blue_[i];
+  cost += PopulationCost(tmp, 256);
+  if (cost > cost_threshold) return cost;
+
+  for (i = 0; i < NUM_DISTANCE_CODES; ++i) {
+    tmp[i] = a->distance_[i] + b->distance_[i];
+  }
+  cost += PopulationCost(tmp, NUM_DISTANCE_CODES);
+  cost += ExtraCost(tmp, NUM_DISTANCE_CODES);
+  if (cost > cost_threshold) return cost;
+
+  for (i = 0; i < 256; ++i) tmp[i] = a->alpha_[i] + b->alpha_[i];
+  cost += PopulationCost(tmp, 256);
+
+  return cost;
+}
+
+// -----------------------------------------------------------------------------
+
 static void HistogramBuildImage(int xsize, int histo_bits,
                                const VP8LBackwardRefs* const backward_refs,
                                VP8LHistogramSet* const image) {
@ -273,7 +383,7 @@ static int HistogramCombine(const VP8LHistogramSet* const in,
  // Collapse similar histograms in 'out'.
  for (iter = 0; iter < outer_iters && out_size >= min_cluster_size; ++iter) {
    double best_cost_diff = 0.;
-    int best_idx1 = 0, best_idx2 = 1;
+    int best_idx1 = -1, best_idx2 = 1;
    int j;
    const int num_tries = (num_pairs < out_size) ? num_pairs : out_size;
    seed += iter;
@ -281,20 +391,17 @@ static int HistogramCombine(const VP8LHistogramSet* const in,
      double curr_cost_diff;
      // Choose two histograms at random and try to combine them.
      const uint32_t idx1 = MyRand(&seed) % out_size;
-      const uint32_t tmp = ((j & 7) + 1) % (out_size - 1);
+      const uint32_t tmp = (j & 7) + 1;
      const uint32_t diff = (tmp < 3) ? tmp : MyRand(&seed) % (out_size - 1);
      const uint32_t idx2 = (idx1 + diff + 1) % out_size;
      if (idx1 == idx2) {
        continue;
      }
-      *cur_combo = *out->histograms[idx1];
-      VP8LHistogramAdd(cur_combo, out->histograms[idx2]);
-      cur_combo->bit_cost_ = VP8LHistogramEstimateBits(cur_combo);
      // Calculate cost reduction on combining.
-      curr_cost_diff = cur_combo->bit_cost_
-                     - out->histograms[idx1]->bit_cost_
-                     - out->histograms[idx2]->bit_cost_;
-      if (best_cost_diff > curr_cost_diff) {    // found a better pair?
+      curr_cost_diff = HistogramAddEval(out->histograms[idx1],
+                                        out->histograms[idx2],
+                                        cur_combo, best_cost_diff);
+      if (curr_cost_diff < best_cost_diff) {    // found a better pair?
        {     // swap cur/best combo histograms
          VP8LHistogram* const tmp_histo = cur_combo;
          cur_combo = best_combo;
@ -306,7 +413,7 @@ static int HistogramCombine(const VP8LHistogramSet* const in,
      }
    }

-    if (best_cost_diff < 0.0) {
+    if (best_idx1 >= 0) {
      *out->histograms[best_idx1] = *best_combo;
      // swap best_idx2 slot with last one (which is now unused)
      --out_size;
@ -331,20 +438,11 @@ static int HistogramCombine(const VP8LHistogramSet* const in,
 // -----------------------------------------------------------------------------
 // Histogram refinement

-// What is the bit cost of moving square_histogram from
-// cur_symbol to candidate_symbol.
-// TODO(skal): we don't really need to copy the histogram and Add(). Instead
-// we just need VP8LDualHistogramEstimateBits(A, B) estimation function.
+// What is the bit cost of moving square_histogram from cur_symbol to candidate.
 static double HistogramDistance(const VP8LHistogram* const square_histogram,
-                                const VP8LHistogram* const candidate) {
-  const double previous_bit_cost = candidate->bit_cost_;
-  double new_bit_cost;
-  VP8LHistogram modified_histo;
-  modified_histo = *candidate;
-  VP8LHistogramAdd(&modified_histo, square_histogram);
-  new_bit_cost = VP8LHistogramEstimateBits(&modified_histo);
-
-  return new_bit_cost - previous_bit_cost;
+                                const VP8LHistogram* const candidate,
+                                double cost_threshold) {
+  return HistogramAddThresh(candidate, square_histogram, cost_threshold);
 }

 // Find the best 'out' histogram for each of the 'in' histograms.
@ -355,11 +453,12 @@ static void HistogramRemap(const VP8LHistogramSet* const in,
  int i;
  for (i = 0; i < in->size; ++i) {
    int best_out = 0;
-    double best_bits = HistogramDistance(in->histograms[i], out->histograms[0]);
+    double best_bits =
+        HistogramDistance(in->histograms[i], out->histograms[0], 1.e38);
    int k;
    for (k = 1; k < out->size; ++k) {
      const double cur_bits =
-          HistogramDistance(in->histograms[i], out->histograms[k]);
+          HistogramDistance(in->histograms[i], out->histograms[k], best_bits);
      if (cur_bits < best_bits) {
        best_bits = cur_bits;
        best_out = k;
@ -373,7 +472,7 @@ static void HistogramRemap(const VP8LHistogramSet* const in,
    HistogramClear(out->histograms[i]);
  }
  for (i = 0; i < in->size; ++i) {
-    VP8LHistogramAdd(out->histograms[symbols[i]], in->histograms[i]);
+    HistogramAdd(in->histograms[i], out->histograms[symbols[i]]);
  }
 }

--- a/src/enc/histogram.h
+++ b/src/enc/histogram.h
@ -80,22 +80,6 @@ double VP8LHistogramEstimateBits(const VP8LHistogram* const p);
 // represent the entropy code itself.
 double VP8LHistogramEstimateBitsBulk(const VP8LHistogram* const p);

-static WEBP_INLINE void VP8LHistogramAdd(VP8LHistogram* const p,
-                                         const VP8LHistogram* const a) {
-  int i;
-  for (i = 0; i < PIX_OR_COPY_CODES_MAX; ++i) {
-    p->literal_[i] += a->literal_[i];
-  }
-  for (i = 0; i < NUM_DISTANCE_CODES; ++i) {
-    p->distance_[i] += a->distance_[i];
-  }
-  for (i = 0; i < 256; ++i) {
-    p->red_[i] += a->red_[i];
-    p->blue_[i] += a->blue_[i];
-    p->alpha_[i] += a->alpha_[i];
-  }
-}
-
 static WEBP_INLINE int VP8LHistogramNumCodes(const VP8LHistogram* const p) {
  return 256 + NUM_LENGTH_CODES +
      ((p->palette_code_bits_ > 0) ? (1 << p->palette_code_bits_) : 0);
--- a/src/enc/syntax.c
+++ b/src/enc/syntax.c
@ -12,7 +12,8 @@
 #include <assert.h>

 #include "../utils/utils.h"
-#include "../webp/format_constants.h"
+#include "../webp/format_constants.h"  // RIFF constants
+#include "../webp/mux_types.h"         // ALPHA_FLAG
 #include "./vp8enci.h"

 #if defined(__cplusplus) || defined(c_plusplus)
--- a/src/enc/vp8enci.h
+++ b/src/enc/vp8enci.h
@ -27,8 +27,8 @@ extern "C" {

 // version numbers
 #define ENC_MAJ_VERSION 0
-#define ENC_MIN_VERSION 2
-#define ENC_REV_VERSION 1
+#define ENC_MIN_VERSION 3
+#define ENC_REV_VERSION 0

 // intra prediction modes
 enum { B_DC_PRED = 0,   // 4x4 modes
@ -55,8 +55,9 @@ enum { NUM_MB_SEGMENTS = 4,
       NUM_BANDS = 8,
       NUM_CTX = 3,
       NUM_PROBAS = 11,
-       MAX_LF_LEVELS = 64,      // Maximum loop filter level
-       MAX_VARIABLE_LEVEL = 67  // last (inclusive) level with variable cost
+       MAX_LF_LEVELS = 64,       // Maximum loop filter level
+       MAX_VARIABLE_LEVEL = 67,  // last (inclusive) level with variable cost
+       MAX_LEVEL = 2047          // max level (note: max codable is 2047 + 67)
     };

 typedef enum {   // Rate-distortion optimization levels
--- a/src/enc/vp8l.c
+++ b/src/enc/vp8l.c
@ -563,6 +563,9 @@ static int EncodeImageInternal(VP8LBitWriter* const bw,
      !GetHuffBitLengthsAndCodes(histogram_image, huffman_codes)) {
    goto Error;
  }
+  // Free combined histograms.
+  free(histogram_image);
+  histogram_image = NULL;

  // Color Cache parameters.
  VP8LWriteBits(bw, 1, use_color_cache);
@ -609,9 +612,6 @@ static int EncodeImageInternal(VP8LBitWriter* const bw,
      ClearHuffmanTreeIfOnlyOneSymbol(codes);
    }
  }
-  // Free combined histograms.
-  free(histogram_image);
-  histogram_image = NULL;

  // Store actual literals.
  StoreImageToBitMask(bw, width, histogram_bits, &refs,
@ -619,7 +619,7 @@ static int EncodeImageInternal(VP8LBitWriter* const bw,
  ok = 1;

 Error:
-  if (!ok) free(histogram_image);
+  free(histogram_image);

  VP8LClearBackwardRefs(&refs);
  if (huffman_codes != NULL) {
--- a/src/libwebp.pc.in
+++ b/src/libwebp.pc.in
@ -8,4 +8,4 @@ Description: Library for the WebP graphics format
 Version: @PACKAGE_VERSION@
 Cflags: -I${includedir}
 Libs: -L${libdir} -lwebp
-Libs.private: -lm
+Libs.private: -lm @PTHREAD_CFLAGS@ @PTHREAD_LIBS@
--- a/src/libwebpdecoder.pc.in
+++ b/src/libwebpdecoder.pc.in
@ -8,4 +8,4 @@ Description: Library for the WebP graphics format (decode only)
 Version: @PACKAGE_VERSION@
 Cflags: -I${includedir}
 Libs: -L${libdir} -lwebpdecoder
-Libs.private: -lm
+Libs.private: -lm @PTHREAD_CFLAGS@ @PTHREAD_LIBS@
--- a/src/mux/Makefile.am
+++ b/src/mux/Makefile.am
@ -9,7 +9,10 @@ libwebpmux_la_SOURCES += muxread.c

 libwebpmuxinclude_HEADERS =
 libwebpmuxinclude_HEADERS += ../webp/mux.h
+libwebpmuxinclude_HEADERS += ../webp/mux_types.h
 libwebpmuxinclude_HEADERS += ../webp/types.h

-libwebpmux_la_LDFLAGS = -version-info 0:0:0
+libwebpmux_la_LIBADD = ../libwebp.la
+libwebpmux_la_LDFLAGS = -no-undefined -version-info 0:0:0
 libwebpmuxincludedir = $(includedir)/webp
+pkgconfig_DATA = libwebpmux.pc
--- a/src/mux/libwebpmux.pc.in
+++ b/src/mux/libwebpmux.pc.in
@ -0,0 +1,11 @@
+prefix=@prefix@
+exec_prefix=@exec_prefix@
+libdir=@libdir@
+includedir=@includedir@
+
+Name: libwebpmux
+Description: Library for manipulating the WebP graphics format container
+Version: @PACKAGE_VERSION@
+Requires: libwebp >= 0.2.0
+Cflags: -I${includedir}
+Libs: -L${libdir} -lwebpmux
--- a/src/mux/muxedit.c
+++ b/src/mux/muxedit.c
@ -306,6 +306,11 @@ WebPMuxError WebPMuxPushFrame(WebPMux* mux, const WebPMuxFrameInfo* frame,
  if (!(is_frame || (frame->id == WEBP_CHUNK_FRGM))) {
    return WEBP_MUX_INVALID_ARGUMENT;
  }
+#ifndef WEBP_EXPERIMENTAL_FEATURES
+  if (frame->id == WEBP_CHUNK_FRGM) {     // disabled for now.
+    return WEBP_MUX_INVALID_ARGUMENT;
+  }
+#endif

  if (bitstream->bytes == NULL || bitstream->size > MAX_CHUNK_PAYLOAD) {
    return WEBP_MUX_INVALID_ARGUMENT;
--- a/src/mux/muxi.h
+++ b/src/mux/muxi.h
@ -121,7 +121,8 @@ WebPMuxError ChunkAssignData(WebPChunk* chunk, const WebPData* const data,

 // Sets 'chunk' at nth position in the 'chunk_list'.
 // nth = 0 has the special meaning "last of the list".
-WebPMuxError ChunkSetNth(const WebPChunk* chunk, WebPChunk** chunk_list,
+// On success ownership is transferred from 'chunk' to the 'chunk_list'.
+WebPMuxError ChunkSetNth(WebPChunk* chunk, WebPChunk** chunk_list,
                         uint32_t nth);

 // Releases chunk and returns chunk->next_.
--- a/src/mux/muxinternal.c
+++ b/src/mux/muxinternal.c
@ -96,7 +96,7 @@ CHUNK_INDEX ChunkGetIndexFromFourCC(const char fourcc[4]) {

 // Returns next chunk in the chunk list with the given tag.
 static WebPChunk* ChunkSearchNextInList(WebPChunk* chunk, uint32_t tag) {
-  while (chunk && chunk->tag_ != tag) {
+  while (chunk != NULL && chunk->tag_ != tag) {
    chunk = chunk->next_;
  }
  return chunk;
@ -105,7 +105,7 @@ static WebPChunk* ChunkSearchNextInList(WebPChunk* chunk, uint32_t tag) {
 WebPChunk* ChunkSearchList(WebPChunk* first, uint32_t nth, uint32_t tag) {
  uint32_t iter = nth;
  first = ChunkSearchNextInList(first, tag);
-  if (!first) return NULL;
+  if (first == NULL) return NULL;

  while (--iter != 0) {
    WebPChunk* next_chunk = ChunkSearchNextInList(first->next_, tag);
@ -121,10 +121,10 @@ WebPChunk* ChunkSearchList(WebPChunk* first, uint32_t nth, uint32_t tag) {
 static int ChunkSearchListToSet(WebPChunk** chunk_list, uint32_t nth,
                                WebPChunk*** const location) {
  uint32_t count = 0;
-  assert(chunk_list);
+  assert(chunk_list != NULL);
  *location = chunk_list;

-  while (*chunk_list) {
+  while (*chunk_list != NULL) {
    WebPChunk* const cur_chunk = *chunk_list;
    ++count;
    if (count == nth) return 1;  // Found.
@ -149,27 +149,18 @@ WebPMuxError ChunkAssignData(WebPChunk* chunk, const WebPData* const data,
  ChunkRelease(chunk);

  if (data != NULL) {
-    if (copy_data) {
-      // Copy data.
-      chunk->data_.bytes = (uint8_t*)malloc(data->size);
-      if (chunk->data_.bytes == NULL) return WEBP_MUX_MEMORY_ERROR;
-      memcpy((uint8_t*)chunk->data_.bytes, data->bytes, data->size);
-      chunk->data_.size = data->size;
-
-      // Chunk is owner of data.
-      chunk->owner_ = 1;
-    } else {
-      // Don't copy data.
+    if (copy_data) {        // Copy data.
+      if (!WebPDataCopy(data, &chunk->data_)) return WEBP_MUX_MEMORY_ERROR;
+      chunk->owner_ = 1;    // Chunk is owner of data.
+    } else {                // Don't copy data.
      chunk->data_ = *data;
    }
  }
-
  chunk->tag_ = tag;
-
  return WEBP_MUX_OK;
 }

-WebPMuxError ChunkSetNth(const WebPChunk* chunk, WebPChunk** chunk_list,
+WebPMuxError ChunkSetNth(WebPChunk* chunk, WebPChunk** chunk_list,
                         uint32_t nth) {
  WebPChunk* new_chunk;

@ -180,6 +171,7 @@ WebPMuxError ChunkSetNth(const WebPChunk* chunk, WebPChunk** chunk_list,
  new_chunk = (WebPChunk*)malloc(sizeof(*new_chunk));
  if (new_chunk == NULL) return WEBP_MUX_MEMORY_ERROR;
  *new_chunk = *chunk;
+  chunk->owner_ = 0;
  new_chunk->next_ = *chunk_list;
  *chunk_list = new_chunk;
  return WEBP_MUX_OK;
@ -199,7 +191,7 @@ WebPChunk* ChunkDelete(WebPChunk* const chunk) {

 size_t ChunksListDiskSize(const WebPChunk* chunk_list) {
  size_t size = 0;
-  while (chunk_list) {
+  while (chunk_list != NULL) {
    size += ChunkDiskSize(chunk_list);
    chunk_list = chunk_list->next_;
  }
@ -220,7 +212,7 @@ static uint8_t* ChunkEmit(const WebPChunk* const chunk, uint8_t* dst) {
 }

 uint8_t* ChunkListEmit(const WebPChunk* chunk_list, uint8_t* dst) {
-  while (chunk_list) {
+  while (chunk_list != NULL) {
    dst = ChunkEmit(chunk_list, dst);
    chunk_list = chunk_list->next_;
  }
@ -281,7 +273,7 @@ static int SearchImageToGetOrDelete(WebPMuxImage** wpi_list, uint32_t nth,
    if (nth == 0) return 0;  // Not found.
  }

-  while (*wpi_list) {
+  while (*wpi_list != NULL) {
    WebPMuxImage* const cur_wpi = *wpi_list;
    ++count;
    if (count == nth) return 1;  // Found.
@ -327,7 +319,7 @@ WebPMuxImage* MuxImageDelete(WebPMuxImage* const wpi) {
 }

 void MuxImageDeleteAll(WebPMuxImage** const wpi_list) {
-  while (*wpi_list) {
+  while (*wpi_list != NULL) {
    *wpi_list = MuxImageDelete(*wpi_list);
  }
 }
@ -370,7 +362,7 @@ size_t MuxImageDiskSize(const WebPMuxImage* const wpi) {

 size_t MuxImageListDiskSize(const WebPMuxImage* wpi_list) {
  size_t size = 0;
-  while (wpi_list) {
+  while (wpi_list != NULL) {
    size += MuxImageDiskSize(wpi_list);
    wpi_list = wpi_list->next_;
  }
@ -409,7 +401,7 @@ uint8_t* MuxImageEmit(const WebPMuxImage* const wpi, uint8_t* dst) {
 }

 uint8_t* MuxImageListEmit(const WebPMuxImage* wpi_list, uint8_t* dst) {
-  while (wpi_list) {
+  while (wpi_list != NULL) {
    dst = MuxImageEmit(wpi_list, dst);
    wpi_list = wpi_list->next_;
  }
--- a/src/mux/muxread.c
+++ b/src/mux/muxread.c
@ -223,7 +223,9 @@ WebPMux* WebPMuxCreateInternal(const WebPData* bitstream, int copy_data,
        MuxImageInit(wpi);  // Reset for reading next image.
        break;
      case WEBP_CHUNK_ANMF:
+#ifdef WEBP_EXPERIMENTAL_FEATURES
      case WEBP_CHUNK_FRGM:
+#endif
        if (wpi->is_partial_) goto Err;  // Previous wpi is still incomplete.
        if (!MuxImageParse(&chunk, copy_data, wpi)) goto Err;
        ChunkRelease(&chunk);
@ -378,6 +380,9 @@ static WebPMuxError MuxGetFrameFragmentInternal(const WebPMuxImage* const wpi,
  const int is_frame = (wpi->header_->tag_ == kChunks[IDX_ANMF].tag);
  const CHUNK_INDEX idx = is_frame ? IDX_ANMF : IDX_FRGM;
  const WebPData* frame_frgm_data;
+#ifndef WEBP_EXPERIMENTAL_FEATURES
+  if (!is_frame) return WEBP_MUX_INVALID_ARGUMENT;
+#endif
  assert(wpi->header_ != NULL);  // Already checked by WebPMuxGetFrame().
  // Get frame/fragment chunk.
  frame_frgm_data = &wpi->header_->data_;
--- a/src/utils/Makefile.am
+++ b/src/utils/Makefile.am
@ -5,7 +5,10 @@ if BUILD_LIBWEBPDECODER
  noinst_LTLIBRARIES += libwebputilsdecode.la
 endif

-COMMON_SOURCES  =
+common_HEADERS = ../webp/types.h
+commondir = $(includedir)/webp
+
+COMMON_SOURCES =
 COMMON_SOURCES += bit_reader.c
 COMMON_SOURCES += bit_reader.h
 COMMON_SOURCES += color_cache.c
@ -23,7 +26,7 @@ COMMON_SOURCES += thread.h
 COMMON_SOURCES += utils.c
 COMMON_SOURCES += utils.h

-ENC_SOURCES  =
+ENC_SOURCES =
 ENC_SOURCES += bit_writer.c
 ENC_SOURCES += bit_writer.h
 ENC_SOURCES += huffman_encode.c
@ -31,15 +34,8 @@ ENC_SOURCES += huffman_encode.h
 ENC_SOURCES += quant_levels.c
 ENC_SOURCES += quant_levels.h

-
 libwebputils_la_SOURCES = $(COMMON_SOURCES) $(ENC_SOURCES)

-libwebputilsinclude_HEADERS = ../webp/types.h
-libwebputilsincludedir = $(includedir)/webp
-
 if BUILD_LIBWEBPDECODER
  libwebputilsdecode_la_SOURCES = $(COMMON_SOURCES)
-
-  libwebputilsdecodeinclude_HEADERS = $(libwebputilsinclude_HEADERS)
-  libwebputilsdecodeincludedir = $(libwebputilsincludedir)
 endif
--- a/src/utils/bit_reader.h
+++ b/src/utils/bit_reader.h
@ -59,7 +59,7 @@ extern "C" {
 // The right-justify strategy tends to use less shifts and is often faster.

 //------------------------------------------------------------------------------
-// BITS can be either 32, 24, 16 or 8.
+// BITS can be any multiple of 8 from 8 to 56 (inclusive).
 // Pick values that fit natural register size.

 #if !defined(WEBP_REFERENCE_IMPLEMENTATION)
@ -68,7 +68,9 @@ extern "C" {

 #if defined(__i386__) || defined(_M_IX86)      // x86 32bit
 #define BITS 16
-#elif defined(__arm__) || defined(_M_ARM)     // ARM
+#elif defined(__x86_64__) || defined(_M_X64)   // x86 64bit
+#define BITS 56
+#elif defined(__arm__) || defined(_M_ARM)      // ARM
 #define BITS 24
 #else                      // reasonable default
 #define BITS 24
@ -84,9 +86,15 @@ extern "C" {
 //------------------------------------------------------------------------------
 // Derived types and constants

-#if (BITS == 32)
-typedef uint64_t bit_t;   // natural register type
-typedef uint32_t lbit_t;  // natural type for memory I/O
+// bit_t = natural register type
+// lbit_t = natural type for memory I/O
+
+#if (BITS > 32)
+typedef uint64_t bit_t;
+typedef uint64_t lbit_t;
+#elif (BITS == 32)
+typedef uint64_t bit_t;
+typedef uint32_t lbit_t;
 #elif (BITS == 24)
 typedef uint32_t bit_t;
 typedef uint32_t lbit_t;
@ -148,19 +156,36 @@ static WEBP_INLINE void VP8LoadNewBytes(VP8BitReader* const br) {
    lbit_t in_bits = *(lbit_t*)br->buf_;
    br->buf_ += (BITS) >> 3;
 #if !defined(__BIG_ENDIAN__)
-#if (BITS == 32) || (BITS == 24)
+#if (BITS > 32)
+// gcc 4.3 has builtin functions for swap32/swap64
+#if defined(__GNUC__) && \
+           (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 3))
+    bits = (bit_t)__builtin_bswap64(in_bits);
+#elif defined(_MSC_VER)
+    bits = (bit_t)_byteswap_uint64(in_bits);
+#elif defined(__x86_64__)
+    __asm__ volatile("bswapq %0" : "=r"(bits) : "0"(in_bits));
+#else  // generic code for swapping 64-bit values (suggested by bdb@)
+    bits = (bit_t)in_bits;
+    bits = ((bits & 0xffffffff00000000ull) >> 32) |
+           ((bits & 0x00000000ffffffffull) << 32);
+    bits = ((bits & 0xffff0000ffff0000ull) >> 16) |
+           ((bits & 0x0000ffff0000ffffull) << 16);
+    bits = ((bits & 0xff00ff00ff00ff00ull) >> 8) |
+           ((bits & 0x00ff00ff00ff00ffull) << 8);
+#endif
+    bits >>= 64 - BITS;
+#elif (BITS >= 24)
 #if defined(__i386__) || defined(__x86_64__)
    __asm__ volatile("bswap %k0" : "=r"(in_bits) : "0"(in_bits));
    bits = (bit_t)in_bits;   // 24b/32b -> 32b/64b zero-extension
 #elif defined(_MSC_VER)
-    bits = _byteswap_ulong(in_bits);
+    bits = (bit_t)_byteswap_ulong(in_bits);
 #else
    bits = (bit_t)(in_bits >> 24) | ((in_bits >> 8) & 0xff00)
         | ((in_bits << 8) & 0xff0000)  | (in_bits << 24);
 #endif  // x86
-#if (BITS == 24)
-    bits >>= 8;
-#endif
+    bits >>= (32 - BITS);
 #elif (BITS == 16)
    // gcc will recognize a 'rorw $8, ...' here:
    bits = (bit_t)(in_bits >> 8) | ((in_bits & 0xff) << 8);
@ -248,7 +273,7 @@ static WEBP_INLINE int VP8GetBit(VP8BitReader* const br, int prob) {
 }

 static WEBP_INLINE int VP8GetSigned(VP8BitReader* const br, int v) {
-  const bit_t split = (br->range_ >> 1);
+  const range_t split = (br->range_ >> 1);
  const int bit = VP8BitUpdate(br, split);
  VP8Shift(br);
  return bit ? -v : v;
--- a/src/utils/filters.c
+++ b/src/utils/filters.c
@ -26,8 +26,7 @@ extern "C" {
  assert(out != NULL);                                      \
  assert(width > 0);                                        \
  assert(height > 0);                                       \
-  assert(bpp > 0);                                          \
-  assert(stride >= width * bpp);
+  assert(stride >= width);

 static WEBP_INLINE void PredictLine(const uint8_t* src, const uint8_t* pred,
                                    uint8_t* dst, int length, int inverse) {
@ -43,7 +42,8 @@ static WEBP_INLINE void PredictLine(const uint8_t* src, const uint8_t* pred,
 // Horizontal filter.

 static WEBP_INLINE void DoHorizontalFilter(const uint8_t* in,
-    int width, int height, int bpp, int stride, int inverse, uint8_t* out) {
+                                           int width, int height, int stride,
+                                           int inverse, uint8_t* out) {
  int h;
  const uint8_t* preds = (inverse ? out : in);
  SANITY_CHECK(in, out);
@ -52,11 +52,11 @@ static WEBP_INLINE void DoHorizontalFilter(const uint8_t* in,
  for (h = 0; h < height; ++h) {
    // Leftmost pixel is predicted from above (except for topmost scanline).
    if (h == 0) {
-      memcpy((void*)out, (const void*)in, bpp);
+      out[0] = in[0];
    } else {
-      PredictLine(in, preds - stride, out, bpp, inverse);
+      PredictLine(in, preds - stride, out, 1, inverse);
    }
-    PredictLine(in + bpp, preds, out + bpp, bpp * (width - 1), inverse);
+    PredictLine(in + 1, preds, out + 1, width - 1, inverse);
    preds += stride;
    in += stride;
    out += stride;
@ -64,46 +64,46 @@ static WEBP_INLINE void DoHorizontalFilter(const uint8_t* in,
 }

 static void HorizontalFilter(const uint8_t* data, int width, int height,
-                             int bpp, int stride, uint8_t* filtered_data) {
-  DoHorizontalFilter(data, width, height, bpp, stride, 0, filtered_data);
+                             int stride, uint8_t* filtered_data) {
+  DoHorizontalFilter(data, width, height, stride, 0, filtered_data);
 }

-static void HorizontalUnfilter(const uint8_t* data, int width, int height,
-                               int bpp, int stride, uint8_t* recon_data) {
-  DoHorizontalFilter(data, width, height, bpp, stride, 1, recon_data);
+static void HorizontalUnfilter(int width, int height, int stride,
+                               uint8_t* data) {
+  DoHorizontalFilter(data, width, height, stride, 1, data);
 }

 //------------------------------------------------------------------------------
 // Vertical filter.

 static WEBP_INLINE void DoVerticalFilter(const uint8_t* in,
-    int width, int height, int bpp, int stride, int inverse, uint8_t* out) {
+                                         int width, int height, int stride,
+                                         int inverse, uint8_t* out) {
  int h;
  const uint8_t* preds = (inverse ? out : in);
  SANITY_CHECK(in, out);

  // Very first top-left pixel is copied.
-  memcpy((void*)out, (const void*)in, bpp);
+  out[0] = in[0];
  // Rest of top scan-line is left-predicted.
-  PredictLine(in + bpp, preds, out + bpp, bpp * (width - 1), inverse);
+  PredictLine(in + 1, preds, out + 1, width - 1, inverse);

  // Filter line-by-line.
  for (h = 1; h < height; ++h) {
    in += stride;
    out += stride;
-    PredictLine(in, preds, out, bpp * width, inverse);
+    PredictLine(in, preds, out, width, inverse);
    preds += stride;
  }
 }

 static void VerticalFilter(const uint8_t* data, int width, int height,
-                           int bpp, int stride, uint8_t* filtered_data) {
-  DoVerticalFilter(data, width, height, bpp, stride, 0, filtered_data);
+                           int stride, uint8_t* filtered_data) {
+  DoVerticalFilter(data, width, height, stride, 0, filtered_data);
 }

-static void VerticalUnfilter(const uint8_t* data, int width, int height,
-                             int bpp, int stride, uint8_t* recon_data) {
-  DoVerticalFilter(data, width, height, bpp, stride, 1, recon_data);
+static void VerticalUnfilter(int width, int height, int stride, uint8_t* data) {
+  DoVerticalFilter(data, width, height, stride, 1, data);
 }

 //------------------------------------------------------------------------------
@ -111,19 +111,19 @@ static void VerticalUnfilter(const uint8_t* data, int width, int height,

 static WEBP_INLINE int GradientPredictor(uint8_t a, uint8_t b, uint8_t c) {
  const int g = a + b - c;
-  return (g < 0) ? 0 : (g > 255) ? 255 : g;
+  return ((g & ~0xff) == 0) ? g : (g < 0) ? 0 : 255;  // clip to 8bit
 }

 static WEBP_INLINE
 void DoGradientFilter(const uint8_t* in, int width, int height,
-                      int bpp, int stride, int inverse, uint8_t* out) {
+                      int stride, int inverse, uint8_t* out) {
  const uint8_t* preds = (inverse ? out : in);
  int h;
  SANITY_CHECK(in, out);

  // left prediction for top scan-line
-  memcpy((void*)out, (const void*)in, bpp);
-  PredictLine(in + bpp, preds, out + bpp, bpp * (width - 1), inverse);
+  out[0] = in[0];
+  PredictLine(in + 1, preds, out + 1, width - 1, inverse);

  // Filter line-by-line.
  for (h = 1; h < height; ++h) {
@ -132,24 +132,23 @@ void DoGradientFilter(const uint8_t* in, int width, int height,
    in += stride;
    out += stride;
    // leftmost pixel: predict from above.
-    PredictLine(in, preds - stride, out, bpp, inverse);
-    for (w = bpp; w < width * bpp; ++w) {
-      const int pred = GradientPredictor(preds[w - bpp],
+    PredictLine(in, preds - stride, out, 1, inverse);
+    for (w = 1; w < width; ++w) {
+      const int pred = GradientPredictor(preds[w - 1],
                                         preds[w - stride],
-                                         preds[w - stride - bpp]);
+                                         preds[w - stride - 1]);
      out[w] = in[w] + (inverse ? pred : -pred);
    }
  }
 }

 static void GradientFilter(const uint8_t* data, int width, int height,
-                           int bpp, int stride, uint8_t* filtered_data) {
-  DoGradientFilter(data, width, height, bpp, stride, 0, filtered_data);
+                           int stride, uint8_t* filtered_data) {
+  DoGradientFilter(data, width, height, stride, 0, filtered_data);
 }

-static void GradientUnfilter(const uint8_t* data, int width, int height,
-                             int bpp, int stride, uint8_t* recon_data) {
-  DoGradientFilter(data, width, height, bpp, stride, 1, recon_data);
+static void GradientUnfilter(int width, int height, int stride, uint8_t* data) {
+  DoGradientFilter(data, width, height, stride, 1, data);
 }

 #undef SANITY_CHECK
@ -215,7 +214,7 @@ const WebPFilterFunc WebPFilters[WEBP_FILTER_LAST] = {
  GradientFilter     // WEBP_FILTER_GRADIENT
 };

-const WebPFilterFunc WebPUnfilters[WEBP_FILTER_LAST] = {
+const WebPUnfilterFunc WebPUnfilters[WEBP_FILTER_LAST] = {
  NULL,                // WEBP_FILTER_NONE
  HorizontalUnfilter,  // WEBP_FILTER_HORIZONTAL
  VerticalUnfilter,    // WEBP_FILTER_VERTICAL
--- a/src/utils/filters.h
+++ b/src/utils/filters.h
@ -30,18 +30,19 @@ typedef enum {
 } WEBP_FILTER_TYPE;

 typedef void (*WebPFilterFunc)(const uint8_t* in, int width, int height,
-                               int bpp, int stride, uint8_t* out);
+                               int stride, uint8_t* out);
+typedef void (*WebPUnfilterFunc)(int width, int height, int stride,
+                                 uint8_t* data);

 // Filter the given data using the given predictor.
 // 'in' corresponds to a 2-dimensional pixel array of size (stride * height)
 // in raster order.
-// 'bpp' is number of bytes per pixel, and
 // 'stride' is number of bytes per scan line (with possible padding).
 // 'out' should be pre-allocated.
 extern const WebPFilterFunc WebPFilters[WEBP_FILTER_LAST];

-// Reconstruct the original data from the given filtered data.
-extern const WebPFilterFunc WebPUnfilters[WEBP_FILTER_LAST];
+// In-place reconstruct the original data from the given filtered data.
+extern const WebPUnfilterFunc WebPUnfilters[WEBP_FILTER_LAST];

 // Fast estimate of a potentially good filter.
 extern WEBP_FILTER_TYPE EstimateBestFilter(const uint8_t* data,
--- a/src/webp/decode.h
+++ b/src/webp/decode.h
@ -18,7 +18,7 @@
 extern "C" {
 #endif

-#define WEBP_DECODER_ABI_VERSION 0x0200    // MAJOR(8b) + MINOR(8b)
+#define WEBP_DECODER_ABI_VERSION 0x0201    // MAJOR(8b) + MINOR(8b)

 typedef struct WebPRGBABuffer WebPRGBABuffer;
 typedef struct WebPYUVABuffer WebPYUVABuffer;
@ -392,9 +392,10 @@ WEBP_EXTERN(const WebPDecBuffer*) WebPIDecodedArea(

 // Features gathered from the bitstream
 struct WebPBitstreamFeatures {
-  int width;        // Width in pixels, as read from the bitstream.
-  int height;       // Height in pixels, as read from the bitstream.
-  int has_alpha;    // True if the bitstream contains an alpha channel.
+  int width;          // Width in pixels, as read from the bitstream.
+  int height;         // Height in pixels, as read from the bitstream.
+  int has_alpha;      // True if the bitstream contains an alpha channel.
+  int has_animation;  // True if the bitstream is an animation.

  // Unused for now:
  int bitstream_version;        // should be 0 for now. TODO(later)
@ -402,7 +403,7 @@ struct WebPBitstreamFeatures {
                                // recommended.
  int rotate;                   // TODO(later)
  int uv_sampling;              // should be 0 for now. TODO(later)
-  uint32_t pad[3];              // padding for later use
+  uint32_t pad[2];              // padding for later use
 };

 // Internal, version-checked, entry point
--- a/src/webp/demux.h
+++ b/src/webp/demux.h
@ -45,7 +45,6 @@
 #ifndef WEBP_WEBP_DEMUX_H_
 #define WEBP_WEBP_DEMUX_H_

-#include "./format_constants.h"
 #include "./mux_types.h"

 #if defined(__cplusplus) || defined(c_plusplus)
@ -81,14 +80,13 @@ enum WebPDemuxState {
 WEBP_EXTERN(WebPDemuxer*) WebPDemuxInternal(
    const WebPData*, int, WebPDemuxState*, int);

-// Parses the WebP file given by 'data'.
-// A complete WebP file must be present in 'data' for the function to succeed.
+// Parses the full WebP file given by 'data'.
 // Returns a WebPDemuxer object on successful parse, NULL otherwise.
 static WEBP_INLINE WebPDemuxer* WebPDemux(const WebPData* data) {
  return WebPDemuxInternal(data, 0, NULL, WEBP_DEMUX_ABI_VERSION);
 }

-// Parses the WebP file given by 'data'.
+// Parses the possibly incomplete WebP file given by 'data'.
 // If 'state' is non-NULL it will be set to indicate the status of the demuxer.
 // Returns a WebPDemuxer object on successful parse, NULL otherwise.
 static WEBP_INLINE WebPDemuxer* WebPDemuxPartial(
--- a/src/webp/format_constants.h
+++ b/src/webp/format_constants.h
@ -12,6 +12,9 @@
 #ifndef WEBP_WEBP_FORMAT_CONSTANTS_H_
 #define WEBP_WEBP_FORMAT_CONSTANTS_H_

+// Create fourcc of the chunk from the chunk tag characters.
+#define MKFOURCC(a, b, c, d) ((uint32_t)(a) | (b) << 8 | (c) << 16 | (d) << 24)
+
 // VP8 related constants.
 #define VP8_SIGNATURE 0x9d012a              // Signature in VP8 data.
 #define VP8_MAX_PARTITION0_SIZE (1 << 19)   // max size of mode partition
@ -70,19 +73,6 @@ typedef enum {
 #define FRGM_CHUNK_SIZE    6     // Size of a FRGM chunk.
 #define VP8X_CHUNK_SIZE    10    // Size of a VP8X chunk.

-// VP8X Feature Flags.
-#if !(defined(__cplusplus) || defined(c_plusplus))
-typedef enum WebPFeatureFlags WebPFeatureFlags;
-#endif
-enum WebPFeatureFlags {
-  FRAGMENTS_FLAG  = 0x00000001,
-  ANIMATION_FLAG  = 0x00000002,
-  XMP_FLAG        = 0x00000004,
-  EXIF_FLAG       = 0x00000008,
-  ALPHA_FLAG      = 0x00000010,
-  ICCP_FLAG       = 0x00000020
-};
-
 #define MAX_CANVAS_SIZE     (1 << 24)     // 24-bit max for VP8X width/height.
 #define MAX_IMAGE_AREA      (1ULL << 32)  // 32-bit max for width x height.
 #define MAX_LOOP_COUNT      (1 << 16)     // maximum value for loop-count
--- a/src/webp/mux.h
+++ b/src/webp/mux.h
@ -45,7 +45,6 @@
 #ifndef WEBP_WEBP_MUX_H_
 #define WEBP_WEBP_MUX_H_

-#include "./format_constants.h"
 #include "./mux_types.h"

 #if defined(__cplusplus) || defined(c_plusplus)
--- a/src/webp/mux_types.h
+++ b/src/webp/mux_types.h
@ -5,7 +5,7 @@
 //  Additional IP Rights Grant:  http://www.webmproject.org/license/additional/
 // -----------------------------------------------------------------------------
 //
-// Internal header for some common data-types used by mux and demux libraries.
+// Data-types common to the mux and demux libraries.
 //
 // Author: Urvang (urvang@google.com)

@ -20,14 +20,23 @@
 extern "C" {
 #endif

-// Create fourcc of the chunk from the chunk tag characters.
-#define MKFOURCC(a, b, c, d) ((uint32_t)(a) | (b) << 8 | (c) << 16 | (d) << 24)
+#if !(defined(__cplusplus) || defined(c_plusplus))
+typedef enum WebPFeatureFlags WebPFeatureFlags;
+typedef enum WebPMuxAnimDispose WebPMuxAnimDispose;
+#endif
+
+// VP8X Feature Flags.
+enum WebPFeatureFlags {
+  FRAGMENTS_FLAG  = 0x00000001,
+  ANIMATION_FLAG  = 0x00000002,
+  XMP_FLAG        = 0x00000004,
+  EXIF_FLAG       = 0x00000008,
+  ALPHA_FLAG      = 0x00000010,
+  ICCP_FLAG       = 0x00000020
+};

 // Dispose method (animation only). Indicates how the area used by the current
 // frame is to be treated before rendering the next frame on the canvas.
-#if !(defined(__cplusplus) || defined(c_plusplus))
-typedef enum WebPMuxAnimDispose WebPMuxAnimDispose;
-#endif
 enum WebPMuxAnimDispose {
  WEBP_MUX_DISPOSE_NONE,       // Do not dispose.
  WEBP_MUX_DISPOSE_BACKGROUND  // Dispose to background color.