|
1 libteora 1.2.0alpha1 (2010 September 23) |
|
2 |
|
3 - New 'ptalarbvorm' encoder with better rate/distortion optimization |
|
4 - New th_encode_ctl option for copying configuration from an existing |
|
5 setup header, useful for splicing streams. |
|
6 - Returns TH_DUPFRAME in more cases. |
|
7 - Add ARM optimizations |
|
8 - Add TI C64x+ DSP optimizations |
|
9 - Other performance improvements |
|
10 - Rename speedlevel 2 to 3 and provide a new speedlevel 2 |
|
11 - Various minor bug fixes |
|
12 |
|
13 libtheora 1.1.2 (unreleased snapshot) |
|
14 |
|
15 - Fix Huffman table decoding with OC_HUFF_SLUSH is set to 0 |
|
16 - Fix a frame size bug in player_example |
|
17 - Add support for passing a buffer the size of the picture |
|
18 region, rather than a full padded frame to th_encode_ycbcr_in() |
|
19 as was possible with the legacy pre-1.0 API. |
|
20 - 4:4:4 support in player_example using software yuv->rgb |
|
21 - Better rgb->yuv conversion in png2theora |
|
22 - Clean up warnings and local variables |
|
23 - Build and documentation fixes |
|
24 |
|
25 libtheora 1.1.1 (2009 October 1) |
|
26 |
|
27 - Fix problems with MSVC inline assembly |
|
28 - Add the missing encoder_disabled.c to the distribution |
|
29 - build updates: autogen.sh should work better after switching systems |
|
30 and the MSVC project now defaults to the dynamic runtime library |
|
31 - Namespace some variables to avoid conflicts on wince. |
|
32 |
|
33 libtheora 1.1.0 (2009 September 24) |
|
34 |
|
35 - Fix various small issues with the example and telemetry code |
|
36 - Fix handing a zero-byte packet as the first frame |
|
37 - Documentation cleanup |
|
38 - Two minor build fixes |
|
39 |
|
40 libtheora 1.1beta3 (2009 August 22) |
|
41 |
|
42 - Rate control fixes to smooth quality |
|
43 - MSVC build now exports all of the 1.0 api |
|
44 - Assorted small bug fixes |
|
45 |
|
46 libtheora 1.1beta2 (2009 August 12) |
|
47 |
|
48 - Fix a rate control problem with difficult input |
|
49 - Build fixes for OpenBSD and Apple Xcode |
|
50 - Examples now all use the 1.0 api |
|
51 - TH_ENCCTL_SET_SPLEVEL works again |
|
52 - Various bug fixes and source tree rearrangement |
|
53 |
|
54 libtheora 1.1beta1 (2009 August 5) |
|
55 |
|
56 - Support for two-pass encoding |
|
57 - Performance optimization of both encoder and decoder |
|
58 - Encoder supports dynamic adjustment of quality and |
|
59 bitrate targets |
|
60 - Encoder is generally more configurable, and all |
|
61 rate control modes perform better |
|
62 - Encoder now accepts 4:2:2 and 4:4:4 chroma sampling |
|
63 - Decoder telemetry output shows quantization choice |
|
64 and a breakdown of bitrate usage in the frame |
|
65 - MSVC assembly optimizations up to date and functional |
|
66 |
|
67 libtheora 1.1alpha2 (2009 May 26) |
|
68 |
|
69 - Reduce lambda for small quantizers. |
|
70 - New encoder fDCT does better on smooth gradients |
|
71 - Use SATD for mode decisions (1-2% bitrate reduction) |
|
72 - Assembly rewrite for new features and general speed up |
|
73 - Share code between the encoder and decoder for performance |
|
74 - Fix 4:2:2 decoding and telemetry |
|
75 - MSVC project files updated, but assembly is disabled. |
|
76 - New configure option --disable-spec to work around toolchain |
|
77 detection failures. |
|
78 - Limit symbol exports on MacOS X. |
|
79 - Port remaining unit tests from the 1.0 release. |
|
80 |
|
81 libtheora 1.1alpha1 (2009 March 27) |
|
82 |
|
83 - Encoder rewrite with much improved vbr quality/bitrate and |
|
84 better tracking of the target rate in cbr mode. |
|
85 - MSVC project files do not work in this release. |
|
86 |
|
87 libtheora 1.0 (2008 November 3) |
|
88 |
|
89 - Merge x86 assembly for forward DCT from Thusnelda branch. |
|
90 - Update 32 bit MMX with loop filter fix. |
|
91 - Check for an uninitialized state before dereferencing in propagating |
|
92 decode calls. |
|
93 - Remove all TH_DEBUG statements. |
|
94 - Rename the bitpacker source files copied from libogg to avoid |
|
95 confusing simple build systems using both libraries. |
|
96 - Declare bitfield entries to be explicitly signed for Solaris cc. |
|
97 - Set quantization parameters to default values when an empty buffer is |
|
98 passed with TH_ENCCTL_SET_QUANT_PARAMS. |
|
99 - Split encoder and decoder tests depending on configure settings. |
|
100 - Return lstylex.sty to the distribution. |
|
101 - Disable inline assembly on gcc versions prior to 3.1. |
|
102 - Remove extern references for OC_*_QUANT_MIN. |
|
103 - Make various data tables static const so they can be read-only. |
|
104 - Remove ENCCTL codes from the old encoder API. |
|
105 - Implement TH_ENCCTL_SET_KEYFRAME_FREQUENCY_FORCE ctl. |
|
106 - Fix segfault when exactly one of the width or height is not a multiple |
|
107 of 16, but the other is. |
|
108 - Compute the correct vertical offset for chroma. |
|
109 - cpuid assembly fix for MSVC. |
|
110 - Add VS2008 project files. |
|
111 - Build updates for 64-bit platforms, Mingw32, VS and XCode. |
|
112 - Do not clobber the cropping rectangle. |
|
113 - Declare ourselves 1.0final to pkg-config to sort after beta releases. |
|
114 - Fix the scons build to include asm in libtheoradec/enc. |
|
115 |
|
116 libtheora 1.0beta3 (2008 April 16) |
|
117 |
|
118 - Build new libtheoradec and libtheoraenc libraries |
|
119 supporting the new API from theora-exp. This API should |
|
120 not be considered stable yet. |
|
121 - Change granule_frame() to return an index as documented. |
|
122 This is a change of behaviour from 1.0beta1. |
|
123 - Document that granule_time() returns the end of the |
|
124 presentation interval. |
|
125 - Use a custom copy of the libogg bitpacker in the decoder |
|
126 to avoid function call overhead. |
|
127 - MMX code improved and ported to MSVC. |
|
128 - Fix a problem with the MMX code on SELinux. |
|
129 - Fix a problem with decoder quantizer initialization. |
|
130 - Fix a page queue problem with png2theora. |
|
131 - Improved robustness. |
|
132 - Updated VS2005 project files. |
|
133 - Dropped build support for Microsoft VS2003. |
|
134 - Dropped build support for the unreleased libogg2. |
|
135 - Added the specification to the autotools build. |
|
136 - Specification corrections. |
|
137 |
|
138 libtheora 1.0beta2 (2007 October 12) |
|
139 |
|
140 - Fix a crash bug on char-is-unsigned architectures (PowerPC) |
|
141 - Fix a buffer sizing issue that caused rare encoder crashes |
|
142 - Fix a buffer alignment issue |
|
143 - Build fixes for MingW32, MSVC |
|
144 - Improved format documentation. |
|
145 |
|
146 libtheora 1.0beta1 (2007 September 22) |
|
147 |
|
148 - Granulepos scheme modified to match other codecs. This bumps |
|
149 the bitstream revision to 3.2.1. Bitstreams marked 3.2.0 are |
|
150 handled correctly by this decoder. Older decoders will show |
|
151 a one frame sync error in the less noticeable direction. |
|
152 |
|
153 libtheora 1.0alpha8 (2007 September 18) |
|
154 |
|
155 - Switch to new spec compliant decoder from theora-exp branch. |
|
156 Written by Dr. Timothy Terriberry. |
|
157 - Add support to the encoder for using quantization settings |
|
158 provided by the application. |
|
159 - more assembly optimizations |
|
160 |
|
161 libtheora 1.0alpha7 (2006 June 20) |
|
162 |
|
163 - Enable mmx assembly by default |
|
164 - Avoid some relocations that caused problems on SELinux |
|
165 - Other build fixes |
|
166 - time testing mode (-f) for the dump_video example |
|
167 |
|
168 libtheora 1.0alpha6 (2006 May 30) |
|
169 |
|
170 * Merge theora-mmx simd acceleration (x86_32 and x86_64) |
|
171 * Major RTP payload specification update |
|
172 * Minor format specification updates |
|
173 * Fix some spurious calls to free() instead of _ogg_free() |
|
174 * Fix invalid array indexing in PixelLineSearch() |
|
175 * Improve robustness against invalid input |
|
176 * General warning cleanup |
|
177 * The offset_y member now means what every application thought it meant |
|
178 (offset from the top). This will mean some old files (those with a |
|
179 non-centered image created with a buggy encoder) will display differently. |
|
180 |
|
181 libtheora 1.0alpha5 (2005 August 20) |
|
182 |
|
183 * Fixed bitrate management bugs that caused popping and encode |
|
184 errors |
|
185 * Fixed a crash problem with the theora_state internals not |
|
186 being intialized properly. |
|
187 * new utility function: |
|
188 - theora_granule_shift() |
|
189 * dump_video example now makes YUV4MPEG files by default, so |
|
190 the results can be fed back to encoder_example and similar |
|
191 tools. The old behavior is restored through the '-r' switch. |
|
192 * ./configure now prints a summary |
|
193 * simple unit test of the comment api under 'make check' |
|
194 * misc code cleanup, warning and leak fixes |
|
195 |
|
196 libtheora 1.0alpha4 (2004 December 15) |
|
197 |
|
198 * first draft of the Theora I Format Specification |
|
199 * API documentation generated from theora.h with Doxygen |
|
200 * fix a double-update bug in the motion analysis |
|
201 * apply the loop filter before filling motion vector border |
|
202 in the reference frame |
|
203 * new utility functions: |
|
204 - theora_packet_isheader(), |
|
205 - theora_packet_iskeyframe() |
|
206 - theora_granule_frame() |
|
207 * optional support for building without floating point |
|
208 * optional support for building without encode support |
|
209 * various build and packaging fixes |
|
210 * pkg-config support |
|
211 * SymbianOS build support |
|
212 |
|
213 libtheora 1.0alpha3 (2004 March 20) |
|
214 |
|
215 UPDATE: on 2004 July 1 the Theora I bitstream format was frozen. Files |
|
216 produced by the libtheora 1.0alpha3 reference encoder will always be |
|
217 decodable by the Theora I spec. |
|
218 |
|
219 * Bitstream info header FORMAT CHANGES: |
|
220 - move the granulepos shift field to maintain byte alignment longer. |
|
221 - reserve 5 additional bits for subsampling and interlace flags. |
|
222 * Bitstream setup header FORMAT CHANGES: |
|
223 - support for a range of interpolated quant matricies. |
|
224 - include the in-loop block filter coeff. |
|
225 * Bitsteam data packet FORMAT CHANGES: |
|
226 - Reserve a bit for per-block Q index selection. |
|
227 - Flip the coded image orientation for compatibility with VP3. |
|
228 This allows lossless transcoding of VP3 content, but files |
|
229 encoded with earlier theora releases would play upside down. |
|
230 * example VP3 lossless transcoder |
|
231 * optional support for libogg2 |
|
232 * timing improvements in the example player |
|
233 * packaging and build system updates and fixes |
|
234 |
|
235 libtheora 1.0alpha2 (2003 June 9) |
|
236 |
|
237 * bitstream FORMAT CHANGES: |
|
238 - store the quant tables in a third setup header for |
|
239 future encoder flexibility |
|
240 - store the huffman tables in the third setup header |
|
241 - add a field for marking the colorspace to the info header |
|
242 - add crop parameters for non-multiple-of-16 frame sizes |
|
243 - add a second vorbiscomment-style metadata header |
|
244 * API changes to handle multiple headers with a single |
|
245 theora_decode_header() call, like libvorbis |
|
246 * code cleanup and minor fixes |
|
247 * new dump_video code example/utility |
|
248 * experimental win32 code examples |
|
249 |
|
250 libtheora 1.0alpha1 (2002 September 25) |
|
251 |
|
252 * First release of the theora reference implementation |
|
253 * Port of the newly opened VP3 code to the Ogg container |
|
254 * Rewrite of the code for portability and to use the libogg bitpacker |
|
255 |