5 |
# relevant. |
# relevant. |
6 |
# |
# |
7 |
|
|
8 |
|
2003-12-05 14:35:22 GMT patch-122 |
9 |
|
|
10 |
|
Summary: |
11 |
|
Cap quants correctly (the best we can at least) |
12 |
|
Revision: |
13 |
|
xvidcore--devapi4--1.0--patch-122 |
14 |
|
|
15 |
|
Cap quants correctly (the best we can at least) |
16 |
|
|
17 |
|
|
18 |
|
modified files: |
19 |
|
src/plugins/plugin_single.c |
20 |
|
|
21 |
|
|
22 |
|
2003-12-05 14:33:48 GMT patch-121 |
23 |
|
|
24 |
|
Summary: |
25 |
|
Small glitch |
26 |
|
Revision: |
27 |
|
xvidcore--devapi4--1.0--patch-121 |
28 |
|
|
29 |
|
Small glitch |
30 |
|
|
31 |
|
|
32 |
|
modified files: |
33 |
|
src/motion/vop_type_decision.c |
34 |
|
|
35 |
|
|
36 |
|
2003-12-05 14:06:19 GMT patch-120 |
37 |
|
|
38 |
|
Summary: |
39 |
|
KFthresholding changes. |
40 |
|
Revision: |
41 |
|
xvidcore--devapi4--1.0--patch-120 |
42 |
|
|
43 |
|
As user reports proved, the logic behind the min_key_interval was |
44 |
|
1/ misleading because the parameter is kfthreshold indeed and not |
45 |
|
a minimum keyframe interval |
46 |
|
2/ the formula was a bit too aggressive (removing 20% of bitrate |
47 |
|
per frame until distance to next iframe was 1) |
48 |
|
|
49 |
|
I posted a RFC to try to settle a decision on what behavior this |
50 |
|
setting should have. We have still have no clear answer so i prefer |
51 |
|
just fixing the misleading name right now and wait for a common |
52 |
|
position about its behavior later. |
53 |
|
|
54 |
|
Libraries are *binary* compatible, but *source code* compatibility |
55 |
|
is broken (rename rc_2pass2_t->min_key_interval to kfthreshold). |
56 |
|
This is probably the last API change. |
57 |
|
|
58 |
|
NB: fixes a type problem during scaling parameter computing which |
59 |
|
was causing insane pb_iboost_tax_ratio values. |
60 |
|
|
61 |
|
modified files: |
62 |
|
src/plugins/plugin_2pass2.c src/xvid.h vfw/src/codec.c |
63 |
|
vfw/src/config.c vfw/src/config.h |
64 |
|
|
65 |
|
|
66 |
|
2003-12-05 00:20:28 GMT patch-119 |
67 |
|
|
68 |
|
Summary: |
69 |
|
ivop decision tuning. |
70 |
|
Revision: |
71 |
|
xvidcore--devapi4--1.0--patch-119 |
72 |
|
|
73 |
|
ivop decision tuning from sysKin. |
74 |
|
|
75 |
|
modified files: |
76 |
|
src/motion/vop_type_decision.c |
77 |
|
|
78 |
|
|
79 |
|
2003-12-03 18:55:29 GMT patch-118 |
80 |
|
|
81 |
|
Summary: |
82 |
|
VOSH header always written. |
83 |
|
Revision: |
84 |
|
xvidcore--devapi4--1.0--patch-118 |
85 |
|
|
86 |
|
* profile is set to sane default value in BitstreamWriteVolHeaders |
87 |
|
* VOSH is now always written (note that the ending code is never |
88 |
|
written) |
89 |
|
* doubled variable removed from bvop estimation file. |
90 |
|
|
91 |
|
bitstream version set to 23 |
92 |
|
Version 22 was used in CVS by a fix from michael for VOSH, it was just |
93 |
|
lacking the sane default value setting when profile is 0x00 from user. |
94 |
|
0x00 is a reserved profile ID in the spec so it wasn't making much sense |
95 |
|
to write 0x00. |
96 |
|
|
97 |
|
modified files: |
98 |
|
src/bitstream/bitstream.c src/motion/estimation_bvop.c |
99 |
|
src/xvid.h |
100 |
|
|
101 |
|
|
102 |
|
2003-12-03 15:29:30 GMT patch-117 |
103 |
|
|
104 |
|
Summary: |
105 |
|
VFW GUI fixes. |
106 |
|
Revision: |
107 |
|
xvidcore--devapi4--1.0--patch-117 |
108 |
|
|
109 |
|
min key was misleading because it was legacy code from dev-api-3. |
110 |
|
|
111 |
|
modified files: |
112 |
|
vfw/src/codec.c vfw/src/config.c vfw/src/resource.rc |
113 |
|
|
114 |
|
|
115 |
|
2003-12-03 15:22:25 GMT patch-116 |
116 |
|
|
117 |
|
Summary: |
118 |
|
ME fixes. |
119 |
|
Revision: |
120 |
|
xvidcore--devapi4--1.0--patch-116 |
121 |
|
|
122 |
|
From syskin: |
123 |
|
- small typo in chroma sad reset |
124 |
|
- code tweaking |
125 |
|
+ adv diamond search used intead of mainsearch |
126 |
|
+ and some other stuff |
127 |
|
- thresholds tuned. |
128 |
|
|
129 |
|
modified files: |
130 |
|
src/motion/estimation.h src/motion/estimation_bvop.c |
131 |
|
src/motion/estimation_pvop.c |
132 |
|
|
133 |
|
|
134 |
|
2003-12-01 11:17:20 GMT patch-115 |
135 |
|
|
136 |
|
Summary: |
137 |
|
Small changes and interlacing bugfix. |
138 |
|
Revision: |
139 |
|
xvidcore--devapi4--1.0--patch-115 |
140 |
|
|
141 |
|
* Interlacing bugfix, code got swapped in a very old patch |
142 |
|
(back in March) |
143 |
|
* xvid_encraw forces VOSH writing. |
144 |
|
* VFW GUI fixes. |
145 |
|
|
146 |
|
modified files: |
147 |
|
examples/Makefile examples/xvid_encraw.c |
148 |
|
src/utils/mbtransquant.c vfw/src/config.c vfw/src/resource.rc |
149 |
|
|
150 |
|
|
151 |
|
2003-11-30 15:47:41 GMT patch-114 |
152 |
|
|
153 |
|
Summary: |
154 |
|
Thread safe MPEG4 quantization functions + xvid_bench update |
155 |
|
Revision: |
156 |
|
xvidcore--devapi4--1.0--patch-114 |
157 |
|
|
158 |
|
* Thread safe MPEG4 quantization functions. |
159 |
|
Cleaned up version of patch provided by Michael |
160 |
|
- fixed compiling problems on gcc |
161 |
|
- added const qualifiers every where it was possible to help C compiler |
162 |
|
optimization. |
163 |
|
- added the mpeg_quant_matrices param to all ASM function prototype |
164 |
|
in comments (even if it's not used, that shows we do it deliberatly) |
165 |
|
- forces m[intra][0][0] = 8, otherwise XviD could write invalid streams. |
166 |
|
* Added real CRC computing in xvid_bench.c |
167 |
|
|
168 |
|
modified files: |
169 |
|
TODO examples/xvid_bench.c src/bitstream/bitstream.c |
170 |
|
src/decoder.c src/decoder.h src/encoder.c src/encoder.h |
171 |
|
src/motion/estimation.h src/motion/estimation_pvop.c |
172 |
|
src/motion/estimation_rd_based.c src/quant/quant.h |
173 |
|
src/quant/quant_h263.c src/quant/quant_matrix.c |
174 |
|
src/quant/quant_matrix.h src/quant/quant_mpeg.c |
175 |
|
src/quant/x86_asm/quantize_h263_3dne.asm |
176 |
|
src/quant/x86_asm/quantize_h263_mmx.asm |
177 |
|
src/quant/x86_asm/quantize_mpeg_mmx.asm |
178 |
|
src/quant/x86_asm/quantize_mpeg_xmm.asm |
179 |
|
src/utils/mbtransquant.c src/xvid.c |
180 |
|
|
181 |
|
|
182 |
|
2003-11-29 18:10:25 GMT patch-113 |
183 |
|
|
184 |
|
Summary: |
185 |
|
Fixed csp asm rules for real? |
186 |
|
Revision: |
187 |
|
xvidcore--devapi4--1.0--patch-113 |
188 |
|
|
189 |
|
Fixed csp asm rules for real? |
190 |
|
|
191 |
|
|
192 |
|
modified files: |
193 |
|
build/win32/libxvidcore.dsp |
194 |
|
|
195 |
|
|
196 |
|
2003-11-29 17:58:09 GMT patch-112 |
197 |
|
|
198 |
|
Summary: |
199 |
|
TODO/ChangeLog updated |
200 |
|
Revision: |
201 |
|
xvidcore--devapi4--1.0--patch-112 |
202 |
|
|
203 |
|
TODO/ChangeLog updated |
204 |
|
|
205 |
|
|
206 |
|
modified files: |
207 |
|
ChangeLog TODO |
208 |
|
|
209 |
|
|
210 |
|
######################################################################### |
211 |
|
# 1.0.0 beta1 |
212 |
|
######################################################################### |
213 |
|
|
214 |
|
2003-11-29 17:21:08 GMT patch-111 |
215 |
|
|
216 |
|
Summary: |
217 |
|
First beta marking |
218 |
|
Revision: |
219 |
|
xvidcore--devapi4--1.0--patch-111 |
220 |
|
|
221 |
|
First beta marking |
222 |
|
|
223 |
|
|
224 |
|
modified files: |
225 |
|
build/generic/configure.in src/xvid.c src/xvid.h |
226 |
|
|
227 |
|
|
228 |
|
2003-11-29 16:59:14 GMT patch-110 |
229 |
|
|
230 |
|
Summary: |
231 |
|
Catching up with CVS. |
232 |
|
Revision: |
233 |
|
xvidcore--devapi4--1.0--patch-110 |
234 |
|
|
235 |
|
Changes from sysKin: |
236 |
|
* dquant optimization. |
237 |
|
* CBR fix |
238 |
|
|
239 |
|
modified files: |
240 |
|
src/encoder.c src/plugins/plugin_single.c |
241 |
|
|
242 |
|
|
243 |
|
2003-11-24 22:05:38 GMT patch-109 |
244 |
|
|
245 |
|
Summary: |
246 |
|
Big level handling in trellis. |
247 |
|
Revision: |
248 |
|
xvidcore--devapi4--1.0--patch-109 |
249 |
|
|
250 |
|
Trellis was treating big levels exactly the same way as lower ones. |
251 |
|
In some cases, trellis was doing wild optimizations favoring a 0 |
252 |
|
because the distortion introduced by that big coeff change was acceptable. |
253 |
|
But visually this could result in some nasty blocks with wrong chroma |
254 |
|
information or similar brutal changes in other planes as well. |
255 |
|
|
256 |
|
Skal added big levels handling where trellis just tries to minimize |
257 |
|
the cost varying the run value only. No level modification is done |
258 |
|
anymore. |
259 |
|
|
260 |
|
modified files: |
261 |
|
TODO src/utils/mbtransquant.c |
262 |
|
|
263 |
|
|
264 |
|
2003-11-23 16:42:55 GMT patch-108 |
265 |
|
|
266 |
|
Summary: |
267 |
|
Trellis for MPEG. |
268 |
|
Revision: |
269 |
|
xvidcore--devapi4--1.0--patch-108 |
270 |
|
|
271 |
|
* Added trellis support for MPEg quantization type. |
272 |
|
* Changed RD fixed point precision, should help avoiding overflow |
273 |
|
(see the constant TL_SHIFT) |
274 |
|
|
275 |
|
NB: we still have some problems when trellis optimizes DC for big DC |
276 |
|
values. |
277 |
|
|
278 |
|
modified files: |
279 |
|
src/utils/mbtransquant.c |
280 |
|
|
281 |
|
|
282 |
|
2003-11-22 00:53:59 GMT patch-107 |
283 |
|
|
284 |
|
Summary: |
285 |
|
Win32 lib project fix (bis) |
286 |
|
Revision: |
287 |
|
xvidcore--devapi4--1.0--patch-107 |
288 |
|
|
289 |
|
* nasm >= 0.98.37 support in project file got reverted in a previous |
290 |
|
patch, push it back. Nota bene /O3 changed to /O2 for proper |
291 |
|
compilation with msvc compiler (everyone is not supposed to compile |
292 |
|
stuff with icc) |
293 |
|
* TODO update. |
294 |
|
|
295 |
|
modified files: |
296 |
|
TODO build/win32/libxvidcore.dsp |
297 |
|
|
298 |
|
|
299 |
|
2003-11-19 21:26:34 GMT patch-106 |
300 |
|
|
301 |
|
Summary: |
302 |
|
updated bench crc |
303 |
|
Revision: |
304 |
|
xvidcore--devapi4--1.0--patch-106 |
305 |
|
|
306 |
|
updated bench crc |
307 |
|
|
308 |
|
|
309 |
|
modified files: |
310 |
|
examples/xvid_bench.c |
311 |
|
|
312 |
|
|
313 |
|
2003-11-19 16:00:00 GMT patch-105 |
314 |
|
|
315 |
|
Summary: |
316 |
|
Lumimasking fixes. |
317 |
|
Revision: |
318 |
|
xvidcore--devapi4--1.0--patch-105 |
319 |
|
|
320 |
|
from sysKin: |
321 |
|
* New plugin hook entry XVID_PLG_FRAME that happens inside FrameCodeIPB |
322 |
|
when both type and quant are known. Added hook handling in all plugins. |
323 |
|
* Fixed lumimasking. |
324 |
|
|
325 |
|
from me: |
326 |
|
* small reverse commit in pvop estimation fixed. |
327 |
|
|
328 |
|
modified files: |
329 |
|
src/bitstream/bitstream.c src/bitstream/bitstream.h |
330 |
|
src/encoder.c src/motion/estimation_pvop.c |
331 |
|
src/plugins/plugin_2pass1.c src/plugins/plugin_2pass2.c |
332 |
|
src/plugins/plugin_dump.c src/plugins/plugin_lumimasking.c |
333 |
|
src/plugins/plugin_psnr.c src/plugins/plugin_single.c |
334 |
|
src/xvid.h |
335 |
|
|
336 |
|
|
337 |
|
2003-11-19 15:37:16 GMT patch-104 |
338 |
|
|
339 |
|
Summary: |
340 |
|
Removed indirections from SearchData structure. |
341 |
|
Revision: |
342 |
|
xvidcore--devapi4--1.0--patch-104 |
343 |
|
|
344 |
|
Patch from sysKin: |
345 |
|
* removed indirections in SearchData structure. CheckCandidate functions |
346 |
|
don't use a const SearchData pointer anymore, but they should be a bit |
347 |
|
faster because of the less numerous indirections. |
348 |
|
|
349 |
|
modified files: |
350 |
|
src/motion/estimation.h src/motion/estimation_bvop.c |
351 |
|
src/motion/estimation_common.c src/motion/estimation_gmc.c |
352 |
|
src/motion/estimation_pvop.c src/motion/estimation_rd_based.c |
353 |
|
src/motion/vop_type_decision.c |
354 |
|
|
355 |
|
|
356 |
|
2003-11-19 15:33:55 GMT patch-103 |
357 |
|
|
358 |
|
Summary: |
359 |
|
Formula error in twopass code. |
360 |
|
Revision: |
361 |
|
xvidcore--devapi4--1.0--patch-103 |
362 |
|
|
363 |
|
The reversing bframe formula in 2pass 2 was not right. This was in fact |
364 |
|
a test code i used when i was maintaining the code on its own branch. |
365 |
|
I should not have commited it :\ |
366 |
|
|
367 |
|
Fixed :-) |
368 |
|
|
369 |
|
modified files: |
370 |
|
src/plugins/plugin_2pass2.c |
371 |
|
|
372 |
|
|
373 |
|
2003-11-18 21:41:08 GMT patch-102 |
374 |
|
|
375 |
|
Summary: |
376 |
|
Another problem with mis/unitialized reads. |
377 |
|
Revision: |
378 |
|
xvidcore--devapi4--1.0--patch-102 |
379 |
|
|
380 |
|
Michael introduced a fast subpel refine that uses a (iMinSAD2, currentQMV2) |
381 |
|
couple of data. The problem is that he plugged this in CheckCandidate16_qpel |
382 |
|
that is used outside this context, thus a if statement was traversed with |
383 |
|
garbage data in the standard subpel case. For perfection sake, using a |
384 |
|
iMinSAD=256*4096 value collects correct data even if it will not be used in |
385 |
|
the normal subpel case. |
386 |
|
|
387 |
|
modified files: |
388 |
|
src/motion/estimation_pvop.c |
389 |
|
|
390 |
|
|
391 |
|
2003-11-16 17:29:39 GMT patch-101 |
392 |
|
|
393 |
|
Summary: |
394 |
|
The MEanalysis patch assumed bvops were always used |
395 |
|
Revision: |
396 |
|
xvidcore--devapi4--1.0--patch-101 |
397 |
|
|
398 |
|
The MEanalysis patch assumed bvops were always used |
399 |
|
|
400 |
|
|
401 |
|
modified files: |
402 |
|
src/encoder.c |
403 |
|
|
404 |
|
|
405 |
|
2003-11-16 15:12:15 GMT patch-100 |
406 |
|
|
407 |
|
Summary: |
408 |
|
MEanalysis using wrong mvs + bframe search using wrong mvs |
409 |
|
Revision: |
410 |
|
xvidcore--devapi4--1.0--patch-100 |
411 |
|
|
412 |
|
Valgrind reported lot of unitialized reads. These unitialized reads |
413 |
|
helped sysKin finding three bugs: |
414 |
|
- ZeroMacroblock did not reset the cbp field. So for some skipped |
415 |
|
blocks, a test was done on the cbp value... |
416 |
|
- MEanalysis was using wrong mvs from the current bvop (unitialized |
417 |
|
or just wrong in current context). That's because in devapi3, bframes |
418 |
|
used to share the same mvs array wheras now, it's one array a bvop. |
419 |
|
- Collocated skipped MBs for a bvop didn't reset mvs[0] and b_mvs[0]. |
420 |
|
|
421 |
|
modified files: |
422 |
|
src/encoder.c src/motion/estimation_bvop.c src/motion/motion.h |
423 |
|
src/motion/motion_inlines.h src/motion/vop_type_decision.c |
424 |
|
|
425 |
|
|
426 |
|
2003-11-15 15:21:09 GMT patch-99 |
427 |
|
|
428 |
|
Summary: |
429 |
|
Small fixes |
430 |
|
Revision: |
431 |
|
xvidcore--devapi4--1.0--patch-99 |
432 |
|
|
433 |
|
Small fixes |
434 |
|
|
435 |
|
|
436 |
|
modified files: |
437 |
|
src/encoder.c src/motion/vop_type_decision.c |
438 |
|
|
439 |
|
|
440 |
|
2003-11-15 15:02:47 GMT patch-98 |
441 |
|
|
442 |
|
Summary: |
443 |
|
DShow update + libxvidcore project file update |
444 |
|
Revision: |
445 |
|
xvidcore--devapi4--1.0--patch-98 |
446 |
|
|
447 |
|
From peter: |
448 |
|
* DShow now links against libxvidcore.lib |
449 |
|
* Some updates to the libxvidcore project file |
450 |
|
|
451 |
|
modified files: |
452 |
|
TODO build/win32/libxvidcore.dsp dshow/dshow.dsp |
453 |
|
dshow/src/CXvidDecoder.cpp dshow/src/CXvidDecoder.h |
454 |
|
|
455 |
|
|
456 |
|
2003-11-15 01:51:28 GMT patch-97 |
457 |
|
|
458 |
|
Summary: |
459 |
|
Small updates to doc files |
460 |
|
Revision: |
461 |
|
xvidcore--devapi4--1.0--patch-97 |
462 |
|
|
463 |
|
Small updates to doc files |
464 |
|
|
465 |
|
|
466 |
|
modified files: |
467 |
|
AUTHORS CodingStyle TODO doc/INSTALL doc/README |
468 |
|
|
469 |
|
renamed files: |
470 |
|
.arch-ids/authors.txt.id |
471 |
|
==> .arch-ids/AUTHORS.id |
472 |
|
.arch-ids/todo.txt.id |
473 |
|
==> .arch-ids/TODO.id |
474 |
|
authors.txt |
475 |
|
==> AUTHORS |
476 |
|
todo.txt |
477 |
|
==> TODO |
478 |
|
|
479 |
|
|
480 |
|
2003-11-14 11:23:55 GMT patch-96 |
481 |
|
|
482 |
|
Summary: |
483 |
|
Updated ChangeLog |
484 |
|
Revision: |
485 |
|
xvidcore--devapi4--1.0--patch-96 |
486 |
|
|
487 |
|
Updated ChangeLog |
488 |
|
|
489 |
|
|
490 |
|
modified files: |
491 |
|
ChangeLog |
492 |
|
|
493 |
|
|
494 |
2003-11-13 23:09:34 GMT patch-95 |
2003-11-13 23:09:34 GMT patch-95 |
495 |
|
|