gmane.comp.video.ffmpeg.devel

Re: FFmpeg wiki

compn — 2012-05-26T00:59:46

true, but editing docs with texi formatting is really slow and painful.
ffmpeg has a history of not updating docs. of course, some devels arent
interested in wiki web-interfaces either. pros/cons to both.

i got tired of updating mplayer docs and just started dumping to wiki:
http://wiki.multimedia.cx/index.php?title=MPlayer_FAQ

-compn

Re: FFmpeg wiki

Stefano Sabatini — 2012-05-26T00:00:31

On date Friday 2012-05-25 16:18:32 +0200, Burek Pekaric encoded:

We already have a wiki:
http://ffmpeg.org/trac/ffmpeg/wiki

although it is still very incomplete and not much popular / renown.


Regarding docs -> wiki conversions, I'm not sure it is a good idea, in
theory we should keep the official docs separated from the wiki,
otherwise we'll have diverging versions and confused users.

Thanks for your interest.

Re: [PATCH] lavc/raw: add test utility for listingfourcc/pix_fmt map

Stefano Sabatini — 2012-05-25T23:56:28

On date Saturday 2012-05-26 01:47:18 +0200, Stefano Sabatini encoded:

And this is the output of libavcodec/raw-test -L:

yuv420p: I420 IYUV YV12 
yuyv422: YUY2 Y422 V422 VYUY YUNV yuv2 yuvs DVOO 
rgb24: RGB[24] 
bgr24: BGR[24] 24BG 
yuv422p: Y42B P422 YV16 422P 
yuv444p: 444P YV24 
yuv410p: YUV9 YVU9 
yuv411p: Y41B 411P 
gray: Y800 Y8   GREY 
monow: B1W0 
monob: B0W1 
pal8: 
yuvj420p: I420 IYUV YV12 
yuvj422p: Y42B P422 422P 
yuvj444p: 444P 
uyvy422: UYVY HDYC UYNV UYNY uyv1 2Vu1 AVRn AV1x AVup VDTZ auv2 cyuv 2vuy 2Vuy AVUI 
uyyvyy411: Y411 
bgr8: BGR[8] 
bgr4: BGR[4] 
bgr4_byte: R4BY 
rgb8: RGB[8] 
rgb4: RGB[4] 
rgb4_byte: B4BY 
nv12: NV12 
nv21: NV21 
argb: ARGB 
rgba: RGBA RGBA 
abgr: ABGR ABGR 
bgra: BGRA BGRA 
gray16be: [16][0]1Y b16g 
gray16le: Y1[0][16] 
yuv440p: 440P 
yuvj440p: 440P 
yuva420p: Y4[11][8] 
rgb48be: 0RGB b48r 
rgb48le: RGB0 
rgb565be: [16]BGR B565 
rgb565le: RGB[16] L565 [3][0][0][0] 
rgb555be: [15]BGR 
rgb555le: RGB[15] L555 
bgr565be: [16]RGB 
bgr565le: BGR[16] 
bgr555be: [15]RGB 
bg

[PATCH] lavc: prettify printing of some codec tagswhich contains non alphanumeric char

Stefano Sabatini — 2012-05-25T23:47:32

Make av_get_codec_tag_string() show codec tag string characters in a more
intelligible ways. For example the ascii char "< at >" is used as a number, so
should be displayed like a "[64]" rather than as a printable character.

Apart alphanumeric chars, only the characters ' ' and '.' are used
literally in codec tags, all the other characters represent numbers.
---
 libavcodec/utils.c |    7 ++++++-
 1 files changed, 6 insertions(+), 1 deletions(-)

diff --git a/libavcodec/utils.c b/libavcodec/utils.c
index 15dd05a..74329e1 100644
--- a/libavcodec/utils.c
+++ b/libavcodec/utils.c
< at >< at > -1774,9 +1774,14 < at >< at > size_t av_get_codec_tag_string(char *buf, size_t buf_size, unsigned int codec_ta
 {
     int i, len, ret = 0;
 
+#define IS_PRINT(x)                                               \
+    (((x) >= '0' && (x) <= '9') ||                                \
+     ((x) >= 'a' && (x) <= 'z') || ((x) >= 'A' && (x) <= 'Z') ||  \
+     ((x) == '.' || (x) == ' '))
+
     for (i = 0; i < 4; i++) {
         len = snprintf(buf, buf_s

[PATCH] lavc/raw: add test utility for listingfourcc/pix_fmt map

Stefano Sabatini — 2012-05-25T23:47:18

The test program is useful for checking which raw formats are currently
identified by a codec tag.
---
 libavcodec/Makefile |    1 +
 libavcodec/raw.c    |   96 +++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 97 insertions(+), 0 deletions(-)

diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index da22704..add73bb 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
< at >< at > -783,6 +783,7 < at >< at > TESTPROGS = cabac                                                       \
             golomb                                                      \
             iirfilter                                                   \
             rangecoder                                                  \
+            raw                                                         \
             snowenc                                                     \
 
 TESTPROGS-$(HAVE_MMX) += motion
diff --git a/libavcodec/raw.c b/libavcodec/raw.c
index 2a6d7d0..45a6d24 100644
--- a/libavcodec/raw.c
+++ b/libavcode

[PATCH] ffprobe: reduce the number of AVBPrintresets.

Clément Bœsch — 2012-05-25T23:28:45

---
 ffprobe.c |   26 ++++++++++----------------
 1 file changed, 10 insertions(+), 16 deletions(-)

diff --git a/ffprobe.c b/ffprobe.c
index d53ec11..cbcc708 100644
--- a/ffprobe.c
+++ b/ffprobe.c
< at >< at > -662,19 +662,16 < at >< at > static void compact_show_tags(WriterContext *wctx, AVDictionary *dict)
     AVDictionaryEntry *tag = NULL;
     AVBPrint buf;
 
+    av_bprint_init(&buf, 1, AV_BPRINT_SIZE_UNLIMITED);
     while ((tag = av_dict_get(dict, "", tag, AV_DICT_IGNORE_SUFFIX))) {
         if (wctx->nb_item) printf("%c", compact->item_sep);
-
-        if (!compact->nokey) {
-            av_bprint_init(&buf, 1, AV_BPRINT_SIZE_UNLIMITED);
+        av_bprint_clear(&buf);
+        if (!compact->nokey)
             printf("tag:%s=", compact->escape_str(&buf, tag->key, compact->item_sep, wctx));
-            av_bprint_finalize(&buf, NULL);
-        }
-
-        av_bprint_init(&buf, 1, AV_BPRINT_SIZE_UNLIMITED);
+        av_bprint_clear(&buf);
         printf("%s", compact->escape_str(&buf, tag->value, compact->item_sep, wc

[PATCH] ffprobe: add flat output format.

Clément Bœsch — 2012-05-25T23:14:50

---
 doc/ffprobe.texi |    6 ++++
 ffprobe.c        |   81 +++++++++++++++++++++++++++++++++++++++++++++++++++++-
 2 files changed, 86 insertions(+), 1 deletion(-)

diff --git a/doc/ffprobe.texi b/doc/ffprobe.texi
index ed96575..2cab961 100644
--- a/doc/ffprobe.texi
+++ b/doc/ffprobe.texi
< at >< at > -269,6 +269,12 < at >< at > CSV format.
 This writer is equivalent to
 < at >code{compact=item_sep=,:nokey=1:escape=csv}.
 
+< at >section flat
+Flat format.
+
+A free-form output with each line contains an explicit key=value, such as
+"stream.3.tag.25=foobar"
+
 < at >section json
 JSON based format.
 
diff --git a/ffprobe.c b/ffprobe.c
index d53ec11..43770cb 100644
--- a/ffprobe.c
+++ b/ffprobe.c
< at >< at > -710,6 +710,84 < at >< at > static const Writer csv_writer = {
     .flags = WRITER_FLAG_DISPLAY_OPTIONAL_FIELDS,
 };
 
+/* Flat output */
+
+typedef struct FlatContext {
+    const AVClass *class;
+    const char *section;
+    int section_entry_id;
+} FlatContext;
+
+#undef OFFSET
+#define OFFSET(x) offsetof(FlatContext, x)
+
+static const char *flat_get_n

Re: [PATCH 0/2] RFC: TEXT subtitle decoder

Clément Bœsch — 2012-05-25T22:07:32

HTML-like markup


Well I'd rather add one actually. Like CODEC_ID_SRT (complete subtitles
entry as it is now, with the X/Y positionning hack), and
CODEC_ID_MUXED_SRT or CODEC_ID_PURE_SRT with only the text markup.

Re: [PATCH 0/2] RFC: TEXT subtitle decoder

Michael Niedermayer — 2012-05-25T22:05:02


patch? did i miss some patch



why?

[...]

Re: [PATCH 0/2] RFC: TEXT subtitle decoder

Nicolas George — 2012-05-25T21:59:59

Le septidi 7 prairial, an CCXX, Hendrik Leppkes a écrit :

There is pseudo-HTML markup. We need to decide what to do about it.

Regard,

Re: [PATCH 0/2] RFC: TEXT subtitle decoder

Hendrik Leppkes — 2012-05-25T21:58:29

Once you strip all timing metadata from the srt packets, there is nothing
special about the packet anymore, they are just pure text.
I would rather recommend to drop/deprecate CODEC_ID_SRT (its really more of
a container to store text with timing info in a plain-text file), and use
CODEC_ID_TEXT for all those cases.

Note that CODEC_ID_TEXT is not ttxt (as in MPEG4 Timed Text), thats
CODEC_ID_MOV_TEXT.

Re: Best Android Configuration?

Kevin Kowalewski — 2012-05-25T21:31:22

That's true... I took out the two extra-cflags like you mentioned, but no
performance gain was realized.

Do you have any advice about the GPU acceleration? Some Android video
players like BSPlayer make use of the local GPU.

Kevin

On Fri, May 25, 2012 at 5:14 PM, Reimar Döffinger
<Reimar.Doeffinger< at >gmx.de>wrote:

Re: Best Android Configuration?

Reimar Döffinger — 2012-05-25T21:14:33

This is not at all ARM-specific (neither the options, nor their effect),
it is general "don't shoot yourself in the foot on purpose" advice.
And unfortunately the mere existence of some way to set different CFLAGS
seems to invariably cause people to use it extensively to shoot
themselves in the foot, I have no idea why people insist not just
using the default config when I don't see why they would even think
to know better than the developers of that software and worse testing
would reveal that what they are doing is a bad idea.

Re: Best Android Configuration?

compn — 2012-05-25T20:58:09

it may be useful information for android arm docs, if anyone writes
them.

-compn
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel< at >ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [PATCH 0/2] RFC: TEXT subtitle decoder

Clément Bœsch — 2012-05-25T20:50:31

Ah I'm sorry I just read the thread right now.


OK, sorry for the confusion.

Still, it seems to me your first approach was a better one.

Or to be more correct, I think it should be done like this:
 - fix the subrip demuxer (lavf/srtdec.c) to set pkt->duration (it already
   set pkt->pts) and strip out the timing text information from the
   packet.
 - change the srt "raw" muxer into a smarter one using pts & duration (so
   introduce a lavf/srtenc.c and drop the raw one from lavf/rawenc.c)
 - make the subrip decoder (lavc/srtdec.c) use these new fields instead or
   reconstructing them from the packet.
 - then just drop the {"S_TEXT/UTF8", CODEC_ID_TEXT} in matroska like you
   did the first time.

This is almost what is currently done in the MicroDVD design: the demuxer
set the pts/dts packet duration, and the decoder use them. The difference
is that the decoder receives the complete timing text information (and
just ignore them). I think the benefit of this is that it can be remuxed
easily and exactly

Re: [PATCH] Optimization of AMR NB and WB decoders for MIPS

Vitor Sessak — 2012-05-25T20:46:11

Indeed, it looks like the compiler do a mostly good job. So I'm fine 
with leaving the loops in C.


How can does exactly ff_acelp_lspd2lpc() benefit from FP MADD? By saving 
a single add in every call of the function? Unless madds are faster than 
plain muls, it doesn't looks like a benchmarkable gain. I'd say it has 
more to do with allowing the compiler to inline the call to the mips 
version of ff_lsp2polyf()...

Can you do a test? Copy the unmodified C version of ff_acelp_lspd2lpc to 
mips/amrwb_lsp2lpc.h and see if it is any slower than the madd version.

-Vitor

Re: [PATCH 0/2] RFC: TEXT subtitle decoder

Wolfram Gloger — 2012-05-25T20:04:51


Yes, I transcode them with -scodec dvdsub and that works well.  Or did
you mean converting from the bitmaps into text using OCR?


I have been using a teletext decoder with ffmpeg, based on libzvbi, for
many years.  Just recently I added an option to generate plain UTF-8
text subtitles rather than bitmaps.  My patch from earlier today was a
necessary step to mux this into .mkv and test on my TV.  It works :)

Oh well, I'm afraid it might take years to get this merged. :-/

Regards,
Wolfram.

Re: Best Android Configuration?

Kevin Kowalewski — 2012-05-25T19:49:22

Thank you for the response. What about hardware acceleration? I noticed one
of the video players in particular opens the GPU OpenGL library for
interaction with FFMpeg.

D/libEGL  (28389): loaded /system/lib/egl/libEGL_mali.so
D/libEGL  (28389): loaded /system/lib/egl/libGLESv1_CM_mali.so
D/libEGL  (28389): loaded /system/lib/egl/libGLESv2_mali.so

Thanks,
Kevin

On Fri, May 25, 2012 at 3:33 PM, Reimar Döffinger
<Reimar.Doeffinger< at >gmx.de>wrote:

Re: Best Android Configuration?

Reimar Döffinger — 2012-05-25T19:33:03

This really does not belong on this list, but if you have to ask
you generally shouldn't use stuff like --extra-cflags.
Particularly ones like -fpic -fno-strict-aliasing which are
just asking for bad performance for no good reason.

Re: Multi-threaded decoding via C++?

Reimar Döffinger — 2012-05-25T19:29:08

I believe there is hardly any ARM/NEON optimization for them either,
so that is something else contributions would be welcome and helpful
for...

Best Android Configuration?

Kevin Kowalewski — 2012-05-25T19:06:56

Hi,

We're currently using the following script to generate the appropriate
configuration for Android, but find the performance is still not enough on
a dual core Cortex A9 CPU under Android.

We suspect no hardware acceleration is possible yet with FFMpeg and Android.

Does anyone know of better configuration options?

Kevin

######################################################
# FFmpeg builds script for Android+ARM platform
#
# This script is released under term of
#   CDDL (http://www.opensource.org/licenses/cddl1)
# Wrote by pinxue (~< at >gmail.com) from RockPlayer.com
#                                   2010-8 ~ 2011-4
######################################################

#NDK=~/android/android-ndk-r4
#PLATFORM=$NDK/build/platforms/android-8/arch-arm/
#PREBUILT=$NDK/build/prebuilt/darwin-x86/arm-eabi-4.4.0

PLATFORM=$NDK/platforms/android-8/arch-arm
PREBUILT=$NDK/toolchains/arm-linux-androideabi-4.4.3/prebuilt/darwin-x86

function build_one
{

# -fasm : required. Android header file uses asm keyword ins