Commit graph

1061 commits

Author SHA1 Message Date
Mark Nauwelaerts
b9e32291a9 avdemux: only activate in pull mode if upstream is seekable 2012-09-11 17:52:57 +02:00
Wim Taymans
1e5bfa5215 avviddec: add support for codecs that use reget_buffer
Add support for codec that use reget_buffer. In this mode, it reuses the picture
and we need to attach the corresponding input frame to it or else we get the
timestamps wrong.
2012-09-11 13:53:23 +02:00
Edward Hervey
90a8b3fb9f ext: ffmpeg/gstffmpeg*.[ch] => libav/gstav*.[ch] 2012-09-10 17:53:54 +02:00
Edward Hervey
f4532d9135 remove postproc element
libpostproc is gone from upstream libav.
2012-09-10 17:18:30 +02:00
Edward Hervey
0e40eb11bc configure.ac: Import top-level include directory
And fixup all includes
2012-09-10 17:08:00 +02:00
Edward Hervey
9f8e46d3d2 avdemux: Really ignore demuxers we want to ignore 2012-09-10 11:17:11 +02:00
Tim-Philipp Müller
22f67f7b9e viddec: blacklist *crystalhd* decoder wrappers
There are gstreamer crystalhd elements as well,
so shouldn't be a big loss.

https://bugzilla.gnome.org/show_bug.cgi?id=683435
2012-09-07 17:56:43 +01:00
Sjoerd Simons
ed1ea6f097 ffmpegenc: Don't try to set caps to NULL
Fixes https://bugzilla.gnome.org/show_bug.cgi?id=682736
2012-08-30 20:03:10 +02:00
Wim Taymans
4b62101f44 viddec: use the right pointers
Use the plane pointers and strides.
Improve some debug
2012-08-10 17:03:26 +02:00
Sebastian Dröge
fe0a8d005d avviddec: Properly align and pad buffers for libav
https://bugzilla.gnome.org/show_bug.cgi?id=680796
2012-08-08 16:02:32 +02:00
Sebastian Dröge
3a4a139a28 av: Use */x-gst-av instead of */x-gst-ff for unknown caps 2012-08-07 10:19:16 +02:00
Sebastian Dröge
0764e9600c avdemux: Remove unused variable 2012-08-06 14:35:07 +02:00
Sebastian Dröge
1af5b5601d avdemux: Add stream-id to stream-start events 2012-08-06 14:05:28 +02:00
Edward Hervey
9b45987ef5 ffmpegviddec: Return NOT_NEGOTIATED when not configured 2012-08-06 11:55:31 +02:00
Tim-Philipp Müller
241419b46d gst_tag_list_free -> gst_tag_list_unref 2012-08-04 16:19:29 +01:00
Sebastian Dröge
3d2d1beb77 tag: Update for taglist/tag event API changes 2012-07-28 00:36:40 +02:00
Wim Taymans
3c47f584e2 avdec: require alignment=au for h264
Because we don't use a parser anymore we need to make sure that we feed the
decoder with full frames. For h264, require that we are aligned on access
units.

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=680540
2012-07-25 11:55:21 +02:00
Tim-Philipp Müller
2d458ca951 audio encoders: use context default bitrate if no bitrate has been set
Fixes 'bitrate too low: got 0, need 24000 or higher'
error when doing audiotstsrc ! avenc_wmav1 ! fakesink

https://bugzilla.gnome.org/show_bug.cgi?id=680487

https://bugzilla.gnome.org/show_bug.cgi?id=680487
2012-07-23 23:29:26 +01:00
Mark Nauwelaerts
0489f5eb78 ffmpeg: avoid assertion failure on duplicate frame allocation 2012-07-23 15:02:11 +02:00
Robert Swain
1638344081 ffmpegviddec: Correctly flag interlaced state
If the context interlaced state is indicated, the picture interlaced
state can be used to correctly indicate GST_VIDEO_BUFFER_FLAG_INTERLACED
on output buffers.
2012-07-23 11:37:57 +02:00
Sebastian Dröge
899fd3d76e ffvidenc: Allocate output buffers with the base class functions 2012-07-23 10:40:14 +02:00
Sebastian Dröge
941f4c331b ffmpegviddec: Update for video base classes API changes 2012-07-23 10:35:38 +02:00
Edward Hervey
c77db02d6e viddec: Call gst_video_decoder_negotiate() 2012-07-18 18:30:53 +02:00
Tim-Philipp Müller
4e784f0d0e avdeinterlace: fix up output caps for new field as well
And presumably our output is always progressive and not
interlaced any more?
2012-07-17 12:35:03 +01:00
Tim-Philipp Müller
ad04636c6e avdeinterlace: detect interlaced input properly
It's not a boolean field any more, and called "interlaced-mode" now.
2012-07-17 12:31:52 +01:00
Edward Hervey
93ef1b687e demux: Push STREAM_START when needed 2012-07-13 14:01:44 +02:00
Wim Taymans
a3b0ae22d7 avdec: ignore AAC errors instead of erroring out
Also ignore decode errors for AAC and carry on decoding like we do for all
other formats.

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=679639
2012-07-10 16:10:14 +02:00
Wim Taymans
0e1d6c0011 avcodecmap: only restrict caps on encoders
We only want to restrict the channels/rate and width/height on the encoder pads.
For decoders can't assume that the input caps contain channels/rate or
width/height so we need to accept everything.
2012-07-10 15:47:01 +02:00
Wim Taymans
dc7bf19858 update for query api changes 2012-07-06 11:50:57 +02:00
Wim Taymans
f8475ad22a update for query api changes 2012-07-06 11:27:33 +02:00
Wim Taymans
3ae9eb158c update for allocation query changes 2012-07-06 11:05:19 +02:00
Tim-Philipp Müller
ab7b7cbff3 cfg: remove some deprecated properties: statsfile, mb-qmin, mb-qmax
Use multipass-cache-file, qmin, qmax instead.
2012-07-06 09:49:57 +01:00
Sebastian Dröge
06d24b47b6 avdemux: Implement segment-done event 2012-07-05 13:21:33 +02:00
Wim Taymans
cdcaa584a4 avdec: handle_frame should always unref the input frame
Unref the input frame in all cases. Previously it was only unreffed when the
frame was used for output.
When we produce an output frame, take an extra ref because the frame is now
freed elsewhere.
2012-07-03 17:13:00 +02:00
Wim Taymans
c9878e9f86 avdec: small cleanup 2012-07-03 17:13:00 +02:00
Wim Taymans
41ad632275 avdec: avoid some refs
We can let the ffmpeg structure take ownership of the buffer.
2012-07-03 17:13:00 +02:00
Руслан Ижбулатов
3ff6f53f24 ffvidenc: Fix a format string to avoid warnings
istffmpegvidenc.c:412:3: error: format '%lu' expects argument of type 'long unsigned int', but argument 8 has type 'gint' [-Werror=format]
2012-07-03 08:38:28 +02:00
Sebastian Dröge
7b8ffe6169 ffaudenc: Fix format string compiler warning
gstffmpegenc.c:266:3: error: format '%lu' expects argument of type 'long unsigned int', but argument 8 has type 'gint'
2012-07-02 15:51:17 +02:00
Wim Taymans
103f8f9395 avdec: don't wait for keyframe
Don't wait for a keyframe but try to decode as soon as possible. In most cases,
the demuxer will send a keyframe first and things will work fine, in other
cases, like RTP, we try to decode ASAP to provide quick feedback. Some formats
also have no keyframes (h264 with intra refresh).
2012-06-28 09:33:28 +02:00
Wim Taymans
932bd7f79e avenc: don't leak frames 2012-06-26 12:36:56 +02:00
Wim Taymans
eb52e46b07 avdemux: update for task api change 2012-06-20 10:36:25 +02:00
Matej Knopp
17ee8113d7 ffmpegviddec: Don't reopen decoder on same caps, fix memory leaks, only drain on EOS 2012-06-20 07:39:40 +01:00
Wim Taymans
db53c6e5ec viddec: avoid renegotiation
Keep track of the context veriables and renegotiate when something changes. We
need to use separate values for this because not all values can be copied into a
GstVideoInfo.
Set the output interlacing based on the libav context.
Use the demuxer and decoder framerate to construct an output framerate.
Use demuxer and decoder par to construct an output framerate.
2012-06-19 12:23:32 +02:00
Sebastian Dröge
83f57519b4 ffmpegvidenc: Handle allocation queries and improve the padding/direct-rendering handling 2012-06-15 16:37:19 +02:00
Sebastian Dröge
6ff10a922a ffmpegviddec: Port to 0.11 again
No support for video meta, cropping, etc. yet
2012-06-15 15:18:47 +02:00
Sebastian Dröge
85f0dcb1e8 Merge remote-tracking branch 'origin/0.10' 2012-06-15 15:18:41 +02:00
Edward Hervey
1db8779252 ffmpegviddec: Properly return if negotiation fails
Instead of crashing ....
2012-06-15 12:19:58 +02:00
Sebastian Dröge
107c00a761 ffvidenc: Port to 0.11 2012-06-14 16:29:05 +02:00
Sebastian Dröge
de82bfbbba Merge remote-tracking branch 'origin/0.10' 2012-06-14 15:44:18 +02:00
Sebastian Dröge
4a58fd4be4 Merge remote-tracking branch 'origin/0.10'
Conflicts:
	common
	ext/ffmpeg/Makefile.am
	ext/ffmpeg/gstffmpegcfg.c
	ext/ffmpeg/gstffmpegcodecmap.c
	ext/ffmpeg/gstffmpegcodecmap.h
	ext/ffmpeg/gstffmpegdec.c
	ext/ffmpeg/gstffmpegenc.c
	ext/ffmpeg/gstffmpegenc.h
	tests/check/Makefile.am

Porting of the new video elements to 0.11 still pending.
2012-06-14 15:42:06 +02:00
Edward Hervey
ce7f65f752 ffmpegviddec: Properly update the interlacing
Fixes #678080
2012-06-14 15:13:31 +02:00
Edward Hervey
1ca115c44b ffmpegviddec: Properly error out when codec doesn't open
Avoids ending up using an un-set codec
2012-06-14 15:04:33 +02:00
Matej Knopp
a3c7cd4643 gstffmpegdec: only drain on EOS
See https://bugzilla.gnome.org/show_bug.cgi?id=675359
2012-06-11 16:27:35 +02:00
Tim-Philipp Müller
4ffec05c50 avenc: make buffer-size property read-only
Fixes FIXME.
2012-06-09 16:49:54 +01:00
Tim-Philipp Müller
7956478011 avenc: change properties of type ulong to int 2012-06-09 16:41:28 +01:00
Tim-Philipp Müller
c4ca1e86f8 avmux: blacklist "segment" muxer
Causes run-time warnings.
2012-06-09 14:37:09 +01:00
Wim Taymans
ee6c8bd498 update for audio api changes 2012-06-08 10:11:46 +02:00
Wim Taymans
8ae3ffa3d3 update for tag event change 2012-06-06 13:03:18 +02:00
Wim Taymans
c44eff6afb avdec: fix paletted formats
Pass the palette as the side data for AVPacket so that it is written in the
second plane of output AVFrame.
2012-05-31 13:55:44 +02:00
Wim Taymans
e38b00489f libav: add support for 10 bits I420
See https://bugzilla.gnome.org/show_bug.cgi?id=665034
2012-05-25 17:03:53 +02:00
Nicolas Dufresne
35cf2e84f2 codecmap: correctly set height in AVCodecContext
https://bugzilla.gnome.org/show_bug.cgi?id=674899
2012-04-27 16:53:28 +02:00
Edward Hervey
f0a411235a ffmpegviddec: Always keep a reference to the frame
Ensures no data gets cleared before the decoders are really done
with it.
2012-04-26 18:56:35 +02:00
Edward Hervey
6d2b483e15 ffmpegviddec: Flush and drain when needed
we need to flush avcodec internal state on hard resets
we need to drain out pending buffers on EOS and soft resets
2012-04-26 18:53:51 +02:00
Edward Hervey
252c5e0a92 ffmpegviddec: Add video/x-raw-gray to src pad template 2012-04-24 15:36:30 +02:00
Edward Hervey
8e42e0b6a4 codecmap: Add more GstVideoInfo<=>PixelFormat mappings 2012-04-24 15:36:00 +02:00
Edward Hervey
272823296f ffmpegviddec: Port to GstVideoDecoder 2012-04-24 11:37:02 +02:00
Edward Hervey
134f7d7058 gstffmpegvidenc: Port to -base video classes 2012-04-24 11:37:02 +02:00
Edward Hervey
bdf7ebf411 ffmpegenc/dec: Remove audio/video specific code
Makes each file more manageable, second step to porting to base classes
2012-04-24 11:36:31 +02:00
Edward Hervey
6310ed46dd ffmpeg: Clone encoder and decoder files
First step in porting to base video classes
2012-04-24 11:35:56 +02:00
Edward Hervey
04a2d01877 codecmap: Add utility for using with GstVideoFormat 2012-04-24 11:35:56 +02:00
Tim-Philipp Müller
c69da6d988 ffmpegdemux: skip ffmpeg matroska,webm typefinder again
Causes warnings, and shouldn't be needed.
2012-04-22 11:52:18 +01:00
Wim Taymans
cd2e53aa9f Merge remote-tracking branch 'origin/0.10' 2012-04-20 16:23:41 +02:00
Wim Taymans
630790f50d ffdec: don't flush buffers on DISCONT
Don't flush the buffers that ffmpeg has on DISCONT but instead let it recover.
This gives a much better image in the case of packet loss.

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=635020
2012-04-20 16:16:25 +02:00
Wim Taymans
2e3dd525c8 Merge remote-tracking branch 'origin/0.10' 2012-04-19 15:37:06 +02:00
Wim Taymans
e746746120 postproc: fix compilation when ORC is missing
Fix 'set but not used' compiler error when ORC is missing.
2012-04-19 15:35:48 +02:00
Wim Taymans
8fee156175 Merge remote-tracking branch 'origin/0.10'
Conflicts:
	ext/libswscale/gstffmpegscale.c
2012-04-19 14:49:18 +02:00
Wim Taymans
b053a50508 ffscale: fix compilation when ORC is missing
Fix set but not used compiler error when ORC is missing.
2012-04-19 14:45:42 +02:00
Mark Nauwelaerts
5c9d454517 collectpads2: rename to collectpads 2012-04-17 15:36:12 +02:00
Sebastian Dröge
e49c069995 Rename everything from gst-ffmpeg to gst-libav
Also change elements factory names to avenc_*, avdec_*,
avmux_* and avdemux_*. The actual filenames, type names,
function/variable names are not touched to make merging
with 0.10 easier at this point. Once 0.10 is not supported
anymore everything should be renamed.
2012-04-12 19:41:52 +02:00
Tim-Philipp Müller
c1ae78ecea ffmpeg: use set_metadata() for generated strings
Reverts parts of previous commit.
2012-04-10 01:01:00 +01:00
Tim-Philipp Müller
b06b11809b Use new gst_element_class_set_static_metadata() 2012-04-10 00:53:43 +01:00
Sebastian Dröge
1879c67d58 gst: Update for GST_PLUGIN_DEFINE() API changes 2012-04-05 17:41:11 +02:00
Sebastian Dröge
37fcf4009f ffmpeg: Update versioning 2012-04-04 14:53:51 +02:00
Wim Taymans
e438f20c93 ffdec: handle DTS/PTS
Now that we can know about PTS and DTS on incomming buffers, track them and set
the correct output timestamps.
We can remove the code to estimate if input buffers have pts or dts.
2012-04-03 11:52:05 +02:00
Wim Taymans
182af2df48 update for buffer api change 2012-03-30 18:14:00 +02:00
Wim Taymans
2c881eb752 update for buffer changes 2012-03-28 12:50:19 +02:00
Wim Taymans
894be47d5f ffenc: fix caps handling
Avoid copies and leaks
2012-03-27 16:39:09 +02:00
Wim Taymans
b0cc0cf55d Merge branch 'master' into 0.11
Conflicts:
	gst-libs/ext/libav
2012-03-22 15:31:40 +01:00
Wim Taymans
c9538928b6 ffdec: ensure padding is big enough 2012-03-19 11:06:40 +01:00
Wim Taymans
b7b587bce0 ffdec: check if input is already 0-padded
When the input buffer is already 0-padded we don't need to copy. This can happen
because in the ALLOCATION query we suggest this.
Only do temporary padding when we are using a parser.
2012-03-19 10:29:34 +01:00
Wim Taymans
3d2527c0a2 update for bufferpool changes 2012-03-15 22:11:54 +01:00
Wim Taymans
427ad305c5 update for allocation query changes 2012-03-15 20:39:10 +01:00
Wim Taymans
d9e16a498c ffdec: propose some padding in the allocation query
Hopefully upstream can add some padding and then we don't have to memcpy.
2012-03-14 20:04:53 +01:00
Wim Taymans
d606efc141 take padding into account 2012-03-14 19:53:06 +01:00
Wim Taymans
78e0187282 ffdec: remove unused crop property 2012-03-14 17:35:28 +01:00
Wim Taymans
dab06c0e5d ffdec: remove padding property
Users should not mess with this
2012-03-14 17:33:40 +01:00
Wim Taymans
75ba435552 ffdec: make bufferpool setup recalculate direct rendering
We need to recalculate if we can use direct rendering after we have setup the
bufferpool.
2012-03-14 12:58:38 +01:00
Wim Taymans
1ba905c031 ffdec: add some PERFORMANCE logging 2012-03-14 11:53:42 +01:00
Wim Taymans
12c0db9091 ffcodec: add voc demuxer 2012-03-13 13:24:39 +01:00
Wim Taymans
c2469e82f9 ffdemux: activate pad before setting caps 2012-03-13 13:24:23 +01:00