The speed-level property, which allows callers to trade of encoding
quality for speed in the libtheora api, has a version-dependent
maximum and default values. Instead of hardcoding the acceptable
range for the theoraenc element's presentation of this setting,
we query the library directly at class initialization time and
set the maximum and default values from that. If the query fails,
we fall back to the previous default setting.
To keep the values reported by gst-inspect (which I'm told use
the spec values from the class) with those available on an\
instantiated element, we remove to setting of enc->speed_level
from the initializer and instead pass G_PARAM_CONSTRUCT to
the property spec flags, asking g_object to set this property
when theoraenc objects are constructed.
NB in theory the maximum speed-level could depend on the actual
video caps. If later versions of libtheoraenc do this, a second
call will need to be made from theora_enc_reset to update the
property, since this function is mostly useful for realtime
adjustment of performance while the pipeline is running.
libtheora has two encoding modes, CBR, where it tries to hit a target
bitrate and VBR where it tries to achieve a target quality.
Internally if the target bitrate is set to anything other then 0 the
encoding-mode is CBR.
This means that the gstreamer element can leave the video_quality
setting alone as long as the user is tweaking the bitrate. Which has the
nice side-effect that if the user explicitely sets the bitrate to 0
(which is actually the default), the quality value doesn't get reset and
one ends up encoding VBR at quality-level 0...
Remove "This property requires libtheora version >= 1.1" qualifiers
from property descriptions. They aren't needed any longer now that
we require libtheora >= 1.1.
Since this is just a debugging feature and libtheora will usually not be
compiled with that option enabled, we should maybe just hide these properties,
since they won't work anyway, and avoid confusing warnings.
Also rename properties to make them less cryptic.
https://bugzilla.gnome.org/show_bug.cgi?id=628488
Do some additional checks on the granulpos timestamp before using it for
calculating the duration because oggdemux generates wrong granulepos now.
Fixes seeking somewhat again.
Previously, the code always rounded to even sizes. Now it only ensures
that pic_x and pic_y are multiples of 2 if the output format requires
it.
Also inlcudes fixes to take pic_x/y into account properly when copying
the buffer.
https://bugzilla.gnome.org/show_bug.cgi?id=594729
This guarantees that people who use theoraenc without modifying any
properties will end up with a reasonably good quality output.
48 is also the default of the encoder_example application shipped with
libtheora.
For this to work properly, theoradec and vorbisdec need to put
tag events received from upstream into the pending_events list
so they get pushed out after any newsegment event, not before.
Add property "speed-level" to control the amount of motion searching
the encoder does. This is only available in libtheora >= 1.0 and
will silently fail with earlier libraries. Fixes: #572275.
Signed-off-by: David Schleef <ds@schleef.org>
Original commit message from CVS:
* ext/theora/gsttheoraenc.h:
* ext/theora/theoraenc.c: (gst_theora_enc_init),
(theora_buffer_from_packet), (theora_push_packet),
(theora_enc_sink_event), (theora_enc_is_discontinuous),
(theora_enc_chain):
Parse segment events.
Pass incomming buffer timestamps to outgoing buffers.
Use the running_time to construct the granulepos.
Fixes#562163.
Original commit message from CVS:
Based on patch by: Robin Stocker <robin at nibor dot org>
* ext/theora/gsttheoradec.h:
* ext/theora/theoradec.c: (gst_theora_dec_init),
(theora_dec_setcaps), (theora_handle_type_packet),
(theora_dec_decode_buffer), (theora_dec_change_state):
Parse input caps and make the PAR override the encoded PAR when
specified by a container. Fixes#555699.
Original commit message from CVS:
Patch by: ogg.k.ogg.k <ogg dot k dot ogg dot k at googlemail dot com>
* ext/theora/theoraparse.c: (theora_parse_set_streamheader):
Set the BOS flag on the BOS packet. Fixes#553244.
Original commit message from CVS:
* docs/plugins/Makefile.am:
* docs/plugins/gst-plugins-base-plugins-docs.sgml:
* docs/plugins/gst-plugins-base-plugins-sections.txt:
* ext/theora/theoraparse.c:
Add theoraparse to the docs and fix some docs.
Original commit message from CVS:
* ext/theora/theoradec.c: (theora_handle_type_packet),
(theora_dec_chain):
Don't try to do anything fancy with the return code from pushing an
event, it does not have enough information to turn it into a
GST_FLOW_ERROR.
Original commit message from CVS:
* ext/theora/gsttheoradec.h:
* ext/theora/gsttheoraparse.h:
* ext/theora/theoradec.c:
* ext/theora/theoraparse.c:
Take a 2nd stab at handling libtheora granulepos changes in the decoder
and parser by inspecting the bitstream version of the incoming data.
Original commit message from CVS:
* ext/theora/theoradec.c: (gst_theora_dec_class_init),
(_theora_granule_frame), (_theora_granule_start_time),
(theora_dec_sink_convert), (theora_dec_decode_buffer):
Adapt for post-alpha meaning of granulepos, when we
have a newer version of libtheora.
* ext/theora/theoraenc.c: (gst_theora_enc_class_init),
(theora_enc_get_ogg_packet_end_time), (theora_enc_sink_event),
(theora_enc_is_discontinuous), (theora_enc_chain):
Likewise.
* tests/check/Makefile.am:
Link libtheora into theoraenc test so we can check which version of
libtheora we're testing against.
* tests/check/pipelines/theoraenc.c: (check_libtheora),
(check_buffer_granulepos),
(check_buffer_granulepos_from_starttime), (GST_START_TEST),
(theoraenc_suite):
Adapt tests to check the values that are now defined for theora; make
the tests backwards-adapt the passed values if we're running against an
old libtheora.
Fixes#497964
Original commit message from CVS:
* docs/design/draft-keyframe-force.txt:
* ext/theora/theoraenc.c: (theora_enc_sink_event),
(theora_enc_chain):
Add draft design for forcing keyframes in encoders and implement in
theoraenc.
Original commit message from CVS:
* ext/cdparanoia/gstcdparanoiasrc.c:
(gst_cd_paranoia_src_read_sector):
* gst-libs/gst/audio/gstbaseaudiosrc.c:
(gst_base_audio_src_create):
Small cleanups.
* ext/theora/theoradec.c: (theora_dec_sink_event):
Fix typo.
* gst-libs/gst/rtp/gstbasertpdepayload.c:
(gst_base_rtp_depayload_set_gst_timestamp):
Add some FIXME
* gst/playback/gstdecodebin.c: (queue_underrun_cb):
And some debug info when a FIXME path is hit.
Original commit message from CVS:
* ext/theora/gsttheoradec.h:
* ext/theora/theoradec.c: (gst_theora_dec_reset),
(theora_dec_sink_event), (theora_handle_comment_packet),
(theora_handle_type_packet), (theora_dec_change_state):
Don't push events (newsegment, tags) before initialising the
decoder.
This is neccesary for seeking to work correctly in gnonlin.
Original commit message from CVS:
* ext/theora/gsttheoraenc.h:
* ext/theora/theoraenc.c: (theora_enc_sink_setcaps),
(theora_enc_sink_event), (theora_enc_change_state):
Track initialisation state; don't try to use encoder state if we're
not initialised (it'll segfault).
Original commit message from CVS:
* ext/theora/theoraenc.c (theora_buffer_from_packet, theora_enc_chain):
Don't use pad_alloc_buffer_and_set_caps to create a small header
packet, or, worse, to create a big temporary video buffer using the
src pad.
Original commit message from CVS:
* ext/theora/theoradec.c: (theora_handle_type_packet):
Since the plugin doesn't support anything other than 4:2:0 right
now, post an error and fail if we get something else. Won't matter
until libtheora supports the other pixel formats, but hopefully
that'll be soon...
Original commit message from CVS:
* ext/theora/theoraenc.c: (theora_enc_chain):
Check return value of theora_encode_header(), or we might try to
allocate a random number of bytes. theora_encode_header() can fail
if libtheora has been compiled with encoding support disabled.
Fixes#398110.
Original commit message from CVS:
* tests/check/pipelines/theoraenc.c: (check_buffer_granulepos),
(GST_START_TEST):
It would be very bad if, after a discont buffer, we thought every
single following buffer was also discont. So, add to the test to
ensure that this isn't the case.
* ext/theora/theoraenc.c: (theora_enc_is_discontinuous):
... it was the case. So fix it.
Original commit message from CVS:
* ext/theora/theoradec.c: (gst_theora_dec_reset),
(theora_dec_push_forward), (theora_dec_push_reverse),
(theora_handle_data_packet), (theora_dec_decode_buffer),
(theora_dec_flush_decode), (theora_dec_chain_reverse),
(theora_dec_chain_forward), (theora_dec_chain):
Implement reverse playback.
* ext/vorbis/vorbisdec.c: (gst_vorbis_dec_reset),
(vorbis_dec_decode_buffer), (vorbis_dec_flush_decode),
(vorbis_dec_chain_forward):
Clear buffers used for reverse playback in _reset.
No need to set the eos flag, we clip samples using the segment.
Original commit message from CVS:
* ext/theora/gsttheoradec.h:
* ext/theora/theoradec.c: (gst_theora_dec_init),
(theora_dec_sink_event), (theora_dec_chain_forward),
(theora_dec_flush_decode), (theora_dec_chain_reverse),
(theora_dec_chain):
Document and partially implement an algorithm for doing reverse playback
of theora video.
Original commit message from CVS:
* ext/theora/theoradec.c: (theora_dec_chain):
Zero byte theora packets are valid and well-defined; don't warn on
them.
Original commit message from CVS:
2006-07-14 Andy Wingo <wingo@pobox.com>
* ext/theora/gsttheoraparse.h:
* ext/theora/theoraparse.c (theora_parse_drain_event_queue)
(theora_parse_push_headers, theora_parse_clear_queue)
(theora_parse_drain_queue_prematurely, )
(theora_parse_sink_event, theora_parse_change_state): Queue events
until we initialized our state, like in vorbisparse.
Original commit message from CVS:
2006-07-08 Andy Wingo <wingo@pobox.com>
* theoraparse.c (theora_parse_push_buffer)
(theora_parse_drain_queue_prematurely, theora_parse_drain_queue):
Add some more debugging. Fix granulepos reconstruction in the face
of discontinuities.
Original commit message from CVS:
* ext/theora/theoraparse.c: (theora_parse_drain_queue_prematurely),
(theora_parse_drain_queue):
Mark DELTA_UNIT on non-keyframes.
Original commit message from CVS:
* ext/theora/theoradec.c:
* ext/theora/theoraenc.c:
* ext/theora/theoraparse.c: (gst_theora_parse_class_init):
* gst/audiorate/gstaudiorate.c:
make more debug catagories static
* tests/check/Makefile.am:
* tests/check/elements/adder.c: (message_received),
(test_event_message_received), (GST_START_TEST),
(test_play_twice_message_received), (adder_suite):
added test case for using element twice, extra bonus points for anyone
who can make these test run reliably
Original commit message from CVS:
* ext/theora/theoradec.c: (theora_dec_chain):
Make work with time-stamped input buffers that do not
have a granulepos in BUFFER_OFFSET_END (like theora
buffers coming from matroskademux). Fixes#342448.
Original commit message from CVS:
* ext/theora/theoradec.c: (gst_theora_dec_reset),
(theora_dec_src_query), (theora_dec_src_event),
(theora_dec_sink_event), (theora_handle_comment_packet),
(theora_handle_data_packet), (theora_dec_change_state):
Cleanups, add some G_LIKELY.
Use segment helpers instead of our own wrong code.
Clear queued buffers on seek and READY.
* ext/vorbis/vorbisdec.c: (gst_vorbis_dec_reset),
(vorbis_dec_convert), (vorbis_dec_src_query),
(vorbis_dec_src_event), (vorbis_dec_sink_event),
(vorbis_handle_comment_packet), (vorbis_dec_push),
(vorbis_handle_data_packet), (vorbis_dec_chain),
(vorbis_dec_change_state):
* ext/vorbis/vorbisdec.h:
Remove old useless packetno variable.
Do position query properly.
Add some G_LIKELY.
Do cleanup of queued buffers in new helper function
and use it.
Original commit message from CVS:
2006-05-05 Andy Wingo <wingo@pobox.com>
* ext/theora/theoraparse.c (gst_theora_parse_init)
(theora_parse_src_convert, theora_parse_src_query):
* ext/vorbis/vorbisparse.c (gst_vorbis_parse_init)
(vorbis_parse_convert, vorbis_parse_src_query): Add convert and
query functions on the source pads of the theora and vorbis parse
elements. Fixes position querying when doing a remux.
Original commit message from CVS:
* ext/theora/theoraparse.c: (parse_granulepos),
(theora_parse_drain_queue_prematurely),
(theora_parse_queue_buffer), (theora_parse_sink_event):
Fix flushing.
Fix invalid granulepos outputs when starting with a non-keyframe.
Original commit message from CVS:
* ext/ogg/gstoggdemux.c: (gst_ogg_demux_chain_peer),
(gst_ogg_chain_mark_discont), (gst_ogg_chain_new_stream),
(gst_ogg_demux_activate_chain), (gst_ogg_demux_perform_seek):
Mark buffers with DISCONT after seek and after activating new
chains.
* ext/theora/gsttheoradec.h:
* ext/theora/theoradec.c: (gst_theora_dec_reset),
(theora_get_query_types), (theora_dec_sink_event),
(theora_dec_push), (theora_handle_data_packet), (theora_dec_chain),
(theora_dec_change_state):
Fix frame counter.
Detect and mark DISCONT buffers.
* ext/vorbis/vorbisdec.c: (vorbis_dec_src_query),
(vorbis_dec_sink_event), (vorbis_dec_push), (vorbis_dec_chain),
(vorbis_dec_change_state):
* ext/vorbis/vorbisdec.h:
Use GstSegment.
Detect and mark DISCONT buffers.
Don't crash on 0 sized buffers.
Original commit message from CVS:
* ext/theora/theoradec.c: (theora_dec_src_event),
(theora_handle_data_packet):
Some more debug info.
* tests/examples/seek/seek.c: (start_seek), (main):
Print element messages too.
Original commit message from CVS:
2006-04-04 Andy Wingo <wingo@pobox.com>
* ext/theora/theoradec.c (theora_handle_data_packet): This value
is often negative -- make it signed so as not to wrap around.
Fixes segfaults introduced on 9 March.
Original commit message from CVS:
* ext/theora/gsttheoradec.h:
* ext/theora/theoradec.c: (theora_dec_src_event):
Don't try to store a gdouble in a gboolean.
Small cleanups.
Original commit message from CVS:
2006-04-03 Andy Wingo <wingo@pobox.com>
* ext/theora/theora.c (plugin_init): Register theoraparse.
* ext/theora/gsttheoraparse.h:
* ext/theora/theoraparse.c: New files implementing a theora
parser. Now we can properly remux ogg/theora+vorbis, yay.
Original commit message from CVS:
ogg muxing of vorbis and theora now has pages ordered correctly again,
even with delays.
* ext/ogg/README:
updated with some examples
* ext/theora/theoraenc.c: (granulepos_to_timestamp),
(granulepos_add), (theora_buffer_from_packet):
* ext/vorbis/vorbisenc.c: (granulepos_to_timestamp_offset),
(granulepos_to_timestamp), (gst_vorbisenc_buffer_from_packet),
(gst_vorbisenc_chain):
implement strategy from ext/ogg/README
* ext/ogg/gstoggmux.c: (gst_ogg_mux_buffer_from_page),
(gst_ogg_mux_push_buffer), (gst_ogg_mux_dequeue_page),
(gst_ogg_mux_pad_queue_page), (gst_ogg_mux_compare_pads),
(gst_ogg_mux_queue_pads), (gst_ogg_mux_collected):
Fix muxer so that oggz-validate is happy with all streams;
except for no eos mark, and the BOS page ordering
* tests/check/pipelines/theoraenc.c: (check_buffer_is_header),
(check_buffer_granulepos):
* tests/check/pipelines/vorbisenc.c: (check_buffer_granulepos):
update tests to check for OFFSET being set as requested
fixed type of granulepos, it's not a ClockTime
Original commit message from CVS:
2006-03-03 Thomas Vander Stichele <thomas at apestaart dot org>
* ext/theora/theoraenc.c: (theora_set_header_on_caps):
* tests/check/pipelines/theoraenc.c: (check_buffer_is_header),
(GST_START_TEST):
Fix for http://bugzilla.gnome.org/show_bug.cgi?id=333254
Set IN_CAPS on header buffers
Original commit message from CVS:
* ext/theora/theoraenc.c: (gst_theora_enc_class_init),
(theora_enc_finalize), (theora_enc_sink_setcaps),
(theora_set_header_on_caps), (theora_enc_chain),
(theora_enc_change_state):
* tests/check/pipelines/theoraenc.c: (GST_START_TEST):
Make theoraenc and the tests leak free. Like, really.
Original commit message from CVS:
(theora_enc_finalize), (theora_enc_sink_setcaps):
Add a finalize method to ensure we clean up state even if
someone omitted the state change back to NULL.
* ext/vorbis/vorbisenc.c: (gst_vorbisenc_metadata_set1),
(gst_vorbisenc_chain):
Free some more leaked bits.
* tests/check/pipelines/theoraenc.c: (start_pipeline),
(stop_pipeline):
Wait for state changes to happen if they're ASYNC.
This ought to teach those fancy pants buildbots a lesson.
Original commit message from CVS:
* ext/theora/theoradec.c: (_theora_granule_time),
(theora_dec_src_convert), (theora_dec_sink_convert):
Call right _scale functions.
Use parameter instead of some other random value.
Original commit message from CVS:
* ext/theora/theoradec.c: (_theora_granule_frame),
(_theora_granule_time), (_inc_granulepos),
(theora_dec_src_convert), (theora_dec_sink_convert),
(theora_handle_type_packet), (theora_handle_data_packet),
(theora_dec_chain):
Use higher precision timestamps calculation.
Convert some other conversions to _scale.
Original commit message from CVS:
2006-01-30 Andy Wingo <wingo@pobox.com>
* ext/theora/theoraenc.c (theora_enc_sink_setcaps)
(gst_theora_enc_init): Pull the granule shift out of the encoder.
(granulepos_add): New function, handles the messiness of adjusting
granulepos values.
(theora_buffer_from_packet):
(theora_enc_chain):
(theora_enc_sink_event): Use granulepos_add, not +.
* tests/check/pipelines/theoraenc.c
(check_buffer_granulepos_from_starttime): Just check the frame
count, not the actual granulepos -- we can't dictate to the
encoder when it should be placing keyframes.
Original commit message from CVS:
2006-01-30 Andy Wingo <wingo@pobox.com>
* ext/theora/gsttheoraenc.h:
* ext/theora/theoraenc.c: Same changes as were done to vorbisenc,
although theoraenc was timestamping correctly. Added handling of
streams that start with nonzero timestamps.
* tests/check/Makefile.am:
* tests/check/pipelines/theoraenc.c: New file, basically does same
tests as vorbisenc.
* tests/check/pipelines/vorbisenc.c: I claim these bugs.
Original commit message from CVS:
* ext/theora/theoradec.c: (theora_handle_comment_packet):
Post taglist actually on bus instead of just freeing it
(fixes#327114 and totem bug #327080).
* ext/vorbis/vorbisdec.c: (vorbis_handle_comment_packet):
Use gst_element_found_tags_for_pad(), so that the tags
are sent downstream as an event as well.
Original commit message from CVS:
* ext/theora/theoradec.c: (theora_dec_src_query),
(theora_dec_sink_event):
* ext/vorbis/vorbisdec.c: (vorbis_dec_src_query),
(vorbis_dec_sink_event), (vorbis_handle_identification_packet),
(vorbis_handle_data_packet):
* ext/vorbis/vorbisdec.h:
Fix old naming.
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_render):
Don't try to sync on buffers without a timestamp.
Original commit message from CVS:
2005-10-17 Andy Wingo <wingo@pobox.com>
* ext/theora/theoraenc.c (theora_buffer_from_packet): Pass the
alloc_buffer flow return to callers.
(theora_enc_chain, theora_enc_chain): Adapt to buffer_from_packet
change. Fix some memleaks in theoraenc.
Original commit message from CVS:
* common/gtk-doc-plugins.mak:
* docs/plugins/Makefile.am:
* docs/plugins/gst-plugins-base-plugins-sections.txt:
* ext/theora/Makefile.am:
* ext/theora/gsttheoraenc.h:
* ext/theora/theoraenc.c:
* ext/vorbis/vorbisenc.c:
pick up signals and args for vorbis; add some docs for vorbis
Original commit message from CVS:
make GST_PLUGIN_LDFLAGS only be flags; GST_LIBS should be
added manually to each Makefile.am so we are sure it goes
*last* and doesn't add -L flags before linking in libs of our
own, like, say, internal .la libs, that then accidentally pick
up the installed copy.
Original commit message from CVS:
* ext/theora/theoradec.c: (theora_dec_src_getcaps),
(theora_dec_push), (theora_handle_data_packet):
Prepare for better timestamp fix later.
* gst/audioconvert/gstaudioconvert.c:
List most accurate caps first
* gst/videotestsrc/gstvideotestsrc.c: (gst_videotestsrc_loop):
Use proper pad task function.
* sys/xvimage/xvimagesink.c: (gst_xvimagesink_xvimage_new),
(gst_xvimagesink_show_frame):
Fix deadlock when alloc failed.
Original commit message from CVS:
2005-07-01 Andy Wingo <wingo@pobox.com>
* ext/theora/theoradec.c (theora_dec_src_getcaps): Implement a
getcaps to do explicit caps. Needs to be done in all decoders,
possibly via a base class.
* configure.ac (GST_PLUGIN_LDFLAGS): Add videoscale.
* ext/ogg/gstoggdemux.c (gst_ogg_pad_typefind): No need to set
caps on the sink pad, just rely on the pad template. Also, setting
ANY caps on a pad is not valid because the caps are not fixed.
* sys/ximage/ximagesink.c (gst_ximagesink_buffer_alloc): Set the
caps on the buffer, and get the width from the desired_caps if
they're set.
(gst_ximagesink_renegotiate_size): Implement via setting the
desired_caps on the ximagesink.
(gst_ximagesink_setcaps): Only reset the width of the player if it
wasn't already set. Not sure if this is right.
(gst_ximagesink_show_frame): Memcpy only for normal buffers.
* sys/ximage/ximagesink.h (desired_caps): New field, is the caps
that the user wants. NULL unless the window has been resized.
* gst/volume/gstvolume.c (volume_transform): Adapt to
basetransform refcount changes.
