Commit graph

1117 commits

Author SHA1 Message Date
Gwenole Beauchesne
5975def8ab mpeg2: allocate dummy picture for first field based I-frame.
In P-pictures, prediction shall be made from the two most recently
decoded reference fields. However, when the first I-frame is a field,
the next field of the current picture could be a P-picture but only a
single field was decoded so far. In this case, create a dummy picture
with POC = -1 that will be used as reference.

Some VA drivers would error out if P-pictures don't have a forward
reference picture. This is true in general but not in this very specific
initial case.
2012-04-02 11:36:56 +02:00
Gwenole Beauchesne
99932049e4 mpeg2: fix simple to main profile fallback.
Allow fallback from simple to main profile when the HW decoder does
not support the former profile and that no sequence_header_extension()
is available to point out this.
2012-04-02 11:36:56 +02:00
Wind Yuan
b33dd69300 mpeg4: improve error checking while decoding packets.
decode_picture() could return an error when an MPEG-4 profile is not
supported for example. In this case, the underlying VA context is not
allocated and no other proper action can be taken. Likewise on exit
from decode_slice().

Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2012-03-30 17:23:38 +02:00
Gwenole Beauchesne
75538bbc9b decoder: use POC to maintain the DPB.
Introduce a POC field in GstVaapiPicture so that to store simpler sequential
numbers. A signed 32-bit integer should be enough for 1 year of continuous
video streaming at 60 Hz.

Use this new POC value to maintain the DPB, instead of 64-bit timestamps.
This also aligns with H.264 that will be migrated to GstVaapiDpb infrastructure.
2012-03-30 17:08:37 +02:00
Gwenole Beauchesne
d51bb87482 mpeg2: rework generation of presentation timestamps.
Always prefer PTS from the demuxer layer for GOP times. If this is invalid,
i.e. demuxer could not determine the PTS or the generated PTS is lower than
max PTS from past pictures, then try to fix it up based on the duration of
a frame.

For picture PTS, simply use the GOP PTS formerly computed then use TSN to
reconstruct a current time. Also now handle wrapped TSN correctly.
2012-03-30 17:08:37 +02:00
Gwenole Beauchesne
6eeb20a953 mpeg2: ignore empty user-data packets.
Fix tcela-8.bits conformance test.
2012-03-30 17:08:37 +02:00
Gwenole Beauchesne
7af40475d5 mpeg2: review and report errors accordingly.
Use GST_ERROR() to report real errors instead of hiding them into
GST_DEBUG().
2012-03-29 11:13:20 +02:00
Gwenole Beauchesne
b003a2bab2 mpeg2: fix invalid interlaced frame in progressive sequence.
Some streams, badly constructed, could have signaled an interlaced
frame while the sequence was meant to be progressive. Warn and force
frame to be progressive in this case.
2012-03-28 19:41:28 +02:00
Gwenole Beauchesne
7a9410f826 mpeg2: add support for interlaced streams.
Pictures are submitted to the HW for rendering only when both fields
are decoded or current picture is a full frame.
2012-03-28 19:40:27 +02:00
Gwenole Beauchesne
ac4fc0d36c mpeg2: add new decoded picture buffer infrastructure.
Decoded pictures are now maintained into DPB, similarly to H.264.
The same mechanism could be re-used for VC-1 and MPEG-4:2 codecs.
2012-03-28 19:36:52 +02:00
Gwenole Beauchesne
b56ac16224 decoder: add first-field flag to picture.
Add first-field (FF) flag to GstVaapiPicture, thus not requiring is_first_field
member in each decoder. Rather, when a GstVaapiPicture is created, it is considered
as the first field. Any subsequent allocated field will become the second field.
2012-03-28 19:25:49 +02:00
Gwenole Beauchesne
195a61fa02 decoder: allow pictures to be cloned for field decoding.
Add gst_vaapi_picture_new_field() function that clones a picture, while
preserving the parent picture surface. i.e. the surface proxy reference
count is increased and other fields copied as is. Besides, the picture
is reset into a "non-output" mode.
2012-03-28 18:58:27 +02:00
Gwenole Beauchesne
c27385aa8d decoder: rework picture field flags.
Add top-field-first (TFF) and interlaced flags to GstVaapiPicture so they
could be propagated to the surface proxy when it is pushed for rendering.
Besides, top and bottom fields are now expressed with picture structure flags
from GstVaapiSurfaceRenderFlags.
2012-03-28 18:24:39 +02:00
Gwenole Beauchesne
e8756ae427 decoder: add OUTPUT flag to pictures.
Allow pictures to be marked as output gst_vaapi_picture_output().
2012-03-28 17:59:19 +02:00
Gwenole Beauchesne
a7ec623279 decoder: fix semantics of SKIPPED pictures.
If GstVaapiPicture has flag SKIPPED set, this means gst_vaapi_picture_output()
will not push the underlying surface for rendering. Besides, VC-1 skipped P-frame
has nothing to do with rendering. This only means that the currently decoded
picture is just a copy of its reference picture.
2012-03-28 17:59:19 +02:00
Gwenole Beauchesne
b98d334dce vaapipostproc: get "interlaced" attribute from surface proxy.
Add new "interlaced" attribute to GstVaapiSurfaceProxy. Use this in
vaapipostproc so that to handles cases where bitstream is interlaced
but almost only frame pictures are generated. In this case, we should
not be alternating between top/bottom fields.
2012-03-28 17:59:19 +02:00
Gwenole Beauchesne
894d65b81a vaapipostproc: add new element for video postprocessing.
Add vaapipostproc element for video postprocessing. So far, only basic
bob deinterlacing is implemented. Interlaced mode is automatically
detected based on sink caps ("interlaced" field).
2012-03-26 14:37:24 +02:00
Gwenole Beauchesne
a5144358d1 videobuffer: add surface render flags.
Allow rendering flags, as a combination of GstVaapiSurfaceRenderFlags,
to be set to the video buffer. In particular, this is mostly useful for
basic deinterlacing.
2012-03-26 12:01:36 +02:00
Gwenole Beauchesne
3b67533d3d compat: add compatibility glue with VA-API 0.34+ (WIP). 2012-03-23 17:13:58 +01:00
Gwenole Beauchesne
7277f0e220 h264: skip all Filler Data NALs. 2012-03-23 17:11:18 +01:00
Wind Yuan
81ab8dad37 mpeg4: check for decoder status prior to decoding packet.
Make sure there is a VA surface free prior to decoding the current frame.

Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2012-03-22 14:44:42 +01:00
Wind Yuan
d43f3dc50a decode: delay NEWSEGMENT event if vaapidecode element was not linked.
Rationale: playbin2 links all elements at run-time. Once vaapidecode
is created and a NEWSEGMENT event arrives, downstream element may not
be ready yet. So, delay this event until next element is chained in,
otherwise basesink could output "Received buffer without a new-segment.
Assuming timestamps start from 0".

Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2012-03-22 14:33:17 +01:00
Wind Yuan
baa80d9753 mpeg2: fix GOP timestamps when incorrect data is received.
Some streams have incorrect GOP timestamps, or nothing set at all.
i.e. GOP time is 00:00:00 for all GOPs. Try to recover in this case
from demuxer timestamps, which are monotonic.

Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2012-03-22 13:45:47 +01:00
Wind Yuan
fc3edde05c mpeg2: don't decode anything before the first sequence_header().
Skip all pictures prior to the first sequence_header(). Besides,
skip all picture_data() if there was no prior picture_header().

Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2012-03-21 17:38:22 +01:00
Holger Kaelberer
2ebfa66ade ffmpeg: add support for interlaced streams.
Evaluate interlaced stream properties.

Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2012-03-19 18:43:06 +01:00
Holger Kaelberer
0d0cff695f vaapidecode: propagate interlaced and TFF properties downstream.
Propagate "interlaced" caps downstream and set "tff" buffer flag
appropriately to output buffers for interlaced pictures.

Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2012-03-19 18:43:06 +01:00
Holger Kaelberer
82f11ffb30 decoder: maintain caps for interlaced streams.
Extend GstVaapiDecoder base object to maintain caps with "interlaced"
property.

Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2012-03-19 18:43:06 +01:00
Holger Kaelberer
8593fb619c surfaceproxy: add TFF property.
Add TFF (top-field-first) property to GstVaapiSurfaceProxy.

Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2012-03-19 18:43:06 +01:00
Gwenole Beauchesne
c94390be71 AUTHORS: update to match current authors. 2012-03-16 14:21:36 +01:00
Gwenole Beauchesne
d6ab5e1210 mpeg2: fix is_first_field calculation.
Reset is_first_field for frame pictures. Factor out locations where
the flag is updated.
2012-03-16 13:56:48 +01:00
Gwenole Beauchesne
10dfc67cca mpeg2: catch incorrect picture_structure from bitstreams.
Assume "frame" picture structure if the syntax element was zero or if
progressive_frame is set.
2012-03-16 13:56:26 +01:00
Gwenole Beauchesne
8bbe22f831 mpeg2: fix slice_vertical_position calculation (again).
VA-API expects slice_vertical_position as the initial position from the
bitstream. i.e. the direct slice() information. VA drivers will be fixed
accordingly.
2012-03-16 13:55:28 +01:00
Gwenole Beauchesne
6c8b052bd3 Revert "vaapidecode: fix another pad template ref leak" (Holger Kaelberer)
This reverts commit 2f127d6af4.

For gst_element_class_get_pad_template(), no unreferencing is necessary
according to the GStreamer documentation.
2012-03-02 15:03:57 +01:00
Gwenole Beauchesne
63e29adbf5 h264: fix slice_data_bit_offset calculation.
Unlike what VA-API documentation defines, the slice_data_bit_offset
represents the offset to the first macroblock in the slice data, minus
any emulation prevention bytes in the slice_header().

This fix copes with binary-only VA drivers that won't be fixed any
time soon. Besides, this aligns with the current FFmpeg behaviour
that was based on those proprietary drivers implementing the API
incorrectly.
2012-03-02 13:41:16 +01:00
Wind Yuan
c979d51da6 h264: skip all Access Unit (AU) NALs.
Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2012-03-01 10:35:58 +01:00
Wind Yuan
6b56c84c9b h264: fix modification process of reference picture lists.
Construction of RefPicList0/1 could be off by one element.

Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2012-03-01 10:35:58 +01:00
Sreerenj Balachandran
033a675bcc mpeg2: fix size calculation from sequence_extension().
Original values from sequence_header() are 12-bit and the remaining
2 most significant bits are coming from sequence_extension().

Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2012-02-24 11:09:05 +01:00
Gwenole Beauchesne
f1883b4a4e mpeg2: fix slice_vertical_position calculation.
Make sure to adjust slice_vertical_position if picture structure
is a top or bottom field.
2012-02-23 16:41:11 +01:00
Gwenole Beauchesne
124cd8a3d3 decoder: add picture structure flags. 2012-02-23 16:39:51 +01:00
Gwenole Beauchesne
3bc68b5b32 mpeg2: fix decoding at end-of-sequence. 2012-02-23 16:15:23 +01:00
Gwenole Beauchesne
60e515e0b7 mpeg2: fix slice_horizontal_position calculation. 2012-02-23 16:15:22 +01:00
Gwenole Beauchesne
b36d2c810e mpeg2: drop useless mb_y and mb_height members. 2012-02-23 16:14:02 +01:00
Gwenole Beauchesne
d47031b8ce mpeg2: fix decoding of multiple slices with same slice_vertical_position.
6.3.15 says that "some slices may have the same slice_vertical_position,
since slices may start and finish anywhere". So, we can't submit the current
picture to the HW right away since subsequent slices would be missing.
2012-02-23 11:19:48 +01:00
Gwenole Beauchesne
de7432f712 image: fix source stride in picture copy. 2012-02-15 14:08:44 +01:00
Gwenole Beauchesne
ab94048ea0 decoder: fix double buffer free with some VA drivers.
vaRenderPicture() implicitly disposes VA buffers. Some VA drivers would
push the VA buffer object into a list of free buffers to be re-used. However,
reference pictures (and data) that was kept would explicitly release the VA
buffer object later on, thus possibly destroying a valid (re-used) object.

Besides, some other VA drivers don't support correctly the vaRenderPicture()
semantics for VA buffers disposal and would leak memory if there is no explicit
vaDestroyBuffer(). The temporary workaround is to explcitily destroy VA buffers
right after vaRenderPicture(). All VA drivers need to be aligned.
2012-02-13 10:10:35 +01:00
Gwenole Beauchesne
82dbd6f5be h264: complete any current picture decoder before SPS / PPS change.
This ensures the VA context is clear when the encoded resolution
changes. i.e. make sure older picture is decoded with the older
VA context before it changes.
2012-02-08 18:08:49 +01:00
Gwenole Beauchesne
6c5054e840 h264: create VA context earlier when SPS is parsed. 2012-02-08 18:07:14 +01:00
Gwenole Beauchesne
26c105a6ca h264: don't allocate too big data structures on stack. 2012-02-08 17:57:29 +01:00
Gwenole Beauchesne
830efb3fbd glib: map deprecated API to glib >= 2.32 equivalents.
GStaticMutex and GStaticRecMutex are now replaced with GMutex and
GRecMutex, which no longer require any prior call to g_thread_init().
2012-02-07 11:07:15 +01:00
Gwenole Beauchesne
a698766a8b glib: fix includes. 2012-02-07 10:06:15 +01:00