- 22 Dec, 2015 14 commits
-
-
Rémi Denis-Courmont authored
-
Rémi Denis-Courmont authored
-
Francois Cartegnie authored
Does reinject xps NAL on I Frame. Still not optimal and duplicate prone, but that's what current AVC muxer does.
-
Francois Cartegnie authored
Content for layer >= 63 is unspecified
-
Francois Cartegnie authored
-
Thomas Guillem authored
This fixes a regression from d3db4b7f
-
Steve Lhomme authored
Signed-off-by:
Thomas Guillem <thomas@gllm.fr>
-
Steve Lhomme authored
Signed-off-by:
Thomas Guillem <thomas@gllm.fr>
-
Steve Lhomme authored
when seeking during stream output we need to be able to flush internal buffers Signed-off-by:
Thomas Guillem <thomas@gllm.fr>
-
Steve Lhomme authored
similar to vout_IsEmpty() for stream output Signed-off-by:
Thomas Guillem <thomas@gllm.fr>
-
Francois Cartegnie authored
-
Francois Cartegnie authored
-
Francois Cartegnie authored
Converts any AnnexB prefix to any xVC prefix. Adds zero copy optimizations for single NAL.
-
Thomas Guillem authored
-
- 21 Dec, 2015 24 commits
-
-
Sean McGovern authored
It was mistakenly added in 315f8e47. Signed-off-by:
Rémi Denis-Courmont <remi@remlab.net>
-
Rémi Denis-Courmont authored
-
Rémi Denis-Courmont authored
-
Rémi Denis-Courmont authored
-
Rémi Denis-Courmont authored
-
Rémi Denis-Courmont authored
If there are no interrupt contexts in the whole process, then the thread variable is uninitialized. This adds a check. Using thread_local would be much simpler and faster (but C11 threads).
-
Rémi Denis-Courmont authored
-
Thomas Guillem authored
-
Thomas Guillem authored
When draining (pp_block == NULL), queue a dummy input buffer with a EOS flag, and wait for the output thread to process all output frames. This patch also add extra check when processing output buffers. Indeed, in some case, the last output buffer with a EOS flag can be invalid.
-
Thomas Guillem authored
Drain when pp_block is NULL.
-
Thomas Guillem authored
The packetizer set the *pp_block to NULL, therefore p_block was NULL too. This fixes a regression from 7ada6eae Closes #16128
-
Thomas Guillem authored
Audio, video, and sub decoders were not drained. Indeed, pf_decode_* functions were called with *pp_block == NULL instead of pp_block == NULL. There was no way to distinguish a drain from an other call from the decoder loop. There was also an issue with the packetizer. Indeed, a successful call to pf_packetize leaded to a drain.
-
Thomas Guillem authored
-
Thomas Guillem authored
This fixes a deadlock when seeking paused. It happened when avcodec needed more pictures than VLC. It was easily reproducible with h264 in a TS file.
-
Thomas Guillem authored
In case of ENOMEN, the current avcodec worker thread will be stopped. As advised by lu_zero.
-
Thomas Guillem authored
-
Thomas Guillem authored
Add a boolean to reset the cancel state to false
-
Thomas Guillem authored
Add a boolean to reset the cancel state to false.
-
Hugo Beauzée-Luyssen authored
The file has been renamed by 4441e96 in latest libgpg-error
-
Thomas Guillem authored
Can be used if the surface given to mediacodec has a punch hole. Configuring a punch hole depends of the hardware, therefore, this option is only for developers that use libvlc with a known hardware.
-
Francois Cartegnie authored
-
Steve Lhomme authored
Signed-off-by:
Jean-Baptiste Kempf <jb@videolan.org>
-
Petri Hintukainen authored
First sectors of DVD are filled with zeros. Checking boot sector avoids seeking when probing "regular" media files. Signed-off-by:
Jean-Baptiste Kempf <jb@videolan.org>
-
Felix Paul Kühne authored
-
- 20 Dec, 2015 2 commits
-
-
Rémi Denis-Courmont authored
-
Rémi Denis-Courmont authored
-