At least show the extra frames even if correct-pts mode is disabled.
They cannot be timed properly though; the most practical way to fix
that is to just move to correct-pts mode instead.
OSD contents such as video position and non-libass subtitles were
updated just before feeding a video frame through filters. The updates
were done at that point mainly for the benefit of vf_expand OSD
rendering functionality, which draws the OSD contents when the frame
passes through that filter. However this gave the wrong results for
VO-drawn OSD if the filter chain or VO had any delay or timestamp
manipulation: the OSD contents should reflect the most recent contents
that were _output_ after any filtering, not the last frame that was
fed _into_ the filtering machinery. This issue became more important
after vo_vdpau started buffering frames by default.
Move the OSD updates to be done just before the OSD is drawn, using
the most accurate available information. This fixes the common case
but worsens vf_expand OSD behavior (adding extra latency). A special
case could be added to fall back to the previous behavior when
vf_expand OSD is being used; however I don't consider that a high
priority at the moment especially when other problems with vf_expand
OSD would still remain.
This has little effect on libass-rendered subtitles either way,
because both vf_ass and VO libass rendering use timestamps from the
filter chain and do not rely on separate position updates.
Add a mode where libavcodec's reordered_opaque feature is used to
associate container packet timestamps with decoded frames. This should
improve behavior at least for MPEG files with interlaced h264; the
previous code does not cope well with the libavformat demuxer
producing two field packets with separate timestamps but the
libavcodec h264 decoder only producing a single output frame for those
two packets (so half the timestamps have no associated output frame).
The current libavformat mpeg demuxer seems to finally work with
interlaced h264 files and produce valid timestamps which are useful
with a mode like this.
By default MPlayer now selects between this new mode and the old one
automatically based on the number of timestamp problems they cause; by
default the new mode is used if both seem to work. The new option
-pts-association-mode can be used to force a particular mode. If
correct-pts mode is disabled this has no effect on timing.
Also remove the "EXPERIMENTAL" marker from the manpage description of
-correct-pts.
Main things added are custom frame dropping for VDPAU to work around
the display FPS limit, frame timing adjustment to avoid jitter when
video frame times keep falling near vsyncs, and use of VDPAU's timing
feature to keep one future frame queued in advance.
NVIDIA's VDPAU implementation refuses to change the displayed frame
more than once per vsync. This set a limit on how much video could be
sped up, and caused problems for nearly all videos on low-FPS video
projectors (playing 24 FPS video on a 24 FPS projector would not work
reliably as MPlayer may need to slightly speed up the video for AV
sync). This commit adds a framedrop mechanism that drops some frames
so that no more than one is sent for display per vsync. The code
tries to select the dropped frames smartly, selecting the best one to
show for each vsync. Because of the timing features needed the drop
functionality currently does not work if the correct-pts option is
disabled.
The code also adjusts frame timing slightly to avoid jitter. If you
for example play 24 FPS video content on a 72 FPS display then
normally a frame would be shown for 3 vsyncs, but if the frame times
happen to fall near vsyncs and change between just before and just
after then there could be frames alternating between 2 and 4
vsyncs. The code changes frame timing by up to one quarter vsync
interval to avoid this.
The above functionality depends on having reliable vsync timing
information available. The display refresh rate is not directly
provided by the VDPAU API. The current code uses information from the
XF86VidMode extension if available; I'm not sure how common cases
where that is inaccurate are. The refresh rate can be specified
manually if necessary.
After the changes in this commit MPlayer now always tries to keep one
frame queued for future display using VDPAU's internal timing
mechanism (however no more than 50 ms to the future). This should make
video playback somewhat more robust against timing inaccuracies caused
by system load.
Make audio uninit consistent with e.g. the demuxer uninit code and
also avoids a possible crash.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@29908 b3059339-0415-0410-9bf9-f77b7e298cf2
being paused (e.g. because of a "pausing_keep_force pt_step 1").
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@29898 b3059339-0415-0410-9bf9-f77b7e298cf2
and if it is unchanged re-apply -slang on stream reset.
This makes -slang work when used with menu navigation or in general
when the subtitle is not available for parts of the playback or the
subtitle stream ID changes during playback.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@29897 b3059339-0415-0410-9bf9-f77b7e298cf2
This seems to leave the ts demuxer unaffected, but fixes -tsprog with
the lavf demuxer.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@29846 b3059339-0415-0410-9bf9-f77b7e298cf2
decoder is used but the hardware does not support hardware MPEG audio).
Otherwise this will lead to a crash later on when the decode code tries to access
the audio filter chain.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@29836 b3059339-0415-0410-9bf9-f77b7e298cf2
selecting the subtitle via -slang or -sid.
It is quite possible that removing it will cause other issues, though
I could not see any with the DVDs I had available for testing.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@29827 b3059339-0415-0410-9bf9-f77b7e298cf2
Move av_log callback handling from vd_ffmpeg.c to a new file
av_log.c and install the callback immediately when starting the
program. Main functionality improvements of the new code:
- The old version only installed the callback when opening an FFmpeg
video decoder. If nothing had triggered that then av_log() messages
from other sources (libavformat, audio decoding, swscale usage)
bypassed MPlayer's output system completely. Now the callback is
always installed.
- Current av_log message severity levels are handled correctly. The
old code used MSGL_ERR for some messages that should be MSGL_V.
- Message type is now set for libavformat contexts
(MSGT_DEMUXER / MSGT_MUXER).
- The old code did "mp_msg_test(type, mp_level)" before actually
determining the type, so that it always used MSGT_FIXME. This led
to some messages being incorrectly dropped in case the user
had specified module-specific verbosity levels. The old check in
question was originally motivated by performance problems when
there were a lot of callbacks; however it's not clear whether the
part about it skipping the type determination was intentional (most
of the performance problems must have come from the way the
original code used snprintf) and in my tests current FFmpeg
libraries have not generated unreasonable amounts of callbacks
anyway.
As part of merging subtitle-in-terminal changes make
update_subtitles() only clear existing subtitles if called with the
reset argument, and not try to set new ones. Later calls should set
the needed new subtitles, and this change avoids some problems with
trying to set subtitles when mp_property_sub() in command.c gets
called from initialization code before full initialization.
This still needs some additional checking that subtitle selection via dvdnav works.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@29732 b3059339-0415-0410-9bf9-f77b7e298cf2
This fixes a crash with e.g. auto-enabled subtitles and -novideo due to
command.c calling update_subtitles even without video and is a step
toward subtitle support for audio-only files.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@29710 b3059339-0415-0410-9bf9-f77b7e298cf2
* vdpau: (22 commits)
VO: Prefer vo_vdpau over vo_xv again
vo_vdpau: Fix X event handling bugs
vo_vdpau: Fix memory corruption bug with MP_IMGTYPE_NUMBERED
core/VO: Allow VO drivers to add/modify frames
video_out.h: Cosmetics
VO interface: Remove obsolete draw_frame() from new interface
vo_vdpau: Support recovering from VDPAU display preemption
vo_vdpau: Support updating OSD while paused
vo_vdpau.c: Reindent control() switch statement
vo_vdpau: Allocate one large surface for EOSD content
vo_vdpau.c: cosmetics
vo_vdpau: reindent after GUI code removal
vo_vpdau: Clean up uninit logic
vo_vdpau: Make CHECK_ST macro safer
vo_vdpau: Move all remaining static/global variables to context
vo_vdpau: Move things to context struct
vo_vdpau: Make info struct const
vo_vdpau: Replace global function table with context variable
vo_vdpau: Move VDPAU interface pointers into one struct
vo_vdpau: Add template file for VDPAU functions
...
Add interfaces to allow VO drivers to add or remove frames from the
video stream and to alter timestamps. Currently this functionality
only works with in correct-pts mode. Use the new functionality in
vo_vdpau to properly support frame-adding deinterlace modes.
Frames added by the VDPAU deinterlacing code are now properly timed.
Before every second frame was always shown immediately (probably next
monitor refresh) after the previous one, even if you were watching
things in slow motion, and framestepping didn't stop at them at all.
When seeking the deinterlace algorithm is no longer fed a mix of
frames from old and new positions.
As a side effect of the changes a problem with resize events was also
fixed. Resizing calls video_to_output_surface() to render the frame at
the new resolution, but before this function also changed the list of
history frames, so resizing could give an image different from the
original one, and also corrupt next frames due to them seeing the
wrong history. Now the function has no such side effects. There are
more resize-related problems though that will be fixed in a later
commit.
The deint_mpi[] list of reserved frames is increased from 2 to 3
entries for reasons related to the above. Having 2 entries is enough
when you initially get a new frame in draw_image() because then you'll
have those two entries plus the new one for a total of 3 (the code
relied on the oldest mpi implicitly staying reserved for the duration
of the call even after usage count was decreased). However if you want
to be able to reproduce the rendering outside draw_image(), relying on
the explicitly reserved list only, then it needs to store 3 entries.
This should really be done via the NXCOMPAT PE flag, but looks like
binutils will not get support for it any time soon and not having this
security feature enabled is just ridiculous.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@29683 b3059339-0415-0410-9bf9-f77b7e298cf2
current file name against full name to be added, not current file name
against new base path.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@29637 b3059339-0415-0410-9bf9-f77b7e298cf2
file name strings that might come from arbitrary playlists, use PATH_MAX instead.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@29636 b3059339-0415-0410-9bf9-f77b7e298cf2
dvbin.h #includes some system DVB headers that may be unavailable.
Fixes compilation with DVB disabled.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@29474 b3059339-0415-0410-9bf9-f77b7e298cf2
The GUI is badly designed and too closely coupled to the internal
details of other code. The GUI code is in bad shape and unmaintained
for years. There is no indication that anyone would maintain it in the
future either. Even if someone did volunteer to implement a better
integrated GUI having the current code in the tree probably wouldn't
help much. So get rid of it.
instead of only depending on the stream.
This makes the value correctly 0 e.g. for AVI files without index.
Patch by Jason Tackaberry [tack urandom ca]
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@29331 b3059339-0415-0410-9bf9-f77b7e298cf2
Code that was supposed to merge two timeline parts if the second one
started exactly where the first one stopped didn't work because it
mixed timestamps in different units (I had changed the units of some
variables after originally writing that code but forgotten to update
it). As a result of the bug there were unnecessary part switches, but
it was unlikely to cause any serious problems.
The uninit code called talloc_free(mpctx->chapters) but didn't clear
the pointer. As a result playing a file with ordered chapters and one
without almost certainly triggered an abort when closing the second
file. Fix by setting the pointer to NULL after freeing.
If there is no timestamp available for a video frame at all, use
timestamp from the previous frame (if any) instead of leaving it as
MP_NOPTS_VALUE. The main practical difference for playback is that
MP_NOPTS_VALUE caused timing to be reset, moving current video
position immediately to time of the next frame that had a timestamp
and thus likely advancing the video too fast. Now the next frame with
a timestamp will be shown at its proper position relative to the
previous timestamp.
Remove the "num_chapters" and "mode" parameters that aren't needed by
any callers. Change "float *seek_pts" to "double *". Allocate the
string returned via "chapter_name" with talloc.
Many VOs kept track of pause status, but reset the value when their
config() function was called. However it can be called while playback
stays in pause mode. Modify the VOs to not change anything in
config(). Also send the VO either VOCTRL_PAUSE or VOCTRL_RESUME when
the playback of a new file is starting to make sure they have the
right status.
Add basic support for Matroska ordered chapters. The switching between
segments is implemented as a general edit timeline that could also be
used for other purposes.
Some things still need improvement. In particular the current code
does not try to do any proper mapping between audio/video/subtitle
streams of different files and there should be options for better
control of how MPlayer searches other files for the required content.
Clean up the code and make the behavior more consistent. Before
bits of the OSD information were triggered in different places, and
various property commands that affect playback position only showed
the seek bar while the main seek command also triggered showing the
percentage in OSD text. Now only the seek and chapter commands trigger
all information and others nothing (which is consistent with most
property behavior).
When the new mode is active relative seeks are converted to absolute
ones (current video pts + relative seek amount) and forward/backward
flag before being sent to the demuxer. This mode is used if the
demuxer has set the accurate_seek field in the demuxer struct and
there is a video stream. At the moment the mkv and lavf demuxers
enable the flag.
This change is useful for later Matroska ordered chapter support (and
for more general timelime editing), but also fixes problems in
existing functionality. The main problem with the old mode, where
relative seeks are passed directly to the demuxer, is that the user
wants to seek relative to the currently displayed position but the
demuxer does not know what that position is. There can be an arbitrary
amount of buffering between the demuxer read position and what is
displayed on the screen. In some situations this makes small seeks
fail to move backward at all (especially visible at high playback
speed, when audio needs to be demuxed and decoded further ahead to
fill the output buffers after resampling).
Some container formats that can be used with the lavf demuxer do not
always have reliable timestamps that could be used for unambiguous
absolute seeking. However I made the demuxer always enable the new
mode because it already converted all seeks to absolute ones before
sending them to libavformat, so cases without reliable absolute seeks
were failing already and this should only improve the working cases.
Restore accidentally dropped '!' to fixed-vo test in
reinit_video_chain(). It could have caused some issues when switching
video streams. Probably nobody noticed because files with multiple
video streams are rare.
different behaviour is unlikely to make sense but it is better to handle
this in input.c instead of adding special cases to mplayer.c and being
able to override the default behaviour at least should not hurt.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@28870 b3059339-0415-0410-9bf9-f77b7e298cf2
This allows printing the CPU information when verbose mode is
triggered on the command line.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@28691 b3059339-0415-0410-9bf9-f77b7e298cf2
Print CPU information in verbose mode instead of by default.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@28360 b3059339-0415-0410-9bf9-f77b7e298cf2
The libdvdread4 and libdvdnav directories, which are externals in the
svn repository, are at least for now not included in any form. I added
configure checks to automatically disable internal libdvdread and
libdvdnav if the corresponding directories are not present; if they're
added manually then things work the same as in svn.
Move the OSD drawing calls from filter_video() to higher-level code to
ensure that VOs will draw the OSD also in filter-added frames, which
are displayed without a separate call to filter_video().
Rewrite some of the -correct-pts frame generation code. The most
noticeable difference is that video timestamps which jump around but
do not go backwards more than 0.5 seconds will not affect overall
video playback speed. Only larger jumps are treated as timestamp
resets and increase the overall length of the video timeline. This
will hopefully give watchable results for videos which have broken
timestamps or which MPlayer does not demux or otherwise handle
properly.
Rewrite the -nocorrect-pts frame creation code. The new version always
updates the visible frame when seeking while pausing, and supports
filter-added frames. It can not time those properly though.
Now the handling of filter-added frames in MPlayer always uses the new
method independently of the value of correct-pts but MEncoder still
expects the old behavior. Add a global variable that is set under
MEncoder only and change the filters to choose behavior based on that
instead of the correct_pts option.
Handle timing out of OSD messages and set the OSD function symbol to
pause instead of play. The implementation is hackish and should be
cleaned up later with other pause loop changes.
Notable functionality changes:
* Timing change between any two frames is now accurately limited to
1/10 of their nominal distance. Previous code did not always use the
correct duration.
* The status line now keeps showing the same A-V sync value from one
video frame change to the next. Previously it kept recalculating
the value using a new audio position but the same video position
when the status line was updated between video frames. This
incorrectly showed the video losing sync with audio.
* The status line now displays actual measured A-V difference in
autosync mode too. The previous code displayed values that
completely ignored real timing in autosync mode, showing 0 A-V
difference even when video was significantly behind audio due to
inadequate decoding speed. The new behavior can make the shown A-V
values appear more unstable if the audio out has unreliable delay
measurements (the most likely reason to use autosync), but this is a
display change rather than a timing quality change.
* Autosync mode now tries to adjust timing by the amount of time
vo_flip() calls take, so the calls start earlier and finish at the
time the frame should be shown. Previously non-autosync mode
adjusted for this but autosync did not.
* The warning about the system being too slow to decode the video in
realtime is now displayed in autosync mode too.
