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.
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
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.
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.
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.
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 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.
This avoids a pointless indirection that only obscures what is really done.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@27761 b3059339-0415-0410-9bf9-f77b7e298cf2
Replace all USE_ prefixes by CONFIG_ prefixes to indicate
options which are configurable.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@27373 b3059339-0415-0410-9bf9-f77b7e298cf2
The casts were left from the time when the type of sh_video->vfilter
was (unnecessarily) void *. Now that the field has the correct type
the casts are pointless.
This commit creates the struct and passes it to some functions that
needs to access OSD state but does not yet move much data from globals
to it.
vf_expand accesses the OSD state for rendering purposes outside of the
normal OSD draw time. The way this currently works is suboptimal, but
I did not attempt to clean it up now. To keep things working the same
way vf_expand needs to know the address of the state object to be able
to access the data even in the functions that should normally not need
it. For that purpose this commit adds a VFCTRL to tell vf_expand the
address of the object.
