| Commit message (Collapse) | Author | Age |
|---|
| ... | |
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
draw_image_timed is renamed to draw_frame. struct frame_timing is
renamed to vo_frame. flip_page_timed is merged into draw_frame (the
additional parameters are part of struct vo_frame). draw_frame also
deprecates VOCTRL_REDRAW_FRAME, and replaces it with a method that
works for both VOs which can cache the current frame, and VOs which
need to redraw it anyway.
This is preparation to making the interpolation and (work in progress)
display sync code saner.
Lots of other refactoring, and also some simplifications.
|
| |
|
|
|
|
|
|
|
|
| |
Now the VO can request a number of future frames with the last parameter
of vo_set_queue_params(). This will be helpful to fix the interpolation
code.
Note that the first frame (after playback start or seeking) will usually
not have any future frames (to make seeking fast). Near the end of the
file, the number of future frames will become lower as well.
|
| |
|
|
|
| |
I don't think most of these suggestions are overly helpful. Just get rid
of them.
|
| | |
|
| |
|
|
| |
Broken by e00e9d65.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
When showing cover art, the decoding logic pretends that the source has
an infinite number of frames. This slightly simplifies dealing with
filter data flow. It was done by feeding the same packet repeatedly to
the decoder (each decode run produces new output).
Change this by decoding once at the video initialization. This is easier
to follow, and increases robustness in case of broken images. Usually,
we try to tolerate decoding errors, so decoding normally continues, but
in this case it would just burn the CPU for no reason.
Fixes #2056.
|
| |
|
|
|
| |
All this information is already output otherwise. Except the FourCC,
which lost most of its importance in mpv.
|
| | |
|
| |
|
|
|
|
| |
There's no need for this, it just creates more corner cases.
Also always reset it on seeks etc..
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Reduce the default tolerance for timestamp jumps from 60 to 15 seconds.
For .ts files, where ts_resets_possible coming from AVFMT_TS_DISCONT is
set, apply a more sophisticated heuristic. It's clear that such a file
wouldn't have a framerate below, say, 23hz. If the demuxer reports a
lower fps, we allow longer PTS jumps.
This should replace long pauses on discontinuities with .ts files with
at most a short stutter.
Of course, all kinds of things could go wrong anyway if the source is
VFR, or FFmpeg's frame rate detection fails in some other way. I haven't
found such a file yet, though.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fixes PNG cover art not showing up immediately (for example when running
with --pause).
libavformat exports embedded cover art as a single packet. For example,
a PNG attachment simply contains the PNG image, which can be sent to the
decoder. Normally you would expect that the PNG decoder would return 1
frame for 1 packet, without any delays. But this stopped working, and it
incurs a 1 frame delay.
This is perfectly legal (even if unexpected), so let our code feed the
decoder packets until we get something back. (In theory feeding the
packet instead of a real flush packet is still somewhat questionable.)
|
| |
|
|
|
|
|
| |
last_av_difference can be MP_NOPTS_VALUE under certain circumstances
(like no video timestamp yet). This triggered the desync message,
because fabs(MP_NOPTS_VALUE) is quite a large value. We don't want to
show a message in this situation.
|
| |
|
|
|
| |
It was called only in 2 places, one of them redundant (the container FPS
can not change).
|
| |
|
|
|
| |
These are basically MPlayer leftovers, and barely useful due to being
redundant with other messages. The FPS message is used somewhere else.
|
| |
|
|
|
|
|
|
|
|
|
| |
libavcodec makes it impossible to distinguish dropped frames (requested
with AVCodecContext.skip_frame), and cases when the decoder simply does
not return a frame by default (such as with VP9, which has invisible
reference frames).
This confuses users when decoding VP9 video. It's basically a cosmetic
issue, so just paint it over by ignoring them if framedropping is
disabled.
|
| |
|
|
|
|
| |
When playing cover art, it conceptually reaches EOF as soon as the image
was put on the VO, causing the EOF message to be repeated every time new
audio was decoded. Just silence the message.
|
| |
|
|
| |
Signed-off-by: wm4 <wm4@nowhere>
|
| |
|
|
|
|
|
|
|
| |
Use OPT_CHOICE_C() instead of the custom parser. The functionality is
pretty much equivalent.
(On a side note, it seems --video-stereo-mode can't be removed, because
it controls whether to "reduce" stereo video to mono, which is also the
default. In fact I'm not sure how this should be handled at all.)
|
| |
|
|
|
| |
Accidentally broken in 79779616; we really need to check for true EOF,
not just whether there are no frames yet.
|
| | |
|
| |
|
|
|
| |
Instead of touching the 2-entry queue in mpctx->next_frame directly,
move some of it to functions.
|
| |
|
|
|
|
|
| |
"Non-monotonic" isn't even 100% correct; it's missing "strictly" (for
briefness I guess), and also the message is printed if the PTS jumps
forward. So just print something that is likely a bit easier to
understand.
|
| |
|
|
| |
It obviously needs to be updated after the VO was destroyed.
|
| |
|
|
|
|
|
|
| |
For some reason there were two points in the code where it warned
against non-monotonic video PTS. The one in video.c triggered on PTS
going backwards or making large jumps forwards, while dec_video.c
triggered on PTS going backwards or PTS not changing. Merge them into a
single check, which warns against all cases.
|
| |
|
|
| |
Meh.
|
| |
|
|
|
|
| |
Broken drivers are an issue rather often. Maybe this gives the user an
idea that this could be the reason. (We can't dump much more info on a
80x24 terminal.)
|
| |
|
|
|
| |
This value is not necessarily trustworthy (it might change) and can be
0.
|
| |
|
|
|
|
|
| |
In ancient times, this was needed because it was not default, and many
VOs had problems with it. But it was always default in mpv, and all VOs
are required to deal with it. Also, running --fixed-vo=no is not useful
and just creates weird corner cases. Get rid of it.
|
| |
|
|
|
| |
This allows us to plot the difference between video timestamps, and the
adjusted video timestamps due to syncing video to audio speed.
|
| |
|
|
| |
Just minor things.
|
| |
|
|
|
|
|
|
| |
This reverts commit 7b3feecbc23e3e0b0d9cf66f02af53d127a0b681.
It's broken, hr-seek never ends at a video position before seek pts.
Not sure what I was thinking, although it did work anyway when
artificially forcing a video frame to display before seek pts.
|
| |
|
|
|
|
|
|
| |
At least there is _some_ problem if this happens. It would mean that
audio is playing slower than video. Normally, video is synced to audio,
so if audio stops playback completely, video will not advance at all.
But using things like --autosync, it's well possible that this kind of
desync happens.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Move the update_avsync_before_frame() call further down. Moving it
closer to where the time_frame value is used (and which the function
updates) should make the code more readable. With this change, there's
no need anymore to reset the time_frame value on the video reconfig
path.
Move the update_avsync_after_frame() up. Now no meaningful amount of
time passes since the previous get_relative_time() call anymore, and the
second one can be removed.
|
| |
|
|
|
|
| |
We use double for these things everywhere, just this code didn't. It
likely doesn't matter much, and this code is for an optional feature
too.
|
| |
|
|
|
|
| |
mpctx->audio_delay always has the same value as opts->audio_delay. (This
was not the case a long time ago, when the audio-delay property didn't
actually write to opts->audio_delay. I think.)
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This allows seeking audio between two video frames that are relatively
far away.
The implementation of this is a bit subtle. It pretend the audio
position is different, and the actual PTS adjustment happens in audio.c
with this line:
sync_pts -= mpctx->audio_delay - mpctx->delay;
Effectively this is the same as setting sync_pts to hrseek_pts after
this line, though. (I'm actually not sure if this could be written in a
more straightforward way; probably yes.)
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
| |
We still need to send the VO a duration in these cases. Disabling
framedrop has logically absolutely nothing to do with these cases; it
was overlooked in commit 918b06c4.
So we always send the frame duration (or a guess for it), and check
whether framedropping is actually enabled in the VO code. (It would
be cleaner to send framedrop as a flag, but I don't care about that
right now.)
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
The last video frame is another case that has a separate code path,
although it's pretty similar to the one in commit 73e5aa87. Fix this
in a different way, which also takes care of the last frame case,
although without context the code becomes slightly more tricky.
As further cleanup, move the decision about framedropping itself to
the same place, so the check in vo.c becomes much simpler. The check
for the vo->driver->encode flag, which is remvoed completely, was
redundant too.
Fixes #1480.
|
| |
|
|
|
|
|
|
|
|
|
| |
If the video format changes (e.g. different frame size), a special code
path is entered to wait until the currently displayed frame is done.
Otherwise, the frame before the change would be destroyed by the
vo_reconfig() call.
This code path didn't respect --untimed; correct this.
Fixes #1475.
|
| |
|
|
|
|
| |
Using edl or --merge-files with .avi files didn't work, because the DTS
was not offset. Only the PTS was adjusted, which led to nonsense
timestamps.
|
| |
|
|
|
|
|
| |
There are currently 568 pixel formats (actually fewer, but the namespace
is this big), and for each format elaborate synchronization was done to
call it synchronously on the VO. This is completely unnecessary, and we
can do with just a single call.
|
| |
|
|
|
|
|
|
| |
This is basically a hack; but apparently a needed one, since many
vapoursynth filters insist on having a FPS set.
We need to apply the FPS override before creating the filters. Also
change some terminal output related to the FPS value.
|
| |
|
|
|
|
|
|
|
|
| |
Most of this is explained in the code comments. This change should
improve performance with vapoursynth, especially if concurrent requests
are used.
This should change nothing if vf_vapoursynth is not in the filter chain,
since non-threaded filters obviously can not asynchronously finish
filtering of frames.
|
| |
|
|
|
| |
This is simpler than setting the context after VO creation, which
requires the code to check for the context on every entrypoint.
|
| |
|
|
|
| |
Not particularly elegant, but better than adding more and more stuff to
the relevant function signatures.
|
| |
|
|
| |
This typo has been around for over a year. Oops.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This adds API to libmpv that lets host applications use the mpv opengl
renderer. This is a more flexible (and possibly more portable) option to
foreign window embedding (via --wid).
This assumes that methods like context sharing and multithreaded OpenGL
rendering are infeasible, and that a way is needed to integrate it with
an application that uses a single thread to render everything.
Add an example that does this with QtQuick/qml. The example is
relatively lazy, but still shows how relatively simple the integration
is. The FBO indirection could probably be avoided, but would require
more work (and would probably lead to worse QtQuick integration, because
it would have to ignore transformations like rotation).
Because this makes mpv directly use the host application's OpenGL
context, there is no platform specific code involved in mpv, except
for hw decoding interop.
main.qml is derived from some Qt example.
The following things are still missing:
- a way to do better video timing
- expose GL renderer options, allow changing them at runtime
- support for color equalizer controls
- support for screenshots
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
It feels strange that seeking past EOF with --keep-open actually leaves
the player at a random position. You can't even unpause, because the
demuxer is in the EOF state, and what you see on screen is just what was
around before the seek.
Improve this by attempting to seek to the last video frame if EOF
happens. We explicitly don't do this if EOF was reached normally to
increase robustness (if the VO got a frame since the last seek, it
obviously means we had normal playback before EOF).
If an error happens when trying to find the last frame (such as not
actually finding a last frame because e.g. the demuxer misbehaves), this
will probably turn your CPU into a heater. There is no logic to prevent
reinitiating the last-frame search if the last-frame search reached EOF.
(Pausing usually prevents that EOF is reached again after a successful
last-frame search.)
Fixes #819.
|
wm4
Marcin Kurczewski
Julian