Upstream finally caved in to peer pressure and added this filter. Of
course, this also removes the fallback for people on older versions of
libplacebo, but people using mpv git master are probably using
libplacebo git master anyway. It's time to debloat this code.
And make it the default. In libplacebo, this uses internal heuristics to
pick a good size based on the actual ICC characteristics. This is
significantly less wasteful than always generating a 64x64x64 3DLUT (the
old status quo).
In vo_gpu, for simplicity, just default to 65x65x65. Note that this
provides slightly better accuracy than the old default of 64x64x64 for
technical reasons, and matches what libplacebo defaults to for typical
display profiles.
Upstream has moved from passing struct pl_icc_profile to directly
attaching a managed pl_icc_object, plus providing a new function
pl_icc_update to update the ICC profile object parameters (if needed).
To facilitate this move, pull our ICC params back out of pl_options and
update the target ICC object directly.
The original OML sync implementation (which is where this calculation
originally comes from) made now_monotonic and ust_mp_time unsigned. This
is fine except it has the assumption that now_monotonic is always
greater than ust. This actually isn't always the case. In wayland, I
observed instances where the reported timestamp is in the future. Of
course, it's a valid question to wonder if this even makes sense but
these UST values are essentially opaque black boxes to us anyways. It's
entirely plausible that the gpu is expecting the actual presentation of
the last swap to be a bit in the future, the compositor gets this and
reports this to us. So we'll consider such stats as valid. Note that
xorg doesn't have this problem because it's roughly one swap buffer call
behind because of how the event loop works (honestly something that
should be fixed).
Of course, the problem with the unsigned type here is that it overflows
on the subtraction so make it signed and allow the appropriate negative
value to happen if it does. Note that this will simply result in a small
addition to mp_time_us() which is exactly what we want here (timestamp
slightly in the future). Some small style changes included just because.
While this resolves limitations of lavc decoder crop, it also introduces
artifacts with some of the source files or hwdec.
Depending on chroma sampler it is possible to sample outside the decoder
crop area, pulling dirty pixels into the image. Some decoders left them
zeroed, not black. To fix that we would need specifc solution during
mapping of avframes.
As most of the files require the crop only in bottom/right area, the
AVCodecContext::apply_cropping works ok for those.
For all other cases that require more fancy cropping like 1440x1080+240+0
user can manually set `--vd-apply-cropping=no`.
Limitations of the lavc crop are explained here:
https://ffmpeg.org/doxygen/trunk/structAVCodecContext.html#a4745c7455c317272c4e139d6f369936c
Fixes: 826ce82cad
So strangely enough, estimated_vsync_interval is stored as a double but
nominal_vsync_interval is not and neither is the vsync_interval. Those
are stored as int64_t. This loss of precision can matter even in common
cases. For instance, take a typical 60 Hz monitor. Instead of 16666.6666
(repeating) being calculated as the vsync interval, you would get 16666
since the decimals are truncated. This is not really good at all and
affects the calculated speed values you get when using display sync. For
consistency and better precision, these should all be doubles just like
estimated_vsync_interval. Technically this means that we won't be able
to store as high of an integer value but such values would be absurdly
huge and never actually needed. Also estimated_vsync_interval already
can't handle such a case anyway.
Interrim solution, forwards compatible with new and backwards compatible
with old API. Eventually, we will want to discontinue the use of
deprecated pl_icc_params.save/load and pl_renderer_save/load, but that
requires minimum version bump.
Cropping by decoder has limitations with regards to aligment and hwdec.
It doesn't work to offset top left corner with hwdec and even with
software decoding it don't crop fully when resulting data would not be
aligned.
VO cropping is way more robust.
We currently only allow specifying the Vulkan device to use by name. We
did this to avoid confusion around devices being enumerated in an
unpredictable order. However, there is a valid edge case where a system
may contain multiple devices of the same type - which means they will
have the same name, and so you can't control which one is used.
This change implements picking devices by UUID so that if names don't
work, you have some option available. As Vulkan 1.1 is a hard
requirement for libplacebo, we can just use UUIDs without conditional
checks.
Fixes#10898
Causes bad performance with interpolation because the changing hue angle
invalidates the mixing cache, as a result of libplacebo implementations
(specifically, the fact that this graph is drawn during the color
management process, instead of as a separate overlay).
Fix it by just hard-coding a particular, relatively interesting plane
(pi/4 approximately maps onto the red-blue axis).
There are way too many preexisting scripts that rely on this behavior
for video panning, like for example scripts that allow you to use mpv as
an image viewer. If this behavior is desired then it may be better to
add a new option for panning relative to the destination instead.
The restriction of video-pan-x/y being clamped to {-3, 3} also results
in the video being impossible to pan if it was zoomed in beyond a
certain degree.
This reverts commit 7d6f9e3739.
`--vo=gpu-next` no longer uses this option, being replaced entirely by a
luminance-based approach internally. And even for `--vo=gpu`, the values
other than 'hybrid' are universally inferior in quality.
In the interest of gradually reducing the amount of option bloat here,
remove this mostly-pointless option.
Pick a smarter rect, limit it to right half of screen (to allow more
easily using it simultaneously with stats), and also auto-rotate through
the hue plane by default (wrapping once every 10 seconds of playback).
I briefly considered making it based on wallclock time instead of pts,
but this has the rather unfortunate downside of only being updated
sporadically as the user moves the mouse over OSC elements. (Of course,
we could also forcibly redraw the screen continously to avoid it, but
I'd rather not make such an invasive change for no real reason)
This design also allows you to pause and focus on (via framestepping)
individual parts of the hue graph that you're interested in.
This actually fixes a bug that was present in this code even before the
new pl_options system, which is that `video_screenshot` would
permanently leave `peak_detect_params.allow_delayed` as false, until
something else forced options re-application due to options cache
change.
The last piece in the puzzle for doing hardware conversions
automatically is ensuring we only consider valid target formats for the
conversion. Although it is unintuitive, some vaapi drivers can expose a
different set of formats for uploads vs for conversions, and that is
the case on the Intel hardware I have here.
Before this change, we would use the upload target list, and our
selection algorithm would pick a format that doesn't work for
conversions, causing everything to fail. Whoops.
Successfully obtaining the conversion target format list is a bit of a
convoluted process, with only parts of it encapsulated by ffmpeg.
Specifically, ffmpeg understands the concept of hardware configurations
that can affect the constraints of a device, but does not define what
configurations are - that is left up to the specific hwdevice type.
In the case of vaapi, we need to create a config for the video
processing endpoint, and use that when querying for constraints.
I decided to encapsulate creation of the config as part of the hwdec
init process, so that the constraint query can be down in the
hwtransfer code in an opaque way. I don't know if any other hardware
will need this capability, but if so, we'll be able to account for it.
Then, when we look at probing, instead of checking for what formats
are supported for transfers, we use the results of the constraint query
with the conversion config. And as that config doesn't depend on the
source format, we only need to do it once.
Historically, we have not attempted to support hw->hw format conversion
in the autoconvert logic. If a user needed to do these kinds of
conversions, they needed to manually insert the hw format's conversion
filter manually (eg: scale_vaapi).
This was usually fine because the general rule is that any format
supported by the hardware can be used as well as any other. ie: You
would only need to do conversion if you have a specific goal in mind.
However, we now have two situations where we can find ourselves with a
hardware format being produced by a decoder that cannot be accepted by
a VO via hwdec-interop:
* dmabuf-wayland can only accept formats that the Wayland compositor
accepts. In the case of GNOME, it can only accept a handful of RGB
formats.
* When decoding via VAAPI on Intel hardware, some of the more unusual
video encodings (4:2:2, 10bit 4:4:4) lead to packed frame formats
which gpu-next cannot handle, causing rendering to fail.
In both these cases (at least when using VAAPI with dmabuf-wayland), if
we could detect the failure case and insert a `scale_vaapi` filter, we
could get successful output automatically. For `hwdec=drm`, there is
currently no conversion filter, so dmabuf-wayland is still out of luck
there.
The basic approach to implementing this is to detect the case where we
are evaluating a hardware format where the VO can accept the hardware
format itself, but may not accept the underlying sw format. In the
current code, we bypass autoconvert as soon as we see the hardware
format is compatible.
My first observation was that we actually have logic in autoconvert to
detect when the input sw format is not in the list of allowed sw
formats passed into the autoconverter. Unfortunately, we never populate
this list, and the way you would expect to do that is vo-query-format
returning sw format information, which it does not. We could define an
extended vo-query-format-2, but we'd still need to implement the
probing logic to fill it in.
On the other hand, we already have the probing logic in the hwupload
filter - and most recently I used that logic to implement conversion
on upload. So if we could leverage that, we could both detect when
hw->hw conversion is required, and pick the best target format.
This exercise is then primarily one of detecting when we are in this
case and letting that code run in a useful way. The hwupload filter is
a bit awkward to work with today, and so I refactored a bunch of the
set up code to actually make it more encapsulated. Now, instead of the
caller instantiating it and then triggering the probe, we probe on
creation and instantiate the correct underlying filter (hwupload vs
conversion) automatically.
So far all the keypad keys except for `0` and `,` mapped to the same
MP_KEY_* independent of numlock state, even though different key codes
are received.
Now all the alternative functions map to appropriate MP_KEY_* defines,
with missing ones added.
It makes it more usable to virtually move dst rect instead of scaled
src.
The issue with the latter is that it is affected by video-zoom
paramters. For example if we do `--video-pan-x=-0.5` in normal case it
will move video -50%, but if we apply video-zoom/scale the video will
float towards the middle in unintuitive way. Extreme case is when one do
--video-zoom-x=0.01, now it is impossible to move video unless you
specify huge video-pan-x value, but it is limited to [-3, 3]. So you
cannot do anything.
With this changes regardless of video scale/zoom, video-pan will keep
center of scaled video in one place.
It doesn't make sense to have zero here and it breaks, logic below.
Width was still campled to 1, but with broken offset.
This fixes things like video-scale-x=0
- uninit gl before the VO
- destroy EGL surface before context: reverse of the creation order.
I also checked other code which doesn't even call this and leaves
it up to eglTerminate, which frees everything anyway. But this
seems more correct since we're destroying the gbm surface afterwards.
- check against at EGL_NO_DISPLAY instead of 0: there is probably no
EGL implementation where this makes a difference, it is more correct
regardless
- remove pointless eglDestroyContext call
- check render_fd before close
mpv was taking focus when being started with "--window-minimized=yes". This change stops mpv from taking focus when this option is used.
Resolves: #9641
It's possible for systems to have multiple cards, and the first capable
card to not have a connected output. Skip such cards and continue
iterating until we find one with a connected output.
Without doing this, --fs --fs-screen=1 wouldn't actually end up on the
desired screen since the sizehint was never sent. Also check the
screen-name variants for an empty string as well so the option can
properly "undo" itself (not sure if this has any practical effect).
mpv mixes xinerama and randr usage together which gets kind of
confusing and is also pretty stupid. Xinerama is completely unneccesary
today since randr can do everything it can do and much more. Remove it.
This reworks a lot of the window/geometry handling stuff to be centered
completely around xrandr_display plus some other tweaks to the geometry
handling. An important concept is that current_icc_screen is changed
into current_screen and used more generously since it is useful for
things besides just icc profiles.
There really is no reason to not just hard require randr 1.4. xorg added
1.4 support in 2012 and someone somehow using an xorg server older than
that today surely has several other problems.
If we failed to create a gbm surface, we would call drm_egl_uninit
to free up any state we had allocated. However, we would segfault if we
tried to cleanup properties there were never initialized. Null checks
have been added to guard against this.
Before d208284, this was implicitly reset back to 0 at the start of
every update_options(). But we no longer explicitly reset par->params,
so we need to do it manually here for the hooks.
Fixes: https://github.com/mpv-player/mpv/issues/12203
With a backwards compatibility shim for older versions of libplacebo in
which we simply define the relevant subset of this struct ourselves and
initialize it to known-good values, to be able to continue using our
options assigning code.
It's worth pointing out that our use of scalers deviates from how
pl_options was intended to be used, as a consequence of backwards
compatibility with pre-308 versions of libplacebo. But this should work
fine in practice, since we don't care about serializing these custom
scalers correctly. Users can still override them using the built-in
pl_options scalers when loading custom scalers via --libplacebo-options.
(To be added in the next commit)
This was already correctly freed when acquiring a new profile, but never
freed on uninit. Fix by reparenting the profile onto `p`, which is what
vo_gpu also does.
This completely disables all smoothing. Despite what the manual claims,
a decay rate of 1.0 does *not*.
It's worth pointing out that this depends on the following commit to
work properly in --vo=gpu-next, but I don't think working around such a
minor detail is worth the trouble, considering people building nightly
mpv are probably also building nightly libplacebo it should just work
(tm).
See-Also: 1c464baaf4
See-Also: 83af2d4ebd
libplacebo doesn't like it when the queue_params PTS is less than the
actual PTS of the frame for the first frame and skips mixing it during
interpolation. This can happen if you seek while paused because mpv will
always keep the vsync_offset value as if it was still playing. So in
some cases, this can be a negative value and thus the PTS will end up
decreasing and libplacebo interprets this frame as a first frame. This
skips mixing the frame and thus you get a black screen. To fix this
this, just realize that vsync timings are completely meaninglessly in
while paused. If you are not actively pushing frames, there's no reason
to care about vsync_offset. So just clear it and make it zero when the
VO's internal state is paused and we're trying to render a frame. Makes
libplacebo happy and fixes#11958.
It was actually always wrong, and no one ever noticed. This currently
returns the size of the entire display area and not one actual monitor.
Fix this by calling get_monitor_info and then doing the appropriate
subtractions. Checked by @CrendKing and fixes#12172.
There was no known problem with this, but according to the wayland
spec*, "After this event client must assume that no keys are
pressed...", so go ahead and do that.
*: 72da004b3e/protocol/wayland.xml (L2449)
Introduced by 1f8013ff3f. We try to save
the mpkey so it can be used in the modifier event that comes next if
appropriate and also clear it when needed. The problem is that the
condition for clearing is too strict and things like mismatched cases
and so on can make mpkey on the corresponding key release event not
match the saved mpkey even though in reality they were the same key.
Loosen the check by simply always clearing the saved mpkey as long as
there was some key found and the state is up. We don't handle multiple
keys at the same time anyways (they're interpreted in a sequence), so it
should be hopefully OK.
During initialization, the mpv window was not available and wayland
simply just reported nothing. But this can be a nuisance and there are
cases where having a values is better than nothing (vapoursynth). So if
current->output isn't available yet, fallback to find_output instead.
This is influenced by what is set by options like --screen and
--screen-name, but we'll consider that a feature not a bug.
This deliberately wasn't being done when mpv was embedded
(fbccddb48b). There are some applications
that would benefit from mpv setting a title since they don't do so
themselves (such as tabbed), but at the same time some others would
probably rather not have this behavior (like smplayer). Add an option
that allows an embedded mpv to set the title if the user wishes.
Fixes#11528.
mpv has a generic method for getting the display resolution, so we can
save it in vf_vapoursynth without too much trouble. Unfortunately, the
resolution won't actually be available in many cases (like my own)
because the windowing backend doesn't actually know it yet. It looks
like at least windows always returns the default monitor (maybe we
should do something similar for x11 and wayland), so there's at least
some value. Of course, this still has a bunch of pitfalls like not being
able to cope with multi monitor setups at all but so does display_fps.
As an aside, the vapoursynth API this uses apparently requires R26 (an
ancient version anyway), so bump the build to compensate for this.
Fixes#11510
XDestroyWindow() is called immediately after, which also unmaps window
if needed. according to the manpage:
> If the window specified by the w argument is mapped, it is unmapped
> automatically.
983e8f0100 resulted in the correct
dimensions, but it was not actually right because vo_gpu_next still had
the src and dst rects the same. This just needs to work like how vo_gpu
does where the src is the image params and the dst is desired output. So
basically, just copy that code over here. Fixes#12108 and as a bonus,
overriding the aspect ratio now results in correct screenshots
(previously didn't work at now and then with the above commit it had
correct dimensions but still incorrect output).
Using the width and height params directly doesn't actually work if PAR
is something other than 1. Instead, use mp_image_params_get_dsize and
calculate the correct dimensions which matches the vo_gpu behavior.
ffmpeg does not tag yuv levels for GRAY formats, but apparently they
should be infered as full range. Instead of defaulting to limited range
always. Fixes (M)JPEG playback.
This mimic ffmpeg's behaviour.
See: d295b6b693/libswscale/utils.c (L926-L962)Fixes: #12089
1 is not enough to prevent PL_QUEUE_MORE, because the pl_queue is
designed to always know the next frame (in addition to the current).
Before haasn/libplacebo@112bb886, this was was (wrongly) silently
omitted by the pl_queue code, but that fix exposed this.
While it's technically API misuse on mpv side, due to the mpv vo code
having its own internal queueing and timing control, it shouldn't
actually make any difference in practice (and likely, the error message
showing up is the only meaningful bug here - the issue is entirely
cosmetic).
Fixes: https://github.com/mpv-player/mpv/issues/12101
Configuration of filter parameters was moved from pl_filter_function (of
which the user had to make a copy) to pl_filter_config, with the
pl_filter_function remaining immutable.
Implement this new logic in a way that can reasonably exist side-by-side
with the old configuration API. Once we drop support for PL_API_VER
below 303, we can drastically simplify this code.
The win32 code already updates itself on dpi changes. However, it never
signalled mpv's core when this happened which meant that the
display-hidpi-scale property never changed. Simply send the
VO_EVENT_DPI event when appropriate. Fixes#12081.
In the same way that fallback in receive_frame() needs to be repeated
until we get a working decoder, we have to do the same thing in
reinit(). Calling force_fallback() only once can still yield a non
functional decoder.
The reason why we have these multiple paths which each require their
own fallback logic is that we can fail at different stages:
* hwdec init
* decoder init <- repeated fallback was missing here
* frame decoding
Fixes#12084
The manual currently says that if dscale is unset, --scale will be
applied. However, this only works at init time. If you change the dscale
filter to be empty later, vo_gpu will segfault and vo_gpu_next will
throw an error and refuse the changes. That's because when the option is
unset at runtime, the value becomes "" not NULL and the vo's never
accounted for this. Fixes#12031.
This only existed as essentially a workaround for meson's behavior and
to maintain compatibility with the waf build. Since waf put everything
in a generated subdirectory, we had to put make a subdirectory called
"generated" in the source for meson so stuff could go to the right
place. Well now we don't need to do that anymore. Move the meson.build
files around so they go in the appropriate place in the subdirectory of
the source tree and change the paths of the headers accordingly. A
couple of important things to note.
1. mpv.com now gets made in build/player/mpv.com (necessary because of
a meson limitation)
2. The macos icon generation path is shortened to
TOOLS/osxbundle/icon.icns.inc.
This is yet another unfortunate side effect of the width % SLICE_W == 0
special case. While looping through these rectangles, the rc->x1 value
on the final loop can be greater than the actual total width. This will
cause a buffer overflow if using the mp_draw_sub_overlay API. 2 of the
current VOs that use that work around it by adjusting the values
returned, but the better fix is to correct what's actually given in the
rectangles so you can safely use the values. As for the fix, it's simply
ensuring that rc->x1 doesn't extend beyond p->w with a MPCLAMP.
Previously, the code would always naively add SLICE_W (256) to rc->x0
which would easily extend past the actual width in many cases. The
adjustments in vo_vaapi and vo_dmabuf_wayland are dropped and in theory
vo_direct3d should work now since we can just trust the values given to
us in the rectangles. How nice.
reproduction steps: take any lossy WebP screenshot and check the colors
(or JPEG without libjpeg support, or AVIF with tag-colorspace=no)
fixes: 6d9e72cd89
We make the assumption that there is more buffer available
than indicated, this needs to be considered in this specific
location too as mp_image_crop understandably checks whether
we're about to do something unsafe.
minimal reproducer: mpv av://lavfi:testsrc -vf crop=320:239,format=yuv420p -o test.png
fixes#10469
In 1f8013ff3f, I mistakenly thought this
was only used for modifier presses way back in the commit it was
introduced in, but it actually also handles non-english keys/letters.
Instead of returning early, we should try xkb_keysym_to_utf8 first and
then return if that doesn't do anything so the modifier can then be
handled in the appropriate event. Fixes#12009.
This has always been a pet peeve of mine and in fact I named the option
in meson "egl-wayland" with the intention of finally doing this. We call
everything that's egl "egl-foo" internally except for wayland.
We wanted to preserve the libplacebo v5.264.0 requirement for gpu_next
for this release, since this is the what most Linux distributions are shipping.
The VLC 3 <-> libplacebo v6 situation is an additional reason distros are not
likely to ship the newest libplacebo release soon.
This reverts commit b73d96776c.
Since this works by mpctx giving us a dummy image, we can be reasonably
be confident that when we receive a matching imgfmt with no underlying
hw_subfmt that this is simply force window from the player. Just allow
this to go through the usual wayland logic and skip the hwdec checking.
When we get the real image later, it will still error out on the
compositors that don't have proper support.
It's a bit of an edge case, but since we now allow the disabling of the
software fallback it's possible to have a situation where hwdec
completely fails and the mpv window is still lingering from the previous
item in the playlist. What needs to happen is simply that the vo_chain
should uninit itself and handle force_window if needed. In order to do
that, a new VDCTRL is added that checks vd_lavc if force_eof was set.
player/video will then start the uninit process if needed after getting
this.
There's an option that's supposed to stop mpv from falling back to
software decoding if hardware decoding fails. Except that it doesn't
work and can fallback to software decoding anyway. Correct this by
checking if all possible hwdec failed after the loop in
select_and_set_hwdec and that we have this option. If so, flag a bool to
force eof. In decode_frame afterwards, we then simply immediately return
an EOF.
When video width is not a multiple of slice size, the last slice will overflow. Fix this by adjusting the width of the last slice to fit within the video.
Fixes issue #10436
In sum there were three issues:
1. nominal_peak_luminance is an output parameter, not input
2. mpv internally uses MP_REF_WHITE which we need to scale by
3. the value should be left at its default for SDR
Together with the last change this means mpv can sucessfully
take HDR screenshots with --screenshot-sw without resulting
in a brightly colored mess. Though note that zimg does not
perform proper tonemapping.
As a consequence:
- write_avif now respects tag-colorspace=no
- write_lavc additionally sets colorspace and chroma_location
(of no consequence for png or jpeg)
mpv has historically always treated the various tiled states in
xdg-shell as maximized (probably because it was easier). Well it turns
out that there are some tiling compositors (hyprland) that allow tiled
windows to maximize themselves. This can lead to some scenarios where
mpv ends up doing a maximize on hyprland which actually works since it's
not a no-op like on sway. Fix this by separating out the tiled state
from maximize. It works mostly the same, but the main difference is that
there's no request to tile yourself like there is with maximize. Should
fix#11954.
There's a lot of checks that are along the lines of !maximized &&
!minimized or vice versa. Make a locked_size boolean and store the
value of this in here to avoid writing long lines since the next commit
will add yet another condition to this.
There is an additional failure path I didn't account for in my previous
work. While I ensured that a late hwdec failure in receive_frame can be
recovered from by trying the next hwdec, there is a specific
combination where if an hwdec fails in receive_frame, and the next
hwdec is a full decoder (eg: v4l2m2m), and that also fails, we are left
with no decoder and so the entire decoding process ends and playback is
stopped.
Basically, we must keep re-attempting the fallback in receive_frame
until we get a valid decoder (software or hardware). This edge case
will rarely be encountered as there are only a couple of decoder based
hwdecs.
Fixes#11947
I don't know why we've been doing this wrong for so long or how I didn't
notice until now. Wayland specifically has an event for handling
modifiers. We even named it "keyboard_handle_modifiers" in the code.
What we should do is just get the modifier and save it after the xkb
state is updated. Then later if the user does something else (press
another key or clicks the mouse button), the saved modifier key is
applied. If you let go of the modifier at any point, the xkb will just
update its state again and we save a 0 again here (i.e. no modifier).
There is one bit of an edge case however. If a key is pressed BEFORE the
modifier, then we have to handle the mp_input_put_key in the modifier
event instead since the ordering is not guarenteed. What we do here is
keep track of the mpkey as well as the mpmod. However if we are unable
to find a matching mpkey and the key state is pressed down, assume it's
a modifery key that was pressed and don't update mpkey. That way
whenever the modifier event does happen, it can correctly handle this
and we know that the keys must be pressed down if we end up there in the
code path.
As another fun historical note, the xkb_keysym_to_utf8 line was actually
written by wm4 himself in 460ef9c7a4
nearly 10 years ago. As the commit shows, it was clearly intended to
handle modifiers (if lookupkey finds nothing, then try to find a mod
instead). Of course, this is extremely dated and wayland hasn't worked
like that in ages. This branch never actually did anything, and thus
we'll remove it here along with modifier lookup changes.
This solves bizarre issues with modifiers not working with random keys
while working fine with others (don't ask me why). Fixes#10286 and
fixes#11945.
828dd65ef8 started this and put it in the
most common place, but any resize mpv is doing should be accompanied by
that function call to correctly inform the compositor of the window
bounds. Since we also have to set VO_EVENT_RESIZE at the same time, it
makes sense to just put these in the same function. A slight thing to
note is that xdg_surface_set_window_geometry uses surface local
coordinates which means they are divided by the wl->scaling factor. If
possible, we should use surface local coordinates directly if available
(like in the toplevel config) to avoid potential rounding errors.
Otherwise, just calculate it out. The linked commit actually broke
some weston resizes in weird ways (window disappearing or completely
freezing). Don't ask me why and I didn't attempt to find out why either,
but with this it all appears to work normally again (the other
compositors don't appear to be any different, and they shouldn't be), so
we'll go with this since it fixes soemthing and it also is more
conceptually correct.
To remain consistent with our belief that `auto-safe` should be the
recommended mode when turning on hw decoding, let's remap `yes` from
`auto` to `auto-safe`.
We have supported `hwdec=drm` for some time now, and on devices where
this method is present, it is the preferred, and only, method
available. Combine that with the fact that software decoding on
embedded devices will not turn out well, and it's clear that we should
put `drm` and `drm-copy` on the `auto-safe` list to maximise usability
in simple configurations.
Today, if hwdec initialisation succeeds, we move on to attempting
decoding, and if decoding fails, we enter a fallback path that will
only do software decoding.
This is suboptimal because some hwdecs and codec combinations can
pass hwdec init, but then fail at hwaccel init, or even frame decoding.
In these situations, we will never attempt other hwdecs in the `auto`
or manually specified lists which might work.
One good example is someone tries to use `vulkan,auto` and tries to
play av1 content. The vulkan decoding will fail at hwaccel init time,
and then fallback to software instead of going through the auto list
which would allow vaapi, nvdec, etc to play the file.
Instead, let's not give up immediately, and try the next hwdec after
a failure, just like we do if hwdec init fails.
The key part of this change is to keep track of the hwdecs we have
tried, so that each time we run through hwdec selection, we don't try
the same one over and over again. If we really get through the whole
list with no success, we will then fall back to software decoding.
Fixes#11865
This was never implemented so in cases where the compositor didn't have
support for the underlying format, the window would just be black and
quite unhelpful. Well it turns out that fixing this is actually really
easy because mpv has reconfig2 which returns the entire mp_image on
every reconfig. We can just use that mp_image to try and obtain the
format and modifier pair and then use that to check against what we know
the compositor supports. If there are any problems, error out
gracefully. Some code duplication with the vaapi/drmprime importers, but
it's probably better to keep these separate. Fixes#11198 and fixes#11664
It makes more sense as an option at this point. Also libdrm is not
optional at all. You have to get a drm format and modifier for this to
even work (the VO will just fail without DRM). Just hard require it and
remove the guards. vaapi can remain optional.
mpv has never used this because we never really seemed to need it, but
it actually has a purpose. This informs the compositor of the actual
dimensions of the mpv window and avoids errors with sizes not matching
(particularly with weston). It's a better way to fix the "maximizing
causing an error on weston" issue since it also works for dmabuf-wayland
(which always had this problem).
The original reason for this commit was to prevent a compositor error on
weston when going into the maximized state. The configured dimensions of
mpv didn't match its actual size and Weston is super strict about this
so it threw a compositor error which is fatal. It only happened in
opengl and this seemed like an OK workaround, so I went with this.
However, there's actually a far easier way to solve this problem and we
don't need this anymore. This has the benefit of avoiding a harmless
warning message that appears with gpu-next on opengl. The next commit is
the proper solution. Closes#10324.
This reverts commit 661b5542de.
The implementation was copied from vo_gpu/vo_gpu_next but fundamentally
it doesn't make sense for this VO. Hardware decoding is not optional in
vo_dmabuf_wayland. We should be sure to request and load all supported
formats in the preinit and fail if there are any problems. There should
be no functional change from before, but it's more conceptually correct
this way.
This adds osd support via shm buffers using a similar approach that the
normal buffers do, but it differs in a few key areas. One thing to note
is that sway and weston actually handle this extremely differently which
required all the abstractions here. In particular, weston does not cope
well with destroying the wl_buffer from shm outside of the release
handler at all (i.e. it segfaults). The workaround here is to simply
attach a NULL to the osd surface and do a surface commit before we
destroy the buffers. This is reasonable enough and seems to work well
although it's pretty weird. Sway is more straightforward although it
actually releases the osd buffer when the window goes out of sight.
Also, I found that it doesn't always release every buffer before you
close it unlike weston seems to do which is part of the reason all this
bookkeeping is required. I don't know if there's any other compositor
out there that can possibly handle vo_dmabuf_wayland right now, but
suffering through these two is good enough for now I think.
Otherwise, we could fail and skip to uninit without initalizing this
which then will segfault because the list is null and accessed while
trying to destroy buffers.
It's entirely pointless. Not having viewport is already a fatal error
for this VO as it cannot possibly work without that protocol. Just drop
all these redundant if's.
make sure that fwrite is error checked. and if any data was still
remaining on the buffer, it will be flushed - and errors checked - via
the fclose() call below.
mpv already guesses when the window is hidden so plugging in a proper
event that actually tells us this is really trivial. Note that there's
some redundancy with setting wl->hidden in a few spots, but nothing can
really be done about that as long as the crappy hack is still in place.
This protocol no longer requires us to draw a separate cursor surface
and all of that horrible stuff. We can just ask the compositor for the
default cursor instead since that's literally all mpv cares about.
In case there are no packets from demuxer we cannot send EAGAIN, because
we will not proceed and get stuck with one frame in queue and never
output it. Just respect avcodec_receive_frame ret code and act
accordingly. The only case to care about is EOF when we have to drain
already queued frames.
Fixes playback of 1-2 frame videos.
It can happen during initialization and of course nothing good will
happen if we let this go through (i.e. segfault). Return and wait for
geometry to finish setting up in the wayland stuff before doing the
initial resize.
Very dumb. I can't remember if it was always like this or if I broke it
at some point, but clearly each wl_output should just be freed in
remove_output. Freeing it if it happens to be wl->current_output only
works for that one monitor, so remove that whole line. This has to
happen before we close the wayland connection so reorder the uninit a
little bit.
--no-config should prevent loading user files of any type: configs,
cache, etc. For cache files, this case wasn't properly handled and it
was assumed they would always get something. vo_gpu's shader cache
actually already handles this, so it was left untouched. In theory,
demuxer cache should never have this issue because saving it to disk is
disabled by default (and likely that will never change), but go ahead
and change it for consistency's sake. Fixes some segfaults with
--no-config and various combinations of settings (particularly
--vo=gpu-next).
4502522a7a changed the way mpv handled and
saved cached files. In particular, it made a separate boolean option for
actually enabling cache and left the *-dir options as purely just a path
(i.e. having a dir set didn't mean you save cache). This technically
regressed people's configs, so let's just turn the cache on by default.
Linux users already expect random stuff in ~/.cache and well everyone
else can just live with some files possibly appearing in their config
directory.
Add an option for allowing pointer events to pass through the mpv
window. This could be useful in cases where a user wants to display
transparent images/video with mpv and interact with applications beneath
the window. This commit implements this functionality for x11 and
wayland. Note that whether or not this actually works likely depends on
your window manager and/or compositor. E.g. sway ignores pointer events
but the entire window becomes draggable when you float it (nothing under
the mpv window receives events). Weston behaves as expected however so
that is a compositor bug. Excuse the couple of completely unrelated
style fixes (both were originally done by me).
Notes:
- converts the (image) write() api to filenames, because using avio
with FILE* is a pain.
- adds more debug logs for screenshots.
build: rename av1 dependency to avif_muxer
wscript: unify lavf dependency with meson
Today, the only way to make mpv consider multiple hwdecs and pick the
first one that works is to use one of the `auto` modes. But the list
that is considered in those cases is hard-coded. If the user wants to
provide their own list, they are out of luck.
And I think that there is now a significant reason to support this -
the new Vulkan hwdec is definitely not ready to be in the auto list,
but if you want to use it by default, it will not work with many codecs
that are normally hardware decodable (only h.264, hevc and av1 if you
are very lucky). Everything else will fall back to software decoding.
Instead, what you really want to say is: use Vulkan for whatever it
supports, and fall back to my old hwdec for everything else.
One side-effect of this implementation is that you can freely mix
hwdec names and special values like `auto` and `no`. The behaviour will
be correct, so I didn't try and prohibit any combinations. However,
some combinations will be silly - eg: sticking any further values after
`no` will result in them being ignored. On the other hand, a
combination like `vulkan,auto` could be very useful as that will use
Vulkan if possible, and if not, run the normal auto routine.
Fixes#11797
this changes mp_image_new_ref() to handle allocation failure itself
instead of doing it at its many call-sites (some of which never checked
for failure at all).
also remove MP_HANDLE_OOM() from the call sites since this is not
necessary anymore.
not all the call-sites have been touched, since some of the caller might
be relying on `mp_image_new_ref(NULL)` returning NULL.
Fixes: https://github.com/mpv-player/mpv/issues/11840
when vo_drm_init() fails inside of preinit(), uninit() will be called as
part of cleanup with vo->drm being NULL and thus `drm->fd` would lead to
null dereference.
and since vo_drm_uninit() closes drm->fd, destroy_framebuffer() ends up
using a closed fd.
according to the drm-gem manpage [0]:
> If you close the DRM file-descriptor, all open dumb-buffers are
> automatically destroyed.
so remove the destroy_framebuffer() loop entirely, which fixes both the
issues.
[0]: https://www.systutorials.com/docs/linux/man/7-drm-gem/
I originally left `drmprime_overlay` as higher priority because
`drmprime` was new, and because I didn't have any hardware where both
worked (only one or the other) so I couldn't compare relative
performance, and if only one worked, the priority didn't matter.
But with time and more usage, we've reached a point where we can say we
would recommend using `drmprime` in situations where both work, and
we've also been able to identify hardware where both do indeed work and
it seems that `drmprime` is more reliable.
So, let's flip them.
It was done once before but later reverted for testing reasons. This
time it's permanent though since I can test this VO on ARM and with an
up to date system.
When using a display-* video-sync mode, it is possible for buffers with
a matching id to already have an image associated with them (i.e. the
compositor hasn't released it yet). Previously, it was thought that we
could just unref, return null, and make a new buffer but this eventually
leads to a fatal error that originates from libwayland itself which
stops playback. Admittedly, the reason for the error is a bit nebulous
but likely it seems to be some kind of mismatch between dmabuf params
and the associated image with the buffer.
However, we can simplify this process greatly. Instead when the
previously mentioned edge case happens, the old image can simply be
freed and we give the buffer the new image. This saves creating a new
buffer and also avoids that nasty libwayland error. A nice win-win all
around. Fixes#11773.
vo_dmabuf_wayland has a pool of wl_buffers that it cycles through when
drawing frame. There needs to be at least some minimum number otherwise
a flickering artifact occurs where old frames are mistakenly repeated.
When using display-resample and other similar modes, it seems more
buffers are required (more drawing happens so it makes sense) and the
current minimum of 8 isn't good enough. Let's just bump this to 15. It's
also a random ad hoc number, but as far as I know there's not really a
way to predict how many buffers a random video may need. From testing,
it works fine and overall 15 is still a tiny amount of objects to create
considering the lifetime of a video, so we'll just go with this.
Some platforms (wayland) apparently have a lot of trouble with drag and
drop. The default behavior is still the same which is basically obeying
what we get from the window manager/compositor, but the --drag-and-drop
option allows forcibly overriding the drag and drop behavior. i.e. you
can force it to always replace the playlist or append at the end. This
only implements this in X11 and Wayland but in theory windows and macos
could find this option useful (both hardcode the shift key for
appending). Patches welcome.
In data_offer_actions, it's possible to get the
WL_DATA_DEVICE_MANAGER_DND_ACTION_NONE action which would set
wl->dnd_action to DND_APPEND (did nothing in practice) but also log a
message which is confusing and misleading. Instead, just ignore and
don't do anything when we get this case.
This is not technically necessary, because we never touch the fd again
after passing to cuda, but having it there could lead to future code
accidentally using it.
All hwdecs should respect the probing flag and demote their lgoging to
verbose level, so that initialisation failures during probing do not
spam the user. I forgot to do this for the Vulkan hwdec.
The dmabuf-wayland vo has a stub ra implementation that doesn't
have a swapchain. That means that it's currently not safe to call
ra_vk_ctx_get on that ra_ctx, but it must be safe to call on all ra
implementations as this is how we discover if it is a vulkan ra.
This hasn't been an issue before because no Vulkan code paths would be
triggered when using dmabuf-wayland, but with the new vulkan hwdec, it
becomes possible to trigger when hwdecs are probed.
The libplacebo sync abstraction is deprecated and we should be using
the more explicit Vulkan semaphore helpers instead. This means that
more of the book keeping moves to our side, but it's not too bad.
There are two main things going on here:
1. After a lot of back and forth, I decided to write the new code with
timeline semaphores to streamline things, and that also means all
the variables are separate - which makes the #ifdefs easier to read.
Which is important because:
2. While pl_sync owned the exported fd/handle, pl_vulkan_sem does not,
so we are responsible for managing them. That means reversing the
previous logic - we now can pass an original fd to CUDA and never
think about it again, while we have to clean up a Win32 Handle
because CUDA will not take ownership.
AV1 support in Vulkan is extremely bleeding edge - to the point that
the extension is not present in official Khronos releases, but it has
a reserved identifier and we can look it up with a string literal for
now.
This will be skipped and ignored if the driver doesn't support it, so
it's safe if/when the name changes later (it'll just never be activated
in that case).
I originally wrote this trying to avoid doing an explicit version check
on the headers, but it just makes things more confusing, and the
requirements harder to understand.
So, Vulkan interop now takes a dependency on the header release where
they finalised the video decode headers. VK_EXT_descriptor_buffer was
added in 1.3.235, so that's covered as well.
Along the way I fixed a bug in the waf build where it was depending
on libplacebo-next instead of libplacebo-decode.
ffmpeg was previously allocating images for frames as the code size,
rather than the presentation one (1088 vs 1080 in the most common
example). Using the coded size when wrapping images for libplacebo
resulted in incorrect scaling from 1088 -> 1080, but even using the
presentation size wasn't perfect, as discussed in the original
commit.
However, ffmpeg has now been updated to use the presentation size for
the frame images, after discussions that concluded this must be done
because there is no way for a frame consumer to fix the dimensions
without copying the frame.
With that ffmpeg change, we can just use the normal layout information
like all the other hwdecs.
Although we can support vulkan multiplane images, cuda lacks any such
support, and so cannot natively import such images for interop. It's
possible that we can do separate exports for each plane in the image
and have it work, but for now, we can selectively disable multiplane
when we know that we'll be consuming cuda frames.
As a reminder, even though cuda is the frame source, interop is one way
so the vulkan images have to be imported to cuda before we copy the
frame contents over.
This logic here is slightly more complex than I'd like but you can't
just set the flag blindly, as it will cause hwframes ctx creation to
fail if the format is packed or if it's planar rgb. Oh well.
Vulkan hwdec interop with the ffmpeg 6.1 vulkan code will require
additional features beyond those activated by libplacebo by default.
Enabling these features requires both requesting the features'
extensions and then explicitly turning on the features. libplacebo
handles detecting unsupported features and dropping them, to avoid
failing to create the vulkan device.
We then leave it to ffmpeg to decide if any missing features are
required for functionality, and error out if necessary.
As ffmpeg requires at least one bleeding edge extension (descriptor
buffers), all of this logic is gated on the presence of sufficiently
new Vulkan headers.
Vulkan Video Decoding has finally become a reality, as it's now
showing up in shipping drivers, and the ffmpeg support has been
merged.
With that in mind, this change introduces HW interop support for
ffmpeg Vulkan frames. The implementation is functionally complete - it
can display frames produced by hardware decoding, and it can work with
ffmpeg vulkan filters. There are still various caveats due to gaps and
bugs in drivers, so YMMV, as always.
Primary testing has been done on Intel, AMD, and nvidia hardware on
Linux with basic Windows testing on nvidia.
Notable caveats:
* Due to driver bugs, video decoding on nvidia does not work right now,
unless you use the Vulkan Beta driver. It can be worked around, but
requires ffmpeg changes that are not considered acceptable to merge.
* Even if those work-arounds are applied, Vulkan filters will not work
on video that was decoded by Vulkan, due to additional bugs in the
nvidia drivers. The filters do work correctly on content decoded some
other way, and then uploaded to Vulkan (eg: Decode with nvdec, upload
with --vf=format=vulkan)
* Vulkan filters can only be used with drivers that support
VK_EXT_descriptor_buffer which doesn't include Intel ANV as yet.
There is an MR outstanding for this.
* When dealing with 1080p content, there may be some visual distortion
in the bottom lines of frames due to chroma scaling incorporating the
extra hidden lines at the bottom of the frame (1080p content is
actually stored as 1088 lines), depending on the hardware/driver
combination and the scaling algorithm. This cannot be easily
addressed as the mechanical fix for it violates the Vulkan spec, and
probably requires a spec change to resolve properly.
All of these caveats will be fixed in either drivers or ffmpeg, and so
will not require mpv changes (unless something unexpected happens)
If you want to run on nvidia with the non-beta drivers, you can this
ffmpeg tree with the work-around patches:
* https://github.com/philipl/FFmpeg/tree/vulkan-nvidia-workarounds
We will need the full ra_ctx to be able to look up all the state
required to initialise an ffmpeg vulkan hwcontext, so pass let's
pass the ra_ctx instead of just the ra.
This is motivated by a need to access it from vo_gpu_next's opengl
wrapping code, and justified by it being an inherent property of the GL
context itself,
It was unsafe to return pointer to memory that was freed on another
thread, just copy the string to caller owned sturcture.
Fixes crashes when displaying passes stats with gpu-next.
vo_dmabuf_wayland worked by allocating entries to a pool and then having
a lot of complex logic dealing with releasing buffers, pending entries,
etc. along with some other not so nice things. Instead, we can rewrite
this logic so that the wl_buffers created by the imported dmabuf is
instead stored in a linked list, wl_list. We can simply append our
buffers to the list when needed and destroy everything at the end. On
every frame, we can check the ids of our surfaces and reuse existing
buffers, so in practice there will only ever be a handful at a time.
Some other small changes were made in an attempt to organize the
vaapi/drmprime code a little better as well.
An important change is to always enforce at least a minimum number of
buffers. Certain formats would not make enough unique buffers, and this
results in flickering/artifacts occuring. The old way to attempt to deal
with this was to clear out all the existing buffers and remake them, but
this gets complicated and also didn't always work. An easy solution to
this is just create more buffers which appears to solve this problem.
The actual number needed is not really based on anything solid, but 8
is a reasonable number to create for the lifetime of a file and it seems
to do the trick.
Additionally, seeking/loading new files can result in flicker artificts
due to buffers being reused when they shouldn't. When that happens, we
flip a bool so all the buffers get destroyed in draw_frame to avoid any
visual glitches.
