When #5169 implemented deferred encoder updates, it did not account for
the case when the encoder hadn't started yet. This means the encoder would
start and then immediately call update with the same settings it was
started with, which in the case of some hardware encoders would trigger
a reconfiguration request to the driver.
Reset frame_rate_divisor_counter to 0 on encoder shutdown.
After starting and stopping a GPU encoder session, obs_encoder_shutdown
would set frame_rate_divisor_counter to 1. When the next GPU encoder
session was started, in libobs/obs-video-gpu-encode.c, gpu_encode_thread
would set skip to this value (1), and increment
frame_rate_divisor_counter to 2. This causes the next check to fail, as
frame_rate_divisor is 1 by default (2 == 1 is false), so
frame_rate_divisor_counter retains its value. Since skip is non-zero,
the next check, if(skip), passes, and skip to the next loop iteration.
This will continue forever, because frame_rate_divisor_counter will
continue to increment, so it will never hold the same value as
frame_rate_divisor. This means that send_off_encoder_packet is never
called, so the muxer never receives encoded packets.
To the end-user, this manifests as their second encoder session being
impossible to stop. They then have to force quit OBS and the
obs-ffmpeg-mux process. This change prevents that from occurring and
allows multiple back-to-back GPU encoder sessions to be completed.
Allows rescaling resolution for GPU encoders and allows moving
rescaling for CPU encoders from CPU to GPU
Rescaling is implemented via core video mixes; encoders create
their own core mix with matching width/height/format/colorspace/
range when gpu scaling is enabled and no matching core video
mix exists
It was possible to set the encoder to scale the video to the same size
as the output, resulting in unnecessary passes through swscale and
skipping hardware encoders. This was a common user error in Advanced
Output mode where they would tick the rescale box but leave it at the
same resolution as their main output.
There is currently no way to clear a video_t or audio_t object from an
encoder once applied. `audio_t`/`video_t` objects can be destructed at
any time, and it is dangerous to prevent these object references from
even being cleared.
This does not fix the issue where destroying an audio/video object does
not clear the reference from all subscribed encoders.
Protect `obs_encoder_set_video()` and `obs_encoder_set_audio()` from
being used if the encoder is active. Changing these values while active
is undefined behavior.
Split render_texture and derived fields in obs_core_video into new
obs_core_video_mix struct. Add new APIs to add additional obs_view to the render loop, each with a separate render_texture / obs_core_video_mix.
With this, you can now cast normal obs objects (services, outputs,
sources, encoders) to an obs_object_t, and then use obs_object_*
functions to get references, release references, and similar for weak
object references as well. This allows the ability for the frontend to
use an object of any of those types interchangeably in certain
situations without having to handle each specific type individually.
This is useful because the properties view in particular doesn't care
what type of object it uses, it just needs to be able to hold weak
references to abstract OBS objects.
Fixes a long-standing issue with Dynamic Bitrate where the RTMP output
thread could try to reconfigure an encoder while the encoder is in the
middle of encoding. x264 seems to handle multithreaded calls well, but
NVENC would deadlock in this situation with no error visible to the
user.
Similar to how outputs can pass errors, add the same functionality for
encoders so that if an output encoder has an error, it is made available
to the output and eventually the UI / user.
As os_gettime_ns() gets large the current scaling methods, mostly by casting
to uint64_t, may lead to numerical overflows. Sweep the code and use
util_mul_div64() where applicable.
Signed-off-by: Hans Petter Selasky <hps@selasky.org>
This reverts commit ff22c20019.
This caused a bug in FTL output, which started hitching after this
commit. Presumably due to opus; it's likely you're not supposed to do
this with all audio encoders.
Returns whether rescaling is enabled for an encoder. This will be used
with texture-based encoders to determine whether to fall back to
RAM-based encoding instead.
When an unpause occurs, it takes an audio segment and splits it at the
exact point corresponding to the pause timestamp, and then it's supposed
to only send the ending part of the split. However, the audio pointers
were not being incremented, therefore it was sending the front of the
audio segment to instead of the back of the audio segment by mistake.
When pause has been activated, the video_pause_check() function is used
when receiving raw frames in order to filter out frames that are in the
pause window, that way they aren't sent to the encoder or output.
However, when pause was enabled, it was unintentionally filtering out
some frames before the specified starting timestamp as well, causing
extra video data to get cut out prematurely. This fixes that issue.
Unlike get_properties, there is not reason to not call get_defaults if it is
given in addition to get_defaults2. Additonally this fixes the bug with
'init_encoder' which would only ever call get_defaults, resulting in broken
encoders if those used get_defaults2.
This implements pausing of outputs. To accomplish this, raw audio/video
data is halted to the encoders or raw output. Pausing is as precisely
timed as possible according to the timing of the obs_output_pause call,
and audio data will be spliced down to the exact audio sample in
accordance to that timing at the start/end marks.
Outputs that support this (outputs used for recording) can set the
OBS_OUTPUT_CAN_PAUSE capability flag.
Code submissions have continually suffered from formatting
inconsistencies that constantly have to be addressed. Using
clang-format simplifies this by making code formatting more consistent,
and allows automation of the code formatting so that maintainers can
focus more on the code itself instead of code formatting.
(This commit also modifies the UI, obs-ffmpeg, and obs-output modules)
Fixes a long-time regression where the program would lock up if an
encode call fails. Shuts down all outputs associated with the failing
encoder and displays an error message to the user.
Ideally, it would be best if a more detailed error could be displayed to
the user about the nature of the error, though the primary problem is
the encoder errors are typically not something the user would be able to
understand. The current message is a bit of a generic error message;
improvement is welcome.
Another suggestion is to try to have the encoder restart seamlessly,
though it would take a significant amount of work to be able to make it
do something like that properly, and it sort of assumes that encoder
failures are sporadic, which may not necessarily be the case with some
hardware encoders on some systems. It may be better just to use another
encoder in that case. For now, seamless restart is ruled out.
If the remove_connection call of obs_encoder_stop_internal took too
long, obs_encoder_destroy could get called before that function
completed, causing a race condition.
Allows the ability to encode by passing NV12 textures. This uses a
separate thread for texture-based encoders with a small queue of
textures. An output texture with a keyed mutex shared texture is locked
between OBS and each encoder. A new encoder callback and capability
flag is used to encode with textures.
This splits the "do_encode" function in to "do_encode" and
"send_off_encoder_packet", the latter of which allows the ability for
texture-based encoders to manage their own encoding and just simply send
off a packet to the outputs.
Allows the ability for one encoder to defer to another in case of
failure or unsupported feature. Okay, fine, it's mostly a hack so the
new NVENC encoder can fall back to the FFmpeg encoder if NV12 textures
aren't in use, that way it does not have to implement raw fallback
support itself. The settings and properties are pretty much the same,
so there's no reason not to utilize it in order to save time that could
otherwise be spent more productively.
Reduces GPU usage when encoding is not active. Does not perform color
conversion, frame staging, or frame downloading unless encoding is
explicitly active.
On audio encoder startup, audio encoders paired with a video encoder
would unintentionally discard a single audio data segment, causing it to
be 1024 audio frames out of sync.
