Introduce support for delivering BPM (Broadcast
Performance Metrics) over SEI (for AVC/H.264 and
HEVC/H.265) and OBU (for AV1) unregistered messages.
Metrics being sent are the session frame counters,
per-rendition frame counters, and RFC3339-based
timestamping information to support end-to-end
latency measurement.
SEI/OBU messages are generated and sent with each IDR
frame, and the frame counters are diff-based, meaning
the counts reflect the diff between IDRs, not the running
totals.
BPM documentation is available at [1].
BPM relies on the recently introduced encoder packet timing
support and the packet callback mechanism.
BPM injection is enabled for an output by registering
the `bpm_inject()` callback via `obs_output_add_packet_callback()`
function. The callback must be unregistered using
`obs_output_remove_packet_callback()` and `bpm_destroy()`
must be used by the caller to release the BPM structures.
It is important to measure the number of frames successfully
encoded by the obs_encoder_t instances, particularly for
renditions where the encoded frame rate differs from the
canvas frame rate. The encoded_frames counter and
`obs_encoder_get_encoded_frames()` API is introduced
to measure and report this in the encoded rendition
metrics message.
[1] https://d50yg09cghihd.cloudfront.net/other/20240718-MultitrackVideoIntegrationGuide.pdf
Packet callbacks are invoked in `send_interleaved()` and
are useful for any plugin to extend functionality at the
packet processing level without needing to modify code in
libobs. Closed caption support is one candidate that is
suitable for migration to a packet callback.
The packet callback also supports the new encoder packet
timing feature. This means a registered callback will have
access to both the compressed encoder packet and the associated
encoder packet timing information.
Introduce support for the `encoder_packet_time` struct
to capture timing information for each frame, starting
from the composition of each frame, through the encoder,
to the queueing of the frame data to each output_t.
Timestamps for each of the following events are based on
`os_gettime_ns()`:
CTS: Composition time stamp (in the encoder render threads)
FER: Frame encode request
FERC: Frame encoder request completely
PIR: Packet interleave request (`send_interleaved()`)
Frame times are forwarded through encoder callbacks in the
context that runs on the relevant encoder thread, ensuring
no race conditions with accessing per encoder array happen.
All per-output processing happens on data that is owned by
the output.
Co-authored-by: Ruwen Hahn <haruwenz@twitch.tv>
Deprecated in 28.0, documentation erroneously states 27.2.
The following functions were erroneously not marked as deprecated in
the header:
- obs_sceneitem_set_show_transition()
- obs_sceneitem_set_show_transition_duration()
A recent obs-deps change removed all non-essential x86 deps. This caused
the x86 subproject(s) on Windows to fail to configure due to being
unable to find x86 dependencies that do not exist.
Co-authored-by: PatTheMav <PatTheMav@users.noreply.github.com>
Adds a new encoder cap which tells libobs that rather than scaling
video frames in software, the encoder is capable of scaling them via
its own (presumably more efficient) means.
An encoder may implement this cap by comparing the VOI of its assigned
`video_t` and the results of `obs_encoder_get_width/height()`. If the
width/height values differ, then the encoder is being asked by libobs
to self-scale, and the resolution in VOI will be the raw frame size,
with the `...get_width/height()` being the intended output resolution
of the encoder.
It is important to note that GPU rescaling mode will take priority
over self-scaling. If GPU rescaling is enabled, the encoder will never
be asked to self-scale.
This is useful for discrete hardware encoders, where they might have
fixed-function video scaling logic that is highly efficient and fast.
Additionally, this feature allows a hardware device which is encoding
a full ABR ladder of tracks to be smart and only copy a video frame
from GPU -> Host -> Device once for the entire ladder, rather than
once for every track.
This logic would previously have written any changed scale resolution
set by the encoder in the `.get_video_info` callback back to the
encoder, however this functionality was "broken" by
20d8779d30. In reality, this would have
never worked with texture encoders or with GPU rescaling enabled, and
probably would have had odd side effects for CPU rescaling, too. It's
best just to remove this functionality.
I forgot to update this before tagging 30.2.1, so the updater won't stop
telling people there's an update available, even after updating.
Update this for real now to 30.2.2.
Fixes a crash when the following steps occur:
- Encoder group created and then started with encoders, only the group
owning encoder refs
- Encoder group destroy called: group has `destroy_on_stop` set
without actual destroy
- Outputs holding encoders from stopping are stopped
- `remove_connection()` function destroys encoder group, releasing
encoders and causing the currently processed encoder to be destroyed
early
- parent scopes try to access destroyed encoder pointer and crash
This change moves some logic around to improve the release/destruct
order of `obs_encoder_stop()` to fix the race above.
Note: Cases of encoder errors will not destruct the group if it has
`destroy_on_stop` set, as the encoder is also not destroyed if it also
is set to destroy on stop. This part of the code should be revisited
at a later date and fixed up to prevent memory leaks.
Modifies the encoder group API added previously to better follow the
existing libobs API naming paradigms. This also produces much more
readable code, and allows a few small benefits like only needing to
hold a reference to the encoder group, instead of every encoder.
