Adds support for texture-based AMD encoding, with both H264, HEVC, and
HDR support. Falls back to FFmpeg when texture-based encoding cannot be
used for whatever reason.
(Jim note: This is based upon obsproject/obs-studio#4538 by AMD/Luxoft
with fewer files, FFmpeg fallback for software encoding, and HDR
support. I also went to lengths to ensure that FFmpeg command line
parameters also works with it)
Co-authored-by: Jim <obs.jim@gmail.com>
Similar to the Windows counter part. Check the PCI bus for
installed cards. When found at least one VGA compatible adapter
from NVIDIA that is not in the blacklist we pass the check.
Per FFmpeg commit 337f777f378c [1], FFmpeg removed nvenc_h264_encoder
and nvenc_hevc_encoder after deprecation in FFmpeg commit 888a5c794778
[2]. The names to be used are ff_h264_nvenc_encoder and
ff_hevc_nvenc_encoder. So we must allow alternative search of codec as
h264_nvenc or nvenc_h264 in obs-ffmpeg.c.
[1]: 337f777f37
[2]: 888a5c7947
Add and register an obs_output_info struct called
ffmpeg_hls_muxer. It uses ffmpeg's HLS muxer to stream output.
Add threading and buffer to reduce skipped frames.
Also add frame drop logic when there are too many packets in the
buffer for congestion control.
This replaces the ffmpeg-encoded-output which had serious issues
(missing headers, muxing issues with non compliant mpegts streams) with
an output grafted from obs-ffmpeg-mux.
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.
The quadro P5000 would incorrectly be considered blacklisted because it
used a string search for the P500, which is an earlier quadro variant
that does not have NVENC support.
To fix this, instead of just doing a string search, additionally check
to make sure that there preceding or trailing numbers on the adapter
name.
Adds a texture-based NVENC implementation which passes OBS NV12 output
textures directly to NVENC without downloading them off of the GPU,
increasing NVENC performance by a significant margin.
If NV12 textures are unavailable or the new encoder fails to initialize
for whatever reason, it will fall back to the FFmpeg NVENC
implementation safely.
This implements a blacklist of devices that ship with NVENC libraries,
but do not support NVENC, which would cause the plugin to mistakenly
think that NVENC was available when it was not. If these devices are
the only NVIDIA devices on the system, consider NVENC unavailable.
This is much more efficient than spawning an encoder on startup to see
if NVENC is available, which would incur a ~500 millisecond hit on
startup. Additionally, also much less crash-prone.
The reason this code is being reverted/removed is because this code is a
risk to startup stability. This check will be added again in the
future, however this code needs to be executed from a secondary piped
process instead of directly in the process to reduce risk of
driver/hardware issues impacting program startup. The same needs to be
accomplished for the AMF plugin as well.
This reverts commit 94b5982216.
Reverting this commit because it had some negative side effects, such as
adding 500 milliseconds to the startup time. NVENC detection should
really be done through its proper API, and not via creating an encoder
on startup.
Certain NVIDIA GPUs don't support NVENC, but ship with the NVENC
library, causing OBS to mistakenly think that NVENC is available when it
actually isn't.
Closesjp9000/obs-studio#1087
Another thread could be manipulating the active_log_contexts array while the current thread is trying to read it, resulting in an uninitialized memory crash as the da_push_back call was not protected by the mutex.
In addition to the flv file format, this allows the ability to save to
container formats such as mp4, ts, mkv, and any other containers that
support the current codecs being used.
It pipes the encoded data to the ffmpeg-mux process, which then safely
muxes the file from the encoded data. If the main program unexpectedly
terminates, the ffmpeg-mux piped program will safely close the file and
write trailer data, preventing file corruption.
This functionality can now be handled automatically because locale can
now be freed seaparately from obs_module_unload with
obs_module_free_locale, which is called automatically when the module is
being freed.
- Implement the RTMP output module. This time around, we just use a
simple FLV muxer, then just write to the stream with RTMP_Write.
Easy and effective.
- Fix the FLV muxer, the muxer now outputs proper FLV packets.
- Output API:
* When using encoders, automatically interleave encoded packets
before sending it to the output.
* Pair encoders and have them automatically wait for the other to
start to ensure sync.
* Change 'obs_output_signal_start_fail' to 'obs_output_signal_stop'
because it was a bit confusing, and doing this makes a lot more
sense for outputs that need to stop suddenly (disconnections/etc).
- Encoder API:
* Remove some unnecessary encoder functions from the actual API and
make them internal. Most of the encoder functions are handled
automatically by outputs anyway, so there's no real need to expose
them and end up inadvertently confusing plugin writers.
* Have audio encoders wait for the video encoder to get a frame, then
start at the exact data point that the first video frame starts to
ensure the most accrate sync of video/audio possible.
* Add a required 'frame_size' callback for audio encoders that
returns the expected number of frames desired to encode with. This
way, the libobs encoder API can handle the circular buffering
internally automatically for the encoder modules, so encoder
writers don't have to do it themselves.
- Fix a few bugs in the serializer interface. It was passing the wrong
variable for the data in a few cases.
- If a source has video, make obs_source_update defer the actual update
callback until the tick function is called to prevent threading
issues.
There were a *lot* of warnings, managed to remove most of them.
Also, put warning flags before C_FLAGS and CXX_FLAGS, rather than after,
as -Wall -Wextra was overwriting flags that came before it.