If b-frames were enabled implicitly (if max_b_frames wasn't set by
the caller at all, since a0949d0bcb),
we wouldn't offset dts at all, producing invalid pts/dts combinations
(causing loud warnings by ffmpeg, or muxer errors if passed without
an extra cleanup pass).
Instead use frameIntervalP for offsetting, which should always be
accurate.
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
This commit follows the same logic as 061a0c14bb, but for the encode API: The
new public encoding API will no longer be a wrapper around the old deprecated
one, and the internal API used by the encoders now consists of a single
receive_packet() callback that pulls frames as required.
amf encoders adapted by James Almer
librav1e encoder adapted by James Almer
nvidia encoders adapted by James Almer
MediaFoundation encoders adapted by James Almer
vaapi encoders adapted by Linjie Fu
v4l2_m2m encoders adapted by Andriy Gelman
Signed-off-by: James Almer <jamrial@gmail.com>
The h264_nvenc and hevc_nvenc encoders aren't respecting the framerate in the codec context.
Instead it was using the timebase which in our use-case was 1/1000 so the encoder was behaving
as if we wanted 1000fps. This resulted in poor encoding results due to an extremely low bitrate.
Both the amf and qsv encoders already contain similar logic to first check the framerate before
falling back to the timebase.
Signed-off-by: Zachariah Brown <zachariah@renewedvision.com>
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
The old approach used some highly complex delta computation math and
output-delaying.
I do not remember what the initial reasoning behind that was, but given
that we can just offset the dts by the amount of bframes, it seems wholy
unnecessary.
This leaves open an issue with VFR content, for which some more complex
logic might be needed.
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
This reverts nvenc to old behaviour, which in some super rare edge cases
performs better.
The implication of this is that any potential API user who relies on
nvenc cleaning up every frames device resources after it's done using
them will have to change their usage pattern.
That should not be a problem, since pretty much every normal usage
pattern automatically implies that surfaces are reused from a common
pool, since constant re-allocation is also very expensive.
We have a pattern of wrapping CUDA calls to print errors and
normalise return values that is used in a couple of places. To
avoid duplication and increase consistency, let's put the wrapper
implementation in a shared place and use it everywhere.
Affects:
* avcodec/cuviddec
* avcodec/nvdec
* avcodec/nvenc
* avfilter/vf_scale_cuda
* avfilter/vf_scale_npp
* avfilter/vf_thumbnail_cuda
* avfilter/vf_transpose_npp
* avfilter/vf_yadif_cuda
This reverts commit 7d4e1f7cfb.
Accidentially pushed this with a batch of other patches, and it didn't
seem to break anything, so I went with it.
Except it does, so reverting it it is.
The patch enables dynamic bitrate through ReconfigureEncoder method
from nvenc API.
This is useful for live streaming in case of network congestion.
Signed-off-by: pkviet <pkv.stream@gmail.com>
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
If there is input like DVB-T streams it can change aspect ratio
on-the-fly, so nvenc should respect this change and change aspect ratio
in encoder.
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
