| Message ID | 20231213131536.10242-1-ffmpeg@haasn.xyz |
|---|---|
| Headers | show |
| Series | YUV colorspace filter negotiation | expand |
On Wed, 13 Dec 2023 14:11:57 +0100 Niklas Haas <ffmpeg@haasn.xyz> wrote: > Split off from my YUVJ removal series. This implements all of the > libavfilter changes needed to fully deprecate YUVJ, but does not yet > remove YUVJ, nor add an AVCodec API for advertising colorspace support. > > Update includes all of the feedback that was brought up by Anton. the > major change from v1 is that YUV metadata configured on a link is no > longer strictly required to be consistent with the frames, simply to > avoid breaking users that don't set this link metadata. Effectively, > this allows us to have a grace period where the field functions more > like a hint than an authoritative field. Ping, any further feedback? If not, I will merge this sometime after Christmas.
On Sun, 24 Dec 2023 10:35:22 -0800 Niklas Haas <ffmpeg@haasn.xyz> wrote: > On Wed, 13 Dec 2023 14:11:57 +0100 Niklas Haas <ffmpeg@haasn.xyz> wrote: > > Split off from my YUVJ removal series. This implements all of the > > libavfilter changes needed to fully deprecate YUVJ, but does not yet > > remove YUVJ, nor add an AVCodec API for advertising colorspace support. > > > > Update includes all of the feedback that was brought up by Anton. the > > major change from v1 is that YUV metadata configured on a link is no > > longer strictly required to be consistent with the frames, simply to > > avoid breaking users that don't set this link metadata. Effectively, > > this allows us to have a grace period where the field functions more > > like a hint than an authoritative field. > > Ping, any further feedback? If not, I will merge this sometime after Christmas. Merged as 1b0ca477..44a33fccd
[AMD Official Use Only - General]
I found out this change caused page fault as well.
--------
8c7934f73ab6c568acaa47c821a6833f9145fdbb is the first bad commit
commit 8c7934f73ab6c568acaa47c821a6833f9145fdbb
Author: Niklas Haas <git@haasn.dev>
Date: Sun Dec 31 13:35:03 2023 -0800
avfilter: add negotiation API for color space/range
Motivated by YUVJ removal. This change will allow full negotiation
between color ranges and matrices as needed. By default, all ranges and
matrices are marked as supported.
Because grayscale formats are currently handled very inconsistently (and
in particular, assumed as forced full-range by swscale), we exclude them
from negotiation altogether for the time being, to get this API merged.
After filter negotiation is available, we can relax the
grayscale-is-forced-jpeg restriction again, when it will be more
feasible to do so without breaking a million test cases.
Note that this commit updates one FATE test as a consequence of the
sanity fallback for non-YUV formats. In particular, the test case now
writes rgb24(pc, gbr/unspecified/unspecified) to the matroska file,
instead of rgb24(unspecified/unspecified/unspecified) as before.
doc/APIchanges | 3 +
libavfilter/avfilter.c | 17 ++++-
libavfilter/avfilter.h | 28 +++++++
libavfilter/avfiltergraph.c | 173 +++++++++++++++++++++++++++++++++++++++++++-
libavfilter/formats.c | 121 +++++++++++++++++++++++++++++++
libavfilter/formats.h | 54 ++++++++++++++
libavfilter/internal.h | 6 ++
libavfilter/vaapi_vpp.c | 4 +
libavfilter/version.h | 4 +-
libavfilter/video.c | 2 +
tests/ref/fate/shortest-sub | 4 +-
11 files changed, 406 insertions(+), 10 deletions(-)
ffmpeg -v trace -hwaccel vaapi -vaapi_device /dev/dri/renderD128 -i ./example_720p30.y4m -vf 'format=nv12|vaapi, hwupload' -c:v h264_vaapi ./out_720p30.264 -y
Successfully opened the file.
Stream mapping:
Stream #0:0 -> #0:0 (rawvideo (native) -> h264 (h264_vaapi))
[vost#0:0/h264_vaapi @ 0x55a86a145c70] Starting thread...
[vf#0:0 @ 0x55a86a104290] Starting thread...
[vist#0:0/rawvideo @ 0x55a86a105cc0] Starting thread...
[in#0/yuv4mpegpipe @ 0x55a86a11d250] Starting thread...
Press [q] to stop, [?] for help
[rawvideo @ 0x55a86a144d40] PACKET SIZE: 1382400, STRIDE: 1920
[AVFilterGraph @ 0x7fa5e0000ff0] Setting 'pix_fmts' to value 'nv12|vaapi'
detected 24 logical cores
[graph 0 input from stream 0:0 @ 0x7fa5e0003ab0] Setting 'video_size' to value '1280x720'
[graph 0 input from stream 0:0 @ 0x7fa5e0003ab0] Setting 'pix_fmt' to value '0'
[graph 0 input from stream 0:0 @ 0x7fa5e0003ab0] Setting 'time_base' to value '1/30'
[graph 0 input from stream 0:0 @ 0x7fa5e0003ab0] Setting 'pixel_aspect' to value '0/1'
[graph 0 input from stream 0:0 @ 0x7fa5e0003ab0] Setting 'colorspace' to value 'unknown'
[graph 0 input from stream 0:0 @ 0x7fa5e0003ab0] Setting 'range' to value 'unknown'
[graph 0 input from stream 0:0 @ 0x7fa5e0003ab0] Setting 'frame_rate' to value '30/1'
[rawvideo @ 0x55a86a144d40] PACKET SIZE: 1382400, STRIDE: 1920
[graph 0 input from stream 0:0 @ 0x7fa5e0003ab0] w:1280 h:720 pixfmt:yuv420p tb:1/30 fr:30/1 sar:0/1 csp:unknown range:unknown
[format @ 0x7fa5e0004390] Setting 'pix_fmts' to value 'vaapi'
[auto_scale_0 @ 0x7fa5e00055d0] w:iw h:ih flags:'' interl:0
[Parsed_format_0 @ 0x7fa5e0003450] auto-inserting filter 'auto_scale_0' between the filter 'graph 0 input from stream 0:0' and the filter 'Parsed_format_0'
[AVFilterGraph @ 0x7fa5e0000ff0] query_formats: 5 queried, 9 merged, 3 already done, 0 delayed
Segmentation fault (core dumped)
--------
-----Original Message-----
From: ffmpeg-devel <ffmpeg-devel-bounces@ffmpeg.org> On Behalf Of Niklas Haas
Sent: Sunday, December 31, 2023 4:51 PM
To: ffmpeg-devel@ffmpeg.org
Subject: Re: [FFmpeg-devel] [PATCH v2 00/15] YUV colorspace filter negotiation
On Sun, 24 Dec 2023 10:35:22 -0800 Niklas Haas <ffmpeg@haasn.xyz> wrote:
> On Wed, 13 Dec 2023 14:11:57 +0100 Niklas Haas <ffmpeg@haasn.xyz> wrote:
> > Split off from my YUVJ removal series. This implements all of the
> > libavfilter changes needed to fully deprecate YUVJ, but does not yet
> > remove YUVJ, nor add an AVCodec API for advertising colorspace support.
> >
> > Update includes all of the feedback that was brought up by Anton.
> > the major change from v1 is that YUV metadata configured on a link
> > is no longer strictly required to be consistent with the frames,
> > simply to avoid breaking users that don't set this link metadata.
> > Effectively, this allows us to have a grace period where the field
> > functions more like a hint than an authoritative field.
>
> Ping, any further feedback? If not, I will merge this sometime after Christmas.
Merged as 1b0ca477..44a33fccd