Zhong Li [Fri, 31 May 2019 00:44:47 +0000 (08:44 +0800)]
lavf/qsv_vpp: add frame format option
1. Currently output format is hard-coded as NV12, thus means
CSC is always done for not NV12 input such as P010.
Follow original input format as default output.
2. Add an option to specify output format.
avcodec/mediacodecdec: try to receive a frame after signaling EOF to the codec
Avoids returning EAGAIN after signaling EOF to the codec in
ff_mediacodec_dec_send() so we can try to receive a frame before
returning in mediacodec_receive_frame().
This helps avoiding an extra round-trip between avcodec_send_frame() and
avcodec_receive_frame() while draining the remaining frames.
U. Artie Eoff [Fri, 7 Jun 2019 21:45:33 +0000 (14:45 -0700)]
vaapi_encode_mjpeg: fix bad component id bug
The compound literals assigned to "components"
only exist within the scope of the if/else
block (thanks Mark Thompson for the better
explanation).
Thus, after this if/else block, "components"
ends up pointing to an arbitrary/undefined
array. With some compilers and depending on
optimization settings, these arbitrary values
may end up being the same value (i.e. 0 with
GNU GCC 9.x). Unfortunately, the GNU GCC
compiler, at least, never prints any warnings
about this.
This patch fixes this issue by assigning the
constant arrays to local variables at function
scope and then pointing "components" to those
as necessary.
Fixes #7915
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
avcodec/h264_parse: Use 64bit for expectedpoc and expected_delta_per_poc_cycle
Fixes: signed integer overflow: -2142516591 + -267814575 cannot be represented in type 'int' Fixes: 14450/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_H264_fuzzer-5716105319940096 Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg Reviewed-by: James Almer <jamrial@gmail.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Ruiling Song [Wed, 15 May 2019 09:54:10 +0000 (17:54 +0800)]
avfilter/vf_gblur: add x86 SIMD optimizations
The horizontal pass get ~2x performance with the patch
under single thread.
Tested overall performance using the command(avx2 enabled):
./ffmpeg -i 1080p.mp4 -vf gblur -f null /dev/null
./ffmpeg -i 1080p.mp4 -vf gblur=threads=1 -f null /dev/null
For single thread, the fps improves from 43 to 60, about 40%.
For multi-thread, the fps improves from 110 to 130, about 20%.
Signed-off-by: Ruiling Song <ruiling.song@intel.com>
READ has already been undefined at this point; it is obviously intended
to undef WRITE.
Furthermore, leb128 (in cbs_av1) was undefined too often and
inconsistently.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Matroska EBML IDs can be only four bytes long maximally, so it is
natural to use uint32_t for them. By doing this and rearranging the
elements of the MatroskaLevel1Element structure, one can reduce the size
of said structure.
Notice that this field is not read via the generic reading process for
EBML_UINT, so one is not forced to use an uint64_t for it.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Xuewei Meng [Thu, 30 May 2019 12:35:17 +0000 (20:35 +0800)]
libavfilter: Add derain filter
Remove the rain in the input image/video by applying the derain
methods based on convolutional neural networks. Training scripts
as well as scripts for model generation are provided in the
repository at https://github.com/XueweiMeng/derain_filter.git.
avcodec/dxv: Check op_offset in dxv_decompress_cocg()
Fixes: signed integer overflow: -2147483648 - 12 cannot be represented in type 'int' Fixes: 14732/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_DXV_fuzzer-5735273129836544 Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
avcodec/diracdec: Fix integer overflow in global_mv()
Fixes: signed integer overflow: 16384 * 196607 cannot be represented in type 'int' Fixes: 14810/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_DIRAC_fuzzer-5091232683917312 Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
avcodec/arbc: Skip tiles in fill_tileX() which are completely outside
Fixes: signed integer overflow: 2052526848 + 147237888 cannot be represented in type 'int' Fixes: 14441/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ARBC_fuzzer-5717632944177152 Fixes: 14453/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ARBC_fuzzer-5739679254577152 Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This removes the use of the nonstandard combined structures, which
generated some warnings with clang and will cause alignment problems
with some parameter buffer types.
Ruiling Song [Tue, 7 May 2019 01:46:33 +0000 (09:46 +0800)]
avfilter/vf_unsharp: enable slice threading
benchmarking with a simple command:
ffmpeg -i 1080p.mp4 -vf unsharp=la=3:ca=3 -an -f null /dev/null
with the patch, the fps increase from 50 to 120 on my local machine (i7-6770HQ).
Signed-off-by: Ruiling Song <ruiling.song@intel.com>
Mark Thompson [Mon, 6 May 2019 14:45:44 +0000 (15:45 +0100)]
hwcontext_qsv: Try to select a matching VAAPI device by default
Tries to find a device backed by the i915 kernel driver and loads the iHD
VAAPI driver to use with it. This reduces confusion on machines with
multiple DRM devices and removes the surprising requirement to set the
LIBVA_DRIVER_NAME environment variable to use libmfx at all.
Mark Thompson [Mon, 6 May 2019 14:41:09 +0000 (15:41 +0100)]
hwcontext_vaapi: Try to create devices via DRM before X11
Opening the device via X11 (DRI2/DRI3) rather than opening a DRM render
node directly is only useful if you intend to use the legacy X11 interop
functions. That's never true for the ffmpeg utility, and a library user
who does want this will likely provide their own display instance rather
than making a new one here.
Mark Thompson [Mon, 6 May 2019 14:43:04 +0000 (15:43 +0100)]
hwcontext_vaapi: Make default DRM device selection more helpful
Iterate over available render devices and pick the first one which looks
usable. Adds an option to specify the name of the kernel driver associated
with the desired device, so that it is possible to select a specific type
of device in a multiple-device system without knowing the card numbering.
For example: -init_hw_device vaapi:,kernel_driver=amdgpu will select only
devices using the "amdgpu" driver (as used with recent AMD graphics cards).
Mark Thompson [Mon, 6 May 2019 14:30:24 +0000 (15:30 +0100)]
ffmpeg_hw: Treat empty device string as no device setting
The implementation will use some default in this case. The empty string
is not a meaningful device for any existing hardware type, and indeed
OpenCL treats it identically to no device already to work around the lack
of this setting on the command line.
Mark Thompson [Thu, 28 Feb 2019 00:38:08 +0000 (00:38 +0000)]
lavfi/vaapi: Improve support for colour properties
Attempts to pick the set of supported colour properties best matching the
input. Output is then set with the same values, except for the colour
matrix which may change when converting between RGB and YUV.
Carl Eugen Hoyos [Fri, 31 May 2019 20:30:46 +0000 (22:30 +0200)]
lavfi/showspatial: Remove two write-only variables.
Fixes two warnings:
libavfilter/avf_showspatial.c:157:26: warning: variable ‘w’ set but not used
libavfilter/avf_showspatial.c:157:23: warning: variable ‘h’ set but not used
Nick Renieris [Thu, 30 May 2019 10:42:42 +0000 (13:42 +0300)]
avcodec/tiff: Recognize DNG/CinemaDNG images
Additionally:
- Renamed TIFF_WHITE_LEVEL to DNG_WHITE_LEVEL since it is specified
in the DNG spec.
- Added/changed some comments to be more precise in differentiating
between TIFF, TIFF/EP and DNG values.
Related to ticket: https://trac.ffmpeg.org/ticket/4364
Signed-off-by: Nick Renieris <velocityra@gmail.com>
Nick Renieris [Thu, 30 May 2019 10:42:41 +0000 (13:42 +0300)]
libavcodec/tiff: Process SubIFDs tag with multiple entries
SubIFDs that were part of more than single-sized "SubIFDs" tags were
being ignored due to existing code ignoring that case.
This patch makes is so the first entry is read, which is not ideal
but enough for some DNG images present in the wild to be decodeable
More specifically, the first SubIFD which we would process with this
patch is the main image and the second one is a second thumbnail,
which is not as important to decode.
In DNG images with the .tiff extension, it solves the issue where
the TIFF thumbnail in IFD 0 was incorrectly parsed (related
confusion: [1]).
Embedded thumbnails for DNG images can still be decoded with the
"-thumbnail" option.
Related to ticket: https://trac.ffmpeg.org/ticket/4364
Linjie Fu [Mon, 15 Apr 2019 13:24:06 +0000 (21:24 +0800)]
lavf/qsvvpp: avoid the double-free when working in sys memory mode
Currently, picref will be freed by calling av_frame_free(&picref) in
submit_frame() in qsvvpp.c when working in system memory mode,and normally it
is freed in filter_frame() in vf_vpp_qsv.c when working in other modes.
Double free happens when working in system memory mode, remove to
fix the memory issue.
cbs_mpeg2: Fix storage type for frame_centre_*_offset
The frame_centre_horizontal/vertical_offset values contained in picture
display extensions are actually signed values (i.e. it is possible to
indicate that the display device should add black bars/pillars).
The files sony-ct3.bs and tcela-6.bits (which are both used in fate
tests for mpeg2_metadata) contain picture display extensions; the former
even contains a negative frame_centre_vertical_offset. Fortunately, the
old code did not damage the picture display extensions when one did a
cycle of reading and writing. For the same reason the fate tests needn't
be updated either.
Furthermore these fields now use the trace output for matrices.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
MPEG-2 contains several elements that mustn't be zero according to the
specifications: horizontal/vertical_size_value, aspect_ratio_information,
frame_rate_code, the quantiser matrices, the colour_description
elements, picture_coding_type, the f_code[r][s] values and
quantiser_scale_code. It is now checked that the invalid values don't
occur.
The colour_description elements are treated specially in this regard:
Given that there are files in the wild which use illegal values for the
colour_description elements (some of them created by mpeg2_metadata),
they will be corrected to the value meaning "unknown" (namely 2) during
reading. This has been done in such a way that trace_headers will
nevertheless report the original value, together with a message about
the fixup.
Furthermore, the trace_headers output of user_data has been beautified.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
projects / ffmpeg / log