Both of these two structures were first available with X264_BUILD
163, so make relevant functionality conditional on the version
being at least such.
Keep handle_side_data available in all cases as this way X264_init
does not require additional version based conditions within it.
Finally, add a FATE test which verifies that pass-through of the
MDCV/CLL side data is working during encoding.
These two were added in 28e23d7f348c78d49a726c7469f9d4e38edec341
and 3558c1f2e97455e0b89edef31b9a72ab7fa30550 for version 0.9.0 of
SVT-AV1, which is also our minimum requirement right now.
In other words, no additional version limiting conditions seem
to be required.
Additionally, add a FATE test which verifies that pass-through of
the MDCV/CLL side data is working during encoding.
The second part of this condition is intended to check whether the
current quantisation group is in the first CTU column of the current
tile. The issue is that ctb_to_col_bd gives the x-ordinate of the first
column of the current tile *in CTUs*, while xQg gives the x-ordinate of
the quantisation group *in samples*. Rectify this by shifting xQg by
ctb_log2_size to get xQg in CTUs before comparing.
Fixes FFVVC issues #201 and #203.
In MPEG-2 user data, there can be different types of Closed Captions
formats embedded (A53, SCTE-20, or DVD). The current behavior of the
CC extraction code in the MPEG-2 decoder is to not be aware of
multiple formats if multiple exist, therefore allowing one format
to overwrite the other during the extraction process since the CC
extraction shares one output buffer for the normalized bytes.
This causes sources that have two CC formats to produce flawed output.
There exist real-world samples which contain both A53 and SCTE-20 captions
in the same MPEG-2 stream, and that manifest this problem. Example of symptom:
THANK YOU (expected) --> THTHANANK K YOYOUU (actual)
The solution is to pick only the first CC substream observed with valid bytes,
and ignore the other types. Additionally, provide an option for users
to manually "force" a type in the event that this matters for a particular
source.
Signed-off-by: Marth64 <marth64@proxyid.net>
When all cached frames are drained, the output mfxSyncPoint pointer is
NULL and MFX_ERR_MORE_DATA is returned, hence needn't print warning for
this expected behavior, otherwise the user might think the output from
qsv decoders are wrong.
Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>
ff_aac_coder_init_mips() modifies a static const structure of
function pointers. This will crash if the binary uses relro
and is a data race in any case.
Furthermore it points to a maintainability issue: The
AACCoefficientsEncoder structures have been constified
in commit fd9212f2ed,
a Libav commit merged in 318778de9e.
Libav did not have the MIPS-specific AAC code and so this was
fine for them; yet FFmpeg had them, but this was not recognized.
Commit 75a099fc73 points to another
maintainability issue: Contrary to ordinary DSP code, this code
here is way more complex and needs to be constantly kept in sync
with the ordinary code which it mimicks and replaces. Said commit
is the only commit actually changing aaccoder.c in the last few
years and the same change has not been performed for the MIPS
clone; before that, it even happened several times that the mips
code was broken due to changes of the generic code (see commits
97437bd17a and
de262d018d or
860dbe0275 or
933309a6ca or
b65ffa316e). This might even lead
to scenarios where someone changing non-dsp aacenc code would
have to modify mips inline asm in order to keep them in sync.
This is obviously a significant burden (if the AAC encoder were
actively developed).
Finally, the code does not even compile here due to errors like
"Error: float register should be even, was 1".
Reviewed-by: Lynne <dev@lynne.ee>
Reviewed-by: Jean-Baptiste Kempf <jb@videolan.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
All versions of MSVC that support C11 (namely >= v19.27)
also support the restrict keyword, therefore av_restrict
is no longer necessary since 75697836b1.
Reviewed-by: Martin Storsjö <martin@martin.st>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
h_subpel_filters_inner[i] and h_subpel_filters_outer[i / 2]
belong together and the former allows the range 0..6,
so the latter needs to support 0..3. But it has only three
elements. Add another one.
The value for the last element has been guesstimated
from subpel_filters in libavcodec/vp8dsp.c.
This is also intended to fix FATE-failures with UBSan here:
https://fate.ffmpeg.org/report.cgi?time=20240312011016&slot=ppc-linux-gcc-13.2-ubsan-altivec-qemu
Tested-by: Sean McGovern <gseanmcg@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is accessed via AV_RN64A in ff_simple_idct_put_int32_10bit().
Should fix the UBSan failures in the mpeg4-simple-studio-profile
test here:
https://fate.ffmpeg.org/report.cgi?time=20240312011016&slot=ppc-linux-gcc-13.2-ubsan-altivec-qemu
Reviewed-by: Kieran Kunhya <kierank@obe.tv>
Tested-by: Sean McGovern <gseanmcg@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
HAVE_FAST_UNALIGNED being true does not imply that
one can simply read from any pointer via *(long*).
It is undefined behaviour in case the pointer is not
sufficiently aligned; and even if it is, it is (likely)
a violation of the effective-type rules. Fix both
of these by using the appropriate AV_[RW]N macros.
Also, the current code used sizeof(long) as if this
were the CPU's native arithmetic size, but this is
not true on 64bit Windows. This has been fixed, too.
This affected huffyuv FATE-tests.
Tested-by: Sean McGovern <gseanmcg@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
They are incorrect according to [1]. They also share keys with valid
entries, so that it is unspecified which entry bsearch returns
in this case. Fix this by removing the incorrect values.
[1]: https://www.earthdata.nasa.gov/s3fs-public/imported/19-008r4.pdf
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Instead store all the strings in one continugous string
(with internal \0) and use offsets to access the actual
substrings. This replaces the pointers to the strings
and therefore avoids relocations (and on x64, it actually
shrinks TiffGeoTagNameType by reusing padding to store
the offset field).
This saves 720B of .data.rel.ro and 1080B of .rela.dyn
(containing the relocation records) here while increasing
.rodata by 384B.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The first 32 elements of each row were correct, while the
last 16 were scrambled.
This hasn't been noticed, because the checkasm test erroneously
only checked half of the output (for 8 bit functions), and
apparently none of the samples as part of "fate-hevc" seem to
trigger this specific function.
Signed-off-by: J. Dekker <jdek@itanimul.li>
This is easily possible with an X macro.
Using an enum for the offsets also allows to remove
two arrays which are not really needed and will typically
be optimized away by the compiler: The first just exists
to count the number of syntax elements*, the second one
exists to get offset[CONSTANT]. These constants were
of type enum SyntaxElement and this enum was only used
in hevc_cabac.c (although it was declared in hevcdec.h);
it is now no longer needed at all and has therefore been
removed.
The first of these arrays led to a warning from Clang
which is fixed by this commit:
warning: variable 'num_bins_in_se' is not needed and will
not be emitted [-Wunneeded-internal-declaration]
*: One could also just added a trailing SYNTAX_ELEMENT_NB
to the SyntaxElement enum for this purpose.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Add an AVOption to the libjxl encoder wrapper, which exposes the flag
uses_original_profile in libjxl. For highly unusual ICC profiles where
the target needs to stay in the original space, this can be useful.
Signed-off-by: Leo Izen <leo.izen@gmail.com>
No need to use the generic ff_tget() when we know the type.
Reviewed-by: Stefano Sabatini <stefasab@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
For unknown geokey values, get_geokey_val() returns
"Unknown-%d" with val being used for %d. This string
is allocated and therefore all the known geokey values
(static strings) are strdup'ed. In case this fails
it is either ignored or treated as "Unknown-%d".
(Furthermore it is possible to call av_strdup(NULL),
although this is not documented to be legal.)
This commit changes this by only returning the static strings
in get_geokey_val(); the unknown handling and strdup'ing
is moved out of it.
Reviewed-by: Stefano Sabatini <stefasab@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This is thankfully passed through verbatim by libdav1d, so we can parse
it in our own code.
In theory, taking the DV profile from the packet-level configuration
struct is redundant since there is currently only one possible DV level
for AV1 (and all others would fail parsing), but this marginally
future-proofs it against possible new AV1-specific profiles being added
in the future.
This currently builds files in the libavcodec/x86/{vvc,h26x}
subdirectories, which is somewhat unexpected when building for
another architecture than x86.
The regular arch subdirectories are handled with
-include $(SRC_PATH)/$(1)/$(ARCH)/Makefile
in the toplevel Makefile. Switch this to a similar optional
inclusion, using $(ARCH).
Signed-off-by: Martin Storsjö <martin@martin.st>
In some builds, the following object files could be left behind
after make clean:
./libavfilter/metal/utils.o
./libavfilter/metal/vf_yadif_videotoolbox.metallib.o
./libavcodec/x86/h26x/h2656dsp.o
./libavcodec/neon/mpegvideo.o
./ffbuild/bin2c_host.o
Fixes: http://trac.ffmpeg.org/ticket/10895
Signed-off-by: Martin Storsjö <martin@martin.st>
According to kostya, it is not based on Wassermans encoder
CC: Kostya Shishkov <kostya.shishkov@gmail.com>
CC: Anatoliy Wasserman <anatoliy.wasserman@yandex.ru>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
