Ronald S. Bultje [Sat, 10 Oct 2015 02:35:49 +0000 (22:35 -0400)]
vp9: add itxfm_add eob shortcuts to 10/12bpp functions.
These aren't quite as helpful as the ones in 8bpp, since over there,
we can use pmulhrsw, but here the coefficients have too many bits to
be able to take advantage of pmulhrsw. However, we can still skip
cols for which all coefs are 0, and instead just zero the input data
for the row itx. This helps a few % on overall decoding speed.
Ronald S. Bultje [Mon, 12 Oct 2015 14:16:56 +0000 (10:16 -0400)]
vp9: initial attempt at a idct_idct_4x4 12bpp x86 simd (sse2) impl.
The trouble with this function is that intermediates overflow 31+sign
bits, so I've added some helpers (that will also be used in 10/12bpp
8x8, 16x16 and 32x32) to make that easier, basically emulating a half-
assed pmaddqd using 2xpmaddwd. It's currently sse2-only, if anyone sees
potential in adding ssse3, I'd love to hear it.
On 12 frames of a 444p 12 bits DNxHR sequence, _put function:
C: 78902 decicycles in idct, 262071 runs, 73 skips
avx: 32478 decicycles in idct, 262045 runs, 99 skips
Difference between the 2:
stddev: 0.39 PSNR:104.47 MAXDIFF: 2
This is unavoidable and due to the scale factors used in the x86
version, which cannot match the C ones.
In addition, the trick of adding an initial bias to the input of a
pass can overflow, as the input coefficients are already 15bits,
which is the maximum this function can handle.
Overall, however, the omse on 12 bits samples goes from 0.16916 to
0.16883. Reducing rowshift by 1 improves to 0.0908, but causes
overflows.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Modeled from the prores version. Clips to [0;1023] and is bitexact.
Bitexactness requires to add offsets in different places compared to
prores or C, and makes the function approximately 2% slower.
For 16 frames of a DNxHD 4:2:2 10bits test sequence:
C: 60861 decicycles in idct, 1048205 runs, 371 skips
sse2: 27567 decicycles in idct, 1048216 runs, 360 skips
avx: 26272 decicycles in idct, 1048171 runs, 405 skips
The add version is not implemented, so the corresponding dsp
function is set to NULL to make it clear in a code executing it.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
When the input of a pass has 15 or 16 bits of precision (in particular
the column pass), the addition of a bias to W4 may lead to overflows
in the input to pmaddwd.
This requires postponing the adding of the bias to after the first
butterfly. To do so, the fact that m15, unused although zeroed, is
exploited. In case the pass is safe, an address can be directly used,
and the number of xmm regs can be decreased. Otherwise, the 32bits bias
is loaded into it.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
omse goes from 0.03060703 (which fails for dct-test) to 0.01663750.
This also actually improve the error of decoding the sample generated
by fate-vsynth3-dnxhd1080i-10bit using simple_idct10 to FAANI, which
goes (when resampled to yuv422p) from:
stddev: 0.06 PSNR: 72.28 MAXDIFF: 1
to identical.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Includes escapes that should now be supported and a few features not yet
fully supported, like comments, regions, classes, ruby, and lang.
All were tested with https://quuz.org/webvtt/ for validation, except
regions because the validator doesn't support them yet, and I couldn't
find any other way to validate WebVTT.
aacenc: add support for changing options based on a profile
This commit adds the ability for a profile to set the default
options, as well as for the user to override such options
by simply stating them in the command line while still keeping
the same profile, as long as those options are still permitted by
the profile.
Example: setting the profile to aac_low (the default) will turn
PNS and IS on. They can be disabled by -aac_pns 0 and -aac_is 0,
respectively. Turning on -aac_pred 1 will cause the profile to be
elevated to aac_main, as long as no options forbidding aac_main
have been entered (like AAC-LTP, which will be pushed soon).
A useful feature is that by setting the profile to mpeg2_aac_low,
all MPEG4 features will be disabled and if the user tries to enable
them then the program will exit with an error. This profile is
signalled with the same bitstream as aac_low (MPEG4) but some devices
and decoders will fail if any MPEG4 features have been enabled.
aacenc: add support for encoding 7.1 channel audio
This commit implements support for 7.1 channel audio. There's no
more predefined bitstream channel mappings so going beyond 8 channels
(and 7 channels exactly) will require programmable channel elements,
which is already underway.
Claudio Freire [Mon, 12 Oct 2015 06:56:22 +0000 (03:56 -0300)]
AAC encoder: memoize quantize_band_cost
The bulk of calls to quantize_band_cost are replaced
by a call to a version that memoizes, greatly improving
performance, since during coefficient search there is
a great deal of repeat work.
Memoization cannot always be applied, so do this in a
different function, and leave the original as-is.
Claudio Freire [Sun, 11 Oct 2015 20:29:50 +0000 (17:29 -0300)]
AAC encoder: Extensive improvements
This finalizes merging of the work in the patches in ticket #2686.
Improvements to twoloop and RC logic are extensive.
The non-exhaustive list of twoloop improvments includes:
- Tweaks to distortion limits on the RD optimization phase of twoloop
- Deeper search in twoloop
- PNS information marking to let twoloop decide when to use it
(turned out having the decision made separately wasn't working)
- Tonal band detection and priorization
- Better band energy conservation rules
- Strict hole avoidance
For rate control:
- Use psymodel's bit allocation to allow proper use of the bit
reservoir. Don't work against the bit reservoir by moving lambda
in the opposite direction when psymodel decides to allocate more/less
bits to a frame.
- Retry the encode if the effective rate lies outside a reasonable
margin of psymodel's allocation or the selected ABR.
- Log average lambda at the end. Useful info for everyone, but especially
for tuning of the various encoder constants that relate to lambda
feedback.
Psy:
- Do not apply lowpass with a FIR filter, instead just let the coder
zero bands above the cutoff. The FIR filter induces group delay,
and while zeroing bands causes ripple, it's lost in the quantization
noise.
- Experimental VBR bit allocation code
- Tweak automatic lowpass filter threshold to maximize audio bandwidth
at all bitrates while still providing acceptable, stable quality.
I/S:
- Phase decision fixes. Unrelated to #2686, but the bugs only surfaced
when the merge was finalized. Measure I/S band energy accounting for
phase, and prevent I/S and M/S from being applied both.
PNS:
- Avoid marking short bands with PNS when they're part of a window
group in which there's a large variation of energy from one window
to the next. PNS can't preserve those and the effect is extremely
noticeable.
M/S:
- Implement BMLD protection similar to the specified in
ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision
doesn't conform to section 6.1, a different method had to be
implemented, but should provide equivalent protection.
- Move the decision logic closer to the method specified in
ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically,
make sure M/S needs less bits than dual stereo.
- Don't apply M/S in bands that are using I/S
Now, this of course needed adjustments in the compare targets and
fuzz factors of the AAC encoder's fate tests, but if wondering why
the targets go up (more distortion), consider the previous coder
was using too many bits on LF content (far more than required by
psy), and thus those signals will now be more distorted, not less.
The extra distortion isn't audible though, I carried extensive
ABX testing to make sure.
A very similar patch was also extensively tested by Kamendo2 in
the context of #2686.
FFmpeg already tests for this case in configure_output_filter() and printed a
clearer error message
example:
./ffmpeg -f lavfi -i color -f lavfi -i color -filter_complex "[1]null[x],[0][1]overlay" -f null -
before the merge / after the revert:
Filter null has a unconnected output
after the merge / before the revert:
Output pad "default" with type video of the filter instance "Parsed_null_0" of null not connected to any destination
Error configuring complex filters.
Invalid argument
avutil/mathematics: speed up av_gcd by using Stein's binary GCD algorithm
This uses Stein's binary GCD algorithm:
https://en.wikipedia.org/wiki/Binary_GCD_algorithm
to get a roughly 4x speedup over Euclidean GCD on standard architectures
with a compiler intrinsic for ctzll, and a roughly 2x speedup otherwise.
At the moment, the compiler intrinsic is used on GCC and Clang due to
its easy availability.
Quick note regarding overflow: yes, subtractions on int64_t can, but the
llabs takes care of that. The llabs is also guaranteed to be safe, with
no annoying INT64_MIN business since INT64_MIN being a power of 2, is
shifted down before being sent to llabs.
The binary GCD needs ff_ctzll, an extension of ff_ctz for long long (int64_t). On
GCC, this is provided by a built-in. On Microsoft, there is a
BitScanForward64 analog of BitScanForward that should work; but I can't confirm.
Apparently it is not available on 32 bit builds; so this may or may not
work correctly. On Intel, per the documentation there is only an
intrinsic for _bit_scan_forward and people have posted on forums
regarding _bit_scan_forward64, but often their documentation is
woeful. Again, I don't have it, so I can't test.
As such, to be safe, for now only the GCC/Clang intrinsic is added, the rest
use a compiled version based on the De-Bruijn method of Leiserson et al:
http://supertech.csail.mit.edu/papers/debruijn.pdf.
Tested with FATE, sample benchmark (x86-64, GCC 5.2.0, Haswell)
with a START_TIMER and STOP_TIMER in libavutil/rationsl.c, followed by a
make fate.
aac-am00_88.err:
builtin:
714 decicycles in av_gcd, 4095 runs, 1 skips
de-bruijn:
1440 decicycles in av_gcd, 4096 runs, 0 skips
previous:
2889 decicycles in av_gcd, 4096 runs, 0 skips
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
doc/resampler, swresample/options: use proper capitalization
Proper names should be capitalized in all user facing API as far as
possible. The option names themselves have not been changed since:
1. We consistently keep option names in lower case.
2. Changing them would break existing scripts.
3. I suspect that we want to be similar to Sox and its relevant options.
The converse is also true: improper names should not be capitalized
generally.
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
During a build, a lot of *.o.-hash files are created - had not noticed
this as they are usually dumped in tmpfs on Linux. However, they
sometimes are present during a long build in the project directory, making it
annoying to commit while the project is being built.
These have been observed with Clang, -fsanitize-undefined on Arch Linux,
though other configurations may also generate such temporaries.
The solution here is on lines with the Linux kernel's .gitignore:
https://github.com/torvalds/linux/blob/master/.gitignore.
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
doc/scaler, swscale/options: use proper capitalization
Proper names should be capitalized in all user facing API as far as
possible. The option names themselves have not been changed since:
1. We consistently keep option names in lower case.
2. Changing them would break existing scripts.
The converse is also true: improper names should not be capitalized
generally.
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This should fix the undefined behavior reported in:
https://trac.ffmpeg.org/ticket/4727.
I can reproduce this at runtime: simply stick in an abort call in
asym_quant to check if c < 0 and run FATE. I don't know ac3 so I can't
confirm if negative coefficients are intentional, but at the moment they
clearly are according to FATE.
This resolves the undefined behavior. Tested with FATE.
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
赵宇龙 [Fri, 9 Oct 2015 05:59:56 +0000 (13:59 +0800)]
h264_mp4toannexb: fix the pps offset when there are more than one sps in avcc
the pps offset is used to locate pps in the spspps_buf; however, the
current calc method is wrong because it is the offset of the original
avctx->extradata;
when there is only one sps in the avcc; the value is correct by
coincidence, however, it will fail in avcc with multi sps
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This fixes a warning observed on Clang 3.7:
"warning: attribute 'deprecated' is ignored, place it after "struct" to apply attribute to type declaration [-Wignored-attributes]"
and thus enables deprecation warning for the relevant struct.
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com> Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
Currently only 2 profiles are evaluated because they are the only 2
with distributed test sequences.
- CID 1260: YUV 4:2:2 10 bits with block-adaptive interlace coding,
from ticket 4876;
- CID 1270: YUV 4:4:4 10 bits (HR), 1920x839, from ticket 4581.
They were generated from the ticket sequences by running the
following kind of command-line;
ffmpeg -i $INPUT -an -sn -vcodec copy -vframes 1 -y $OUTPUT.mov
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
On lines 1633,1634 FFABS(pts) is performed. However, if av_stream_get_end_pts
returns AV_NOPTS_VALUE always, pts remains stuck at INT64_MIN, leading
to undefined behavior on FFABS.
One could conceive of a solution using FFNABS. However, such a solution
has to deal with the implementation defined rounding of integer division
with at least one negative operand in ANSI C89. C99 forces truncation to
zero, but I am not sure that all of our platforms compile with full C99
support, and in particular whether we can safely assume a fixed
rounding behavior across all platforms.
This solution is simple, and I doubt changing INT64_MIN to INT64_MIN + 1
has any practical loss - if it is stuck at its initial value, the stream
is messed up anyway.
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
projects / ffmpeg / log