This patch modifies the encode frame function to
retry encoding the frame when the resulting bit count
is too far off target, but only adjusting lambda
in small, incremental step. It also makes the logic
more conservative - otherwise it will contend with
bit reservoir-related variations in bit allocation,
and result in artifacts when frame have to be truncated
(usually at high bit rates transitioning from low
complexity to high complexity).
Reduces the number of times the vbv retry code is used and should have no
effect on quality
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
If cmd_pos is broken, this would just keep accumulating packets in the
reassembly buffer, until it fails and flushes the buffer on overflow.
Since packets are usually rather small, this will take a lot of subtitle
packets. The perceived effect is that subtitles are not displayed
anymore after the faulty packet was passed to the decoder.
I'm not terribly sure about this, but on the other hand this code is
active only when fragmented packets need to be reassembled.
Fixes sample file in trac issue #4872.
Assuming the first and second packets are partial, this would append the
reassembly buffer (ctx->buf) to itself with the second
append_to_cached_buf() call, because buf is set to ctx->buf.
I do not know a valid sample file which triggers this, and do not know
if packets can be split into more than 2 sub-packets, but it triggered
with a (differently) broken sample file in trac issue #4872.
With the move of some functions into templates
in aaccoder_twoloop.h and aaccoder_trellis.h,
make checkheaders started failing. Add them to
SKIPHEADERS as should be.
This resolves implementation defined behavior, and also silences -Wabsolute-value in clang 3.5+.
Moreover, the generated asm is identical to before modulo nop padding.
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
* commit 'e3d4784eb31b3ea4a97f2d4c698a75fab9bf3d86':
d3d11va: WindowsPhone requires a mutex around ID3D11VideoContext
Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>
* commit '777885983533235ccda5145f96317fc8cd0a18ab':
dcadec: set channel layout in a separate function
Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>
* commit '971177f751a6e2931232accceab438bce277bde8':
dcadec: scan for extensions in a separate function
Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>
This patch refactors the AAC coders to reuse code
between the MIPS port and the regular, portable C code.
There were two main functions that had to use
hand-optimized versions of quantization code:
- search_for_quantizers_twoloop
- codebook_trellis_rate
Those two were split into their own template header
files so they can be inlined inside both the MIPS port
and the generic code. In each context, they'll link
to their specialized implementations, and thus be
optimized by the compiler.
This approach I believe is better than maintaining
several copies of each function. As past experience has
proven, having to keep those in sync was error prone.
In this way, they will remain in sync by default.
Also, an implementation of the dequantized output
argument for the optimized quantize_and_encode
functions is included in the patch. While the current
implementation of search_for_pred still isn't using
it, future iterations of main prediction probably will.
It should not imply any measurable performance hit while
not being used.
Failing earlier causes the context to be insufficiently initialized which
can break decoding future frames with threads
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
* commit '95a41311ac3a44773cc4dc407408aca35b1f8e26':
jpeg2000: Factor out band stepsize initialization
Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>
* commit '41bcc3d15204f290400ba02e4e8f87fc07bcc00e':
jpeg2000: Split codeblock decoding from the main tile decoding
Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>
* commit '5d14cf199990cd378904a2618b5c72c4b02290f6':
mpegvideo: Make sure mpegutils.h is included where needed
Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>
* commit '525f58977c93e189fda49a5c4928feaf4d89fac6':
mpegvideo: Move macros to more appropriate headers
Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>
This fixes cases where the shifted number is 64, but we shifted non-
zero numbers away in the shift. The change makes behaviour consistent
with libvpx.
Code in aaccoder_mips.c was not synced with changes in aaccoder.c for
some time.
That was cause for some fate-aac tests failing.
This patch fixes the problems.
Optimizations disabled in 933309a are enabled again.
Signed-off-by: Nedeljko Babic <nedeljko.babic@imgtec.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
"int" is useful in testing because provides accurate results across
different plaftforms, so remove it from the scheduled FF_API_UNUSED_MEMBERS
deprecation.
Deprecate the now unused option, but temporarily retain the capability
to disable the now default behaviour.
Mention this change in the AVPacket documentation.
Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
This also drops setting the frame->pts field. This is usually not set by
decoders, so this would be an inconsistency that's at worst a danger to
the API user.
It appears the buffer->dts field is normally not set by the MMAL
decoder, so don't use it. If it's ever going to be set by MMAL, we
don't know whether the value will be what we want.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
The generic code in utils.c sets the AVFrame.pkt_dts field from the
packet it was supposedly decoded. This does not have to be true for a
fully asynchronous decoder like mmaldec. It could be overwritten with an
incorrect value. Even if the decoder doesn't determine the DTS (but sets
it to AV_NOPTS_VALUE), it's impossible to determine a correct value in
utils.c.
Decoders can now be marked with FF_CODEC_CAP_SETS_PKT_DTS, in which case
utils.c won't overwrite the field. The decoders are expected to set this
field (even if they only set it to AV_NOPTS_VALUE).
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
This MMAL feature fills in missing timestamps from the framerate set on
the input port. This is generally unwanted, since libavcodec decoders
merely pass through timestamps without ever "fixing" them. The framerate
is also unknown, and even the timebase doesn't have to be set.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Don't try to do a blocking wait for MMAL output if we haven't even sent
a single real packet, but only flush packets. Obviously we can't expect
to get anything back.
Additionally, don't send a flush packet to MMAL in the same case. It
appears the MMAL decoder will sometimes hang in mmal_vc_port_disable()
(called from ffmmal_close_decoder()), waiting for a reply from the GPU
which never arrives. Either MMAL disallows sending flush packets without
preceding real data, or it's a MMAL bug.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
I can't come up with a nice way to handle this. It's hard to keep the
lock-stepped input/output in this case. You can't predict whether the
MMAL decoder will output a picture (because it's asynchronous), so
you have to assume in general that any packet could produce 0 or 1
frames. You can't continue to write input packets to the decoder,
because then you might get too many output frames, which you can't
get rid of because the lavc decoding API does not allow the decoder
to return an output frame without consuming an input frame (except
when flushing).
The ideal fix is a M:N decoding API (preferably asynchronous), which
would make this code potentially much cleaner. For now, this hack
will do.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Since TNS was fixed with the recent commits retweak the values
so it's more frequently used.
Still not enabled by default yet, though it's possible that it
will be made enabled by default in the near future.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
This commit was made possible with the earlier commits since the
new quantization method basically means we're working always with
unsigned values. The specifications mention to use compression when
the first 2 bits are identical but they didn't mention if this should
happen before or after the conversion to signed values. Actually
they said nothing about conversion to signed values.
With this commit, coefficient compression usually always happens
which saves a lot of space, especially at extremely low bitrates
and doesn't change the quality at all.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
This finally (and again) gets rid of basically everything the
specifications say about how TNS should be done. The main
problem used to be that a single filter was used for all
coefficients which despite being explicitly recommended by
the specifications usually sounds wrong, therefore it's
a corner case in the current TNS implementation.
This commit also changes the coefficient bit size, as apparently
it's better to use lower precision in case the windows are eight
short. This is apparently what fdk_aac uses, looking at the bit
stream and makes sense. Also the order when 8 SHORT windows happen
is important as 7 was too much and according to PSNR was worse
while 5 is just about correct.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
* commit '5788623d29c3e806a7879210986110aced758dc2':
jpeg2000: Split codeblock decoding from the main tile decoding
Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>
* commit 'db53a2306f62f05faa67e6f3c60ee55a9b8e4776':
jpeg2000: Do not warn about known and skippable markers
Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>
Makes more sense as users usually set the -cutoff option
to low pass filter the signal. The encoder will still over
shoot slightly when encoding normal coefficients however
that's normal.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Signed-off-by: Kevin Wheatley <kevin.j.wheatley@gmail.com>
Reviewed-by: Hendrik Leppkes <h.leppkes@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Kevin Wheatley <kevin.j.wheatley@gmail.com>
Reviewed-by: Hendrik Leppkes <h.leppkes@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Also disable the mmx/iwht optimization when the bitexact flag is set.
With synthetically coded coefficients (i.e. these that lead to a
residual well outside the [-255,255] range), our optimizations will
overflow. It doesn't make sense to fix the overflows, since they can
only occur on synthetic input, not on real fwht-generated input. Thus,
add a bitexact flag that disables this optimization.
File libopenh264enc.c has been modified so that the encoder uses av_log()
to log messages (error, warning, info, etc.) instead of logging them
directly to stderr. At the time the encoder is created, the current
ffmpeg log level is mapped to an equivalent libopenh264 log level. This
log level, and a message logging function that invokes av_log() to
actually log messages, are then set on the encoder.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This commit changes a few things about the noise substitution
logic:
- Brings back the quantization factor (reduced to 3) during
scalefactor index calculations.
- Rejects any zeroed bands. They should be inaudiable and it's
a waste transmitting the scalefactor indices for these.
- Uses swb_offsets instead of incrementing a 'start' with every
window group size.
- Rejects all PNS during short windows.
Overall improves quality. There was a plan to use the lfg system
to create the random numbers instead of using whatever the decoder
uses but for now this works fine. Entropy is far from important here.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
* commit '2268db2cd052674fde55c7d48b7a5098ce89b4ba':
lavu: Drop the {minus,plus}1 suffix from AVComponentDescriptor fields
Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>
Based on a patch by wm4.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Reviewed-by: Hendrik Leppkes <h.leppkes@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Do not make many assumption on the dimension of the slices and just
try to decode additional lines if there is enough data left.
Decodes all the samples kindly provided by ultramage.
This commit once again improves the PNS implementation by scaling the
thresholds with frequency. The thresholds get looser as the frequency
increases since higher frequencies are basically noise to human ears.
Also, this introduces quantization error correction for PNS. Should
the error be too much, no PNS will be used. The energy_ratio is used
to regulate the actual encoded PNS energy: if the generated PNS
energy is higher than the energy from the psy system, energy_ratio
is used to correct it so that hopefully once requantized and
transmitted the value in the decoder will be closer to what the
encoder has.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
This was an oversight when the IS system was being first implemented.
The ener01 part was largely a result of trial and error and the fact
that the sum of coef0 and coef1 could result in a zero was
overlooked. Once ener01 turns to zero it's used to divide the left
channel energy which doesn't turn out so well as it fills IS[]
with -nan's and inf's which in turn confused the quantize_band_cost.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
This commit rewrites the PNS implementation and significantly
improves sonic quality.
The previous implementation marked an incredibly big amount
of SFBs to predict when there was no need for this and this
resulted in quite a large amount of artifacts. Also the
quantization was incorrect (av_clip(4+log2f(...))) which
led to 3x the intensity for PNS values leading to even more
artifacts.
This commit rewrites the PNS search function and introduces
a major change: the PNS values are synthesized and are compared
to the current coefficients in addition to passing through
the revised checks to see whether PNS can be used.
This decreases distortions and makes the current PNS implementation
mainly focused on replacing any low-power non-zero bands as well
as adding any zeroed bands back.
The current encoder's performance is enough (especially with
IS) so PNS isn't really required except to fill in the occasional
few bands as well as extend any zeroed high frequency, so this
combination which is already enabled by default works
to get as much quality as it can within the bits allowed.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Since PNS generates coefficients it doesn't make sense to send
the predicted ones as well. Also the specifications explicitly
state to disable right channel IS predictors.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
It's better to trust that the coefficients generated will be
closer than the coefficients derived, and the new PNS implementation
makes sure that this happens.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
The specifications explicitly state to use roundf() which
also rounds half-integer values away from zero.
This does fix a few IS artifacts.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
