Now coefficients are written to a buffer and are then dequantized by the
new SIMD dequantization functions. For the lower bands without enough
coefficients to fill a register (and hence they overwrite) the C version
of the dequantization function is used.
The buffer is per-thread and will be realloc'd if anything changes.
This prevents regressions and having to limit slice size.
Signed-off-by: Rostislav Pehlivanov <rpehlivanov@obe.tv>
Prevents having to have random magic values in the decoder and a
separate macro in the encoder.
Signed-off-by: Rostislav Pehlivanov <rpehlivanov@obe.tv>
In preparation for the following commits, this commit simplifies the
coefficient parsing and dequantization function. It was needlessly
inlined without much performance gain.
Signed-off-by: Rostislav Pehlivanov <rpehlivanov@obe.tv>
This reverts commit c67aa7f287e791b2c02c0c6471336c8adc88020e, reversing
changes made to 46bd0e8e0dfb4b295e8fbe6854904eccb0cb5c82.
The merge should have been a noop as the support was added in
601c2388549ff395bc99383029e92c2219152af0
* commit '61f168ae348f94f39e7afc6971654455a5de0e4d':
h264: factor out setting the parameter sets for a frame
Michael's changes on top of the merge undo parts of the original diff
that are not factorization:
"The call point is left where it was before. Such a change should be in
a separate commit and has multiple issues, one being null pointer
dereferences the other is that some safety checks would become
conditional.
I tried to split the PPS init between the new and old functions
similarly to the SPS code."
Merged-by: Clément Bœsch <u@pkh.me>
Merged-by: Michael Niedermayer <michael@niedermayer.cc>
a gray channel in exr, is named Y
we admit that the file need to be interpreted as gray
only if no other channel match (except alpha)
to manage RGB and Y in the color conversion part of decode_block,
the color processing is now made with a for loop.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
H264ParamSets has its SPS/PPS stored raw (SODB) and needs to be
converted to NAL units before sending them to MediaCodec.
This patch adds the missing convertion of the SPS/PPS from SOBP to RBSP
which makes the resulting NAL units correct.
Fixes codec initialization on Nexus 4 and Nexus 7.
The default value of -1 indicates that ffmpeg should determine the channel
mapping automatically, which was the behavior before this commit.
Unless the -mapping_family argument is provided, behavior is unchanged.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes use of freed memory
Should fix valgrind failures of fate-h264-skip-nointra
Found-by: logan
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
* commit 'd1f539c97e04e7cebecaf6916c5064f243d39fcf':
h264: merge the two reinit blocks in slice_header_parse()
Merged-by: Clément Bœsch <clement@stupeflix.com>
* commit '3fba16ecd978d5bed338b8da643c3435e62b3437':
h264: factor starting a new field out of parsing the slice header
Merged-by: Clément Bœsch <clement@stupeflix.com>
fix ticket #5674
the size of data to process in piz_uncompress, is now calc
using the pixel type of each channel.
the data reorganization, alos take care about the size of
each channel
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
In order to be able to make SPS const in H264ParamSets,
modify decode_scaling_matrices so that it returns if the scaling
matrix are present in the SPS, instead of altering the input SPS
structure.
