Converting explicit avio_flush() calls helps us to buffer more data and avoid
flushing the IO context too often which causes reduced IO throughput for
non-streamed file output.
The user can control FLUSH_POINT flushing behaviour using the -flush_packets
option, the default typically means to flush unless a non-streamed file output
is used, so this change should have no adverse effect on streaming even if it
is assumed that after an avio_flush() the output buffer is clean so small
seekbacks within the output buffer will work even when the IO context is not
seekable.
Signed-off-by: Marton Balint <cus@passwd.hu>
These instances are simply redundant or present because avio_flush() used to be
required before doing a seekback. That is no longer the case, aviobuf code does
the flush automatically on seek.
This only affects code which is either disabled for streaming IO contexts or
does no seekbacks after the flush, so this change should have no adverse effect
on streaming.
Signed-off-by: Marton Balint <cus@passwd.hu>
Removing explicit avio_flush() calls helps us to buffer more data and avoid
flushing the IO context too often which causes reduced IO throughput for
non-streamed file output.
The user can control flushing behaviour at the end of every packet using the
-flush_packets option, the default typically means to flush unless a
non-streamed file output is used.
Therefore this change should have no adverse effect on streaming, even if it is
assumed that a new packet has a clean buffer so small seekbacks within the
output buffer work even when the IO context is not seekable.
Signed-off-by: Marton Balint <cus@passwd.hu>
To make it consistent with other muxers.
The user can still control the generic flushing behaviour after write_header
(same way as after packets) using the -flush_packets option, the default
typically means to flush unless a non-streamed file output is used.
Therefore this change should have no adverse effect on streaming, even if it is
assumed that the first packet has a clean buffer, so small seekbacks within the
output buffer work even when the IO context is not seekable.
Signed-off-by: Marton Balint <cus@passwd.hu>
The following is a python script to halve the value of the gray
image. It demos how to setup and execute dnn model with python+tensorflow.
It also generates .pb file which will be used by ffmpeg.
import tensorflow as tf
import numpy as np
from skimage import color
from skimage import io
in_img = io.imread('input.jpg')
in_img = color.rgb2gray(in_img)
io.imsave('ori_gray.jpg', np.squeeze(in_img))
in_data = np.expand_dims(in_img, axis=0)
in_data = np.expand_dims(in_data, axis=3)
filter_data = np.array([0.5]).reshape(1,1,1,1).astype(np.float32)
filter = tf.Variable(filter_data)
x = tf.placeholder(tf.float32, shape=[1, None, None, 1], name='dnn_in')
y = tf.nn.conv2d(x, filter, strides=[1, 1, 1, 1], padding='VALID', name='dnn_out')
sess=tf.Session()
sess.run(tf.global_variables_initializer())
graph_def = tf.graph_util.convert_variables_to_constants(sess, sess.graph_def, ['dnn_out'])
tf.train.write_graph(graph_def, '.', 'halve_gray_float.pb', as_text=False)
print("halve_gray_float.pb generated, please use \
path_to_ffmpeg/tools/python/convert.py to generate halve_gray_float.model\n")
output = sess.run(y, feed_dict={x: in_data})
output = output * 255.0
output = output.astype(np.uint8)
io.imsave("out.jpg", np.squeeze(output))
To do the same thing with ffmpeg:
- generate halve_gray_float.pb with the above script
- generate halve_gray_float.model with tools/python/convert.py
- try with following commands
./ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.model:input=dnn_in:output=dnn_out:dnn_backend=native out.native.png
./ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.pb:input=dnn_in:output=dnn_out:dnn_backend=tensorflow out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
do not request AVFrame's format in vf_ddn_processing with 'fmt',
but to add another filter for the format.
command examples:
./ffmpeg -i input.jpg -vf format=bgr24,dnn_processing=model=halve_first_channel.model:input=dnn_in:output=dnn_out:dnn_backend=native -y out.native.png
./ffmpeg -i input.jpg -vf format=rgb24,dnn_processing=model=halve_first_channel.model:input=dnn_in:output=dnn_out:dnn_backend=native -y out.native.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
Fixes: signed integer overflow: 2147483647 + 1 cannot be represented in type 'int'
Fixes: 19788/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_VMDAUDIO_fuzzer-5743379690553344
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The documentation of both avio_open() as well as avio_open2() states
that on failure, the pointer to an AVIOContext given to this function
(via a pointer to a pointer to an AVIOContext) will be set to NULL. Yet
it didn't happen upon failure of ffurl_open_whitelist() or when allocating
the internal buffer failed. This commit changes this.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
"VAProcFilterParameterBufferHDRToneMapping" was defined in libva 2.4.1, which will lead to
build failure for the filter tonemap_vaapi for libva 2.3.0 with current check. This patch
is to fix this build error.
Signed-off-by: Xinpeng Sun <xinpeng.sun@intel.com>
Currently, the decoder checks the 128th value of the 4th quant table during
while deriving the context on each sample, in order to speed itself up. This
is due to relying on the behavior of FFmpeg's FFV1 encoder, in which if that
value is zero, the entire 4th and 5th quant tables are assumed to be entirely
zero.
This does not match the FFV1 spec, which has no such restriction, and after
some discussion, it was decided to fix FFmpeg to abide by the spec, rather
than change the spec.
We will now check whether the 4th and 5th quant tables are zero properly,
by checking the 128th valye of both tables (which means they are zero due
to the way they're coded in the bitstream).
For further context, the FFV1 issue in question is located at:
https://github.com/FFmpeg/FFV1/issues/169
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Fixes: Assertion failure
Fixes: 19753/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_PGSSUB_fuzzer-5688461843759104
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This patch rewrites the innermost loop of ff_yuv2planeX_8_neon to avoid zips and
horizontal adds by using fused multiply adds. The patch also uses ld1r to load
one element and replicate it across all lanes of the vector. The patch also
improves the clipping code by removing the shift right instructions and
performing the shift with the shift-right narrow instructions.
I see 8% difference on an m6g instance with neoverse-n1 CPUs:
$ ffmpeg -nostats -f lavfi -i testsrc2=4k:d=2 -vf bench=start,scale=1024x1024,bench=stop -f null -
before: t:0.014015 avg:0.014096 max:0.015018 min:0.013971
after: t:0.012985 avg:0.013013 max:0.013996 min:0.012818
Tested with `make check` on aarch64-linux.
Signed-off-by: Sebastian Pop <spop@amazon.com>
Reviewed-by: Clément Bœsch <u@pkh.me>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
ff_filter_frame() always frees the frame in case of error, so we don't
need to free the frame after ff_filter_frame() fails.
Fix CID 1457230.
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
In case of failure, all the successfully set entries are stored in
*pm. We need to manually free the created dictionary to avoid
memory leak.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
In case of failure, all the successfully set entries are stored in
*pm. We need to manually free the created dictionary to avoid
memory leak.
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
In case of failure, all the successfully set entries are stored in
*pm. We need to manually free the created dictionary to avoid
memory leak.
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
In case of failure, all the successfully set entries are stored in
*pm. We need to manually free the created dictionary to avoid
memory leak.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
Marton Balint
Guo, Yejun
James Zern
Michael Niedermayer
Limin Wang
Andreas Rheinhardt
Mark Thompson
Xinpeng Sun
Carl Eugen Hoyos
Paul B Mahol
Gyan Doshi
Derek Buitenhuis
Ulrich Spörlein
Sebastian Pop
Jun Zhao