Re: [FFmpeg-devel] [PATCH 2/2] qt-faststart - optimize the offset change loop

[Eran Kornblau]

> 
> 
> -Original Message-
> From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On Behalf Of 
> Michael Niedermayer
> Sent: Wednesday, May 30, 2018 12:37 AM
> To: FFmpeg development discussions and patches 
> Subject: Re: [FFmpeg-devel] [PATCH 2/2] qt-faststart - optimize the offset 
> change loop
> 
> > +*ptr++ = (current_offset >> 56) & 0xFF;
> > +*ptr++ = (current_offset >> 48) & 0xFF;
> > +*ptr++ = (current_offset >> 40) & 0xFF;
> > +*ptr++ = (current_offset >> 32) & 0xFF;
> > +*ptr++ = (current_offset >> 24) & 0xFF;
> > +*ptr++ = (current_offset >> 16) & 0xFF;
> > +*ptr++ = (current_offset >>  8) & 0xFF;
> > +*ptr++ = (current_offset >>  0) & 0xFF;
> 
> can this be simplfified with
> libavcodec/bytestream.h, libavutil/intreadwrite.h or similar ?
> 
> [...]
> 

Yes, I can change it to AV_WB32/AV_WB64, but at the moment this utility is 
completely stand-alone - 
it does not depend on anything from ffmpeg, so maybe it's better to keep it 
this way.

Thanks

Eran

> 
> -- 
> Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
> 
> Modern terrorism, a quick summary: Need oil, start war with country that has 
> oil, kill hundread thousand in war. Let country fall into chaos, be surprised 
> about raise of fundamantalists. Drop more bombs, kill more people, be 
> surprised about them taking revenge and drop even more bombs and strip your 
> own citizens of their rights and freedoms. to be continued
>
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] doc/ffmpeg - rewrite Stream Selection chapter

[Gyan Doshi]

On 30-05-2018 04:57 AM, Carl Eugen Hoyos wrote:

2018-05-27 6:16 GMT+02:00, Gyan Doshi :


v2 attached.



+In the absence of any map options for a particular output file, ffmpeg 
inspects the output
+format to check which type of streams can be included in it, viz. video, audio 
and/or


Sorry, what is "viz."?


"Namely". Commonly seen in English prose. Can change it to 'i.e.' which 
is less correct here.



+subtitles. For each acceptable stream type, ffmpeg will pick one stream, when 
available,
+from among all the inputs.


I don't think this is correct, not every stream type is picked.
Or do I misunderstand?


Yes. The qualifier is at the start, "For each acceptable stream type"

 >> +It will select that stream based upon the following criteria:

+@*
+@*for video, it is the stream with the highest resolution,
+@*for audio, it is the stream with the most channels,
+@*for subtitles, it is the first subtitle stream


Please remove the actual current criteria:
This is just the current state of the implementation, for one
of the above, this is obviously not a good choice, for the
others, we could find better criteria.
Or mention that they may all change at any point.


These have been the criteria for nearly 7 years now. The narrowing of 
the subtitle selection was added by you nearly 4 years ago. This is one 
of the parts I copied from the current version, since it remains valid.



+The output format's default subtitle encoder may be text-based or image-based, 
and only a
+subtitle stream of the same type can be chosen.


I wish that were true but I fear it isn't;-(


Please test. The 2nd example demonstrates it. It's your logic - you 
authored & committed it.

(`dvb teletext` is an exception since it no prop flags set.)


+In the case where several streams of the same type rate equally, the stream 
with the
lowest
+index is chosen.


Please remove this.


Why? Another part, copied from the original. Remains valid


All-in-all, this is far too complicated imo.


The _implementation_ is complicated. The docs now reflect it.

The basic principle I'm aiming to follow for docs, even if execution 
remains uneven, is


If a user consults the relevant parts of the documentation before 
execution, they should be able to predict how the program will behave. 
If they do it afterwards, they should understand what the program did. 
Even though FFmpeg is an open source project, end users of the CLI tools 
aren't expected to understand or dive into the source to grasp how the 
program behaves. It's the job of the docs to convey descriptions of 
behaviour that will affect what the end user expects the program to do. 
Do you disagree?


Regards,
Gyan
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH 2/2] flvenc: Fix sequence header update timestamps

[Alex Converse]

On Sun, May 13, 2018 at 3:24 AM, Michael Niedermayer
 wrote:
> On Thu, May 10, 2018 at 06:40:08PM -0700, Alex Converse wrote:
>> From: Alex Converse 
>>
>> ---
>>  libavformat/flvenc.c | 9 -
>>  1 file changed, 4 insertions(+), 5 deletions(-)
>>
>> diff --git a/libavformat/flvenc.c b/libavformat/flvenc.c
>> index 9b7cdfe7db..7aa2dbf9a6 100644
>> --- a/libavformat/flvenc.c
>> +++ b/libavformat/flvenc.c
>> @@ -485,7 +485,7 @@ static int unsupported_codec(AVFormatContext *s,
>>  return AVERROR(ENOSYS);
>>  }
>>
>> -static void flv_write_codec_header(AVFormatContext* s, AVCodecParameters* 
>> par) {
>> +static void flv_write_codec_header(AVFormatContext* s, AVCodecParameters* 
>> par, unsigned ts) {
>
> It seems jeeb prefers int64_t here instead of unsigned.
> Can you change that before pushing ?
>

Pushed with requested changes.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [GSOC] [PATCH] DNN module introduction and SRCNN filter update

[Guo, Yejun]

For the case that ffmpeg is built with TENSORFLOW_BACKEND enabled, while there 
is no TF at runtime, ff_get_dnn_module always returns valid pointer, and no 
chance for the filter to fall back to native mode. Looks that we need a runtime 
check in function ff_get_dnn_module.


static av_cold int init(AVFilterContext* context)
{
SRCNNContext* srcnn_context = context->priv;

#ifdef TENSORFLOW_BACKEND
srcnn_context->dnn_module = ff_get_dnn_module(DNN_TF);
if (!srcnn_context->dnn_module){
av_log(context, AV_LOG_INFO, "could not load tensorflow backend, using 
native backend instead\n");
srcnn_context->dnn_module = ff_get_dnn_module(DNN_NATIVE);
}
#else
srcnn_context->dnn_module = ff_get_dnn_module(DNN_NATIVE);
#endif



DNNModule* ff_get_dnn_module(DNNBackendType backend_type)
{
...
case DNN_TF:
#ifdef TENSORFLOW_BACKEND
//add a runtime check here, possible?
dnn_module->load_model = _dnn_load_model_tf;
dnn_module->load_default_model = _dnn_load_default_model_tf;
dnn_module->execute_model = _dnn_execute_model_tf;
dnn_module->free_model = _dnn_free_model_tf;
#else
av_freep(dnn_module);
return NULL;
#endif


-Original Message-
From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On Behalf Of Pedro 
Arthur
Sent: Tuesday, May 29, 2018 8:45 PM
To: Sergey Lavrushkin 
Cc: FFmpeg development discussions and patches 
Subject: Re: [FFmpeg-devel] [GSOC] [PATCH] DNN module introduction and SRCNN 
filter update

2018-05-29 5:14 GMT-03:00 Sergey Lavrushkin :
> 2018-05-29 4:08 GMT+03:00 Pedro Arthur :
>>
>> 2018-05-28 19:52 GMT-03:00 Sergey Lavrushkin :
>> > 2018-05-28 9:32 GMT+03:00 Guo, Yejun :
>> >
>> >> looks that no tensorflow dependency is introduced, a new model 
>> >> format is
>> >> created together with some CPU implementation for inference.   With
>> >> this
>> >> idea, Android Neural Network would be a very good reference, see 
>> >> https://developer.android.google.cn/ndk/guides/neuralnetworks/. It 
>> >> defines how the model is organized, and also provided a CPU 
>> >> optimized inference implementation (within the NNAPI runtime, it is open 
>> >> source).
>> >> It
>> >> is still under development but mature enough to run some popular 
>> >> dnn models with proper performance. We can absorb some basic 
>> >> design. Anyway, just a reference fyi.  (btw, I'm not sure about 
>> >> any IP issue)
>> >>
>> >
>> > The idea was to first introduce something to use when tensorflow is 
>> > not available. Here is another patch, that introduces tensorflow backend.
>> I think it would be better for reviewing if you send the second patch 
>> in a new email.
>
>
> Then we need to push the first patch, I think.
Not necessarily, 'git send-email' may give you a glimpse of how it is done.

>
>>
>> >
>> >
>> >> For this patch, I have two comments.
>> >>
>> >> 1. change from "DNNModel* (*load_default_model)(DNNDefaultModel
>> >> model_type);" to " DNNModel* (*load_builtin_model)(DNNBuiltinModel
>> >> model_type);"
>> >> The DNNModule can be invoked by many filters,  default model is a 
>> >> good name at the filter level, while built-in model is better 
>> >> within the DNN scope.
>> >>
>> >> typedef struct DNNModule{
>> >> // Loads model and parameters from given file. Returns NULL if 
>> >> it is not possible.
>> >> DNNModel* (*load_model)(const char* model_filename);
>> >> // Loads one of the default models
>> >> DNNModel* (*load_default_model)(DNNDefaultModel model_type);
>> >> // Executes model with specified input and output. Returns 
>> >> DNN_ERROR otherwise.
>> >> DNNReturnType (*execute_model)(const DNNModel* model);
>> >> // Frees memory allocated for model.
>> >> void (*free_model)(DNNModel** model); } DNNModule;
>> >>
>> >>
>> >> 2. add a new variable 'number' for DNNData/InputParams As a 
>> >> typical DNN concept, the data shape usually is: > >> width, channel> or , the last 
>> >> component denotes its index changes the fastest in the memory. We 
>> >> can add this concept into the API, and decide to support  or  
>> >> or both.
>> >
>> >
>> > I did not add number of elements in batch because I thought, that 
>> > we would not feed more than one element at once to a network in a 
>> > ffmpeg filter.
>> > But it can be easily added if necessary.
>> >
>> > So here is the patch that adds tensorflow backend with the previous 
>> > patch.
>> > I forgot to change include guards from AVUTIL_* to AVFILTER_* in it.
>> You moved the files from libavutil to libavfilter while it was 
>> proposed to move them to libavformat.
>
>
> Not only, it was also proposed to move it to libavfilter if it is 
> going to be used only in filters. I do not know if this module is 
> useful anywhere else besides libavfilter.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avformat/flvenc: Avoid truncating timestamp before avio_write_marker()

[Alex Converse]

On Sun, May 13, 2018 at 3:41 AM, Michael Niedermayer
 wrote:
> Signed-off-by: Michael Niedermayer 
> ---
>  libavformat/flvenc.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/libavformat/flvenc.c b/libavformat/flvenc.c
> index e8af48cb64..168ff9ffb8 100644
> --- a/libavformat/flvenc.c
> +++ b/libavformat/flvenc.c
> @@ -873,7 +873,7 @@ static int flv_write_packet(AVFormatContext *s, AVPacket 
> *pkt)
>  AVCodecParameters *par = s->streams[pkt->stream_index]->codecpar;
>  FLVContext *flv  = s->priv_data;
>  FLVStreamContext *sc = s->streams[pkt->stream_index]->priv_data;
> -unsigned ts;
> +int64_t ts;
>  int size = pkt->size;
>  uint8_t *data = NULL;
>  int flags = -1, flags_size, ret;
> --
> 2.17.0

The put_avc_eos_tag() function signature applies a similar truncation.
Best to be consistent.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

[FFmpeg-devel] [PATCH v3 1/3] lavc,doc: add avs2 codec

[hwren]

Signed-off-by: hwren 
---
 doc/APIchanges  | 3 +++
 libavcodec/avcodec.h| 1 +
 libavcodec/codec_desc.c | 7 +++
 libavcodec/version.h| 4 ++--
 4 files changed, 13 insertions(+), 2 deletions(-)

diff --git a/doc/APIchanges b/doc/APIchanges
index efe15ba..3d08bb9 100644
--- a/doc/APIchanges
+++ b/doc/APIchanges
@@ -15,6 +15,9 @@ libavutil: 2017-10-21
 
 API changes, most recent first:
 
+2018-05-xx - xx - lavc 58.20.100 - avcodec.h
+  Add AV_CODEC_ID_AVS2.
+
 2018-05-xx - xx - lavf 58.15.100 - avformat.h
   Add pmt_version field to AVProgram
 
diff --git a/libavcodec/avcodec.h b/libavcodec/avcodec.h
index fb0c6fa..ce5f307 100644
--- a/libavcodec/avcodec.h
+++ b/libavcodec/avcodec.h
@@ -409,6 +409,7 @@ enum AVCodecID {
 AV_CODEC_ID_DXV,
 AV_CODEC_ID_SCREENPRESSO,
 AV_CODEC_ID_RSCC,
+AV_CODEC_ID_AVS2,
 
 AV_CODEC_ID_Y41P = 0x8000,
 AV_CODEC_ID_AVRP,
diff --git a/libavcodec/codec_desc.c b/libavcodec/codec_desc.c
index 79552a9..e85492e 100644
--- a/libavcodec/codec_desc.c
+++ b/libavcodec/codec_desc.c
@@ -1395,6 +1395,13 @@ static const AVCodecDescriptor codec_descriptors[] = {
 .props = AV_CODEC_PROP_LOSSLESS,
 },
 {
+.id= AV_CODEC_ID_AVS2,
+.type  = AVMEDIA_TYPE_VIDEO,
+.name  = "avs2",
+.long_name = NULL_IF_CONFIG_SMALL("AVS2/IEEE 1857.4"),
+.props = AV_CODEC_PROP_LOSSY,
+},
+{
 .id= AV_CODEC_ID_Y41P,
 .type  = AVMEDIA_TYPE_VIDEO,
 .name  = "y41p",
diff --git a/libavcodec/version.h b/libavcodec/version.h
index f65346a..b9752ce 100644
--- a/libavcodec/version.h
+++ b/libavcodec/version.h
@@ -28,8 +28,8 @@
 #include "libavutil/version.h"
 
 #define LIBAVCODEC_VERSION_MAJOR  58
-#define LIBAVCODEC_VERSION_MINOR  19
-#define LIBAVCODEC_VERSION_MICRO 104
+#define LIBAVCODEC_VERSION_MINOR  20
+#define LIBAVCODEC_VERSION_MICRO 100
 
 #define LIBAVCODEC_VERSION_INT  AV_VERSION_INT(LIBAVCODEC_VERSION_MAJOR, \
LIBAVCODEC_VERSION_MINOR, \
-- 
2.7.4

___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

[FFmpeg-devel] [PATCH v3 3/3] lavf: add avs2 fourcc

[hwren]

Signed-off-by: hwren 
---
 libavformat/riff.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/libavformat/riff.c b/libavformat/riff.c
index 8911725..4153372 100644
--- a/libavformat/riff.c
+++ b/libavformat/riff.c
@@ -369,6 +369,7 @@ const AVCodecTag ff_codec_bmp_tags[] = {
 { AV_CODEC_ID_ZMBV, MKTAG('Z', 'M', 'B', 'V') },
 { AV_CODEC_ID_KMVC, MKTAG('K', 'M', 'V', 'C') },
 { AV_CODEC_ID_CAVS, MKTAG('C', 'A', 'V', 'S') },
+{ AV_CODEC_ID_AVS2, MKTAG('A', 'V', 'S', '2') },
 { AV_CODEC_ID_JPEG2000, MKTAG('m', 'j', 'p', '2') },
 { AV_CODEC_ID_JPEG2000, MKTAG('M', 'J', '2', 'C') },
 { AV_CODEC_ID_JPEG2000, MKTAG('L', 'J', '2', 'C') },
-- 
2.7.4

___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

[FFmpeg-devel] [PATCH v3 2/3] lavc, doc, configure: add libdavs2 video decoder

[hwren]

Add avs2 video decoder via libdavs2 library.

Signed-off-by: hwren 
---
 Changelog  |   1 +
 configure  |   4 +
 doc/decoders.texi  |  10 +++
 doc/general.texi   |   8 ++
 libavcodec/Makefile|   1 +
 libavcodec/allcodecs.c |   1 +
 libavcodec/libdavs2.c  | 204 +
 7 files changed, 229 insertions(+)
 create mode 100644 libavcodec/libdavs2.c

diff --git a/Changelog b/Changelog
index 3d25564..ce1f97c 100644
--- a/Changelog
+++ b/Changelog
@@ -9,6 +9,7 @@ version :
 - aderivative and aintegral audio filters
 - pal75bars and pal100bars video filter sources
 - support mbedTLS based TLS
+- AVS2 video decoder via libdavs2
 
 
 version 4.0:
diff --git a/configure b/configure
index 22eeca2..3c9129f 100755
--- a/configure
+++ b/configure
@@ -226,6 +226,7 @@ External library support:
   --enable-libcelt enable CELT decoding via libcelt [no]
   --enable-libcdio enable audio CD grabbing with libcdio [no]
   --enable-libcodec2   enable codec2 en/decoding using libcodec2 [no]
+  --enable-libdavs2enable AVS2 decoding via libdavs2 [no]
   --enable-libdc1394   enable IIDC-1394 grabbing using libdc1394
and libraw1394 [no]
   --enable-libfdk-aac  enable AAC de/encoding via libfdk-aac [no]
@@ -1636,6 +1637,7 @@ EXTERNAL_LIBRARY_GPL_LIST="
 avisynth
 frei0r
 libcdio
+libdavs2
 librubberband
 libvidstab
 libx264
@@ -3042,6 +3044,7 @@ libaom_av1_encoder_deps="libaom"
 libcelt_decoder_deps="libcelt"
 libcodec2_decoder_deps="libcodec2"
 libcodec2_encoder_deps="libcodec2"
+libdavs2_decoder_deps="libdavs2"
 libfdk_aac_decoder_deps="libfdk_aac"
 libfdk_aac_encoder_deps="libfdk_aac"
 libfdk_aac_encoder_select="audio_frame_queue"
@@ -5992,6 +5995,7 @@ enabled libcelt   && require libcelt celt/celt.h 
celt_decode -lcelt0 &&
die "ERROR: libcelt must be installed and 
version must be >= 0.11.0."; }
 enabled libcaca   && require_pkg_config libcaca caca caca.h 
caca_create_canvas
 enabled libcodec2 && require libcodec2 codec2/codec2.h codec2_create 
-lcodec2
+enabled libdavs2  && require_pkg_config libdavs2 "davs2 >= 1.2.34" 
davs2.h davs2_decoder_decode
 enabled libdc1394 && require_pkg_config libdc1394 libdc1394-2 
dc1394/dc1394.h dc1394_new
 enabled libdrm&& require_pkg_config libdrm libdrm xf86drm.h 
drmGetVersion
 enabled libfdk_aac&& { check_pkg_config libfdk_aac fdk-aac 
"fdk-aac/aacenc_lib.h" aacEncOpen ||
diff --git a/doc/decoders.texi b/doc/decoders.texi
index a551d5d..f9d1b78 100644
--- a/doc/decoders.texi
+++ b/doc/decoders.texi
@@ -47,6 +47,16 @@ top-field-first is assumed
 
 @end table
 
+@section libdavs2
+
+AVS2/IEEE 1857.4 video decoder wrapper.
+
+This decoder allows libavcodec to decode AVS2 streams with libdavs2 library.
+Using it requires the presence of the libdavs2 headers and library during
+configuration. You need to explicitly configure the build with 
@code{--enable-libdavs2}.
+
+libdavs2 uses GPLv2, so you may also need to add @code{--enable-gpl} while 
configuring.
+
 @c man end VIDEO DECODERS
 
 @chapter Audio Decoders
diff --git a/doc/general.texi b/doc/general.texi
index 2583006..d3c1503 100644
--- a/doc/general.texi
+++ b/doc/general.texi
@@ -17,6 +17,14 @@ for more formats. None of them are used by default, their 
use has to be
 explicitly requested by passing the appropriate flags to
 @command{./configure}.
 
+@section libdavs2
+
+FFmpeg can make use of the libdavs2 library for AVS2 decoding.
+
+Go to @url{https://github.com/pkuvcl/davs2} and follow the instructions for
+installing the library. Then pass @code{--enable-libdavs2} to configure to
+enable it.
+
 @section Alliance for Open Media libaom
 
 FFmpeg can make use of the libaom library for AV1 decoding.
diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index 3ab071a..2a845f1 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -944,6 +944,7 @@ OBJS-$(CONFIG_LIBAOM_AV1_ENCODER) += libaomenc.o
 OBJS-$(CONFIG_LIBCELT_DECODER)+= libcelt_dec.o
 OBJS-$(CONFIG_LIBCODEC2_DECODER)  += libcodec2.o codec2utils.o
 OBJS-$(CONFIG_LIBCODEC2_ENCODER)  += libcodec2.o codec2utils.o
+OBJS-$(CONFIG_LIBDAVS2_DECODER)   += libdavs2.o
 OBJS-$(CONFIG_LIBFDK_AAC_DECODER) += libfdk-aacdec.o
 OBJS-$(CONFIG_LIBFDK_AAC_ENCODER) += libfdk-aacenc.o
 OBJS-$(CONFIG_LIBGSM_DECODER) += libgsmdec.o
diff --git a/libavcodec/allcodecs.c b/libavcodec/allcodecs.c
index 90d170b..a59b601 100644
--- a/libavcodec/allcodecs.c
+++ b/libavcodec/allcodecs.c
@@ -667,6 +667,7 @@ extern AVCodec ff_libaom_av1_encoder;
 extern AVCodec ff_libcelt_decoder;
 extern AVCodec ff_libcodec2_encoder;
 extern AVCodec ff_libcodec2_decoder;
+extern AVCodec ff_libdavs2_decoder;
 extern AVCodec ff_libfdk_aac_encoder;
 extern AVCodec ff_libfdk_aac_decoder;
 extern AVCodec 

Re: [FFmpeg-devel] [PATCH] mov: Make sure PTS are both monotonically increasing, and unique

[Michael Niedermayer]

On Tue, May 29, 2018 at 07:20:33PM +0100, Derek Buitenhuis wrote:
> Hi,
> 
> On Tue, May 29, 2018 at 5:04 PM, Sasi Inguva
>  wrote:
> > Hi. sorry for the late reply. I sent a patch similar to this a while back
> > https://patchwork.ffmpeg.org/patch/8227/  but it got lost in the sea. You
> > also want to do,
> 
> Sorry I missed that! I'd prefer to use your patch over mine.

iam fine with sasis original patch too

[...]

-- 
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

It is what and why we do it that matters, not just one of them.


signature.asc
Description: PGP signature
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] lavfi/tests: Fix 16-bit vf_blend test to avoid memory not aligned to 2 bytes

[Michael Niedermayer]

On Tue, May 29, 2018 at 02:41:33PM +0200, Paul B Mahol wrote:
> On 5/29/18, Andrey Semashev  wrote:
> > On 05/24/18 00:07, Andrey Semashev wrote:
> >> Generic C implementation of vf_blend performs reads and writes of 16-bit
> >> elements, which requires the buffers to be aligned to at least 2-byte
> >> boundary.
> >>
> >> Also, the change fixes source buffer overrun caused by src_offset being
> >> added to to test handling of misaligned buffers.
> >>
> >> Fixes: #7226
> >
> > Ping? Any comments?
> 
> Should be OK.

will apply unless someone else does before

thx

[...]
-- 
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Dictatorship: All citizens are under surveillance, all their steps and
actions recorded, for the politicians to enforce control.
Democracy: All politicians are under surveillance, all their steps and
actions recorded, for the citizens to enforce control.


signature.asc
Description: PGP signature
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH 1/2] qt-faststart - stricter input validations

[Michael Niedermayer]

On Tue, May 29, 2018 at 02:35:23PM +, Eran Kornblau wrote:
> Hi,
> 
> The attached patch fixes a couple of input validation issues in fast start 
> that I noticed while going over the code
> 
> Thanks
> 
> Eran

>  qt-faststart.c |   13 +
>  1 file changed, 9 insertions(+), 4 deletions(-)
> 1dad4dfcdd67328ed163440550917a3f8fdcb40d  
> 0001-qt-faststart-stricter-input-validations.patch
> From 26ef40268fce426eea608400f81cf2e4d413fca5 Mon Sep 17 00:00:00 2001
> From: erankor 
> Date: Tue, 29 May 2018 16:18:05 +0300
> Subject: [PATCH 1/2] qt-faststart - stricter input validations
> 
> 1. validate the moov size before checking for cmov atom
> 2. avoid performing arithmetic operations on unvalidated numbers
> 3. verify the stco/co64 offset count does not overflow the stco/co64
> atom (not only the moov atom)
> ---
>  tools/qt-faststart.c | 13 +
>  1 file changed, 9 insertions(+), 4 deletions(-)

will apply

thx

[...]

-- 
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

When you are offended at any man's fault, turn to yourself and study your
own failings. Then you will forget your anger. -- Epictetus


signature.asc
Description: PGP signature
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avcodec/qtrle: Do not output duplicated frames on insufficient input

[Michael Niedermayer]

On Sun, May 27, 2018 at 09:59:58PM +0200, Michael Niedermayer wrote:
> This improves performance and makes qtrle behave more similar to other 
> decoders.
> Libavcodec does generally not output known duplicated frames, instead the 
> calling Application
> can insert them as it needs.
> 
> Fixes: Timeout
> Fixes: 
> 6383/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_QTRLE_fuzzer-6199846902956032
> 
> Found-by: continuous fuzzing process 
> https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
> Signed-off-by: Michael Niedermayer 
> ---
>  libavcodec/qtrle.c|  12 ++---
>  tests/ref/fate/qtrle-8bit | 109 --
>  2 files changed, 6 insertions(+), 115 deletions(-)

will apply

[...]
-- 
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

The worst form of inequality is to try to make unequal things equal.
-- Aristotle


signature.asc
Description: PGP signature
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] doc/ffmpeg - rewrite Stream Selection chapter

[Carl Eugen Hoyos]

2018-05-27 6:16 GMT+02:00, Gyan Doshi :

> v2 attached.

> +In the absence of any map options for a particular output file, ffmpeg 
> inspects the output
> +format to check which type of streams can be included in it, viz. video, 
> audio and/or

Sorry, what is "viz."?

> +subtitles. For each acceptable stream type, ffmpeg will pick one stream, 
> when available,
> +from among all the inputs.

I don't think this is correct, not every stream type is picked.
Or do I misunderstand?

> +It will select that stream based upon the following criteria:
> +@*
> +@*for video, it is the stream with the highest resolution,
> +@*for audio, it is the stream with the most channels,
> +@*for subtitles, it is the first subtitle stream

Please remove the actual current criteria:
This is just the current state of the implementation, for one
of the above, this is obviously not a good choice, for the
others, we could find better criteria.
Or mention that they may all change at any point.

> +The output format's default subtitle encoder may be text-based or 
> image-based, and only a
> +subtitle stream of the same type can be chosen.

I wish that were true but I fear it isn't;-(

[...]

> +In the case where several streams of the same type rate equally, the stream 
> with the
> lowest
> +index is chosen.

Please remove this.

All-in-all, this is far too complicated imo.

Please resend, Carl Eugen
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH]ffplay: Mention codec_name if decoder for codec_id could not be found.

[Carl Eugen Hoyos]

2018-05-29 9:54 GMT+02:00, Marton Balint :
>
>
> On Tue, 29 May 2018, Carl Eugen Hoyos wrote:
>
>> Hi!
>>
>> Attached patch makes debugging a little easier imo.
>>
>> Please comment, Carl Eugen
>>
>> diff --git a/fftools/ffplay.c b/fftools/ffplay.c
>> index dcca9c2..f9571d7 100644
>> --- a/fftools/ffplay.c
>> +++ b/fftools/ffplay.c
>> @@ -2578,7 +2578,7 @@ static int stream_component_open(VideoState *is, int
>> stream_index)
>> if (forced_codec_name) av_log(NULL, AV_LOG_WARNING,
>>   "No codec could be found with name
>> '%s'\n", forced_codec_name);
>> else   av_log(NULL, AV_LOG_WARNING,
>> -  "No codec could be found with id
>> %d\n", avctx->codec_id);
>> +  "No codec could be found with id %d
>> (%s)\n", avctx->codec_id, avcodec_get_name(avctx->codec_id));
>
> Maybe go one step further, and change the error message to
>
> "No decoder could be found for codec %s\n",
> avcodec_get_name(avctx->codec_id)

Patch applied.

Thank you, Carl Eugen
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] doc/ffmpeg - rewrite Stream Selection chapter

[Lou Logan]

On Sun, 27 May 2018 09:46:46 +0530
Gyan Doshi  wrote:

> I was talking about the vertical margins appled to the 'code' element.

That's ugly too. I removed the margin for the code element.

> From 60ed76348e70f1b0a25eadde8d886d47be3fca69 Mon Sep 17 00:00:00 2001
> From: Gyan Doshi 
> Date: Thu, 24 May 2018 19:11:00 +0530
> Subject: [PATCH v2] doc/ffmpeg - rewrite Stream Selection chapter

The subject nit still exists.

doc/ffmpeg: rewrite Stream Selection chapter

> Flesh out with details and examples to show quirks and limitations.
> ---
>  doc/ffmpeg.texi | 187 
> +---
>  1 file changed, 177 insertions(+), 10 deletions(-)
> 
> diff --git a/doc/ffmpeg.texi b/doc/ffmpeg.texi
> index 88dbdeb95a..803490ce7b 100644
> --- a/doc/ffmpeg.texi
> +++ b/doc/ffmpeg.texi
[...]
> +It will select that stream based upon the following criteria:
> +@*
> +@*for video, it is the stream with the highest resolution,
> +@*for audio, it is the stream with the most channels,
> +@*for subtitles, it is the first subtitle stream found but there's a caveat.
> +The output format's default subtitle encoder may be text-based or 
> image-based, and only a
> +subtitle stream of the same type can be chosen.

Using an itemized list here will look better in my opinion.

@itemize
@item
for video, it is the stream with the highest resolution,
@item
...
@end itemize

[...]
> +@subsubheading Example: automatic stream selection

The subsubheadings did not render in the HTML for me, but they did in
man. I didn't investigate why.

[...]
> +@subsubheading Example: unlabeled filtergraph outputs 

Trailing whitespace.

I'm not convinced a verbose, tutorial-style set of examples belongs
here. I tend to put such things in the wiki, but if you think otherwise
that's fine.

That's all of my comments. Everything else LGTM.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

[FFmpeg-devel] [PATCH] lavf/mov.c: Set st->start_time for video streams explicitly.

[Sasi Inguva]

If start_time is not set, ffmpeg takes the duration from the global
movie instead of the per stream duration.
Signed-off-by: Sasi Inguva 
---
 libavformat/mov.c| 20 +---
 tests/fate/mov.mak   |  4 +++
 tests/ref/fate/mov-neg-firstpts-discard  |  2 +-
 tests/ref/fate/mov-stream-shorter-than-movie | 33 
 4 files changed, 54 insertions(+), 5 deletions(-)
 create mode 100644 tests/ref/fate/mov-stream-shorter-than-movie

diff --git a/libavformat/mov.c b/libavformat/mov.c
index f2a540ad50..1915be5fb5 100644
--- a/libavformat/mov.c
+++ b/libavformat/mov.c
@@ -3672,11 +3672,15 @@ static void mov_fix_index(MOVContext *mov, AVStream *st)
 
 // If the minimum pts turns out to be greater than zero after fixing the 
index, then we subtract the
 // dts by that amount to make the first pts zero.
-if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && 
msc->min_corrected_pts > 0) {
-av_log(mov->fc, AV_LOG_DEBUG, "Offset DTS by %"PRId64" to make first 
pts zero.\n", msc->min_corrected_pts);
-for (i = 0; i < st->nb_index_entries; ++i) {
-st->index_entries[i].timestamp -= msc->min_corrected_pts;
+if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
+if (msc->min_corrected_pts > 0) {
+av_log(mov->fc, AV_LOG_DEBUG, "Offset DTS by %"PRId64" to make 
first pts zero.\n", msc->min_corrected_pts);
+for (i = 0; i < st->nb_index_entries; ++i) {
+st->index_entries[i].timestamp -= msc->min_corrected_pts;
+}
 }
+// Start time should be equal to zero or the duration of any empty 
edits.
+st->start_time = empty_edits_sum_duration;
 }
 
 // Update av stream length, if it ends up shorter than the track's media 
duration
@@ -4012,6 +4016,14 @@ static void mov_build_index(MOVContext *mov, AVStream 
*st)
 mov_fix_index(mov, st);
 }
 
+// Update start time of the stream.
+if (st->start_time == AV_NOPTS_VALUE && st->codecpar->codec_type == 
AVMEDIA_TYPE_VIDEO && st->nb_index_entries > 0) {
+st->start_time = st->index_entries[0].timestamp + sc->dts_shift;
+if (sc->ctts_data) {
+st->start_time += sc->ctts_data[0].duration;
+}
+}
+
 mov_estimate_video_delay(mov, st);
 }
 
diff --git a/tests/fate/mov.mak b/tests/fate/mov.mak
index eadee3abfa..c1d399e5c0 100644
--- a/tests/fate/mov.mak
+++ b/tests/fate/mov.mak
@@ -16,6 +16,7 @@ FATE_MOV = fate-mov-3elist \
fate-mov-frag-overlap \
fate-mov-bbi-elst-starts-b \
fate-mov-neg-firstpts-discard-frames \
+  fate-mov-stream-shorter-than-movie \
 
 FATE_MOV_FFPROBE = fate-mov-neg-firstpts-discard \
fate-mov-aac-2048-priming \
@@ -88,6 +89,9 @@ fate-mov-neg-firstpts-discard: CMD = run 
ffprobe$(PROGSSUF)$(EXESUF) -show_entri
 # Makes sure that expected frames are generated for 
mov_neg_first_pts_discard.mov with -vsync 1
 fate-mov-neg-firstpts-discard-frames: CMD = framemd5 -flags +bitexact -i 
$(TARGET_SAMPLES)/mov/mov_neg_first_pts_discard.mov -vsync 1
 
+# Makes sure that no frame is dropped/duplicated with fps filter due to 
start_time / duration miscalculations.
+fate-mov-stream-shorter-than-movie: CMD = framemd5 -flags +bitexact -i 
$(TARGET_SAMPLES)/mov/mov_stream_shorter_than_movie.mov -vf fps=fps=24 -an
+
 fate-mov-aac-2048-priming: CMD = run ffprobe$(PROGSSUF)$(EXESUF) -show_packets 
-print_format compact $(TARGET_SAMPLES)/mov/aac-2048-priming.mov
 
 fate-mov-zombie: CMD = run ffprobe$(PROGSSUF)$(EXESUF) -show_streams 
-show_packets -show_frames -bitexact -print_format compact 
$(TARGET_SAMPLES)/mov/white_zombie_scrunch-part.mov
diff --git a/tests/ref/fate/mov-neg-firstpts-discard 
b/tests/ref/fate/mov-neg-firstpts-discard
index 7c982d3ffe..2e295e3b68 100644
--- a/tests/ref/fate/mov-neg-firstpts-discard
+++ b/tests/ref/fate/mov-neg-firstpts-discard
@@ -1,3 +1,3 @@
 [STREAM]
-start_time=N/A
+start_time=0.00
 [/STREAM]
diff --git a/tests/ref/fate/mov-stream-shorter-than-movie 
b/tests/ref/fate/mov-stream-shorter-than-movie
new file mode 100644
index 00..28f3ef378c
--- /dev/null
+++ b/tests/ref/fate/mov-stream-shorter-than-movie
@@ -0,0 +1,33 @@
+#format: frame checksums
+#version: 2
+#hash: MD5
+#tb 0: 1/24
+#media_type 0: video
+#codec_id 0: rawvideo
+#dimensions 0: 640x480
+#sar 0: 0/1
+#stream#, dts,pts, duration, size, hash
+0,  0,  0,1,   460800, 3a26ddfa53f09d535c701138027e49dc
+0,  1,  1,1,   460800, f09fe0d079ee81eb7db617b48ab5eecf
+0,  2,  2,1,   460800, 40a165b074c7f4d34a41f320400737fc
+0,  3,  3,1,   460800, 8ba73359c89ebc51e29847ef0e27f7c3
+0,  4,  4,1,   460800, 0d783fcf3d37b99e7b41c0450e28f905
+0,  5,  5,1,   460800, 7251de6f3e2ebccc2183aa7090dd59fb
+0,  6,  6,   

Re: [FFmpeg-devel] [PATCH 2/2] qt-faststart - optimize the offset change loop

[Michael Niedermayer]

On Tue, May 29, 2018 at 02:36:28PM +, Eran Kornblau wrote:
> Hi,
> 
> The attached is a slightly more optimized (and IMHO elegant) code for 
> updating the stco/co64 offsets
> 
> Thanks
> 
> Eran

>  qt-faststart.c |   42 +-
>  1 file changed, 25 insertions(+), 17 deletions(-)
> a0f95e960800141a0a666313f2f3d82a87a3309f  
> 0002-qt-faststart-optimize-the-offset-change-loop.patch
> From 776244b79a8bcfb5732f39fbebb9cd7fc0092bcb Mon Sep 17 00:00:00 2001
> From: erankor 
> Date: Tue, 29 May 2018 17:29:09 +0300
> Subject: [PATCH 2/2] qt-faststart - optimize the offset change loop
> 
> ---
>  tools/qt-faststart.c | 42 +-
>  1 file changed, 25 insertions(+), 17 deletions(-)
> 
> diff --git a/tools/qt-faststart.c b/tools/qt-faststart.c
> index d0ae7245f3..2ddaf87e1b 100644
> --- a/tools/qt-faststart.c
> +++ b/tools/qt-faststart.c
> @@ -96,9 +96,11 @@ int main(int argc, char *argv[])
>  int64_t last_offset;
>  unsigned char *moov_atom = NULL;
>  unsigned char *ftyp_atom = NULL;
> +unsigned char *ptr;
> +unsigned char *end;
>  uint64_t moov_atom_size;
>  uint64_t ftyp_atom_size = 0;
> -uint64_t i, j;
> +uint64_t i;
>  uint32_t offset_count;
>  uint64_t current_offset;
>  int64_t start_offset = 0;
> @@ -253,13 +255,16 @@ int main(int argc, char *argv[])
>  printf(" bad atom size/element count\n");
>  goto error_out;
>  }
> -for (j = 0; j < offset_count; j++) {
> -current_offset  = BE_32(_atom[i + 12 + j * 4]);
> +
> +ptr = moov_atom + i + 12;
> +end = ptr + offset_count * 4;
> +while (ptr < end) {
> +current_offset  = BE_32(ptr);
>  current_offset += moov_atom_size;
> -moov_atom[i + 12 + j * 4 + 0] = (current_offset >> 24) & 
> 0xFF;
> -moov_atom[i + 12 + j * 4 + 1] = (current_offset >> 16) & 
> 0xFF;
> -moov_atom[i + 12 + j * 4 + 2] = (current_offset >>  8) & 
> 0xFF;
> -moov_atom[i + 12 + j * 4 + 3] = (current_offset >>  0) & 
> 0xFF;
> +*ptr++ = (current_offset >> 24) & 0xFF;
> +*ptr++ = (current_offset >> 16) & 0xFF;
> +*ptr++ = (current_offset >>  8) & 0xFF;
> +*ptr++ = (current_offset >>  0) & 0xFF;
>  }
>  i += atom_size - 4;
>  } else if (atom_type == CO64_ATOM) {
> @@ -274,17 +279,20 @@ int main(int argc, char *argv[])
>  printf(" bad atom size/element count\n");
>  goto error_out;
>  }
> -for (j = 0; j < offset_count; j++) {
> -current_offset  = BE_64(_atom[i + 12 + j * 8]);
> +
> +ptr = moov_atom + i + 12;
> +end = ptr + offset_count * 8;
> +while (ptr < end) {
> +current_offset  = BE_64(ptr);
>  current_offset += moov_atom_size;
> -moov_atom[i + 12 + j * 8 + 0] = (current_offset >> 56) & 
> 0xFF;
> -moov_atom[i + 12 + j * 8 + 1] = (current_offset >> 48) & 
> 0xFF;
> -moov_atom[i + 12 + j * 8 + 2] = (current_offset >> 40) & 
> 0xFF;
> -moov_atom[i + 12 + j * 8 + 3] = (current_offset >> 32) & 
> 0xFF;
> -moov_atom[i + 12 + j * 8 + 4] = (current_offset >> 24) & 
> 0xFF;
> -moov_atom[i + 12 + j * 8 + 5] = (current_offset >> 16) & 
> 0xFF;
> -moov_atom[i + 12 + j * 8 + 6] = (current_offset >>  8) & 
> 0xFF;
> -moov_atom[i + 12 + j * 8 + 7] = (current_offset >>  0) & 
> 0xFF;
> +*ptr++ = (current_offset >> 56) & 0xFF;
> +*ptr++ = (current_offset >> 48) & 0xFF;
> +*ptr++ = (current_offset >> 40) & 0xFF;
> +*ptr++ = (current_offset >> 32) & 0xFF;
> +*ptr++ = (current_offset >> 24) & 0xFF;
> +*ptr++ = (current_offset >> 16) & 0xFF;
> +*ptr++ = (current_offset >>  8) & 0xFF;
> +*ptr++ = (current_offset >>  0) & 0xFF;

can this be simplfified with
libavcodec/bytestream.h, libavutil/intreadwrite.h
or similar ?

[...]


-- 
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Modern terrorism, a quick summary: Need oil, start war with country that
has oil, kill hundread thousand in war. Let country fall into chaos,
be surprised about raise of fundamantalists. Drop more bombs, kill more
people, be surprised about them taking revenge and drop even more bombs
and strip your own citizens of their rights and freedoms. to be continued


signature.asc
Description: PGP signature
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avcodec/vc1: fix overlap smoothing filter for P frames

[Carl Eugen Hoyos]

2018-05-29 22:02 GMT+02:00, Jerome Borsboom :
> The v_overlap_filter needs to run on the colocated block of the previous
> macroblock. For the luma plane, the colocated block is located two blocks
> on the left instead of one.
>
> Signed-off-by: Jerome Borsboom 
> ---
> This should fix the issue with the SA10100.vc1 test file.
>
>  libavcodec/vc1_loopfilter.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/libavcodec/vc1_loopfilter.c b/libavcodec/vc1_loopfilter.c
> index 4c0de7c025..676922aa18 100644
> --- a/libavcodec/vc1_loopfilter.c
> +++ b/libavcodec/vc1_loopfilter.c
> @@ -145,8 +145,8 @@ void ff_vc1_p_overlap_filter(VC1Context *v)
>
>  if (v->fcm != ILACE_FRAME)
>  for (i = 0; i < block_count; i++) {
> -if (s->mb_x && v->mb_type[0][s->block_index[i] - 1] &&
> -(s->first_slice_line || v->mb_type[0][s->block_index[i] -
> s->block_wrap[i] - 1]))
> +if (s->mb_x && v->mb_type[0][s->block_index[i] - 2 + (i > 3)]
> &&
> +(s->first_slice_line || v->mb_type[0][s->block_index[i] -
> s->block_wrap[i] - 2 + (i > 3)]))

This also fixes the frame number 5 and 6 of SSL0013.rcv (ticket #7171),
the last two frames are still incorrect.

Thank you, Carl Eugen
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH v3] avcodec/vc1: fix out-of-bounds reference pixel replication

[Michael Niedermayer]

On Tue, May 29, 2018 at 02:26:17PM +0200, Jerome Borsboom wrote:
> Out-of-bounds reference pixel replication should take into account the frame
> coding mode of the reference frame(s), not the frame coding mode of the
> current frame.
> 
> Signed-off-by: Jerome Borsboom 
> ---
> Even more corrections. The starting line must also be adjusted by one for an 
> opposite
> refence field.
> 
>  libavcodec/vc1_mc.c | 668 
> ++--
>  1 file changed, 385 insertions(+), 283 deletions(-)

All crashes i saw are gone

thanks

[...]
-- 
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

If you think the mosad wants you dead since a long time then you are either
wrong or dead since a long time.


signature.asc
Description: PGP signature
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

[FFmpeg-devel] [PATCH] avcodec/vc1: fix overlap smoothing filter for P frames

[Jerome Borsboom]

The v_overlap_filter needs to run on the colocated block of the previous
macroblock. For the luma plane, the colocated block is located two blocks
on the left instead of one.

Signed-off-by: Jerome Borsboom 
---
This should fix the issue with the SA10100.vc1 test file.

 libavcodec/vc1_loopfilter.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/libavcodec/vc1_loopfilter.c b/libavcodec/vc1_loopfilter.c
index 4c0de7c025..676922aa18 100644
--- a/libavcodec/vc1_loopfilter.c
+++ b/libavcodec/vc1_loopfilter.c
@@ -145,8 +145,8 @@ void ff_vc1_p_overlap_filter(VC1Context *v)
 
 if (v->fcm != ILACE_FRAME)
 for (i = 0; i < block_count; i++) {
-if (s->mb_x && v->mb_type[0][s->block_index[i] - 1] &&
-(s->first_slice_line || v->mb_type[0][s->block_index[i] - 
s->block_wrap[i] - 1]))
+if (s->mb_x && v->mb_type[0][s->block_index[i] - 2 + (i > 3)] &&
+(s->first_slice_line || v->mb_type[0][s->block_index[i] - 
s->block_wrap[i] - 2 + (i > 3)]))
 vc1_v_overlap_filter(v, s->first_slice_line ? left_blk : 
topleft_blk, left_blk, i);
 if (s->mb_x == s->mb_width - 1)
 if (v->mb_type[0][s->block_index[i]] &&
-- 
2.13.6


___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

[FFmpeg-devel] [PATCH v3 1/3] libavformat/mov: treat udta atoms within trak atoms as stream metadata

[Nik Johnson]

Some muxers produce mp4s with a udta (user data) atom nested within a
trak atom. The nested udta atoms typically contain stream information
such as the title. ffmpeg should treat nested udta atoms as applicable
to the stream instead of globally.

Signed-off-by: Nik Johnson 
---
 libavformat/mov.c | 19 ---
 1 file changed, 16 insertions(+), 3 deletions(-)

diff --git a/libavformat/mov.c b/libavformat/mov.c
index f2a540ad50..b434802207 100644
--- a/libavformat/mov.c
+++ b/libavformat/mov.c
@@ -306,6 +306,8 @@ static int mov_read_udta_string(MOVContext *c, AVIOContext 
*pb, MOVAtom atom)
 int (*parse)(MOVContext*, AVIOContext*, unsigned, const char*) = NULL;
 int raw = 0;
 int num = 0;
+AVStream *st;
+AVDictionary **metadata;
 
 switch (atom.type) {
 case MKTAG( '@','P','R','M'): key = "premiere_version"; raw = 1; break;
@@ -514,12 +516,23 @@ retry:
 }
 str[str_size] = 0;
 }
-c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
-av_dict_set(>fc->metadata, key, str, 0);
+
+// A udta atom may occur inside a trak atom when specifying trak
+// specific user data. For example, some muxers define a trak name.
+if (c->fc->nb_streams > 0 && c->trak_index != -1) {
+st = c->fc->streams[c->fc->nb_streams-1];
+st->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
+metadata = >metadata;
+} else {
+c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
+metadata = >fc->metadata;
+}
+av_dict_set(metadata, key, str, 0);
 if (*language && strcmp(language, "und")) {
 snprintf(key2, sizeof(key2), "%s-%s", key, language);
-av_dict_set(>fc->metadata, key2, str, 0);
+av_dict_set(metadata, key2, str, 0);
 }
+
 if (!strcmp(key, "encoder")) {
 int major, minor, micro;
 if (sscanf(str, "HandBrake %d.%d.%d", , , ) == 
3) {
-- 
2.17.0.windows.1

___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

[FFmpeg-devel] [PATCH v3 3/3] libavformat/mov: add fate tests for parsing of trak titles from mov format

[Nik Johnson]

Create a fate test to verify ffprobe correctly identifies stream titles
in mp4 containers.

Signed-off-by: Nik Johnson 
---
Sample file for fate uploaded to 
https://www.dropbox.com/s/8itks08yf4s1pgs/trak-name.mp4?dl=0. Should be added 
to the fate samples under mov/trak-name.mp4

 tests/fate/mov.mak   | 3 +++
 tests/ref/fate/mov-trak-name | 6 ++
 2 files changed, 9 insertions(+)
 create mode 100644 tests/ref/fate/mov-trak-name

diff --git a/tests/fate/mov.mak b/tests/fate/mov.mak
index eadee3abfa..e6dc72eccb 100644
--- a/tests/fate/mov.mak
+++ b/tests/fate/mov.mak
@@ -26,6 +26,7 @@ FATE_MOV_FFPROBE = fate-mov-neg-firstpts-discard \
fate-mov-guess-delay-1 \
fate-mov-guess-delay-2 \
fate-mov-guess-delay-3 \
+   fate-mov-trak-name \
 
 FATE_SAMPLES_AVCONV += $(FATE_MOV)
 FATE_SAMPLES_FFPROBE += $(FATE_MOV_FFPROBE)
@@ -105,3 +106,5 @@ fate-mov-gpmf-remux: REF = 8f48e435ee1f6b7e173ea756141eabf3
 fate-mov-guess-delay-1: CMD = run ffprobe$(PROGSSUF)$(EXESUF) -show_entries 
stream=has_b_frames -select_streams v 
$(TARGET_SAMPLES)/h264/h264_3bf_nopyramid_nobsrestriction.mp4
 fate-mov-guess-delay-2: CMD = run ffprobe$(PROGSSUF)$(EXESUF) -show_entries 
stream=has_b_frames -select_streams v 
$(TARGET_SAMPLES)/h264/h264_3bf_pyramid_nobsrestriction.mp4
 fate-mov-guess-delay-3: CMD = run ffprobe$(PROGSSUF)$(EXESUF) -show_entries 
stream=has_b_frames -select_streams v 
$(TARGET_SAMPLES)/h264/h264_4bf_pyramid_nobsrestriction.mp4
+
+fate-mov-trak-name: CMD = run ffprobe$(PROGSSUF)$(EXESUF) -show_entries 
stream_tags=title -select_streams a $(TARGET_SAMPLES)/mov/trak-name.mp4
diff --git a/tests/ref/fate/mov-trak-name b/tests/ref/fate/mov-trak-name
new file mode 100644
index 00..36d6c39d82
--- /dev/null
+++ b/tests/ref/fate/mov-trak-name
@@ -0,0 +1,6 @@
+[STREAM]
+TAG:title=System sounds
+[/STREAM]
+[STREAM]
+TAG:title=Microphone
+[/STREAM]
-- 
2.17.0.windows.1

___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

[FFmpeg-devel] [PATCH v3 2/3] libavformat/mov: recognize udta name tag as the stream title

[Nik Johnson]

Some muxers write the stream title in a udta atom with the tag 'name'.
Recognize 'name' tags as the stream title instead of an unknown tag.

Signed-off-by: Nik Johnson 
---
 libavformat/mov.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/libavformat/mov.c b/libavformat/mov.c
index b434802207..c2675d2644 100644
--- a/libavformat/mov.c
+++ b/libavformat/mov.c
@@ -340,6 +340,7 @@ static int mov_read_udta_string(MOVContext *c, AVIOContext 
*pb, MOVAtom atom)
 return mov_metadata_loci(c, pb, atom.size);
 case MKTAG( 'm','a','n','u'): key = "make"; break;
 case MKTAG( 'm','o','d','l'): key = "model"; break;
+case MKTAG( 'n','a','m','e'): key = "title"; raw = 1; break;
 case MKTAG( 'p','c','s','t'): key = "podcast";
 parse = mov_metadata_int8_no_padding; break;
 case MKTAG( 'p','g','a','p'): key = "gapless_playback";
-- 
2.17.0.windows.1

___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] mov: Make sure PTS are both monotonically increasing, and unique

[Derek Buitenhuis]

Hi,

On Tue, May 29, 2018 at 5:04 PM, Sasi Inguva
 wrote:
> Hi. sorry for the late reply. I sent a patch similar to this a while back
> https://patchwork.ffmpeg.org/patch/8227/  but it got lost in the sea. You
> also want to do,

Sorry I missed that! I'd prefer to use your patch over mine.

I'll hold off on a push for a day or two until I figure out how to
handle (or not handle) the midstream PTS.

- Derek
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] mov: Make sure PTS are both monotonically increasing, and unique

[Sasi Inguva]

Hi. sorry for the late reply. I sent a patch similar to this a while back
https://patchwork.ffmpeg.org/patch/8227/  but it got lost in the sea. You
also want to do,
@@ -3579,7 +3579,8 @@ static void mov_fix_index(MOVContext *mov, AVStream
*st)
 frame_duration_buffer[num_discarded_begin - 1] =
frame_duration;

-if (first_non_zero_audio_edit > 0 &&
st->codecpar->codec_id != AV_CODEC_ID_VORBIS) {
+if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO
&&
+first_non_zero_audio_edit > 0 &&
st->codecpar->codec_id != AV_CODEC_ID_VORBIS) {
 st->skip_samples += frame_duration;
 }

so that we only increment skip samples for audio streams. Otherwise patch
looks good to me.

On Thu, May 17, 2018 at 7:03 AM Derek Buitenhuis 
wrote:

> On Tue, May 15, 2018 at 8:44 PM, Derek Buitenhuis
>  wrote:
> > We already did this for audio, but it should be done for video too.
> > If we don't, seeking back to the start of the file, for example, can
> > become quite broken, since the first N packets will have repeating
> > and nonmonotonic PTS, yet they need to be decoded even if they are
> > to be discarded.
> >
> > Signed-off-by: Derek Buitenhuis 
> > ---
> >  libavformat/mov.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
>
> Ping.
>
> Is nobody outside Sasi able to review code in this part of
> mov.c? That is slightly worrying to me.
>
> - Derek
> ___
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] Limited timecode support for lavd/decklink

[Jonathan Morley]

Thank you Marton, That makes sense to me, but can you please clarify which 
context is the most appropriate to use?

Thanks,
Jon

> On May 29, 2018, at 1:41 AM, Marton Balint  wrote:
> 
> 
> 
> On Sat, 26 May 2018, Jonathan Morley wrote:
> 
>> Attaching again from another mail client.
> 
> Thanks. There is one issue I found:
> 
> You are setting >video_st->metadata from the VideoInputFrameArrived 
> callback. That runs in a separate thread from the main thread handling 
> read_packet calls, and you can only invalidate video_st->metadata in the main 
> thread. So I suggest to store the timecode string in a context variable, and 
> do av_dict_set in ff_decklink_read_packet after avpacket_queue_get.
> 
> Regards,
> Marton
> ___
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

[FFmpeg-devel] [PATCH 2/2] qt-faststart - optimize the offset change loop

[Eran Kornblau]

Hi,

The attached is a slightly more optimized (and IMHO elegant) code for updating 
the stco/co64 offsets

Thanks

Eran


0002-qt-faststart-optimize-the-offset-change-loop.patch
Description: 0002-qt-faststart-optimize-the-offset-change-loop.patch
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

[FFmpeg-devel] [PATCH 1/2] qt-faststart - stricter input validations

[Eran Kornblau]

Hi,

The attached patch fixes a couple of input validation issues in fast start that 
I noticed while going over the code

Thanks

Eran


0001-qt-faststart-stricter-input-validations.patch
Description: 0001-qt-faststart-stricter-input-validations.patch
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [GSOC] [PATCH] DNN module introduction and SRCNN filter update

[Pedro Arthur]

Patch 0001 pushed.

Thanks.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] lavfi/tests: Fix 16-bit vf_blend test to avoid memory not aligned to 2 bytes

[Paul B Mahol]

On 5/29/18, Andrey Semashev  wrote:
> On 05/24/18 00:07, Andrey Semashev wrote:
>> Generic C implementation of vf_blend performs reads and writes of 16-bit
>> elements, which requires the buffers to be aligned to at least 2-byte
>> boundary.
>>
>> Also, the change fixes source buffer overrun caused by src_offset being
>> added to to test handling of misaligned buffers.
>>
>> Fixes: #7226
>
> Ping? Any comments?

Should be OK.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [GSOC] [PATCH] DNN module introduction and SRCNN filter update

[Pedro Arthur]

2018-05-29 5:14 GMT-03:00 Sergey Lavrushkin :
> 2018-05-29 4:08 GMT+03:00 Pedro Arthur :
>>
>> 2018-05-28 19:52 GMT-03:00 Sergey Lavrushkin :
>> > 2018-05-28 9:32 GMT+03:00 Guo, Yejun :
>> >
>> >> looks that no tensorflow dependency is introduced, a new model format
>> >> is
>> >> created together with some CPU implementation for inference.   With
>> >> this
>> >> idea, Android Neural Network would be a very good reference, see
>> >> https://developer.android.google.cn/ndk/guides/neuralnetworks/. It
>> >> defines how the model is organized, and also provided a CPU optimized
>> >> inference implementation (within the NNAPI runtime, it is open source).
>> >> It
>> >> is still under development but mature enough to run some popular dnn
>> >> models
>> >> with proper performance. We can absorb some basic design. Anyway, just
>> >> a
>> >> reference fyi.  (btw, I'm not sure about any IP issue)
>> >>
>> >
>> > The idea was to first introduce something to use when tensorflow is not
>> > available. Here is another patch, that introduces tensorflow backend.
>> I think it would be better for reviewing if you send the second patch
>> in a new email.
>
>
> Then we need to push the first patch, I think.
Not necessarily, 'git send-email' may give you a glimpse of how it is done.

>
>>
>> >
>> >
>> >> For this patch, I have two comments.
>> >>
>> >> 1. change from "DNNModel* (*load_default_model)(DNNDefaultModel
>> >> model_type);" to " DNNModel* (*load_builtin_model)(DNNBuiltinModel
>> >> model_type);"
>> >> The DNNModule can be invoked by many filters,  default model is a good
>> >> name at the filter level, while built-in model is better within the DNN
>> >> scope.
>> >>
>> >> typedef struct DNNModule{
>> >> // Loads model and parameters from given file. Returns NULL if it
>> >> is
>> >> not possible.
>> >> DNNModel* (*load_model)(const char* model_filename);
>> >> // Loads one of the default models
>> >> DNNModel* (*load_default_model)(DNNDefaultModel model_type);
>> >> // Executes model with specified input and output. Returns
>> >> DNN_ERROR
>> >> otherwise.
>> >> DNNReturnType (*execute_model)(const DNNModel* model);
>> >> // Frees memory allocated for model.
>> >> void (*free_model)(DNNModel** model);
>> >> } DNNModule;
>> >>
>> >>
>> >> 2. add a new variable 'number' for DNNData/InputParams
>> >> As a typical DNN concept, the data shape usually is: > >> width, channel> or , the last component
>> >> denotes its index changes the fastest in the memory. We can add this
>> >> concept into the API, and decide to support  or  or both.
>> >
>> >
>> > I did not add number of elements in batch because I thought, that we
>> > would
>> > not feed more than one element at once to a network in a ffmpeg filter.
>> > But it can be easily added if necessary.
>> >
>> > So here is the patch that adds tensorflow backend with the previous
>> > patch.
>> > I forgot to change include guards from AVUTIL_* to AVFILTER_* in it.
>> You moved the files from libavutil to libavfilter while it was
>> proposed to move them to libavformat.
>
>
> Not only, it was also proposed to move it to libavfilter if it is going to
> be used only
> in filters. I do not know if this module is useful anywhere else besides
> libavfilter.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] lavfi/tests: Fix 16-bit vf_blend test to avoid memory not aligned to 2 bytes

[Andrey Semashev]

On 05/24/18 00:07, Andrey Semashev wrote:

Generic C implementation of vf_blend performs reads and writes of 16-bit
elements, which requires the buffers to be aligned to at least 2-byte
boundary.

Also, the change fixes source buffer overrun caused by src_offset being
added to to test handling of misaligned buffers.

Fixes: #7226


Ping? Any comments?


---
  tests/checkasm/vf_blend.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tests/checkasm/vf_blend.c b/tests/checkasm/vf_blend.c
index 912f3a2c38..a7578fec39 100644
--- a/tests/checkasm/vf_blend.c
+++ b/tests/checkasm/vf_blend.c
@@ -71,7 +71,7 @@
  w = WIDTH / depth;
 \

 \
  for (i = 0; i < BUF_UNITS - 1; i++) { 
 \
-int src_offset = i * SIZE_PER_UNIT + i; /* Test various alignments 
*/  \
+int src_offset = i * SIZE_PER_UNIT + (BUF_UNITS - 1 - i) * depth; 
/* Test various alignments */  \
  int dst_offset = i * SIZE_PER_UNIT; /* dst must be aligned */ 
 \
  randomize_buffers();  
 \
  call_ref(top1 + src_offset, w, bot1 + src_offset, w,  
 \



___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH v3] avcodec/vc1: fix out-of-bounds reference pixel replication

[Jerome Borsboom]

Out-of-bounds reference pixel replication should take into account the frame
coding mode of the reference frame(s), not the frame coding mode of the
current frame.

Signed-off-by: Jerome Borsboom 
---
Even more corrections. The starting line must also be adjusted by one for an 
opposite
refence field.

 libavcodec/vc1_mc.c | 668 ++--
 1 file changed, 385 insertions(+), 283 deletions(-)

diff --git a/libavcodec/vc1_mc.c b/libavcodec/vc1_mc.c
index 04b359204c..1b8d8799b3 100644
--- a/libavcodec/vc1_mc.c
+++ b/libavcodec/vc1_mc.c
@@ -179,12 +179,17 @@ void ff_vc1_mc_1mv(VC1Context *v, int dir)
 int i;
 uint8_t (*luty)[256], (*lutuv)[256];
 int use_ic;
+int interlace;
+int linesize, uvlinesize;
 
 if ((!v->field_mode ||
  (v->ref_field_type[dir] == 1 && v->cur_field_type == 1)) &&
 !v->s.last_picture.f->data[0])
 return;
 
+linesize = s->current_picture_ptr->f->linesize[0];
+uvlinesize = s->current_picture_ptr->f->linesize[1];
+
 mx = s->mv[dir][0][0];
 my = s->mv[dir][0][1];
 
@@ -220,6 +225,7 @@ void ff_vc1_mc_1mv(VC1Context *v, int dir)
 luty  = v->curr_luty;
 lutuv = v->curr_lutuv;
 use_ic = *v->curr_use_ic;
+interlace = 1;
 } else {
 srcY = s->last_picture.f->data[0];
 srcU = s->last_picture.f->data[1];
@@ -227,6 +233,7 @@ void ff_vc1_mc_1mv(VC1Context *v, int dir)
 luty  = v->last_luty;
 lutuv = v->last_lutuv;
 use_ic = v->last_use_ic;
+interlace = s->last_picture.f->interlaced_frame;
 }
 } else {
 srcY = s->next_picture.f->data[0];
@@ -235,6 +242,7 @@ void ff_vc1_mc_1mv(VC1Context *v, int dir)
 luty  = v->next_luty;
 lutuv = v->next_lutuv;
 use_ic = v->next_use_ic;
+interlace = s->next_picture.f->interlaced_frame;
 }
 
 if (!srcY || !srcU) {
@@ -269,9 +277,9 @@ void ff_vc1_mc_1mv(VC1Context *v, int dir)
 srcV += uvsrc_y * s->uvlinesize + uvsrc_x;
 
 if (v->field_mode && v->ref_field_type[dir]) {
-srcY += s->current_picture_ptr->f->linesize[0];
-srcU += s->current_picture_ptr->f->linesize[1];
-srcV += s->current_picture_ptr->f->linesize[2];
+srcY += linesize;
+srcU += uvlinesize;
+srcV += uvlinesize;
 }
 
 /* for grayscale we should not try to read from unknown area */
@@ -289,112 +297,105 @@ void ff_vc1_mc_1mv(VC1Context *v, int dir)
 const int k = 17 + s->mspel * 2;
 
 srcY -= s->mspel * (1 + s->linesize);
-if (v->fcm == ILACE_FRAME) {
-if (src_y - s->mspel & 1) {
-s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer,
- srcY,
- 2 * s->linesize,
- 2 * s->linesize,
- k,
- k + 1 >> 1,
- src_x - s->mspel,
- src_y - s->mspel >> 1,
- s->h_edge_pos,
- v_edge_pos + 1 >> 1);
-s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer + s->linesize,
- srcY + s->linesize,
- 2 * s->linesize,
- 2 * s->linesize,
- k,
- k >> 1,
- src_x - s->mspel,
- src_y - s->mspel + 1 >> 1,
- s->h_edge_pos,
- v_edge_pos >> 1);
-} else {
-s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer,
- srcY,
- 2 * s->linesize,
- 2 * s->linesize,
- k,
- k + 1 >> 1,
- src_x - s->mspel,
- src_y - s->mspel >> 1,
- s->h_edge_pos,
- v_edge_pos >> 1);
-s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer + s->linesize,
- srcY + s->linesize,
- 2 * s->linesize,
- 2 * s->linesize,
+if (interlace) {
+s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer,
+ srcY,
+ linesize << 1,
+ linesize << 1,
+   

Re: [FFmpeg-devel] [PATCH 2/2] ffmpeg: Use the colour properties from the input stream when doing transcode

[Tobias Rapp]

On 28.05.2018 09:30, Xiang, Haihao wrote:

On Sat, 2018-05-26 at 17:29 +0100, Mark Thompson wrote:

On 25/05/18 07:57, Tobias Rapp wrote:

On 25.05.2018 07:58, Xiang, Haihao wrote:

On Thu, 2018-05-24 at 11:15 +0100, Mark Thompson wrote:


For example:

ffmpeg -i bt709_input.mkv -vf colorspace=bt2020 bt2020_output.mkv

will have the output file marked as BT.709 after this patch, where
previously
it was "unspecified".  (Explicitly setting -color_primaries/-color_trc/-
colorspace on the output works in both cases.)


I agree with you it's not worse than before as we don't get the expected
result
in both cases.


Not quite: When a file says "I don't know this property value" you have a
chance to lookup the value somewhere else or use a default. When it says "I
know the value" and gives a wrong value, you completely loose trust.


Right, that is a compelling argument.  I agree with you, so I definitely won't
apply the patch in this form.



According the comment in avcodec.h, the color properties in AVCodecContext
should be set by user for encoding. I think ffmpeg is the user in the case
below. Where are the color properties set if we don't set the default values in
init_output_stream_encode()?

ffmpeg -hwaccel vaapi -hwaccel_device /dev/dri/renderD128 -hwaccel_output_format
vaapi -i input-with-hdr.mkv -c:v hevc_vaapi -profile:v main10 output.h265


Setting color properties in init_output_stream_encode() basically looks 
OK, but the question "when" seems more important than "where". It should 
be called after some output frame is available, and use that to 
initialize the encoder. Currently the current AVFrame is not available 
in init_output_stream_encode(), only the buffersink properties.


Hopefully somebody with more knowledge about FFmpeg infrastructure will 
make a recommendation on how this should be solved: By adding color 
properties and other frame data to the buffersink interface? Or by 
making the current/last frame available in init_output_stream_encode()? 
Or ...?


It doesn't need to be auto-negotiated like in the mentioned patch series 
for color_range, it just needs to be forwarded properly from filter 
graph output to the encoder.


Regards,
Tobias

___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] lavfi: add opencl tonemap filter.

[Niklas Haas]

I see no obvious issues with the algorithm. (Though I haven't tested it)

So "LGTM"

On Tue, 29 May 2018 13:54:27 +0800, Ruiling Song  wrote:
> This filter does HDR(HDR10/HLG) to SDR conversion with tone-mapping.
> 
> An example command to use this filter with vaapi codecs:
> FFMPEG -init_hw_device vaapi=va:/dev/dri/renderD128 -init_hw_device \
> opencl=ocl@va -hwaccel vaapi -hwaccel_device va -hwaccel_output_format \
> vaapi -i INPUT -filter_hw_device ocl -filter_complex \
> '[0:v]hwmap,tonemap_opencl=t=bt2020:tonemap=linear:format=p010[x1]; \
> [x1]hwmap=derive_device=vaapi:reverse=1' -c:v hevc_vaapi -profile 2 OUTPUT
> 
> v2:
> add peak detection.
> 
> Signed-off-by: Ruiling Song 
> ---
>  configure  |   1 +
>  libavfilter/Makefile   |   2 +
>  libavfilter/allfilters.c   |   1 +
>  libavfilter/colorspace_basic.c |  89 +
>  libavfilter/colorspace_basic.h |  40 ++
>  libavfilter/opencl/colorspace_basic.cl | 187 ++
>  libavfilter/opencl/tonemap.cl  | 278 ++
>  libavfilter/opencl_source.h|   2 +
>  libavfilter/vf_tonemap_opencl.c| 655 
> +
>  9 files changed, 1255 insertions(+)
>  create mode 100644 libavfilter/colorspace_basic.c
>  create mode 100644 libavfilter/colorspace_basic.h
>  create mode 100644 libavfilter/opencl/colorspace_basic.cl
>  create mode 100644 libavfilter/opencl/tonemap.cl
>  create mode 100644 libavfilter/vf_tonemap_opencl.c
> 
> diff --git a/configure b/configure
> index e52f8f8..ee3586b 100755
> --- a/configure
> +++ b/configure
> @@ -3401,6 +3401,7 @@ tinterlace_filter_deps="gpl"
>  tinterlace_merge_test_deps="tinterlace_filter"
>  tinterlace_pad_test_deps="tinterlace_filter"
>  tonemap_filter_deps="const_nan"
> +tonemap_opencl_filter_deps="opencl"
>  unsharp_opencl_filter_deps="opencl"
>  uspp_filter_deps="gpl avcodec"
>  vaguedenoiser_filter_deps="gpl"
> diff --git a/libavfilter/Makefile b/libavfilter/Makefile
> index c68ef05..0915656 100644
> --- a/libavfilter/Makefile
> +++ b/libavfilter/Makefile
> @@ -352,6 +352,8 @@ OBJS-$(CONFIG_TINTERLACE_FILTER) += 
> vf_tinterlace.o
>  OBJS-$(CONFIG_TLUT2_FILTER)  += vf_lut2.o framesync.o
>  OBJS-$(CONFIG_TMIX_FILTER)   += vf_mix.o framesync.o
>  OBJS-$(CONFIG_TONEMAP_FILTER)+= vf_tonemap.o
> +OBJS-$(CONFIG_TONEMAP_OPENCL_FILTER) += vf_tonemap_opencl.o 
> colorspace_basic.o opencl.o \
> +opencl/tonemap.o 
> opencl/colorspace_basic.o
>  OBJS-$(CONFIG_TRANSPOSE_FILTER)  += vf_transpose.o
>  OBJS-$(CONFIG_TRIM_FILTER)   += trim.o
>  OBJS-$(CONFIG_UNPREMULTIPLY_FILTER)  += vf_premultiply.o framesync.o
> diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
> index b44093d..6873bab 100644
> --- a/libavfilter/allfilters.c
> +++ b/libavfilter/allfilters.c
> @@ -343,6 +343,7 @@ extern AVFilter ff_vf_tinterlace;
>  extern AVFilter ff_vf_tlut2;
>  extern AVFilter ff_vf_tmix;
>  extern AVFilter ff_vf_tonemap;
> +extern AVFilter ff_vf_tonemap_opencl;
>  extern AVFilter ff_vf_transpose;
>  extern AVFilter ff_vf_trim;
>  extern AVFilter ff_vf_unpremultiply;
> diff --git a/libavfilter/colorspace_basic.c b/libavfilter/colorspace_basic.c
> new file mode 100644
> index 000..93f9f08
> --- /dev/null
> +++ b/libavfilter/colorspace_basic.c
> @@ -0,0 +1,89 @@
> +/*
> + * This file is part of FFmpeg.
> + *
> + * FFmpeg is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU Lesser General Public
> + * License as published by the Free Software Foundation; either
> + * version 2.1 of the License, or (at your option) any later version.
> + *
> + * FFmpeg is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> + * Lesser General Public License for more details.
> + *
> + * You should have received a copy of the GNU Lesser General Public
> + * License along with FFmpeg; if not, write to the Free Software
> + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 
> USA
> + */
> +
> +#include "colorspace_basic.h"
> +
> +
> +void invert_matrix3x3(const double in[3][3], double out[3][3])
> +{
> +double m00 = in[0][0], m01 = in[0][1], m02 = in[0][2],
> +   m10 = in[1][0], m11 = in[1][1], m12 = in[1][2],
> +   m20 = in[2][0], m21 = in[2][1], m22 = in[2][2];
> +int i, j;
> +double det;
> +
> +out[0][0] =  (m11 * m22 - m21 * m12);
> +out[0][1] = -(m01 * m22 - m21 * m02);
> +out[0][2] =  (m01 * m12 - m11 * m02);
> +out[1][0] = -(m10 * m22 - m20 * m12);
> +out[1][1] =  (m00 * m22 - m20 * m02);
> +out[1][2] = -(m00 * m12 - m10 * m02);
> +out[2][0] =  (m10 * m21 - m20 * m11);
> +out[2][1] = -(m00 * m21 - m20 * 

Re: [FFmpeg-devel] [PATCH v2] avcodec/vc1: fix out-of-bounds reference pixel replication

[Jerome Borsboom]

Out-of-bounds reference pixel replication should take into account the frame
coding mode of the reference frame(s), not the frame coding mode of the
current frame.

Signed-off-by: Jerome Borsboom 
---
Does this resolve the SIGSEGV? I think I made a mistake in the calculation of 
the starting
line for progressive reference pictures when the current picture is a field 
interlaced
picture. Instead of adjusting the edge position, the starting line must be 
adjusted as
the vertical stride for replication is half the stride of the field interlaced 
picture.

 libavcodec/vc1_mc.c | 659 ++--
 1 file changed, 379 insertions(+), 280 deletions(-)

diff --git a/libavcodec/vc1_mc.c b/libavcodec/vc1_mc.c
index 04b359204c..16fc531712 100644
--- a/libavcodec/vc1_mc.c
+++ b/libavcodec/vc1_mc.c
@@ -179,12 +179,17 @@ void ff_vc1_mc_1mv(VC1Context *v, int dir)
 int i;
 uint8_t (*luty)[256], (*lutuv)[256];
 int use_ic;
+int interlace;
+int linesize, uvlinesize;
 
 if ((!v->field_mode ||
  (v->ref_field_type[dir] == 1 && v->cur_field_type == 1)) &&
 !v->s.last_picture.f->data[0])
 return;
 
+linesize = s->current_picture_ptr->f->linesize[0];
+uvlinesize = s->current_picture_ptr->f->linesize[1];
+
 mx = s->mv[dir][0][0];
 my = s->mv[dir][0][1];
 
@@ -220,6 +225,7 @@ void ff_vc1_mc_1mv(VC1Context *v, int dir)
 luty  = v->curr_luty;
 lutuv = v->curr_lutuv;
 use_ic = *v->curr_use_ic;
+interlace = 1;
 } else {
 srcY = s->last_picture.f->data[0];
 srcU = s->last_picture.f->data[1];
@@ -227,6 +233,7 @@ void ff_vc1_mc_1mv(VC1Context *v, int dir)
 luty  = v->last_luty;
 lutuv = v->last_lutuv;
 use_ic = v->last_use_ic;
+interlace = s->last_picture.f->interlaced_frame;
 }
 } else {
 srcY = s->next_picture.f->data[0];
@@ -235,6 +242,7 @@ void ff_vc1_mc_1mv(VC1Context *v, int dir)
 luty  = v->next_luty;
 lutuv = v->next_lutuv;
 use_ic = v->next_use_ic;
+interlace = s->next_picture.f->interlaced_frame;
 }
 
 if (!srcY || !srcU) {
@@ -269,9 +277,9 @@ void ff_vc1_mc_1mv(VC1Context *v, int dir)
 srcV += uvsrc_y * s->uvlinesize + uvsrc_x;
 
 if (v->field_mode && v->ref_field_type[dir]) {
-srcY += s->current_picture_ptr->f->linesize[0];
-srcU += s->current_picture_ptr->f->linesize[1];
-srcV += s->current_picture_ptr->f->linesize[2];
+srcY += linesize;
+srcU += uvlinesize;
+srcV += uvlinesize;
 }
 
 /* for grayscale we should not try to read from unknown area */
@@ -289,112 +297,104 @@ void ff_vc1_mc_1mv(VC1Context *v, int dir)
 const int k = 17 + s->mspel * 2;
 
 srcY -= s->mspel * (1 + s->linesize);
-if (v->fcm == ILACE_FRAME) {
-if (src_y - s->mspel & 1) {
-s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer,
- srcY,
- 2 * s->linesize,
- 2 * s->linesize,
- k,
- k + 1 >> 1,
- src_x - s->mspel,
- src_y - s->mspel >> 1,
- s->h_edge_pos,
- v_edge_pos + 1 >> 1);
-s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer + s->linesize,
- srcY + s->linesize,
- 2 * s->linesize,
- 2 * s->linesize,
- k,
- k >> 1,
- src_x - s->mspel,
- src_y - s->mspel + 1 >> 1,
- s->h_edge_pos,
- v_edge_pos >> 1);
-} else {
-s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer,
- srcY,
- 2 * s->linesize,
- 2 * s->linesize,
- k,
- k + 1 >> 1,
- src_x - s->mspel,
- src_y - s->mspel >> 1,
- s->h_edge_pos,
- v_edge_pos >> 1);
-s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer + s->linesize,
- srcY + s->linesize,
- 2 * s->linesize,
- 2 * s->linesize,
+ 

Re: [FFmpeg-devel] [PATCH] lavfi: add opencl tonemap filter.

[Song, Ruiling]

> -Original Message-
> From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On Behalf
> Of myp...@gmail.com
> Sent: Tuesday, May 29, 2018 3:40 PM
> To: FFmpeg development discussions and patches  de...@ffmpeg.org>
> Subject: Re: [FFmpeg-devel] [PATCH] lavfi: add opencl tonemap filter.
> 
> 2018-05-29 13:54 GMT+08:00 Ruiling Song :
> > This filter does HDR(HDR10/HLG) to SDR conversion with tone-mapping.
> >
> > An example command to use this filter with vaapi codecs:
> > FFMPEG -init_hw_device vaapi=va:/dev/dri/renderD128 -init_hw_device \
> > opencl=ocl@va -hwaccel vaapi -hwaccel_device va -
> hwaccel_output_format \
> > vaapi -i INPUT -filter_hw_device ocl -filter_complex \
> > '[0:v]hwmap,tonemap_opencl=t=bt2020:tonemap=linear:format=p010[x1];
> \
> > [x1]hwmap=derive_device=vaapi:reverse=1' -c:v hevc_vaapi -profile 2
> OUTPUT
> >
> > v2:
> > add peak detection.
> >
> > Signed-off-by: Ruiling Song 
> > +static int tonemap_opencl_config_output(AVFilterLink *outlink)
> > +{
> > +AVFilterContext *avctx = outlink->src;
> > +TonemapOpenCLContext *s = avctx->priv;
> > +int ret;
> > +if (s->format == AV_PIX_FMT_NONE)
> > +av_log(avctx, AV_LOG_WARNING, "format not set, use default
> format NV12\n");
>  I think we can give a default format with AV_PIX_FMT_NV12 in
> tonemap_opencl_options[] for this case
> and I think now we only support NV12/P010 output in current implement.
Sounds good.

> > +{ "format","output pixel format", OFFSET(format),
> AV_OPT_TYPE_PIXEL_FMT, {.i64 = AV_PIX_FMT_NONE},
> AV_PIX_FMT_NONE, AV_PIX_FMT_GBRAP12LE, FLAGS, "fmt" },
> Missing the sub-option nv12 and p010 ?
Seems like using AV_OPT_TYPE_PIXEL_FMT, the framework parsed the user format 
argument correctly.
So I think no need to add sub-options?

Thanks!
Ruiling
> > +{ "peak",  "signal peak override", OFFSET(peak),
> AV_OPT_TYPE_DOUBLE, {.dbl = 0}, 0, DBL_MAX, FLAGS },
> > +{ "param", "tonemap parameter",   OFFSET(param),
> AV_OPT_TYPE_DOUBLE, {.dbl = NAN}, DBL_MIN, DBL_MAX, FLAGS },
> > +{ "desat", "desaturation parameter",   OFFSET(desat_param),
> AV_OPT_TYPE_DOUBLE, {.dbl = 0.5}, 0, DBL_MAX, FLAGS },
> > +{ "threshold", "scene detection threshold",   OFFSET(scene_threshold),
> AV_OPT_TYPE_DOUBLE, {.dbl = 0.2}, 0, DBL_MAX, FLAGS },
> > +{ NULL }
> > +};
> > +
> > +AVFILTER_DEFINE_CLASS(tonemap_opencl);
> > +
> > +static const AVFilterPad tonemap_opencl_inputs[] = {
> > +{
> > +.name = "default",
> > +.type = AVMEDIA_TYPE_VIDEO,
> > +.filter_frame = _opencl_filter_frame,
> > +.config_props = _opencl_filter_config_input,
> > +},
> > +{ NULL }
> > +};
> > +
> > +static const AVFilterPad tonemap_opencl_outputs[] = {
> > +{
> > +.name = "default",
> > +.type = AVMEDIA_TYPE_VIDEO,
> > +.config_props = _opencl_config_output,
> > +},
> > +{ NULL }
> > +};
> > +
> > +AVFilter ff_vf_tonemap_opencl = {
> > +.name   = "tonemap_opencl",
> > +.description= NULL_IF_CONFIG_SMALL("perform HDR to SDR
> conversion with tonemapping"),
> > +.priv_size  = sizeof(TonemapOpenCLContext),
> > +.priv_class = _opencl_class,
> > +.init   = _opencl_filter_init,
> > +.uninit = _opencl_uninit,
> > +.query_formats  = _opencl_filter_query_formats,
> > +.inputs = tonemap_opencl_inputs,
> > +.outputs= tonemap_opencl_outputs,
> > +.flags_internal = FF_FILTER_FLAG_HWFRAME_AWARE,
> > +};
> > --
> > 2.7.4
> >
> > ___
> > ffmpeg-devel mailing list
> > ffmpeg-devel@ffmpeg.org
> > http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
> ___
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [GSOC] [PATCH] DNN module introduction and SRCNN filter update

[Sergey Lavrushkin]

2018-05-29 4:08 GMT+03:00 Pedro Arthur :

> 2018-05-28 19:52 GMT-03:00 Sergey Lavrushkin :
> > 2018-05-28 9:32 GMT+03:00 Guo, Yejun :
> >
> >> looks that no tensorflow dependency is introduced, a new model format is
> >> created together with some CPU implementation for inference.   With this
> >> idea, Android Neural Network would be a very good reference, see
> >> https://developer.android.google.cn/ndk/guides/neuralnetworks/. It
> >> defines how the model is organized, and also provided a CPU optimized
> >> inference implementation (within the NNAPI runtime, it is open source).
> It
> >> is still under development but mature enough to run some popular dnn
> models
> >> with proper performance. We can absorb some basic design. Anyway, just a
> >> reference fyi.  (btw, I'm not sure about any IP issue)
> >>
> >
> > The idea was to first introduce something to use when tensorflow is not
> > available. Here is another patch, that introduces tensorflow backend.
> I think it would be better for reviewing if you send the second patch
> in a new email.


Then we need to push the first patch, I think.


> >
> >
> >> For this patch, I have two comments.
> >>
> >> 1. change from "DNNModel* (*load_default_model)(DNNDefaultModel
> >> model_type);" to " DNNModel* (*load_builtin_model)(DNNBuiltinModel
> >> model_type);"
> >> The DNNModule can be invoked by many filters,  default model is a good
> >> name at the filter level, while built-in model is better within the DNN
> >> scope.
> >>
> >> typedef struct DNNModule{
> >> // Loads model and parameters from given file. Returns NULL if it is
> >> not possible.
> >> DNNModel* (*load_model)(const char* model_filename);
> >> // Loads one of the default models
> >> DNNModel* (*load_default_model)(DNNDefaultModel model_type);
> >> // Executes model with specified input and output. Returns DNN_ERROR
> >> otherwise.
> >> DNNReturnType (*execute_model)(const DNNModel* model);
> >> // Frees memory allocated for model.
> >> void (*free_model)(DNNModel** model);
> >> } DNNModule;
> >>
> >>
> >> 2. add a new variable 'number' for DNNData/InputParams
> >> As a typical DNN concept, the data shape usually is:  >> width, channel> or , the last component
> >> denotes its index changes the fastest in the memory. We can add this
> >> concept into the API, and decide to support  or  or both.
> >
> >
> > I did not add number of elements in batch because I thought, that we
> would
> > not feed more than one element at once to a network in a ffmpeg filter.
> > But it can be easily added if necessary.
> >
> > So here is the patch that adds tensorflow backend with the previous
> patch.
> > I forgot to change include guards from AVUTIL_* to AVFILTER_* in it.
> You moved the files from libavutil to libavfilter while it was
> proposed to move them to libavformat.


Not only, it was also proposed to move it to libavfilter if it is going to
be used only
in filters. I do not know if this module is useful anywhere else besides
libavfilter.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] Limited timecode support for lavd/decklink

[Marton Balint]

On Sat, 26 May 2018, Jonathan Morley wrote:


Attaching again from another mail client.


Thanks. There is one issue I found:

You are setting >video_st->metadata from the VideoInputFrameArrived 
callback. That runs in a separate thread from the main thread 
handling read_packet calls, and you can only invalidate 
video_st->metadata in the main thread. So I suggest to store the timecode 
string in a context variable, and do av_dict_set in 
ff_decklink_read_packet after avpacket_queue_get.


Regards,
Marton
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH]ffplay: Mention codec_name if decoder for codec_id could not be found.

[Marton Balint]

On Tue, 29 May 2018, Carl Eugen Hoyos wrote:


Hi!

Attached patch makes debugging a little easier imo.

Please comment, Carl Eugen

diff --git a/fftools/ffplay.c b/fftools/ffplay.c
index dcca9c2..f9571d7 100644
--- a/fftools/ffplay.c
+++ b/fftools/ffplay.c
@@ -2578,7 +2578,7 @@ static int stream_component_open(VideoState *is, int 
stream_index)
if (forced_codec_name) av_log(NULL, AV_LOG_WARNING,
  "No codec could be found with name 
'%s'\n", forced_codec_name);
else   av_log(NULL, AV_LOG_WARNING,
-  "No codec could be found with id %d\n", 
avctx->codec_id);
+  "No codec could be found with id %d (%s)\n", 
avctx->codec_id, avcodec_get_name(avctx->codec_id));


Maybe go one step further, and change the error message to

"No decoder could be found for codec %s\n", avcodec_get_name(avctx->codec_id)

I don't see any use for dumping the codec_id, it has no usefulness to the 
end user IMHO.


Regards,
Marton
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] lavfi: add opencl tonemap filter.

[myp...@gmail.com]

2018-05-29 13:54 GMT+08:00 Ruiling Song :
> This filter does HDR(HDR10/HLG) to SDR conversion with tone-mapping.
>
> An example command to use this filter with vaapi codecs:
> FFMPEG -init_hw_device vaapi=va:/dev/dri/renderD128 -init_hw_device \
> opencl=ocl@va -hwaccel vaapi -hwaccel_device va -hwaccel_output_format \
> vaapi -i INPUT -filter_hw_device ocl -filter_complex \
> '[0:v]hwmap,tonemap_opencl=t=bt2020:tonemap=linear:format=p010[x1]; \
> [x1]hwmap=derive_device=vaapi:reverse=1' -c:v hevc_vaapi -profile 2 OUTPUT
>
> v2:
> add peak detection.
>
> Signed-off-by: Ruiling Song 
> ---
>  configure  |   1 +
>  libavfilter/Makefile   |   2 +
>  libavfilter/allfilters.c   |   1 +
>  libavfilter/colorspace_basic.c |  89 +
>  libavfilter/colorspace_basic.h |  40 ++
>  libavfilter/opencl/colorspace_basic.cl | 187 ++
>  libavfilter/opencl/tonemap.cl  | 278 ++
>  libavfilter/opencl_source.h|   2 +
>  libavfilter/vf_tonemap_opencl.c| 655 
> +
>  9 files changed, 1255 insertions(+)
>  create mode 100644 libavfilter/colorspace_basic.c
>  create mode 100644 libavfilter/colorspace_basic.h
>  create mode 100644 libavfilter/opencl/colorspace_basic.cl
>  create mode 100644 libavfilter/opencl/tonemap.cl
>  create mode 100644 libavfilter/vf_tonemap_opencl.c
>
> diff --git a/configure b/configure
> index e52f8f8..ee3586b 100755
> --- a/configure
> +++ b/configure
> @@ -3401,6 +3401,7 @@ tinterlace_filter_deps="gpl"
>  tinterlace_merge_test_deps="tinterlace_filter"
>  tinterlace_pad_test_deps="tinterlace_filter"
>  tonemap_filter_deps="const_nan"
> +tonemap_opencl_filter_deps="opencl"
>  unsharp_opencl_filter_deps="opencl"
>  uspp_filter_deps="gpl avcodec"
>  vaguedenoiser_filter_deps="gpl"
> diff --git a/libavfilter/Makefile b/libavfilter/Makefile
> index c68ef05..0915656 100644
> --- a/libavfilter/Makefile
> +++ b/libavfilter/Makefile
> @@ -352,6 +352,8 @@ OBJS-$(CONFIG_TINTERLACE_FILTER) += 
> vf_tinterlace.o
>  OBJS-$(CONFIG_TLUT2_FILTER)  += vf_lut2.o framesync.o
>  OBJS-$(CONFIG_TMIX_FILTER)   += vf_mix.o framesync.o
>  OBJS-$(CONFIG_TONEMAP_FILTER)+= vf_tonemap.o
> +OBJS-$(CONFIG_TONEMAP_OPENCL_FILTER) += vf_tonemap_opencl.o 
> colorspace_basic.o opencl.o \
> +opencl/tonemap.o 
> opencl/colorspace_basic.o
>  OBJS-$(CONFIG_TRANSPOSE_FILTER)  += vf_transpose.o
>  OBJS-$(CONFIG_TRIM_FILTER)   += trim.o
>  OBJS-$(CONFIG_UNPREMULTIPLY_FILTER)  += vf_premultiply.o framesync.o
> diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
> index b44093d..6873bab 100644
> --- a/libavfilter/allfilters.c
> +++ b/libavfilter/allfilters.c
> @@ -343,6 +343,7 @@ extern AVFilter ff_vf_tinterlace;
>  extern AVFilter ff_vf_tlut2;
>  extern AVFilter ff_vf_tmix;
>  extern AVFilter ff_vf_tonemap;
> +extern AVFilter ff_vf_tonemap_opencl;
>  extern AVFilter ff_vf_transpose;
>  extern AVFilter ff_vf_trim;
>  extern AVFilter ff_vf_unpremultiply;
> diff --git a/libavfilter/colorspace_basic.c b/libavfilter/colorspace_basic.c
> new file mode 100644
> index 000..93f9f08
> --- /dev/null
> +++ b/libavfilter/colorspace_basic.c
> @@ -0,0 +1,89 @@
> +/*
> + * This file is part of FFmpeg.
> + *
> + * FFmpeg is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU Lesser General Public
> + * License as published by the Free Software Foundation; either
> + * version 2.1 of the License, or (at your option) any later version.
> + *
> + * FFmpeg is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> + * Lesser General Public License for more details.
> + *
> + * You should have received a copy of the GNU Lesser General Public
> + * License along with FFmpeg; if not, write to the Free Software
> + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 
> USA
> + */
> +
> +#include "colorspace_basic.h"
> +
> +
> +void invert_matrix3x3(const double in[3][3], double out[3][3])
> +{
> +double m00 = in[0][0], m01 = in[0][1], m02 = in[0][2],
> +   m10 = in[1][0], m11 = in[1][1], m12 = in[1][2],
> +   m20 = in[2][0], m21 = in[2][1], m22 = in[2][2];
> +int i, j;
> +double det;
> +
> +out[0][0] =  (m11 * m22 - m21 * m12);
> +out[0][1] = -(m01 * m22 - m21 * m02);
> +out[0][2] =  (m01 * m12 - m11 * m02);
> +out[1][0] = -(m10 * m22 - m20 * m12);
> +out[1][1] =  (m00 * m22 - m20 * m02);
> +out[1][2] = -(m00 * m12 - m10 * m02);
> +out[2][0] =  (m10 * m21 - m20 * m11);
> +out[2][1] = -(m00 * m21 - m20 * m01);
> +out[2][2] =  (m00 * m11 - m10 * m01);
> +
> +det = m00 * out[0][0] + m10 * out[0][1] +