date:20181030

Re: [FFmpeg-devel] [PATCH V2] libavfilter/vaapi: enable vaapi rotation feature via call Intel iHD driver

2018-10-30 Thread Zhou, Zachary



> -Original Message-
> From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On Behalf Of
> Michael Niedermayer
> Sent: Wednesday, October 31, 2018 3:53 AM
> To: FFmpeg development discussions and patches 
> Subject: Re: [FFmpeg-devel] [PATCH V2] libavfilter/vaapi: enable vaapi
> rotation feature via call Intel iHD driver
> 
> On Mon, Oct 29, 2018 at 01:42:55AM +, Zhou, Zachary wrote:
> >
> >
> > > -Original Message-
> > > From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On
> > > Behalf Of Michael Niedermayer
> > > Sent: Saturday, October 27, 2018 5:16 AM
> > > To: FFmpeg development discussions and patches
> > > 
> > > Subject: Re: [FFmpeg-devel] [PATCH V2] libavfilter/vaapi: enable
> > > vaapi rotation feature via call Intel iHD driver
> > >
> > > On Thu, Oct 25, 2018 at 02:52:01PM +0800, Zachary Zhou wrote:
> > > > It supports clockwise rotation by 0/90/180/270 degrees defined in
> > > > va/va_vpp.h, tested following command line on SKL platform
> > > >
> > > > dir=0 for 0   degree
> > > > dir=1 for 90  degree
> > > > dir=2 for 180 degree
> > > > dir=3 for 270 degree
> > > >
> > > > ffmpeg -hwaccel vaapi -vaapi_device /dev/dri/renderD128
> > > > -hwaccel_output_format vaapi -i input.264 -vf "rotation_vaapi=dir=1"
> > > > -c:v h264_vaapi output.h264
> > > >
> > > > Signed-off-by: Zachary Zhou 
> > > > ---
> > > >  configure |   3 +
> > > >  libavfilter/Makefile  |   1 +
> > > >  libavfilter/allfilters.c  |   1 +
> > > >  libavfilter/vaapi_vpp.h   |   1 +
> > > >  libavfilter/vf_rotate_vaapi.c | 252
> > > > ++
> > > >  5 files changed, 258 insertions(+)  create mode 100644
> > > > libavfilter/vf_rotate_vaapi.c
> > >
> > > still fails to build on 32bit x86
> > >
> > > src/libavfilter/vf_rotate_vaapi.c: In function
> > > ‘rotation_vaapi_build_filter_params’:
> > > src/libavfilter/vf_rotate_vaapi.c:69:23: error: ‘VAProcPipelineCaps’
> > > has no member named ‘rotation_flags’
> > >  if (!pipeline_caps.rotation_flags) {
> > >^
> > > src/libavfilter/vf_rotate_vaapi.c:95:33: error: ‘VAProcPipelineCaps’
> > > has no member named ‘rotation_flags’
> > >  support_flag = pipeline_caps.rotation_flags &
> > >  ^
> > > src/libavfilter/vf_rotate_vaapi.c: In function 
> > > ‘rotation_vaapi_filter_frame’:
> > > src/libavfilter/vf_rotate_vaapi.c:152:15: error:
> > > ‘VAProcPipelineParameterBuffer’ has no member named ‘rotation_state’
> > >  params.rotation_state = vpp_ctx->rotation_state;
> > >^
> > > make: *** [libavfilter/vf_rotate_vaapi.o] Error 1
> > > make: *** Waiting for unfinished jobs
> > >
> > > grep VAAPI ffbuild/config.mak
> > > !HAVE_VAAPI_DRM=yes
> > > !HAVE_VAAPI_X11=yes
> > > !HAVE_OPENCL_VAAPI_BEIGNET=yes
> > > !HAVE_OPENCL_VAAPI_INTEL_MEDIA=yes
> > > !CONFIG_VAAPI_ENCODE_EXAMPLE=yes
> > > !CONFIG_VAAPI_TRANSCODE_EXAMPLE=yes
> > > !CONFIG_VAAPI=yes
> > > !CONFIG_VAAPI_1=yes
> > > !CONFIG_VAAPI_ENCODE=yes
> > > !CONFIG_H264_VAAPI_ENCODER=yes
> > > !CONFIG_HEVC_VAAPI_ENCODER=yes
> > > !CONFIG_MJPEG_VAAPI_ENCODER=yes
> > > !CONFIG_MPEG2_VAAPI_ENCODER=yes
> > > !CONFIG_VP8_VAAPI_ENCODER=yes
> > > !CONFIG_VP9_VAAPI_ENCODER=yes
> > > !CONFIG_H263_VAAPI_HWACCEL=yes
> > > !CONFIG_H264_VAAPI_HWACCEL=yes
> > > !CONFIG_HEVC_VAAPI_HWACCEL=yes
> > > !CONFIG_MJPEG_VAAPI_HWACCEL=yes
> > > !CONFIG_MPEG2_VAAPI_HWACCEL=yes
> > > !CONFIG_MPEG4_VAAPI_HWACCEL=yes
> > > !CONFIG_VC1_VAAPI_HWACCEL=yes
> > > !CONFIG_VP8_VAAPI_HWACCEL=yes
> > > !CONFIG_VP9_VAAPI_HWACCEL=yes
> > > !CONFIG_WMV3_VAAPI_HWACCEL=yes
> > > !CONFIG_DEINTERLACE_VAAPI_FILTER=yes
> > > !CONFIG_DENOISE_VAAPI_FILTER=yes
> > > !CONFIG_PROCAMP_VAAPI_FILTER=yes
> > > !CONFIG_SCALE_VAAPI_FILTER=yes
> > > !CONFIG_SHARPNESS_VAAPI_FILTER=yes
> > > CONFIG_ROTATION_VAAPI_FILTER=yes
> > >
> > > [...]
> >
> > Thanks Michael, Can you share your command line for the build script
> configure ?
> > Do I need more setting on the configure file ?
> 
> i simplified it a bit, this can still reproduce it:
> make distclean ; ../configure   --cc='ccache gcc -m32'   && make -j12
> 
> 
> src/libavfilter/vf_rotate_vaapi.c: In function
> ‘rotation_vaapi_build_filter_params’:
> src/libavfilter/vf_rotate_vaapi.c:69:23: error: ‘VAProcPipelineCaps’ has no
> member named ‘rotation_flags’
>  if (!pipeline_caps.rotation_flags) {
>^
> src/libavfilter/vf_rotate_vaapi.c:95:33: error: ‘VAProcPipelineCaps’ has no
> member named ‘rotation_flags’
>  support_flag = pipeline_caps.rotation_flags &
>  ^
> src/libavfilter/vf_rotate_vaapi.c: In function ‘rotation_vaapi_filter_frame’:
> src/libavfilter/vf_rotate_vaapi.c:152:15: error:
> ‘VAProcPipelineParameterBuffer’ has no member named ‘rotation_state’
>  params.rotation_state = vpp_ctx->rotation_state;
>^
> make: *** [libavfilter/vf_rotate_vaapi.o] Error 1
> make: ***

[FFmpeg-devel] [PATCH V3] libavfilter/vaapi: enable vaapi rotation feature via call Intel iHD driver

2018-10-30 Thread Zachary Zhou

It supports clockwise rotation by 0/90/180/270 degrees defined in
va/va_vpp.h, tested following command line on SKL platform

ffmpeg -hwaccel vaapi -vaapi_device /dev/dri/renderD128
-hwaccel_output_format vaapi -i input.264 -vf "rotation_vaapi=angle=180"
-c:v h264_vaapi output.h264

Signed-off-by: Zachary Zhou 
---
 configure |   4 +
 libavfilter/Makefile  |   1 +
 libavfilter/allfilters.c  |   1 +
 libavfilter/vaapi_vpp.h   |   1 +
 libavfilter/vf_rotate_vaapi.c | 252 ++
 5 files changed, 259 insertions(+)
 create mode 100644 libavfilter/vf_rotate_vaapi.c

diff --git a/configure b/configure
index 85d5dd5962..4b5718135e 100755
--- a/configure
+++ b/configure
@@ -3439,6 +3439,7 @@ scale_filter_deps="swscale"
 scale_qsv_filter_deps="libmfx"
 select_filter_select="pixelutils"
 sharpness_vaapi_filter_deps="vaapi"
+rotation_vaapi_filter_deps="vaapi"
 showcqt_filter_deps="avcodec avformat swscale"
 showcqt_filter_suggest="libfontconfig libfreetype"
 showcqt_filter_select="fft"
@@ -6390,6 +6391,9 @@ if enabled vaapi; then
 fi
 
 check_cpp_condition vaapi_1 "va/va.h" "VA_CHECK_VERSION(1, 0, 0)"
+if ! check_struct "va/va.h" "struct _VAProcPipelineCaps" rotation_flags; 
then
+disable rotation_vaapi_filter
+fi
 fi
 
 if enabled_all opencl libdrm ; then
diff --git a/libavfilter/Makefile b/libavfilter/Makefile
index 108a2f87d7..534650364a 100644
--- a/libavfilter/Makefile
+++ b/libavfilter/Makefile
@@ -349,6 +349,7 @@ OBJS-$(CONFIG_SETRANGE_FILTER)   += 
vf_setparams.o
 OBJS-$(CONFIG_SETSAR_FILTER) += vf_aspect.o
 OBJS-$(CONFIG_SETTB_FILTER)  += settb.o
 OBJS-$(CONFIG_SHARPNESS_VAAPI_FILTER)+= vf_misc_vaapi.o vaapi_vpp.o
+OBJS-$(CONFIG_ROTATION_VAAPI_FILTER) += vf_rotate_vaapi.o vaapi_vpp.o
 OBJS-$(CONFIG_SHOWINFO_FILTER)   += vf_showinfo.o
 OBJS-$(CONFIG_SHOWPALETTE_FILTER)+= vf_showpalette.o
 OBJS-$(CONFIG_SHUFFLEFRAMES_FILTER)  += vf_shuffleframes.o
diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
index 557590850b..4b90a7f440 100644
--- a/libavfilter/allfilters.c
+++ b/libavfilter/allfilters.c
@@ -333,6 +333,7 @@ extern AVFilter ff_vf_setrange;
 extern AVFilter ff_vf_setsar;
 extern AVFilter ff_vf_settb;
 extern AVFilter ff_vf_sharpness_vaapi;
+extern AVFilter ff_vf_rotation_vaapi;
 extern AVFilter ff_vf_showinfo;
 extern AVFilter ff_vf_showpalette;
 extern AVFilter ff_vf_shuffleframes;
diff --git a/libavfilter/vaapi_vpp.h b/libavfilter/vaapi_vpp.h
index 0bc31018d4..cfe19b689f 100644
--- a/libavfilter/vaapi_vpp.h
+++ b/libavfilter/vaapi_vpp.h
@@ -44,6 +44,7 @@ typedef struct VAAPIVPPContext {
 int output_width;   // computed width
 int output_height;  // computed height
 
+int rotation_state;
 VABufferID filter_buffers[VAProcFilterCount];
 intnb_filter_buffers;
 
diff --git a/libavfilter/vf_rotate_vaapi.c b/libavfilter/vf_rotate_vaapi.c
new file mode 100644
index 00..82e2a0fce4
--- /dev/null
+++ b/libavfilter/vf_rotate_vaapi.c
@@ -0,0 +1,252 @@
+/*
+ * 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 
+
+#include "libavutil/avassert.h"
+#include "libavutil/mem.h"
+#include "libavutil/opt.h"
+#include "libavutil/pixdesc.h"
+
+#include "avfilter.h"
+#include "formats.h"
+#include "internal.h"
+#include "vaapi_vpp.h"
+
+// Rotation angle values
+enum RotationAngle {
+ROTATION_0   = 0,
+ROTATION_90  = 90,
+ROTATION_180 = 180,
+ROTATION_270 = 270,
+
+ROTATION_MIN = ROTATION_0,
+ROTATION_MAX = ROTATION_270,
+ROTATION_DEFAULT = ROTATION_0,
+};
+
+typedef struct RotationVAAPIContext {
+VAAPIVPPContext vpp_ctx; // must be the first field
+
+int rotation;
+} RotationVAAPIContext;
+
+static int rotation_vaapi_build_filter_params(AVFilterContext *avctx)
+{
+VAAPIVPPContext *vpp_ctx  = avctx->priv;
+RotationVAAPIContext *ctx = avctx->priv;
+
+VAStatus vas;
+int support_flag;
+
+VAProcPipelineCaps pipeline_caps;
+
+memset(_caps, 0, sizeof(pipeline_caps));
+vas = vaQueryVideoProcPipelineCaps(vpp_ctx->hwctx->display,
+   vpp_ctx->va_context,
+

Re: [FFmpeg-devel] [PATCH] doc/fate.texi: Mention that samples should be uploaded before pushing dependant commits

2018-10-30 Thread Michael Niedermayer

On Wed, Oct 24, 2018 at 10:34:39AM +0530, Gyan wrote:
> On Wed, Oct 24, 2018 at 3:09 AM Michael Niedermayer 
> wrote:
> 
> > Signed-off-by: Michael Niedermayer 
> > ---
> >  doc/fate.texi | 2 ++
> >  1 file changed, 2 insertions(+)
> >
> > diff --git a/doc/fate.texi b/doc/fate.texi
> > index a352994230..9ca77f7772 100644
> > --- a/doc/fate.texi
> > +++ b/doc/fate.texi
> > @@ -155,6 +155,8 @@ space on each client, network bandwidth and so on
> > benefit from smaller test case
> >  Also keep in mind older checkouts use existing sample files, that means in
> >  practice generally do not replace, remove or overwrite files as it likely
> > would
> >  break older checkouts or releases.
> > +Also all needed samples for a commit should be uploaded before the commit
> > is
> > +pushed. Ideally 24h before.
> >
> 
> Nit: merge the 2nd sentence into the first.

can you suggest a exact wording for that ?
The sentance seems already a bit long to me and describing a slightly
different area of potential issues


> 
> Also, dependant -> dependent.

locally fixed

thx

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

If you fake or manipulate statistics in a paper in physics you will never
get a job again.
If you fake or manipulate statistics in a paper in medicin you will get
a job for life at the pharma industry.


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

Re: [FFmpeg-devel] [PATCH] Revert "lavc/v4l2_m2m_enc: Add missing braces around initializers."

2018-10-30 Thread Ronak Patel


> On Oct 27, 2018, at 5:25 PM, Carl Eugen Hoyos  wrote:
> 
> 2018-10-27 20:50 GMT+02:00, Mark Thompson :
>> This reverts commit 6dbb64fdccafe846aaec75d3784f7ad49d8af5df.
>> 
>> The additional braces cause build errors with Linux headers earlier
>> than 4.5 because the first element of the structure was not originally
>> a union.
> 
>> Not sure what compiler was warning about these, but it's definitely
>> incorrect for it to do so.
> 
> Must have been NDK clang.

This fix isn’t enough for this. The references to the variables in the union 
fail to compile. This doesn’t fix that.


> 
> Sorry, Carl Eugen
> ___
> 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 4/5] lavc/qsvenc: add an option to set h264 pps for every frame

2018-10-30 Thread Mark Thompson

On 30/10/18 09:21, Li, Zhong wrote:
>> From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On Behalf
>> Of Moritz Barsnick
>> Sent: Friday, October 26, 2018 7:49 PM
>> To: FFmpeg development discussions and patches
>> 
>> Subject: Re: [FFmpeg-devel] [PATCH 4/5] lavc/qsvenc: add an option to set
>> h264 pps for every frame
>>
>> On Thu, Oct 25, 2018 at 20:36:10 +0800, Zhong Li wrote:
>>> RepeatPPS is enabled by default in mfx. It is not necessary mostly and
>>> wasting encoding bits.
>>> Add an option to control it and disable it by default.
>>
>> Could this change in behavior surprise anyone? If it's okay to change,
>> perhaps at least bump lavc's micro version?
>>
>> Moritz
> 
> I doubt how many people would like to repeat every frame's pps. 
> Most codecs (x264/x265, and nvenc) won't do that by default.

Huh, right.  I feel like I must be missing something here - why would anyone 
ever want this option enabled?  It doesn't even help with seeking to arbitrary 
points in streams without global headers because you don't have the SPS as well.

If there isn't some hidden reason for this, it sounds fine to just switch it 
off completely without an option to reenable.

Thanks,

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

Re: [FFmpeg-devel] [PATCH 1/2] lavc/libdavs2: remove incorrect frame settings

2018-10-30 Thread Mark Thompson

On 30/10/18 14:01, hwren wrote:
> Signed-off-by: hwren 
> ---
>  libavcodec/libdavs2.c | 7 ---
>  1 file changed, 7 deletions(-)
> 
> diff --git a/libavcodec/libdavs2.c b/libavcodec/libdavs2.c
> index cadf995..581d568 100644
> --- a/libavcodec/libdavs2.c
> +++ b/libavcodec/libdavs2.c
> @@ -96,13 +96,6 @@ static int davs2_dump_frames(AVCodecContext *avctx, 
> davs2_picture_t *pic,
> pic->widths[plane] * bytes_per_sample);
>  }
>  
> -frame->width = cad->headerset.width;
> -frame->height= cad->headerset.height;
> -frame->pts   = cad->out_frame.pts;
> -frame->pict_type = pic->type;
> -frame->format= avctx->pix_fmt;
> -
> -cad->decoded_frames++;
>  return 1;
>  }
>  
> 

This seems like it loses track of the timestamps?  Output pts is always 
AV_NOPTS_VALUE after this change.

(Patch 3/3 should also be squashed into this one, since this is removing the 
last use of the decoded_frames counter.)

Thanks,

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

Re: [FFmpeg-devel] [PATCH 3/3] avcodec/cbs_vp9: discard empty fragments

2018-10-30 Thread James Almer

On 10/30/2018 8:19 PM, Mark Thompson wrote:
> On 30/10/18 19:21, James Almer wrote:
>> Signed-off-by: James Almer 
>> ---
>>  libavcodec/cbs_vp9.c | 3 +++
>>  1 file changed, 3 insertions(+)
>>
>> diff --git a/libavcodec/cbs_vp9.c b/libavcodec/cbs_vp9.c
>> index 7498be4b73..812be4ddd5 100644
>> --- a/libavcodec/cbs_vp9.c
>> +++ b/libavcodec/cbs_vp9.c
>> @@ -403,6 +403,9 @@ static int cbs_vp9_split_fragment(CodedBitstreamContext 
>> *ctx,
>>  uint8_t superframe_header;
>>  int err;
>>  
>> +if (frag->data_size <= 0)
>> +return 0;
>> +
>>  // Last byte in the packet.
>>  superframe_header = frag->data[frag->data_size - 1];
>>  
>>
> 
> Seems fine, but why would an empty fragment appear here?

I noticed this when i tried to reimplement the vp9 parser using cbs_vp9.
Libavformat passed dummy zero sized packets that resulted in cbs errors
trying to read the frame header.

> 
> (Given that H.26[45] has pretty much the same check, maybe it should be in 
> the generic code.)

And AV1 has a while (size > 0), so yeah, it may be a good idea.

> 
> Thanks,
> 
> - Mark
> ___
> 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 1/5] lavc/qsvenc: add forced_idr opiton

2018-10-30 Thread Mark Thompson

On 30/10/18 09:49, Li, Zhong wrote:
>> From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On Behalf
>> Of Mark Thompson
>> Sent: Tuesday, October 30, 2018 5:06 AM
>> To: ffmpeg-devel@ffmpeg.org
>> Subject: Re: [FFmpeg-devel] [PATCH 1/5] lavc/qsvenc: add forced_idr opiton
>>
>> On 25/10/18 13:36, Zhong Li wrote:
>>> This option can be used to repect original input I/IDR frame type.
>>>
>>> Signed-off-by: Zhong Li 
>>> ---
>>>  libavcodec/qsvenc.c | 7 +++
>>>  libavcodec/qsvenc.h | 2 ++
>>>  2 files changed, 9 insertions(+)
>>>
>>> diff --git a/libavcodec/qsvenc.c b/libavcodec/qsvenc.c index
>>> 948751d..e534dcf 100644
>>> --- a/libavcodec/qsvenc.c
>>> +++ b/libavcodec/qsvenc.c
>>> @@ -1192,6 +1192,13 @@ static int encode_frame(AVCodecContext
>> *avctx, QSVEncContext *q,
>>>  if (qsv_frame) {
>>>  surf = _frame->surface;
>>>  enc_ctrl = _frame->enc_ctrl;
>>> +
>>> +if (q->forced_idr >= 0 && frame->pict_type ==
>> AV_PICTURE_TYPE_I) {
>>> +enc_ctrl->FrameType = MFX_FRAMETYPE_I |
>> MFX_FRAMETYPE_REF;
>>> +if (q->forced_idr || frame->key_frame)
>>> +enc_ctrl->FrameType |= MFX_FRAMETYPE_IDR;
>>> +} else
>>> +enc_ctrl->FrameType = MFX_FRAMETYPE_UNKNOWN;
>>>  }
>>>
>>>  ret = av_new_packet(_pkt, q->packet_size); diff --git
>>> a/libavcodec/qsvenc.h b/libavcodec/qsvenc.h index 055b4a6..1f97f77
>>> 100644
>>> --- a/libavcodec/qsvenc.h
>>> +++ b/libavcodec/qsvenc.h
>>> @@ -87,6 +87,7 @@
>>>  { "adaptive_i", "Adaptive I-frame placement",
>> OFFSET(qsv.adaptive_i), AV_OPT_TYPE_INT, { .i64 = -1 }, -1,
>> 1, VE }, \
>>>  { "adaptive_b", "Adaptive B-frame placement",
>> OFFSET(qsv.adaptive_b), AV_OPT_TYPE_INT, { .i64 = -1 }, -1,
>> 1, VE }, \
>>>  { "b_strategy", "Strategy to choose between I/P/B-frames",
>> OFFSET(qsv.b_strategy),AV_OPT_TYPE_INT, { .i64 = -1 }, -1,
>> 1, VE }, \
>>> +{ "forced_idr", "Forcing I frames as IDR frames",
>> OFFSET(qsv.forced_idr), AV_OPT_TYPE_INT, { .i64 = -1 }, -1,
>> 1, VE }, \
>>>
>>>  typedef int SetEncodeCtrlCB (AVCodecContext *avctx,
>>>   const AVFrame *frame,
>> mfxEncodeCtrl*
>>> enc_ctrl); @@ -168,6 +169,7 @@ typedef struct QSVEncContext {  #endif
>>>  char *load_plugins;
>>>  SetEncodeCtrlCB *set_encode_ctrl_cb;
>>> +int forced_idr;
>>>  } QSVEncContext;
>>>
>>>  int ff_qsv_enc_init(AVCodecContext *avctx, QSVEncContext *q);
>>>
>>
>> This seems confusing, because it doesn't match what forced_idr does in any
>> other encoder.
>>
>> Checking of pict_type for AV_PICTURE_TYPE_I in order to get a key frame
>> (of whatever kind) is always enabled if supported (many encoders).  The
>> "forced_idr" option to H.26[45] encoders (libx264, libx265) then forces that
>> to be an IDR frame, not just an I frame.
>>
>> - Mark
> Yup, I know it doesn’t match other encoders such as libx264/libx265/nvenc. 
> However, it is my intentional behavior. I believe current implement in 
> libx264/libx265 is not complete.
> One case is ignored: user just want to keep the gop structure as input, not 
> to force all I frames as IDR frames.
> So I have an idea:
> Default value = -1, ignore the input gop structure. 
> 0: respect input gop structure but don't force I frame as IDR frame. 
> 1: force all I frame as IDR frame.

This sounds like two independent options.  One is the "forced-idr" option 
implemented by several other encoders (notably libx264, which is the most 
commonly-used external encoder), which looks like a sensible addition to me.  
The second is an "ignore user-set pict_type" option, which I don't understand 
the need for at all - it's never set unless the user deliberately wants to use 
that feature (e.g. by using the force_key_frames option in the ffmpeg utility), 
so why would you want to have a special way to override that?

Thanks,

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

Re: [FFmpeg-devel] [PATCH 1/3] avcodec/cbs_vp9: keep track of reference frames

2018-10-30 Thread Mark Thompson

On 30/10/18 19:21, James Almer wrote:
> Signed-off-by: James Almer 
> ---
>  libavcodec/cbs_vp9.h | 16 
>  libavcodec/cbs_vp9_syntax_template.c | 60 ++--
>  2 files changed, 72 insertions(+), 4 deletions(-)
> 
On 30/10/18 19:21, James Almer wrote:
> Derived from profile_low_bit and profile_high_bit.
> 
> Signed-off-by: James Almer 
> ---
>  libavcodec/cbs_vp9.h |  2 ++
>  libavcodec/cbs_vp9_syntax_template.c | 13 ++---
>  2 files changed, 8 insertions(+), 7 deletions(-)

Both LGTM.

Thanks!

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

Re: [FFmpeg-devel] [PATCH 3/3] avcodec/cbs_vp9: discard empty fragments

2018-10-30 Thread Mark Thompson

On 30/10/18 19:21, James Almer wrote:
> Signed-off-by: James Almer 
> ---
>  libavcodec/cbs_vp9.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/libavcodec/cbs_vp9.c b/libavcodec/cbs_vp9.c
> index 7498be4b73..812be4ddd5 100644
> --- a/libavcodec/cbs_vp9.c
> +++ b/libavcodec/cbs_vp9.c
> @@ -403,6 +403,9 @@ static int cbs_vp9_split_fragment(CodedBitstreamContext 
> *ctx,
>  uint8_t superframe_header;
>  int err;
>  
> +if (frag->data_size <= 0)
> +return 0;
> +
>  // Last byte in the packet.
>  superframe_header = frag->data[frag->data_size - 1];
>  
> 

Seems fine, but why would an empty fragment appear here?

(Given that H.26[45] has pretty much the same check, maybe it should be in the 
generic code.)

Thanks,

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

[FFmpeg-devel] [PATCH] lavf/hlsenc: Fix mixed declarations and code

2018-10-30 Thread Mark Thompson

---
 libavformat/hlsenc.c | 13 +
 1 file changed, 9 insertions(+), 4 deletions(-)

diff --git a/libavformat/hlsenc.c b/libavformat/hlsenc.c
index f8f060d065..73282ed31a 100644
--- a/libavformat/hlsenc.c
+++ b/libavformat/hlsenc.c
@@ -2348,26 +2348,31 @@ static int hls_write_trailer(struct AVFormatContext *s)
 return AVERROR(ENOMEM);
 }
 if ( hls->segment_type == SEGMENT_TYPE_FMP4) {
+int range_length;
+
 if (!vs->init_range_length) {
+uint8_t *buffer;
+int byterange_mode;
+
 av_write_frame(vs->avf, NULL); /* Flush any buffered data */
 avio_flush(oc->pb);
 
-uint8_t *buffer = NULL;
-int range_length = avio_close_dyn_buf(oc->pb, );
+buffer = NULL;
+range_length = avio_close_dyn_buf(oc->pb, );
 avio_write(vs->out, buffer, range_length);
 av_free(buffer);
 vs->init_range_length = range_length;
 avio_open_dyn_buf(>pb);
 vs->packets_written = 0;
 vs->start_pos = range_length;
-int byterange_mode = (hls->flags & HLS_SINGLE_FILE) || 
(hls->max_seg_size > 0);
+byterange_mode = (hls->flags & HLS_SINGLE_FILE) || 
(hls->max_seg_size > 0);
 if (!byterange_mode) {
 ff_format_io_close(s, >out);
 hlsenc_io_close(s, >out, vs->base_output_dirname);
 }
 }
 
-int range_length = 0;
+range_length = 0;
 if (!(hls->flags & HLS_SINGLE_FILE)) {
 ret = hlsenc_io_open(s, >out, vs->avf->url, NULL);
 if (ret < 0) {
-- 
2.19.1
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avfilter: add (a)graphmonitor filter(s)

2018-10-30 Thread Lou Logan

On Tue, Oct 30, 2018, at 12:30 PM, Paul B Mahol wrote:
>
> Will apply soon.

Are you going to add any documentation?
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] doc/hls: fix grammar for HLS options

2018-10-30 Thread Lou Logan

On Tue, 30 Oct 2018 21:07:16 +0100
Werner Robitza  wrote:

> From 7da20791b3d146e49855f975c8b5102e8cbfd9f2 Mon Sep 17 00:00:00 2001
> From: Werner Robitza 
> Date: Tue, 30 Oct 2018 20:22:55 +0100
> Subject: [PATCH] doc/hls: fix grammar for HLS options
> 
> This fixes the grammar of two HLS option descriptions and makes them less
> ambiguous.
> 
> Signed-off-by: Werner Robitza 
> ---
>  doc/muxers.texi | 9 -
>  1 file changed, 4 insertions(+), 5 deletions(-)

Pushed with a few minor changes. Thanks.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avfilter: add (a)graphmonitor filter(s)

2018-10-30 Thread Paul B Mahol

On 10/29/18, Paul B Mahol  wrote:
> Signed-off-by: Paul B Mahol 
> ---
>  libavfilter/Makefile |   2 +
>  libavfilter/allfilters.c |   2 +
>  libavfilter/f_graphmonitor.c | 417 +++
>  3 files changed, 421 insertions(+)
>  create mode 100644 libavfilter/f_graphmonitor.c

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

[FFmpeg-devel] [PATCH] doc/hls: fix grammar for HLS options

2018-10-30 Thread Werner Robitza




0001-doc-hls-fix-grammar-for-HLS-options.patch
Description: Binary data
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH V2] libavfilter/vaapi: enable vaapi rotation feature via call Intel iHD driver

2018-10-30 Thread Michael Niedermayer

On Mon, Oct 29, 2018 at 01:42:55AM +, Zhou, Zachary wrote:
> 
> 
> > -Original Message-
> > From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On Behalf Of
> > Michael Niedermayer
> > Sent: Saturday, October 27, 2018 5:16 AM
> > To: FFmpeg development discussions and patches 
> > Subject: Re: [FFmpeg-devel] [PATCH V2] libavfilter/vaapi: enable vaapi
> > rotation feature via call Intel iHD driver
> > 
> > On Thu, Oct 25, 2018 at 02:52:01PM +0800, Zachary Zhou wrote:
> > > It supports clockwise rotation by 0/90/180/270 degrees defined in
> > > va/va_vpp.h, tested following command line on SKL platform
> > >
> > > dir=0 for 0   degree
> > > dir=1 for 90  degree
> > > dir=2 for 180 degree
> > > dir=3 for 270 degree
> > >
> > > ffmpeg -hwaccel vaapi -vaapi_device /dev/dri/renderD128
> > > -hwaccel_output_format vaapi -i input.264 -vf "rotation_vaapi=dir=1"
> > > -c:v h264_vaapi output.h264
> > >
> > > Signed-off-by: Zachary Zhou 
> > > ---
> > >  configure |   3 +
> > >  libavfilter/Makefile  |   1 +
> > >  libavfilter/allfilters.c  |   1 +
> > >  libavfilter/vaapi_vpp.h   |   1 +
> > >  libavfilter/vf_rotate_vaapi.c | 252
> > > ++
> > >  5 files changed, 258 insertions(+)
> > >  create mode 100644 libavfilter/vf_rotate_vaapi.c
> > 
> > still fails to build on 32bit x86
> > 
> > src/libavfilter/vf_rotate_vaapi.c: In function
> > ‘rotation_vaapi_build_filter_params’:
> > src/libavfilter/vf_rotate_vaapi.c:69:23: error: ‘VAProcPipelineCaps’ has no
> > member named ‘rotation_flags’
> >  if (!pipeline_caps.rotation_flags) {
> >^
> > src/libavfilter/vf_rotate_vaapi.c:95:33: error: ‘VAProcPipelineCaps’ has no
> > member named ‘rotation_flags’
> >  support_flag = pipeline_caps.rotation_flags &
> >  ^
> > src/libavfilter/vf_rotate_vaapi.c: In function 
> > ‘rotation_vaapi_filter_frame’:
> > src/libavfilter/vf_rotate_vaapi.c:152:15: error:
> > ‘VAProcPipelineParameterBuffer’ has no member named ‘rotation_state’
> >  params.rotation_state = vpp_ctx->rotation_state;
> >^
> > make: *** [libavfilter/vf_rotate_vaapi.o] Error 1
> > make: *** Waiting for unfinished jobs
> > 
> > grep VAAPI ffbuild/config.mak
> > !HAVE_VAAPI_DRM=yes
> > !HAVE_VAAPI_X11=yes
> > !HAVE_OPENCL_VAAPI_BEIGNET=yes
> > !HAVE_OPENCL_VAAPI_INTEL_MEDIA=yes
> > !CONFIG_VAAPI_ENCODE_EXAMPLE=yes
> > !CONFIG_VAAPI_TRANSCODE_EXAMPLE=yes
> > !CONFIG_VAAPI=yes
> > !CONFIG_VAAPI_1=yes
> > !CONFIG_VAAPI_ENCODE=yes
> > !CONFIG_H264_VAAPI_ENCODER=yes
> > !CONFIG_HEVC_VAAPI_ENCODER=yes
> > !CONFIG_MJPEG_VAAPI_ENCODER=yes
> > !CONFIG_MPEG2_VAAPI_ENCODER=yes
> > !CONFIG_VP8_VAAPI_ENCODER=yes
> > !CONFIG_VP9_VAAPI_ENCODER=yes
> > !CONFIG_H263_VAAPI_HWACCEL=yes
> > !CONFIG_H264_VAAPI_HWACCEL=yes
> > !CONFIG_HEVC_VAAPI_HWACCEL=yes
> > !CONFIG_MJPEG_VAAPI_HWACCEL=yes
> > !CONFIG_MPEG2_VAAPI_HWACCEL=yes
> > !CONFIG_MPEG4_VAAPI_HWACCEL=yes
> > !CONFIG_VC1_VAAPI_HWACCEL=yes
> > !CONFIG_VP8_VAAPI_HWACCEL=yes
> > !CONFIG_VP9_VAAPI_HWACCEL=yes
> > !CONFIG_WMV3_VAAPI_HWACCEL=yes
> > !CONFIG_DEINTERLACE_VAAPI_FILTER=yes
> > !CONFIG_DENOISE_VAAPI_FILTER=yes
> > !CONFIG_PROCAMP_VAAPI_FILTER=yes
> > !CONFIG_SCALE_VAAPI_FILTER=yes
> > !CONFIG_SHARPNESS_VAAPI_FILTER=yes
> > CONFIG_ROTATION_VAAPI_FILTER=yes
> > 
> > [...]
> 
> Thanks Michael, Can you share your command line for the build script 
> configure ?
> Do I need more setting on the configure file ?

i simplified it a bit, this can still reproduce it:
make distclean ; ../configure   --cc='ccache gcc -m32'   && make -j12


src/libavfilter/vf_rotate_vaapi.c: In function 
‘rotation_vaapi_build_filter_params’:
src/libavfilter/vf_rotate_vaapi.c:69:23: error: ‘VAProcPipelineCaps’ has no 
member named ‘rotation_flags’
 if (!pipeline_caps.rotation_flags) {
   ^
src/libavfilter/vf_rotate_vaapi.c:95:33: error: ‘VAProcPipelineCaps’ has no 
member named ‘rotation_flags’
 support_flag = pipeline_caps.rotation_flags &
 ^
src/libavfilter/vf_rotate_vaapi.c: In function ‘rotation_vaapi_filter_frame’:
src/libavfilter/vf_rotate_vaapi.c:152:15: error: 
‘VAProcPipelineParameterBuffer’ has no member named ‘rotation_state’
 params.rotation_state = vpp_ctx->rotation_state;
   ^
make: *** [libavfilter/vf_rotate_vaapi.o] Error 1
make: *** Waiting for unfinished jobs

[...]

-- 
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 2/4] cbs_h265: Add PTL parsing for Main 10 Still Picture profile

2018-10-30 Thread James Almer

On 10/27/2018 6:39 PM, Mark Thompson wrote:
> This was added in the 2018 version of the standard.
> ---
>  libavcodec/cbs_h265_syntax_template.c | 5 +
>  1 file changed, 5 insertions(+)
> 
> diff --git a/libavcodec/cbs_h265_syntax_template.c 
> b/libavcodec/cbs_h265_syntax_template.c
> index d4e4f7b1c2..e43f3caf99 100644
> --- a/libavcodec/cbs_h265_syntax_template.c
> +++ b/libavcodec/cbs_h265_syntax_template.c
> @@ -130,6 +130,11 @@ static int 
> FUNC(profile_tier_level)(CodedBitstreamContext *ctx, RWContext *rw,
>  fixed(24, general_reserved_zero_34bits, 0);
>  fixed(10, general_reserved_zero_34bits, 0);
>  }
> +} else if (profile_compatible(2)) {
> +fixed(7, general_reserved_zero_7bits, 0);
> +flag(general_one_picture_only_constraint_flag);
> +fixed(24, general_reserved_zero_35bits, 0);
> +fixed(11, general_reserved_zero_35bits, 0);

Fun get_bits() constrain :p

It may be worth looking at enabling CACHED_BITSTREAM_READER. get_bits()
can read up 32 bits with it, and it may be faster overall.

>  } else {
>  fixed(24, general_reserved_zero_43bits, 0);
>  fixed(19, general_reserved_zero_43bits, 0);
> 

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

Re: [FFmpeg-devel] [PATCH 2/2] libx264: Pass the reordered_opaque field through the encoder

2018-10-30 Thread Martin Storsjö


On Tue, 30 Oct 2018, Derek Buitenhuis wrote:


On 29/10/2018 21:06, Martin Storsjö wrote:

As I guess there can be old frames in flight, the only safe option is to
enlarge, not to shrink it. But in case a realloc moves the array, the old
pointers end up pretty useless.


Just always allocate the max (which is known for H.264), and adjust 
nb_reordered_opaque
as need be, on reconfig, no?


Hmm, that might make sense, but with a little twist. The max reordered 
frames for H.264 is known, but onto that you also get more delay due to 
frame threads and other details that this function within x264 knows 
about. So that would make it [H264 max reordering] + [threads] + 
[constant] or something such?


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

Re: [FFmpeg-devel] [PATCH 1/4] cbs_h264: Include SEI type names in trace output

2018-10-30 Thread James Almer

On 10/27/2018 6:39 PM, Mark Thompson wrote:
> ---
>  libavcodec/cbs_h264_syntax_template.c | 16 
>  1 file changed, 16 insertions(+)
> 
> diff --git a/libavcodec/cbs_h264_syntax_template.c 
> b/libavcodec/cbs_h264_syntax_template.c
> index 9e29132fec..dbf9ff1268 100644
> --- a/libavcodec/cbs_h264_syntax_template.c
> +++ b/libavcodec/cbs_h264_syntax_template.c
> @@ -513,6 +513,8 @@ static int 
> FUNC(sei_buffering_period)(CodedBitstreamContext *ctx, RWContext *rw,
>  const H264RawSPS *sps;
>  int err, i, length;
>  
> +HEADER("Buffering Period");
> +
>  ue(seq_parameter_set_id, 0, 31);
>  
>  sps = h264->sps[current->seq_parameter_set_id];
> @@ -605,6 +607,8 @@ static int FUNC(sei_pic_timing)(CodedBitstreamContext 
> *ctx, RWContext *rw,
>  const H264RawSPS *sps;
>  int err;
>  
> +HEADER("Picture Timing");
> +
>  sps = h264->active_sps;
>  if (!sps) {
>  // If there is exactly one possible SPS but it is not yet active
> @@ -674,6 +678,8 @@ static int FUNC(sei_pan_scan_rect)(CodedBitstreamContext 
> *ctx, RWContext *rw,
>  {
>  int err, i;
>  
> +HEADER("Pan-Scan Rectangle");
> +
>  ue(pan_scan_rect_id, 0, UINT32_MAX - 1);
>  flag(pan_scan_rect_cancel_flag);
>  
> @@ -699,6 +705,8 @@ static int 
> FUNC(sei_user_data_registered)(CodedBitstreamContext *ctx, RWContext
>  {
>  int err, i, j;
>  
> +HEADER("User Data Registered ITU-T T.35");
> +
>  u(8, itu_t_t35_country_code, 0x00, 0xff);
>  if (current->itu_t_t35_country_code != 0xff)
>  i = 1;
> @@ -731,6 +739,8 @@ static int 
> FUNC(sei_user_data_unregistered)(CodedBitstreamContext *ctx, RWContex
>  {
>  int err, i;
>  
> +HEADER("User Data Unregistered");
> +
>  #ifdef READ
>  if (*payload_size < 16) {
>  av_log(ctx->log_ctx, AV_LOG_ERROR,
> @@ -758,6 +768,8 @@ static int FUNC(sei_recovery_point)(CodedBitstreamContext 
> *ctx, RWContext *rw,
>  {
>  int err;
>  
> +HEADER("Recovery Point");
> +
>  ue(recovery_frame_cnt, 0, 65535);
>  flag(exact_match_flag);
>  flag(broken_link_flag);
> @@ -771,6 +783,8 @@ static int 
> FUNC(sei_display_orientation)(CodedBitstreamContext *ctx, RWContext *
>  {
>  int err;
>  
> +HEADER("Display Orientation");
> +
>  flag(display_orientation_cancel_flag);
>  if (!current->display_orientation_cancel_flag) {
>  flag(hor_flip);
> @@ -788,6 +802,8 @@ static int 
> FUNC(sei_mastering_display_colour_volume)(CodedBitstreamContext *ctx,
>  {
>  int err, c;
>  
> +HEADER("Mastering Display Colour Volume");
> +
>  for (c = 0; c < 3; c++) {
>  us(16, display_primaries_x[c], 0, 5, 1, c);
>  us(16, display_primaries_y[c], 0, 5, 1, c);

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

[FFmpeg-devel] [PATCH 3/3] avcodec/cbs_vp9: discard empty fragments

2018-10-30 Thread James Almer

Signed-off-by: James Almer 
---
 libavcodec/cbs_vp9.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/libavcodec/cbs_vp9.c b/libavcodec/cbs_vp9.c
index 7498be4b73..812be4ddd5 100644
--- a/libavcodec/cbs_vp9.c
+++ b/libavcodec/cbs_vp9.c
@@ -403,6 +403,9 @@ static int cbs_vp9_split_fragment(CodedBitstreamContext 
*ctx,
 uint8_t superframe_header;
 int err;
 
+if (frag->data_size <= 0)
+return 0;
+
 // Last byte in the packet.
 superframe_header = frag->data[frag->data_size - 1];
 
-- 
2.19.0

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

[FFmpeg-devel] [PATCH 2/3] avcodec/cbs_vp9: store profile in the private context

2018-10-30 Thread James Almer

Derived from profile_low_bit and profile_high_bit.

Signed-off-by: James Almer 
---
 libavcodec/cbs_vp9.h |  2 ++
 libavcodec/cbs_vp9_syntax_template.c | 13 ++---
 2 files changed, 8 insertions(+), 7 deletions(-)

diff --git a/libavcodec/cbs_vp9.h b/libavcodec/cbs_vp9.h
index 12689e51cc..3f1473c9f8 100644
--- a/libavcodec/cbs_vp9.h
+++ b/libavcodec/cbs_vp9.h
@@ -192,6 +192,8 @@ typedef struct VP9ReferenceFrameState {
 } VP9ReferenceFrameState;
 
 typedef struct CodedBitstreamVP9Context {
+int profile;
+
 // Frame dimensions in 8x8 mode info blocks.
 uint16_t mi_cols;
 uint16_t mi_rows;
diff --git a/libavcodec/cbs_vp9_syntax_template.c 
b/libavcodec/cbs_vp9_syntax_template.c
index 7b90775ed6..28f95dcd32 100644
--- a/libavcodec/cbs_vp9_syntax_template.c
+++ b/libavcodec/cbs_vp9_syntax_template.c
@@ -278,15 +278,14 @@ static int 
FUNC(uncompressed_header)(CodedBitstreamContext *ctx, RWContext *rw,
  VP9RawFrameHeader *current)
 {
 CodedBitstreamVP9Context *vp9 = ctx->priv_data;
-int profile, i;
-int err;
+int err, i;
 
 f(2, frame_marker);
 
 f(1, profile_low_bit);
 f(1, profile_high_bit);
-profile = (current->profile_high_bit << 1) + current->profile_low_bit;
-if (profile == 3)
+vp9->profile = (current->profile_high_bit << 1) + current->profile_low_bit;
+if (vp9->profile == 3)
 f(1, profile_reserved_zero);
 
 f(1, show_existing_frame);
@@ -304,7 +303,7 @@ static int FUNC(uncompressed_header)(CodedBitstreamContext 
*ctx, RWContext *rw,
 
 if (current->frame_type == VP9_KEY_FRAME) {
 CHECK(FUNC(frame_sync_code)(ctx, rw, current));
-CHECK(FUNC(color_config)(ctx, rw, current, profile));
+CHECK(FUNC(color_config)(ctx, rw, current, vp9->profile));
 CHECK(FUNC(frame_size)(ctx, rw, current));
 CHECK(FUNC(render_size)(ctx, rw, current));
 
@@ -324,8 +323,8 @@ static int FUNC(uncompressed_header)(CodedBitstreamContext 
*ctx, RWContext *rw,
  if (current->intra_only == 1) {
  CHECK(FUNC(frame_sync_code)(ctx, rw, current));
 
- if (profile > 0) {
- CHECK(FUNC(color_config)(ctx, rw, current, profile));
+ if (vp9->profile > 0) {
+ CHECK(FUNC(color_config)(ctx, rw, current, vp9->profile));
  } else {
  infer(color_space,   1);
  infer(subsampling_x, 1);
-- 
2.19.0

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

[FFmpeg-devel] [PATCH 1/3] avcodec/cbs_vp9: keep track of reference frames

2018-10-30 Thread James Almer

Signed-off-by: James Almer 
---
 libavcodec/cbs_vp9.h | 16 
 libavcodec/cbs_vp9_syntax_template.c | 60 ++--
 2 files changed, 72 insertions(+), 4 deletions(-)

diff --git a/libavcodec/cbs_vp9.h b/libavcodec/cbs_vp9.h
index 5b99c90c2e..12689e51cc 100644
--- a/libavcodec/cbs_vp9.h
+++ b/libavcodec/cbs_vp9.h
@@ -183,6 +183,13 @@ typedef struct VP9RawSuperframe {
 VP9RawSuperframeIndex index;
 } VP9RawSuperframe;
 
+typedef struct VP9ReferenceFrameState {
+int frame_width;// RefFrameWidth
+int frame_height;   // RefFrameHeight
+int subsampling_x;  // RefSubsamplingX
+int subsampling_y;  // RefSubsamplingY
+int bit_depth;  // RefBitDepth
+} VP9ReferenceFrameState;
 
 typedef struct CodedBitstreamVP9Context {
 // Frame dimensions in 8x8 mode info blocks.
@@ -192,6 +199,15 @@ typedef struct CodedBitstreamVP9Context {
 uint16_t sb64_cols;
 uint16_t sb64_rows;
 
+int frame_width;
+int frame_height;
+
+uint8_t subsampling_x;
+uint8_t subsampling_y;
+int bit_depth;
+
+VP9ReferenceFrameState ref[VP9_NUM_REF_FRAMES];
+
 // Write buffer.
 uint8_t *write_buffer;
 size_t write_buffer_size;
diff --git a/libavcodec/cbs_vp9_syntax_template.c 
b/libavcodec/cbs_vp9_syntax_template.c
index b4a7f65e85..7b90775ed6 100644
--- a/libavcodec/cbs_vp9_syntax_template.c
+++ b/libavcodec/cbs_vp9_syntax_template.c
@@ -43,10 +43,14 @@ static int FUNC(frame_sync_code)(CodedBitstreamContext 
*ctx, RWContext *rw,
 static int FUNC(color_config)(CodedBitstreamContext *ctx, RWContext *rw,
   VP9RawFrameHeader *current, int profile)
 {
+CodedBitstreamVP9Context *vp9 = ctx->priv_data;
 int err;
 
-if (profile >= 2)
+if (profile >= 2) {
 f(1, ten_or_twelve_bit);
+vp9->bit_depth = current->ten_or_twelve_bit ? 12 : 10;
+} else
+vp9->bit_depth = 8;
 
 f(3, color_space);
 
@@ -69,6 +73,9 @@ static int FUNC(color_config)(CodedBitstreamContext *ctx, 
RWContext *rw,
 }
 }
 
+vp9->subsampling_x = current->subsampling_x;
+vp9->subsampling_y = current->subsampling_y;
+
 return 0;
 }
 
@@ -81,8 +88,11 @@ static int FUNC(frame_size)(CodedBitstreamContext *ctx, 
RWContext *rw,
 f(16, frame_width_minus_1);
 f(16, frame_height_minus_1);
 
-vp9->mi_cols = (current->frame_width_minus_1  + 8) >> 3;
-vp9->mi_rows = (current->frame_height_minus_1 + 8) >> 3;
+vp9->frame_width  = current->frame_width_minus_1  + 1;
+vp9->frame_height = current->frame_height_minus_1 + 1;
+
+vp9->mi_cols = (vp9->frame_width  + 7) >> 3;
+vp9->mi_rows = (vp9->frame_height + 7) >> 3;
 vp9->sb64_cols = (vp9->mi_cols + 7) >> 3;
 vp9->sb64_rows = (vp9->mi_rows + 7) >> 3;
 
@@ -107,15 +117,33 @@ static int FUNC(render_size)(CodedBitstreamContext *ctx, 
RWContext *rw,
 static int FUNC(frame_size_with_refs)(CodedBitstreamContext *ctx, RWContext 
*rw,
   VP9RawFrameHeader *current)
 {
+CodedBitstreamVP9Context *vp9 = ctx->priv_data;
 int err, i;
 
 for (i = 0; i < VP9_REFS_PER_FRAME; i++) {
 fs(1, found_ref[i], 1, i);
-if (current->found_ref[i])
+if (current->found_ref[i]) {
+VP9ReferenceFrameState *ref =
+>ref[current->ref_frame_idx[i]];
+
+vp9->frame_width   = ref->frame_width;
+vp9->frame_height  = ref->frame_height;
+
+vp9->subsampling_x = ref->subsampling_x;
+vp9->subsampling_y = ref->subsampling_y;
+vp9->bit_depth = ref->bit_depth;
+
 break;
+}
 }
 if (i >= VP9_REFS_PER_FRAME)
 CHECK(FUNC(frame_size)(ctx, rw, current));
+else {
+vp9->mi_cols = (vp9->frame_width  + 7) >> 3;
+vp9->mi_rows = (vp9->frame_height + 7) >> 3;
+vp9->sb64_cols = (vp9->mi_cols + 7) >> 3;
+vp9->sb64_rows = (vp9->mi_rows + 7) >> 3;
+}
 CHECK(FUNC(render_size)(ctx, rw, current));
 
 return 0;
@@ -249,6 +277,7 @@ static int FUNC(tile_info)(CodedBitstreamContext *ctx, 
RWContext *rw,
 static int FUNC(uncompressed_header)(CodedBitstreamContext *ctx, RWContext *rw,
  VP9RawFrameHeader *current)
 {
+CodedBitstreamVP9Context *vp9 = ctx->priv_data;
 int profile, i;
 int err;
 
@@ -301,6 +330,10 @@ static int FUNC(uncompressed_header)(CodedBitstreamContext 
*ctx, RWContext *rw,
  infer(color_space,   1);
  infer(subsampling_x, 1);
  infer(subsampling_y, 1);
+ vp9->bit_depth = 8;
+
+ vp9->subsampling_x = current->subsampling_x;
+ vp9->subsampling_y = current->subsampling_y;
  }
 
  f(8, refresh_frame_flags);
@@ -339,6 +372,25 @@ static int FUNC(uncompressed_header)(CodedBitstreamContext 
*ctx, RWContext *rw,
 
 f(16, header_size_in_bytes);
 
+for (i =

Re: [FFmpeg-devel] [PATCH] lavc/mjpegdec: fix VA-API MJPEG decoding uninitialized huffman table

2018-10-30 Thread Eoff, Ullysses A



> -Original Message-
> From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On Behalf Of 
> myp...@gmail.com
> Sent: Monday, October 29, 2018 10:52 PM
> To: FFmpeg development discussions and patches 
> Cc: Zhao, Jun ; Lin, Decai 
> Subject: Re: [FFmpeg-devel] [PATCH] lavc/mjpegdec: fix VA-API MJPEG decoding 
> uninitialized huffman table
> 
> On Tue, Oct 30, 2018 at 11:41 AM Eoff, Ullysses A
>  wrote:
> >
> > > -Original Message-
> > > From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On Behalf Of 
> > > myp...@gmail.com
> > > Sent: Monday, October 29, 2018 8:10 PM
> > > To: FFmpeg development discussions and patches 
> > > Cc: Zhao, Jun ; Lin, Decai 
> > > Subject: Re: [FFmpeg-devel] [PATCH] lavc/mjpegdec: fix VA-API MJPEG 
> > > decoding uninitialized huffman table
> > >
> > > On Tue, Oct 30, 2018 at 10:41 AM Li, Zhong  wrote:
> > > >
> > > > > From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On Behalf
> > > > > Of myp...@gmail.com
> > > > > Sent: Tuesday, October 30, 2018 9:02 AM
> > > > > To: FFmpeg development discussions and patches
> > > > > 
> > > > > Cc: Zhao, Jun ; Lin, Decai 
> > > > > Subject: Re: [FFmpeg-devel] [PATCH] lavc/mjpegdec: fix VA-API MJPEG
> > > > > decoding uninitialized huffman table
> > > > >
> > > > > On Mon, Oct 29, 2018 at 6:39 PM Li, Zhong  wrote:
> > > > > >
> > > > > > > From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On
> > > > > > > Behalf Of Jun Zhao
> > > > > > > Sent: Monday, October 29, 2018 6:26 PM
> > > > > > > To: ffmpeg-devel@ffmpeg.org
> > > > > > > Cc: Zhao, Jun ; Lin, Decai 
> > > > > > > 
> > > > > > > Subject: [FFmpeg-devel] [PATCH] lavc/mjpegdec: fix VA-API MJPEG
> > > > > > > decoding uninitialized huffman table
> > > > > > >
> > > > > > > From: Jun Zhao 
> > > > > > >
> > > > > > > Now VA-API HWAccel MJPEG decoding uninitialized huffman table, 
> > > > > > > it's
> > > > > > > will lead to the decoding error like"Failed to sync surface 0x5: 
> > > > > > > 23
> > > > > > > (internal decoding error)." in iHD open source driver.
> > > > > > >
> > > > > > > Signed-off-by: dlin2 
> > > > > > > Signed-off-by: Jun Zhao 
> > > > > > > ---
> > > > > > >  libavcodec/mjpegdec.c |   19 +++
> > > > > > >  1 files changed, 19 insertions(+), 0 deletions(-)
> > > > > > >
> > > > > > > diff --git a/libavcodec/mjpegdec.c b/libavcodec/mjpegdec.c index
> > > > > > > b0cb3ff..89effb6 100644
> > > > > > > --- a/libavcodec/mjpegdec.c
> > > > > > > +++ b/libavcodec/mjpegdec.c
> > > > > > > @@ -75,6 +75,25 @@ static int build_vlc(VLC *vlc, const uint8_t
> > > > > > > *bits_table, static int build_basic_mjpeg_vlc(MJpegDecodeContext 
> > > > > > > *s)
> > > > > {
> > > > > > >  int ret;
> > > > > > > +int i;
> > > > > > > +
> > > > > > > +/* initialize default huffman tables */
> > > > > > > +for (i = 0; i < 16; i++)
> > > > > > > +s->raw_huffman_lengths[0][0][i] =
> > > > > > > avpriv_mjpeg_bits_dc_luminance[i + 1];
> > > > > > > +for (i = 0; i < 12; i++)
> > > > > > > +s->raw_huffman_values[0][0][i] = avpriv_mjpeg_val_dc[i];
> > > > > > > +for (i = 0; i < 16; i++)
> > > > > > > +s->raw_huffman_lengths[0][1][i] =
> > > > > > > avpriv_mjpeg_bits_dc_chrominance[i + 1];
> > > > > > > +for (i = 0; i < 12; i++)
> > > > > > > +s->raw_huffman_values[0][1][i] = avpriv_mjpeg_val_dc[i];
> > > > > > > +for (i = 0; i < 16; i++)
> > > > > > > +s->raw_huffman_lengths[1][0][i] =
> > > > > > > avpriv_mjpeg_bits_ac_luminance[i + 1];
> > > > > > > +for (i = 0; i < 162; i++)
> > > > > > > +s->raw_huffman_values[1][0][i] =
> > > > > > > avpriv_mjpeg_val_ac_luminance[i];
> > > > > > > +for (i = 0; i < 16; i++)
> > > > > > > +s->raw_huffman_lengths[1][1][i] =
> > > > > > > avpriv_mjpeg_bits_ac_chrominance[i + 1];
> > > > > > > +for (i = 0; i < 162; i++)
> > > > > > > +s->raw_huffman_values[1][1][i] =
> > > > > > > + avpriv_mjpeg_val_ac_chrominance[i];
> > > > > > >
> > > > > > >  if ((ret = build_vlc(>vlcs[0][0],
> > > > > avpriv_mjpeg_bits_dc_luminance,
> > > > > > >   avpriv_mjpeg_val_dc, 12, 0, 0)) < 0)
> > > > > > > --
> > > > > > > 1.7.1
> > > > > >
> > > > > > Should this be fixed in iHD driver instead of ffmpeg?
> > > > > No, I don't think driver is a good place to initialize the huffman 
> > > > > table.
> > > >
> > > > For the default Huffman table, thus should be initialized by driver 
> > > > itself.
> > > > For non-default case, thus should be passed from ffmpeg.
> > > > So, what is the case you want to fix? Both (if so, no need to mention 
> > > > iHD driver)?
> > > Both, now HWAccel MJPEG always setting the huffman table and
> > > can't distinguish the HWaccel JPEG whether default Huffman table.
> >
> > If your VA-API HWAccel MJPEG decoding case(s) work with i965 driver
> > and not iHD then I would assume iHD needs fixing.
> >
> i965 must be have some issue if MJPEG decoding, when

Re: [FFmpeg-devel] [PATCH 1/5] lavc/qsvenc: add forced_idr opiton

2018-10-30 Thread Rogozhkin, Dmitry V

On Tue, 2018-10-30 at 09:49 +, Li, Zhong wrote:
> > From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On
> > Behalf
> > Of Mark Thompson
> > Sent: Tuesday, October 30, 2018 5:06 AM
> > To: ffmpeg-devel@ffmpeg.org
> > Subject: Re: [FFmpeg-devel] [PATCH 1/5] lavc/qsvenc: add forced_idr
> > opiton
> > 
> > On 25/10/18 13:36, Zhong Li wrote:
> > > This option can be used to repect original input I/IDR frame
> > > type.
> > > 
> > > Signed-off-by: Zhong Li 
> > > ---
> > >  libavcodec/qsvenc.c | 7 +++
> > >  libavcodec/qsvenc.h | 2 ++
> > >  2 files changed, 9 insertions(+)
> > > 
> > > diff --git a/libavcodec/qsvenc.c b/libavcodec/qsvenc.c index
> > > 948751d..e534dcf 100644
> > > --- a/libavcodec/qsvenc.c
> > > +++ b/libavcodec/qsvenc.c
> > > @@ -1192,6 +1192,13 @@ static int encode_frame(AVCodecContext
> > 
> > *avctx, QSVEncContext *q,
> > >  if (qsv_frame) {
> > >  surf = _frame->surface;
> > >  enc_ctrl = _frame->enc_ctrl;
> > > +
> > > +if (q->forced_idr >= 0 && frame->pict_type ==
> > 
> > AV_PICTURE_TYPE_I) {
> > > +enc_ctrl->FrameType = MFX_FRAMETYPE_I |
> > 
> > MFX_FRAMETYPE_REF;
> > > +if (q->forced_idr || frame->key_frame)
> > > +enc_ctrl->FrameType |= MFX_FRAMETYPE_IDR;
> > > +} else
> > > +enc_ctrl->FrameType = MFX_FRAMETYPE_UNKNOWN;
> > >  }
> > > 
> > >  ret = av_new_packet(_pkt, q->packet_size); diff --git
> > > a/libavcodec/qsvenc.h b/libavcodec/qsvenc.h index
> > > 055b4a6..1f97f77
> > > 100644
> > > --- a/libavcodec/qsvenc.h
> > > +++ b/libavcodec/qsvenc.h
> > > @@ -87,6 +87,7 @@
> > >  { "adaptive_i", "Adaptive I-frame placement",
> > 
> > OFFSET(qsv.adaptive_i), AV_OPT_TYPE_INT, { .i64 = -1 }, -1,
> > 1, VE }, \
> > >  { "adaptive_b", "Adaptive B-frame placement",
> > 
> > OFFSET(qsv.adaptive_b), AV_OPT_TYPE_INT, { .i64 = -1 }, -1,
> > 1, VE }, \
> > >  { "b_strategy", "Strategy to choose between I/P/B-frames",
> > 
> > OFFSET(qsv.b_strategy),AV_OPT_TYPE_INT, { .i64 = -1 }, -1,
> > 1, VE }, \
> > > +{ "forced_idr", "Forcing I frames as IDR frames",
> > 
> > OFFSET(qsv.forced_idr), AV_OPT_TYPE_INT, { .i64 = -1 }, -1,
> > 1, VE }, \
> > > 
> > >  typedef int SetEncodeCtrlCB (AVCodecContext *avctx,
> > >   const AVFrame *frame,
> > 
> > mfxEncodeCtrl*
> > > enc_ctrl); @@ -168,6 +169,7 @@ typedef struct QSVEncContext
> > > {  #endif
> > >  char *load_plugins;
> > >  SetEncodeCtrlCB *set_encode_ctrl_cb;
> > > +int forced_idr;
> > >  } QSVEncContext;
> > > 
> > >  int ff_qsv_enc_init(AVCodecContext *avctx, QSVEncContext *q);
> > > 
> > 
> > This seems confusing, because it doesn't match what forced_idr does
> > in any
> > other encoder.
> > 
> > Checking of pict_type for AV_PICTURE_TYPE_I in order to get a key
> > frame
> > (of whatever kind) is always enabled if supported (many
> > encoders).  The
> > "forced_idr" option to H.26[45] encoders (libx264, libx265) then
> > forces that
> > to be an IDR frame, not just an I frame.
> > 
> > - Mark
> 
> Yup, I know it doesn’t match other encoders such as
> libx264/libx265/nvenc. 
> However, it is my intentional behavior. I believe current implement
> in libx264/libx265 is not complete.
> One case is ignored: user just want to keep the gop structure as
> input, not to force all I frames as IDR frames.
> So I have an idea:
> Default value = -1, ignore the input gop structure. 

I see the idea now, but I very much would like to see this at least
somehow reflected in ffmpeg option documentation. To me your
interpretation of -1, 0, 1 is quite non-intuitive. As you can see from
my other comments I did not guess your intent from the patch. End-users 
will have even more problems with this option since a little of them
will take care to look into the sources themselves.

Besides, I am still feeling that you try to mix 2 different options
together: one which will permit to follow (or not) input stream gop
structure and another which forces IDR frames for each I frame.

> 0: respect input gop structure but don't force I frame as IDR frame. 
> 1: force all I frame as IDR frame.
> 
> Since this is a qsv encoder private option, I just changed qsv
> implementation.
> But I believe it should be a good to make other encoders follow such
> a way with separated patches.
> 
> 
> ___
> 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 1/5] lavc/qsvenc: add forced_idr opiton

2018-10-30 Thread Rogozhkin, Dmitry V

On Tue, 2018-10-30 at 18:05 +0800, Li, Zhong wrote:
> > > +} else
> > > +enc_ctrl->FrameType = MFX_FRAMETYPE_UNKNOWN;
> > 
> > "else" block don't make much sense to me. You eventually already
> > had
> > enc_ctrl structure passed to the encoder. Thus, it should be
> > initialized to
> > default (already). And you don't make anything specific/new in the
> > "else".
> > From my perspective "else" just obscures the code and should be
> > dropped.
> 
> This was a case I had concern. I doubt the default initialization is
> always zero 
> (you know MFX_FRAMETYPE_UNKNOWN is zero). Isn't it possible? 
> Please check the regression case I fixed: https://patchwork.ffmpeg.or
> g/patch/10517/ 

Patch 10517 deals with unitialized variable on a compilation level. As
for the enc_ctrl I very much hope that ffmpeg-qsv code takes care to
memset the mfcEncodeCtrl (as well as all other mediasdk structures)
_before_ usage. I.e. there should be a code somewhere similar to:
  memset(enc_ctrl, 0, sizeof(mfxEncodeCtrl));
If this is missing, there is a VERY big problem in the QSV code since
indeed compiler may initialize structures to everything it wants and
there will be very bad consequences.

As for the usage of mfxEncodeCtrl, the idea is the following. User
allocates this control and memsets it to 0. If this will be passed in
that way mediasdk will change nothing to encode the frame. If user
wants to change some parameter, it can do this changing only the
parameter it wants to take effect. So, the "else" should really not be
needed.

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

Re: [FFmpeg-devel] [PATCH 2/2] libx264: Pass the reordered_opaque field through the encoder

2018-10-30 Thread Derek Buitenhuis

On 29/10/2018 21:06, Martin Storsjö wrote:
> As I guess there can be old frames in flight, the only safe option is to
> enlarge, not to shrink it. But in case a realloc moves the array, the old
> pointers end up pretty useless.

Just always allocate the max (which is known for H.264), and adjust 
nb_reordered_opaque
as need be, on reconfig, no?

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

[FFmpeg-devel] [PATCH v3] fate/api-h264-slice-test: use cleaner error handling

2018-10-30 Thread joshdk

From: Josh de Kock 

Fix ticket #7521
---
 tests/api/api-h264-slice-test.c | 88 +++--
 1 file changed, 51 insertions(+), 37 deletions(-)

diff --git a/tests/api/api-h264-slice-test.c b/tests/api/api-h264-slice-test.c
index 57e7dc79c3..ebc90b5b57 100644
--- a/tests/api/api-h264-slice-test.c
+++ b/tests/api/api-h264-slice-test.c
@@ -48,7 +48,7 @@
 
 static int header = 0;
 
-static void decode(AVCodecContext *dec_ctx, AVFrame *frame,
+static int decode(AVCodecContext *dec_ctx, AVFrame *frame,
AVPacket *pkt)
 {
 static uint64_t frame_cnt = 0;
@@ -57,20 +57,20 @@ static void decode(AVCodecContext *dec_ctx, AVFrame *frame,
 ret = avcodec_send_packet(dec_ctx, pkt);
 if (ret < 0) {
 fprintf(stderr, "Error sending a packet for decoding: %s\n", 
av_err2str(ret));
-exit(1);
+return ret;
 }
 
 while (ret >= 0) {
 const AVPixFmtDescriptor *desc;
-char *sum;
+char sum[AV_HASH_MAX_SIZE * 2 + 1];
 struct AVHashContext *hash;
 
 ret = avcodec_receive_frame(dec_ctx, frame);
 if (ret == AVERROR(EAGAIN) || ret == AVERROR_EOF) {
-return;
+return 0;
 } else if (ret < 0) {
 fprintf(stderr, "Error during decoding: %s\n", av_err2str(ret));
-exit(1);
+return ret;
 }
 
 if (!header) {
@@ -87,9 +87,10 @@ static void decode(AVCodecContext *dec_ctx, AVFrame *frame,
 header = 1;
 }
 desc = av_pix_fmt_desc_get(dec_ctx->pix_fmt);
-av_hash_alloc(, "md5");
+if ((ret = av_hash_alloc(, "md5")) < 0) {
+return ret;
+}
 av_hash_init(hash);
-sum = av_mallocz(av_hash_get_size(hash) * 2 + 1);
 
 for (int i = 0; i < frame->height; i++)
 av_hash_update(hash, >data[0][i * frame->linesize[0]], 
frame->width);
@@ -104,25 +105,25 @@ static void decode(AVCodecContext *dec_ctx, AVFrame 
*frame,
 (frame->width * frame->height + 2 * (frame->height >> 
desc->log2_chroma_h) * (frame->width >> desc->log2_chroma_w)), sum);
 frame_cnt += 1;
 av_hash_freep();
-av_free(sum);
 }
+return 0;
 }
 
 int main(int argc, char **argv)
 {
-const AVCodec *codec;
+const AVCodec *codec = NULL;
 AVCodecContext *c = NULL;
-AVFrame *frame;
+AVFrame *frame = NULL;
 unsigned int threads;
 AVPacket *pkt;
-FILE *fd;
+FILE *file = NULL;
 char nal[MAX_SLICES * UINT16_MAX + AV_INPUT_BUFFER_PADDING_SIZE];
-int nals = 0;
+int nals = 0, ret = 0;
 char *p = nal;
 
 if (argc < 4) {
 fprintf(stderr, "Usage: %s   \n", 
argv[0]);
-exit(1);
+return -1;
 }
 
 if (!(threads = strtoul(argv[1], NULL, 0)))
@@ -134,17 +135,20 @@ int main(int argc, char **argv)
 setmode(fileno(stdout), O_BINARY);
 #endif
 
-if (!(pkt = av_packet_alloc()))
-exit(1);
+if (!(pkt = av_packet_alloc())) {
+return -1;
+}
 
 if (!(codec = avcodec_find_decoder(AV_CODEC_ID_H264))) {
 fprintf(stderr, "Codec not found\n");
-exit(1);
+ret = -1;
+goto err;
 }
 
 if (!(c = avcodec_alloc_context3(codec))) {
 fprintf(stderr, "Could not allocate video codec context\n");
-exit(1);
+ret = -1;
+goto err;
 }
 
 c->width  = 352;
@@ -154,15 +158,16 @@ int main(int argc, char **argv)
 c->thread_type = FF_THREAD_SLICE;
 c->thread_count = threads;
 
-if (avcodec_open2(c, codec, NULL) < 0) {
+if ((ret = avcodec_open2(c, codec, NULL)) < 0) {
 fprintf(stderr, "Could not open codec\n");
-exit(1);
+goto err;
 }
 
 #if HAVE_THREADS
 if (c->active_thread_type != FF_THREAD_SLICE) {
 fprintf(stderr, "Couldn't activate slice threading: %d\n", 
c->active_thread_type);
-exit(1);
+ret = -1;
+goto err;
 }
 #else
 fprintf(stderr, "WARN: not using threads, only checking decoding slice 
NALUs\n");
@@ -170,34 +175,39 @@ int main(int argc, char **argv)
 
 if (!(frame = av_frame_alloc())) {
 fprintf(stderr, "Could not allocate video frame\n");
-exit(1);
+ret = -1;
+goto err;
 }
 
-if (!(fd = fopen(argv[2], "rb"))) {
+if (!(file = fopen(argv[2], "rb"))) {
 fprintf(stderr, "Couldn't open NALU file: %s\n", argv[2]);
-exit(1);
+ret = -1;
+goto err;
 }
 
 while(1) {
 uint16_t size = 0;
-ssize_t ret = fread(, 1, sizeof(uint16_t), fd);
-if (ret < 0) {
-perror("Couldn't read size");
-exit(1);
-} else if (ret != sizeof(uint16_t))
+
+size_t ret = fread(, 1, sizeof(uint16_t), file);
+if (ret != sizeof(uint16_t))
 break;
+
 size = ntohs(size);
-ret = fread(p, 1, size, fd);
-if (ret < 0 || ret != size) {
+
+ret = fread(p, 1, size, file);

[FFmpeg-devel] [PATCH 3/3] lavc/libdavs2: remove unused context parameter

2018-10-30 Thread hwren

Signed-off-by: hwren 
---
 libavcodec/libdavs2.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/libavcodec/libdavs2.c b/libavcodec/libdavs2.c
index b05bdef..ee52043 100644
--- a/libavcodec/libdavs2.c
+++ b/libavcodec/libdavs2.c
@@ -32,8 +32,6 @@ typedef struct DAVS2Context {
 davs2_param_tparam;  // decoding parameters
 davs2_packet_t   packet; // input bitstream
 
-int decoded_frames;
-
 davs2_picture_t  out_frame;  // output data, frame data
 davs2_seq_info_t headerset;  // output data, sequence header
 
-- 
2.7.4

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

[FFmpeg-devel] [PATCH 1/2] lavc/libdavs2: remove incorrect frame settings

2018-10-30 Thread hwren

Signed-off-by: hwren 
---
 libavcodec/libdavs2.c | 7 ---
 1 file changed, 7 deletions(-)

diff --git a/libavcodec/libdavs2.c b/libavcodec/libdavs2.c
index cadf995..581d568 100644
--- a/libavcodec/libdavs2.c
+++ b/libavcodec/libdavs2.c
@@ -96,13 +96,6 @@ static int davs2_dump_frames(AVCodecContext *avctx, 
davs2_picture_t *pic,
pic->widths[plane] * bytes_per_sample);
 }
 
-frame->width = cad->headerset.width;
-frame->height= cad->headerset.height;
-frame->pts   = cad->out_frame.pts;
-frame->pict_type = pic->type;
-frame->format= avctx->pix_fmt;
-
-cad->decoded_frames++;
 return 1;
 }
 
-- 
2.7.4

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

[FFmpeg-devel] [PATCH 2/2] lavc/libdavs2: fix sequence incomplete output error

2018-10-30 Thread hwren

Signed-off-by: hwren 
---
 libavcodec/libdavs2.c | 11 ++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/libavcodec/libdavs2.c b/libavcodec/libdavs2.c
index 581d568..b05bdef 100644
--- a/libavcodec/libdavs2.c
+++ b/libavcodec/libdavs2.c
@@ -122,7 +122,16 @@ static int davs2_decode_frame(AVCodecContext *avctx, void 
*data,
 int   ret  = DAVS2_DEFAULT;
 
 if (!buf_size) {
-return 0;
+ret = davs2_decoder_flush(cad->decoder, >headerset, 
>out_frame);
+if (ret == DAVS2_END) {
+return 0;
+} else if (ret == DAVS2_GOT_FRAME) {
+*got_frame = davs2_dump_frames(avctx, >out_frame, 
>headerset, ret, frame);
+davs2_decoder_frame_unref(cad->decoder, >out_frame);
+return ret;
+} else {
+return AVERROR_EXTERNAL;
+}
 }
 
 cad->packet.data = buf_ptr;
-- 
2.7.4

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

[FFmpeg-devel] [PATCH] avfilter: add tpad filter

2018-10-30 Thread Paul B Mahol

Signed-off-by: Paul B Mahol 
---
 doc/filters.texi |  40 +++
 libavfilter/Makefile |   1 +
 libavfilter/allfilters.c |   1 +
 libavfilter/vf_tpad.c| 218 +++
 4 files changed, 260 insertions(+)
 create mode 100644 libavfilter/vf_tpad.c

diff --git a/doc/filters.texi b/doc/filters.texi
index 9b84b1145b..76eec2e918 100644
--- a/doc/filters.texi
+++ b/doc/filters.texi
@@ -16676,6 +16676,46 @@ embedded peak information in display metadata is not 
reliable or when tone
 mapping from a lower range to a higher range.
 @end table
 
+@section tpad
+
+Temporarily pad video frames.
+
+The filter accepts the following options:
+
+@table @option
+@item start
+Specify number of delay frames before input video stream.
+
+@item stop
+Specify number of padding frames after input video stream.
+Set to -1 to pad indefinitely.
+
+@item start_mode
+Set kind of frames added to beginning of stream.
+Can be either @var{add} or @var{clone}.
+With @var{add} frames of solid-color are added.
+With @var{clone} frames are clones of first frame.
+
+@item stop_mode
+Set kind of frames added to end of stream.
+Can be either @var{add} or @var{clone}.
+With @var{add} frames of solid-color are added.
+With @var{clone} frames are clones of last frame.
+
+@item start_duration, stop_duration
+Specify the duration of the start/stop delay. See
+@ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) 
manual,ffmpeg-utils}
+for the accepted syntax.
+These options override @var{start} and @var{stop}.
+
+@item color
+Specify the color of the padded area. For the syntax of this option,
+check the @ref{color syntax,,"Color" section in the ffmpeg-utils
+manual,ffmpeg-utils}.
+
+The default value of @var{color} is "black".
+@end table
+
 @anchor{transpose}
 @section transpose
 
diff --git a/libavfilter/Makefile b/libavfilter/Makefile
index c35cd8f422..51e48efc2e 100644
--- a/libavfilter/Makefile
+++ b/libavfilter/Makefile
@@ -383,6 +383,7 @@ OBJS-$(CONFIG_TMIX_FILTER)   += vf_mix.o 
framesync.o
 OBJS-$(CONFIG_TONEMAP_FILTER)+= vf_tonemap.o colorspace.o
 OBJS-$(CONFIG_TONEMAP_OPENCL_FILTER) += vf_tonemap_opencl.o 
colorspace.o opencl.o \
 opencl/tonemap.o 
opencl/colorspace_common.o
+OBJS-$(CONFIG_TPAD_FILTER)   += vf_tpad.o
 OBJS-$(CONFIG_TRANSPOSE_FILTER)  += vf_transpose.o
 OBJS-$(CONFIG_TRANSPOSE_NPP_FILTER)  += vf_transpose_npp.o
 OBJS-$(CONFIG_TRIM_FILTER)   += trim.o
diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
index d5a211bda5..6052cb8c3c 100644
--- a/libavfilter/allfilters.c
+++ b/libavfilter/allfilters.c
@@ -365,6 +365,7 @@ 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_tpad;
 extern AVFilter ff_vf_transpose;
 extern AVFilter ff_vf_transpose_npp;
 extern AVFilter ff_vf_trim;
diff --git a/libavfilter/vf_tpad.c b/libavfilter/vf_tpad.c
new file mode 100644
index 00..86e063090b
--- /dev/null
+++ b/libavfilter/vf_tpad.c
@@ -0,0 +1,218 @@
+/*
+ * Copyright (c) 2018 Paul B Mahol
+ *
+ * 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 "libavutil/avassert.h"
+#include "libavutil/channel_layout.h"
+#include "libavutil/opt.h"
+#include "avfilter.h"
+#include "audio.h"
+#include "filters.h"
+#include "internal.h"
+#include "formats.h"
+#include "drawutils.h"
+
+typedef struct TPadContext {
+const AVClass *class;
+int pad_start;
+int pad_stop;
+int start_mode;
+int stop_mode;
+int64_t start_duration;
+int64_t stop_duration;
+uint8_t rgba_color[4];  ///< color for the padding area
+
+FFDrawContext draw;
+FFDrawColor color;
+int64_t pts;
+int eof;
+AVFrame *cache_start;
+AVFrame *cache_stop;
+} TPadContext;
+
+#define OFFSET(x) offsetof(TPadContext, x)
+#define VF AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
+
+static const AVOption tpad_options[] = {
+{ "start", "set the number of frames to delay input",  
OFFSET(pad_start),  AV_OPT_TYPE_INT,   {.i64=0},0,   INT_MAX, VF },
+{ "stop",  "set the number of

Re: [FFmpeg-devel] [PATCH] correct the max value of swscale option with type AV_OPT_TYPE_PIXEL_FMT

2018-10-30 Thread Paul B Mahol

On 10/30/18, Guo, Yejun  wrote:
> Signed-off-by: Guo, Yejun 
> ---
>  libswscale/options.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/libswscale/options.c b/libswscale/options.c
> index 7eb2752..db8210b 100644
> --- a/libswscale/options.c
> +++ b/libswscale/options.c
> @@ -57,8 +57,8 @@ static const AVOption swscale_options[] = {
>  { "srch","source height", OFFSET(srcH),
>  AV_OPT_TYPE_INT,{ .i64 = 16 }, 1,   INT_MAX,
> VE },
>  { "dstw","destination width", OFFSET(dstW),
>  AV_OPT_TYPE_INT,{ .i64 = 16 }, 1,   INT_MAX,
> VE },
>  { "dsth","destination height",OFFSET(dstH),
>  AV_OPT_TYPE_INT,{ .i64 = 16 }, 1,   INT_MAX,
> VE },
> -{ "src_format",  "source format",
> OFFSET(srcFormat), AV_OPT_TYPE_PIXEL_FMT,{ .i64 = DEFAULT  }, 0,
>   INT_MAX, VE },
> -{ "dst_format",  "destination format",
> OFFSET(dstFormat), AV_OPT_TYPE_PIXEL_FMT,{ .i64 = DEFAULT  }, 0,
>   INT_MAX, VE },
> +{ "src_format",  "source format",
> OFFSET(srcFormat), AV_OPT_TYPE_PIXEL_FMT,{ .i64 = DEFAULT  }, 0,
>   AV_PIX_FMT_NB,  VE },
> +{ "dst_format",  "destination format",
> OFFSET(dstFormat), AV_OPT_TYPE_PIXEL_FMT,{ .i64 = DEFAULT  }, 0,
>   AV_PIX_FMT_NB,  VE },
>  { "src_range",   "source is full range",  OFFSET(srcRange),
>  AV_OPT_TYPE_BOOL,   { .i64 = DEFAULT}, 0,   1,
> VE },
>  { "dst_range",   "destination is full range", OFFSET(dstRange),
>  AV_OPT_TYPE_BOOL,   { .i64 = DEFAULT}, 0,   1,
> VE },
>  { "param0",  "scaler param 0",OFFSET(param[0]),
>  AV_OPT_TYPE_DOUBLE, { .dbl = SWS_PARAM_DEFAULT  }, INT_MIN, INT_MAX,
> VE },
> --
> 2.7.4


This is wrong, First it does not uses NB-1 and second it is not needed at all.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

[FFmpeg-devel] [PATCH] correct the max value of swscale option with type AV_OPT_TYPE_PIXEL_FMT

2018-10-30 Thread Guo, Yejun

Signed-off-by: Guo, Yejun 
---
 libswscale/options.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/libswscale/options.c b/libswscale/options.c
index 7eb2752..db8210b 100644
--- a/libswscale/options.c
+++ b/libswscale/options.c
@@ -57,8 +57,8 @@ static const AVOption swscale_options[] = {
 { "srch","source height", OFFSET(srcH),  
AV_OPT_TYPE_INT,{ .i64 = 16 }, 1,   INT_MAX,VE 
},
 { "dstw","destination width", OFFSET(dstW),  
AV_OPT_TYPE_INT,{ .i64 = 16 }, 1,   INT_MAX,VE 
},
 { "dsth","destination height",OFFSET(dstH),  
AV_OPT_TYPE_INT,{ .i64 = 16 }, 1,   INT_MAX,VE 
},
-{ "src_format",  "source format", OFFSET(srcFormat), 
AV_OPT_TYPE_PIXEL_FMT,{ .i64 = DEFAULT  }, 0,   INT_MAX, VE },
-{ "dst_format",  "destination format",OFFSET(dstFormat), 
AV_OPT_TYPE_PIXEL_FMT,{ .i64 = DEFAULT  }, 0,   INT_MAX, VE },
+{ "src_format",  "source format", OFFSET(srcFormat), 
AV_OPT_TYPE_PIXEL_FMT,{ .i64 = DEFAULT  }, 0,   AV_PIX_FMT_NB,  VE 
},
+{ "dst_format",  "destination format",OFFSET(dstFormat), 
AV_OPT_TYPE_PIXEL_FMT,{ .i64 = DEFAULT  }, 0,   AV_PIX_FMT_NB,  VE 
},
 { "src_range",   "source is full range",  OFFSET(srcRange),  
AV_OPT_TYPE_BOOL,   { .i64 = DEFAULT}, 0,   1,  VE 
},
 { "dst_range",   "destination is full range", OFFSET(dstRange),  
AV_OPT_TYPE_BOOL,   { .i64 = DEFAULT}, 0,   1,  VE 
},
 { "param0",  "scaler param 0",OFFSET(param[0]),  
AV_OPT_TYPE_DOUBLE, { .dbl = SWS_PARAM_DEFAULT  }, INT_MIN, INT_MAX,VE 
},
-- 
2.7.4

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

Re: [FFmpeg-devel] [PATCH] avfilter: add tpad filter

2018-10-30 Thread Gyan

On Tue, Oct 30, 2018 at 3:27 PM Paul B Mahol  wrote:

> Signed-off-by: Paul B Mahol 
> ---
>  doc/filters.texi |  22 ++
>  libavfilter/Makefile |   1 +
>  libavfilter/allfilters.c |   1 +
>  libavfilter/vf_tpad.c| 166 +++
>  4 files changed, 190 insertions(+)
>  create mode 100644 libavfilter/vf_tpad.c
>
> diff --git a/doc/filters.texi b/doc/filters.texi
> index 9b84b1145b..6dda780aa0 100644
> --- a/doc/filters.texi
> +++ b/doc/filters.texi
> @@ -16676,6 +16676,28 @@ embedded peak information in display metadata is
> not reliable or when tone
>  mapping from a lower range to a higher range.
>  @end table
>
> +@section tpad
> +
> +Temporarily pad video frames.
> +
> +The filter accepts the following options:
> +
> +@table @option
> +@item start
> +Specify number of delay frames before input video stream.
> +
> +@item stop
> +Specify number of padding frames after input video stream.
> +Set to -1 to pad indefinitely.
> +
> +@item color
> +Specify the color of the padded area. For the syntax of this option,
> +check the @ref{color syntax,,"Color" section in the ffmpeg-utils
> +manual,ffmpeg-utils}.
> +
> +The default value of @var{color} is "black".
> +@end table
> +
>  @anchor{transpose}
>  @section transpose
>
> diff --git a/libavfilter/Makefile b/libavfilter/Makefile
> index c35cd8f422..51e48efc2e 100644
> --- a/libavfilter/Makefile
> +++ b/libavfilter/Makefile
> @@ -383,6 +383,7 @@ OBJS-$(CONFIG_TMIX_FILTER)   +=
> vf_mix.o framesync.o
>  OBJS-$(CONFIG_TONEMAP_FILTER)+= vf_tonemap.o colorspace.o
>  OBJS-$(CONFIG_TONEMAP_OPENCL_FILTER) += vf_tonemap_opencl.o
> colorspace.o opencl.o \
>  opencl/tonemap.o
> opencl/colorspace_common.o
> +OBJS-$(CONFIG_TPAD_FILTER)   += vf_tpad.o
>  OBJS-$(CONFIG_TRANSPOSE_FILTER)  += vf_transpose.o
>  OBJS-$(CONFIG_TRANSPOSE_NPP_FILTER)  += vf_transpose_npp.o
>  OBJS-$(CONFIG_TRIM_FILTER)   += trim.o
> diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
> index d5a211bda5..6052cb8c3c 100644
> --- a/libavfilter/allfilters.c
> +++ b/libavfilter/allfilters.c
> @@ -365,6 +365,7 @@ 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_tpad;
>  extern AVFilter ff_vf_transpose;
>  extern AVFilter ff_vf_transpose_npp;
>  extern AVFilter ff_vf_trim;
> diff --git a/libavfilter/vf_tpad.c b/libavfilter/vf_tpad.c
> new file mode 100644
> index 00..e59037ca56
> --- /dev/null
> +++ b/libavfilter/vf_tpad.c
> @@ -0,0 +1,166 @@
> +/*
> + * Copyright (c) 2018 Paul B Mahol
> + *
> + * 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 "libavutil/avassert.h"
> +#include "libavutil/channel_layout.h"
> +#include "libavutil/opt.h"
> +#include "avfilter.h"
> +#include "audio.h"
> +#include "filters.h"
> +#include "internal.h"
> +#include "formats.h"
> +#include "drawutils.h"
> +
> +typedef struct TPadContext {
> +const AVClass *class;
> +int pad_start;
> +int pad_stop;
> +uint8_t rgba_color[4];  ///< color for the padding area
> +
> +FFDrawContext draw;
> +FFDrawColor color;
> +int64_t pts;
> +int eof;
> +} TPadContext;
> +
> +#define OFFSET(x) offsetof(TPadContext, x)
> +#define VF AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
> +
> +static const AVOption tpad_options[] = {
> +{ "start", "set the number of frames to delay input",
> OFFSET(pad_start),  AV_OPT_TYPE_INT,   {.i64=0},0, INT_MAX, VF },
> +{ "stop",  "set the number of frames to add after input finished",
> OFFSET(pad_stop),   AV_OPT_TYPE_INT,   {.i64=0},   -1, INT_MAX, VF },
> +{ "color", "set the color of the added frames",
> OFFSET(rgba_color), AV_OPT_TYPE_COLOR, {.str="black"},  0,   0, VF },
> +{ NULL }
> +};
> +
> +AVFILTER_DEFINE_CLASS(tpad);
> +
> +static int query_formats(AVFilterContext *ctx)
> +{
> +return ff_set_common_formats(ctx, ff_draw_supported_pixel_formats(0));
> +}
> +
> +static int activate(AVFilterContext *ctx)
> +{
> +AVFilterLink *inlink = ctx->inputs[0];
> +

Re: [FFmpeg-devel] [PATCH 1/5] lavc/qsvenc: add forced_idr opiton

2018-10-30 Thread Li, Zhong

> From: Rogozhkin, Dmitry V
> Sent: Tuesday, October 30, 2018 5:07 AM
> To: ffmpeg-devel@ffmpeg.org
> Cc: Li, Zhong 
> Subject: Re: [FFmpeg-devel] [PATCH 1/5] lavc/qsvenc: add forced_idr opiton
> 
> On Thu, 2018-10-25 at 20:36 +0800, Zhong Li wrote:
> > This option can be used to repect original input I/IDR frame type.
> >
> > Signed-off-by: Zhong Li 
> > ---
> >  libavcodec/qsvenc.c | 7 +++
> >  libavcodec/qsvenc.h | 2 ++
> >  2 files changed, 9 insertions(+)
> >
> > diff --git a/libavcodec/qsvenc.c b/libavcodec/qsvenc.c index
> > 948751d..e534dcf 100644
> > --- a/libavcodec/qsvenc.c
> > +++ b/libavcodec/qsvenc.c
> > @@ -1192,6 +1192,13 @@ static int encode_frame(AVCodecContext
> *avctx,
> > QSVEncContext *q,
> >  if (qsv_frame) {
> >  surf = _frame->surface;
> >  enc_ctrl = _frame->enc_ctrl;
> > +
> > +if (q->forced_idr >= 0 && frame->pict_type ==
> > AV_PICTURE_TYPE_I) {
> > +enc_ctrl->FrameType = MFX_FRAMETYPE_I |
> > MFX_FRAMETYPE_REF;
> > +if (q->forced_idr || frame->key_frame)
> > +enc_ctrl->FrameType |= MFX_FRAMETYPE_IDR;
> 
> Hm. There is another option "-force_key_frames" which looks unhandled
> here. At least I don't understand "|| frame->key_frame"...
> 
> 
> > +} else
> > +enc_ctrl->FrameType = MFX_FRAMETYPE_UNKNOWN;
> 
> "else" block don't make much sense to me. You eventually already had
> enc_ctrl structure passed to the encoder. Thus, it should be initialized to
> default (already). And you don't make anything specific/new in the "else".
> From my perspective "else" just obscures the code and should be dropped.

This was a case I had concern. I doubt the default initialization is always 
zero 
(you know MFX_FRAMETYPE_UNKNOWN is zero). Isn't it possible? 
Please check the regression case I fixed: 
https://patchwork.ffmpeg.org/patch/10517/ 

> >  }
> >
> >  ret = av_new_packet(_pkt, q->packet_size); diff --git
> > a/libavcodec/qsvenc.h b/libavcodec/qsvenc.h index 055b4a6..1f97f77
> > 100644
> > --- a/libavcodec/qsvenc.h
> > +++ b/libavcodec/qsvenc.h
> > @@ -87,6 +87,7 @@
> >  { "adaptive_i", "Adaptive I-frame placement",
> > OFFSET(qsv.adaptive_i), AV_OPT_TYPE_INT, { .i64 = -1 },
> -1,
> > 1, VE }, \
> >  { "adaptive_b", "Adaptive B-frame
> placement",
> > OFFSET(qsv.adaptive_b), AV_OPT_TYPE_INT, { .i64 = -1 },
> -1,
> > 1, VE }, \
> >  { "b_strategy", "Strategy to choose between I/P/B-frames",
> > OFFSET(qsv.b_strategy),AV_OPT_TYPE_INT, { .i64 = -1 },
> -1,
> > 1, VE }, \
> > +{ "forced_idr", "Forcing I frames as IDR
> > frames", OFFSET(qsv.forced_idr), AV_OPT_TYPE_INT,
> { .i64 =
> > -1 }, -1,  1, VE }, \
> 
> As pointed out in other mail, I think this should be "force_idr" option and
> "Force I frames as IDR frames" as an option description. Secondly, why
> there are 3 values accepted: -1, 0, 1? I can understand 1 as enable the
> feature and 0 as don't enable, but what is -1? 

Please check my reply to Mark. And grep forced_idr implementation in nvenc.

>Also, how the option correlates with "-force_key_frames"?
> 
> From this perspective shouldn't the code be:
> 
> { "force_idr", "Force I frames as IDR
> frames", OFFSET(qsv.force_idr), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, VE },
> 
> if (frame->pict_type == AV_PICTURE_TYPE_I && (frame->key_frame || q-
> >force_idr)) {
>    enc_ctrl->FrameType = MFX_FRAMETYPE_I | MFX_FRAMETYPE_REF;
>    if (q->force_idr)
>   enc_ctrl->FrameType |= MFX_FRAMETYPE_IDR; }
> 
> This assumes that frame->key_frame corresponds to "-force_key_frames"
> in which I am not quite sure...
> 
> >
> >  typedef int SetEncodeCtrlCB (AVCodecContext *avctx,
> >   const AVFrame *frame,
> mfxEncodeCtrl*
> > enc_ctrl); @@ -168,6 +169,7 @@ typedef struct QSVEncContext {
> >  #endif
> >  char *load_plugins;
> >  SetEncodeCtrlCB *set_encode_ctrl_cb;
> > +int forced_idr;
> 
> int force_idr;
> 
> if agreed on the above...
> 
> >  } QSVEncContext;
> >
> >  int ff_qsv_enc_init(AVCodecContext *avctx, QSVEncContext *q);

I assume the reset of your comments have been replied by others. No? 
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH 1/5] lavc/qsvenc: add forced_idr opiton

2018-10-30 Thread Li, Zhong

> From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On Behalf
> Of Moritz Barsnick
> Sent: Friday, October 26, 2018 7:46 PM
> To: FFmpeg development discussions and patches
> 
> Subject: Re: [FFmpeg-devel] [PATCH 1/5] lavc/qsvenc: add forced_idr opiton
> 
> On Thu, Oct 25, 2018 at 20:36:07 +0800, Zhong Li wrote:
> > +{ "forced_idr", "Forcing I frames as IDR frames",
> OFFSET(qsv.forced_idr), AV_OPT_TYPE_INT, { .i64 = -1 }, -1,
> 1, VE }, \
> 
> ffmpeg uses imperative (mostly): "Force I frames as IDR frames".
> 
> Moritz

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

[FFmpeg-devel] [PATCH] avfilter: add tpad filter

2018-10-30 Thread Paul B Mahol

Signed-off-by: Paul B Mahol 
---
 doc/filters.texi |  22 ++
 libavfilter/Makefile |   1 +
 libavfilter/allfilters.c |   1 +
 libavfilter/vf_tpad.c| 166 +++
 4 files changed, 190 insertions(+)
 create mode 100644 libavfilter/vf_tpad.c

diff --git a/doc/filters.texi b/doc/filters.texi
index 9b84b1145b..6dda780aa0 100644
--- a/doc/filters.texi
+++ b/doc/filters.texi
@@ -16676,6 +16676,28 @@ embedded peak information in display metadata is not 
reliable or when tone
 mapping from a lower range to a higher range.
 @end table
 
+@section tpad
+
+Temporarily pad video frames.
+
+The filter accepts the following options:
+
+@table @option
+@item start
+Specify number of delay frames before input video stream.
+
+@item stop
+Specify number of padding frames after input video stream.
+Set to -1 to pad indefinitely.
+
+@item color
+Specify the color of the padded area. For the syntax of this option,
+check the @ref{color syntax,,"Color" section in the ffmpeg-utils
+manual,ffmpeg-utils}.
+
+The default value of @var{color} is "black".
+@end table
+
 @anchor{transpose}
 @section transpose
 
diff --git a/libavfilter/Makefile b/libavfilter/Makefile
index c35cd8f422..51e48efc2e 100644
--- a/libavfilter/Makefile
+++ b/libavfilter/Makefile
@@ -383,6 +383,7 @@ OBJS-$(CONFIG_TMIX_FILTER)   += vf_mix.o 
framesync.o
 OBJS-$(CONFIG_TONEMAP_FILTER)+= vf_tonemap.o colorspace.o
 OBJS-$(CONFIG_TONEMAP_OPENCL_FILTER) += vf_tonemap_opencl.o 
colorspace.o opencl.o \
 opencl/tonemap.o 
opencl/colorspace_common.o
+OBJS-$(CONFIG_TPAD_FILTER)   += vf_tpad.o
 OBJS-$(CONFIG_TRANSPOSE_FILTER)  += vf_transpose.o
 OBJS-$(CONFIG_TRANSPOSE_NPP_FILTER)  += vf_transpose_npp.o
 OBJS-$(CONFIG_TRIM_FILTER)   += trim.o
diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
index d5a211bda5..6052cb8c3c 100644
--- a/libavfilter/allfilters.c
+++ b/libavfilter/allfilters.c
@@ -365,6 +365,7 @@ 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_tpad;
 extern AVFilter ff_vf_transpose;
 extern AVFilter ff_vf_transpose_npp;
 extern AVFilter ff_vf_trim;
diff --git a/libavfilter/vf_tpad.c b/libavfilter/vf_tpad.c
new file mode 100644
index 00..e59037ca56
--- /dev/null
+++ b/libavfilter/vf_tpad.c
@@ -0,0 +1,166 @@
+/*
+ * Copyright (c) 2018 Paul B Mahol
+ *
+ * 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 "libavutil/avassert.h"
+#include "libavutil/channel_layout.h"
+#include "libavutil/opt.h"
+#include "avfilter.h"
+#include "audio.h"
+#include "filters.h"
+#include "internal.h"
+#include "formats.h"
+#include "drawutils.h"
+
+typedef struct TPadContext {
+const AVClass *class;
+int pad_start;
+int pad_stop;
+uint8_t rgba_color[4];  ///< color for the padding area
+
+FFDrawContext draw;
+FFDrawColor color;
+int64_t pts;
+int eof;
+} TPadContext;
+
+#define OFFSET(x) offsetof(TPadContext, x)
+#define VF AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
+
+static const AVOption tpad_options[] = {
+{ "start", "set the number of frames to delay input",  
OFFSET(pad_start),  AV_OPT_TYPE_INT,   {.i64=0},0, INT_MAX, VF },
+{ "stop",  "set the number of frames to add after input finished", 
OFFSET(pad_stop),   AV_OPT_TYPE_INT,   {.i64=0},   -1, INT_MAX, VF },
+{ "color", "set the color of the added frames",
OFFSET(rgba_color), AV_OPT_TYPE_COLOR, {.str="black"},  0,   0, VF },
+{ NULL }
+};
+
+AVFILTER_DEFINE_CLASS(tpad);
+
+static int query_formats(AVFilterContext *ctx)
+{
+return ff_set_common_formats(ctx, ff_draw_supported_pixel_formats(0));
+}
+
+static int activate(AVFilterContext *ctx)
+{
+AVFilterLink *inlink = ctx->inputs[0];
+AVFilterLink *outlink = ctx->outputs[0];
+TPadContext *s = ctx->priv;
+AVFrame *frame = NULL;
+int ret, status;
+int64_t pts;
+
+FF_FILTER_FORWARD_STATUS_BACK(outlink, inlink);
+
+if (s->pad_start > 0 && ff_outlink_frame_wanted(outlink)) {
+frame =

Re: [FFmpeg-devel] [PATCH 1/5] lavc/qsvenc: add forced_idr opiton

2018-10-30 Thread Li, Zhong

> From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On Behalf
> Of Mark Thompson
> Sent: Tuesday, October 30, 2018 5:06 AM
> To: ffmpeg-devel@ffmpeg.org
> Subject: Re: [FFmpeg-devel] [PATCH 1/5] lavc/qsvenc: add forced_idr opiton
> 
> On 25/10/18 13:36, Zhong Li wrote:
> > This option can be used to repect original input I/IDR frame type.
> >
> > Signed-off-by: Zhong Li 
> > ---
> >  libavcodec/qsvenc.c | 7 +++
> >  libavcodec/qsvenc.h | 2 ++
> >  2 files changed, 9 insertions(+)
> >
> > diff --git a/libavcodec/qsvenc.c b/libavcodec/qsvenc.c index
> > 948751d..e534dcf 100644
> > --- a/libavcodec/qsvenc.c
> > +++ b/libavcodec/qsvenc.c
> > @@ -1192,6 +1192,13 @@ static int encode_frame(AVCodecContext
> *avctx, QSVEncContext *q,
> >  if (qsv_frame) {
> >  surf = _frame->surface;
> >  enc_ctrl = _frame->enc_ctrl;
> > +
> > +if (q->forced_idr >= 0 && frame->pict_type ==
> AV_PICTURE_TYPE_I) {
> > +enc_ctrl->FrameType = MFX_FRAMETYPE_I |
> MFX_FRAMETYPE_REF;
> > +if (q->forced_idr || frame->key_frame)
> > +enc_ctrl->FrameType |= MFX_FRAMETYPE_IDR;
> > +} else
> > +enc_ctrl->FrameType = MFX_FRAMETYPE_UNKNOWN;
> >  }
> >
> >  ret = av_new_packet(_pkt, q->packet_size); diff --git
> > a/libavcodec/qsvenc.h b/libavcodec/qsvenc.h index 055b4a6..1f97f77
> > 100644
> > --- a/libavcodec/qsvenc.h
> > +++ b/libavcodec/qsvenc.h
> > @@ -87,6 +87,7 @@
> >  { "adaptive_i", "Adaptive I-frame placement",
> OFFSET(qsv.adaptive_i), AV_OPT_TYPE_INT, { .i64 = -1 }, -1,
> 1, VE }, \
> >  { "adaptive_b", "Adaptive B-frame placement",
> OFFSET(qsv.adaptive_b), AV_OPT_TYPE_INT, { .i64 = -1 }, -1,
> 1, VE }, \
> >  { "b_strategy", "Strategy to choose between I/P/B-frames",
> OFFSET(qsv.b_strategy),AV_OPT_TYPE_INT, { .i64 = -1 }, -1,
> 1, VE }, \
> > +{ "forced_idr", "Forcing I frames as IDR frames",
> OFFSET(qsv.forced_idr), AV_OPT_TYPE_INT, { .i64 = -1 }, -1,
> 1, VE }, \
> >
> >  typedef int SetEncodeCtrlCB (AVCodecContext *avctx,
> >   const AVFrame *frame,
> mfxEncodeCtrl*
> > enc_ctrl); @@ -168,6 +169,7 @@ typedef struct QSVEncContext {  #endif
> >  char *load_plugins;
> >  SetEncodeCtrlCB *set_encode_ctrl_cb;
> > +int forced_idr;
> >  } QSVEncContext;
> >
> >  int ff_qsv_enc_init(AVCodecContext *avctx, QSVEncContext *q);
> >
> 
> This seems confusing, because it doesn't match what forced_idr does in any
> other encoder.
> 
> Checking of pict_type for AV_PICTURE_TYPE_I in order to get a key frame
> (of whatever kind) is always enabled if supported (many encoders).  The
> "forced_idr" option to H.26[45] encoders (libx264, libx265) then forces that
> to be an IDR frame, not just an I frame.
> 
> - Mark
Yup, I know it doesn’t match other encoders such as libx264/libx265/nvenc. 
However, it is my intentional behavior. I believe current implement in 
libx264/libx265 is not complete.
One case is ignored: user just want to keep the gop structure as input, not to 
force all I frames as IDR frames.
So I have an idea:
Default value = -1, ignore the input gop structure. 
0: respect input gop structure but don't force I frame as IDR frame. 
1: force all I frame as IDR frame.

Since this is a qsv encoder private option, I just changed qsv implementation.
But I believe it should be a good to make other encoders follow such a way with 
separated patches.


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

Re: [FFmpeg-devel] [PATCH 4/5] lavc/qsvenc: add an option to set h264 pps for every frame

2018-10-30 Thread Li, Zhong

> From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On Behalf
> Of Moritz Barsnick
> Sent: Friday, October 26, 2018 7:49 PM
> To: FFmpeg development discussions and patches
> 
> Subject: Re: [FFmpeg-devel] [PATCH 4/5] lavc/qsvenc: add an option to set
> h264 pps for every frame
> 
> On Thu, Oct 25, 2018 at 20:36:10 +0800, Zhong Li wrote:
> > RepeatPPS is enabled by default in mfx. It is not necessary mostly and
> > wasting encoding bits.
> > Add an option to control it and disable it by default.
> 
> Could this change in behavior surprise anyone? If it's okay to change,
> perhaps at least bump lavc's micro version?
> 
> Moritz

I doubt how many people would like to repeat every frame's pps. 
Most codecs (x264/x265, and nvenc) won't do that by default.
Bumping lavc micro version should be a good suggestion.

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

Re: [FFmpeg-devel] [PATCH] add an option to forbid the fallback to sw when hardware init fails

2018-10-30 Thread Hendrik Leppkes

On Tue, Oct 30, 2018 at 10:08 AM Linjie Fu  wrote:
>
> Currently ff_get_format will go through all usable choices if the
> chosen format was not supported. It will fallback to software if
> the hardware init fails.
>
> According to the comment in ticket #7519, provided an option
> "-fallback_forbid 1" to return directly if hardware init fails
> and forbid the fallback to software.
>
> Signed-off-by: Linjie Fu 
> ---
>  libavcodec/avcodec.h   |  8 
>  libavcodec/decode.c| 11 +--
>  libavcodec/options_table.h |  1 +
>  3 files changed, 18 insertions(+), 2 deletions(-)
>
> diff --git a/libavcodec/avcodec.h b/libavcodec/avcodec.h
> index 705a3ce4f3..fac3c6acb2 100644
> --- a/libavcodec/avcodec.h
> +++ b/libavcodec/avcodec.h
> @@ -3312,6 +3312,14 @@ typedef struct AVCodecContext {
>   * used as reference pictures).
>   */
>  int extra_hw_frames;
> +
> +/**
> + * - forbid the fallback to software path in ff_get_format
> + * - when the hardware init fails. (0 -> disabled)
> + * - encoding: unused.
> + * - decoding: Set by user.
> + */
> +int fallback_forbid;
>  } AVCodecContext;
>

This is really something user-code should implement, its easy enough
to detect in eg. get_buffer2, and return an error from there.

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

[FFmpeg-devel] [PATCH] add an option to forbid the fallback to sw when hardware init fails

2018-10-30 Thread Linjie Fu

Currently ff_get_format will go through all usable choices if the
chosen format was not supported. It will fallback to software if
the hardware init fails.

According to the comment in ticket #7519, provided an option
"-fallback_forbid 1" to return directly if hardware init fails
and forbid the fallback to software.

Signed-off-by: Linjie Fu 
---
 libavcodec/avcodec.h   |  8 
 libavcodec/decode.c| 11 +--
 libavcodec/options_table.h |  1 +
 3 files changed, 18 insertions(+), 2 deletions(-)

diff --git a/libavcodec/avcodec.h b/libavcodec/avcodec.h
index 705a3ce4f3..fac3c6acb2 100644
--- a/libavcodec/avcodec.h
+++ b/libavcodec/avcodec.h
@@ -3312,6 +3312,14 @@ typedef struct AVCodecContext {
  * used as reference pictures).
  */
 int extra_hw_frames;
+
+/**
+ * - forbid the fallback to software path in ff_get_format
+ * - when the hardware init fails. (0 -> disabled)
+ * - encoding: unused.
+ * - decoding: Set by user.
+ */
+int fallback_forbid;
 } AVCodecContext;
 
 #if FF_API_CODEC_GET_SET
diff --git a/libavcodec/decode.c b/libavcodec/decode.c
index 4607e9f318..edadbd7e03 100644
--- a/libavcodec/decode.c
+++ b/libavcodec/decode.c
@@ -1441,8 +1441,15 @@ int ff_get_format(AVCodecContext *avctx, const enum 
AVPixelFormat *fmt)
 av_log(avctx, AV_LOG_DEBUG, "Format %s requires hwaccel "
"initialisation.\n", desc->name);
 err = hwaccel_init(avctx, hw_config);
-if (err < 0)
-goto try_again;
+if (err < 0) {
+if (avctx->fallback_forbid) {
+av_log(avctx, AV_LOG_ERROR, "Format %s not usable, 
fallback "
+"was forbidden.\n", desc->name);
+ret = AV_PIX_FMT_NONE;
+break;
+} else
+goto try_again;
+}
 }
 ret = user_choice;
 break;
diff --git a/libavcodec/options_table.h b/libavcodec/options_table.h
index 099261e168..73f0333eeb 100644
--- a/libavcodec/options_table.h
+++ b/libavcodec/options_table.h
@@ -479,6 +479,7 @@ static const AVOption avcodec_options[] = {
 {"allow_high_depth", "allow to output YUV pixel formats with a different 
chroma sampling than 4:2:0 and/or other than 8 bits per component", 0, 
AV_OPT_TYPE_CONST, {.i64 = AV_HWACCEL_FLAG_ALLOW_HIGH_DEPTH }, INT_MIN, 
INT_MAX, V | D, "hwaccel_flags"},
 {"allow_profile_mismatch", "attempt to decode anyway if HW accelerated 
decoder's supported profiles do not exactly match the stream", 0, 
AV_OPT_TYPE_CONST, {.i64 = AV_HWACCEL_FLAG_ALLOW_PROFILE_MISMATCH }, INT_MIN, 
INT_MAX, V | D, "hwaccel_flags"},
 {"extra_hw_frames", "Number of extra hardware frames to allocate for the 
user", OFFSET(extra_hw_frames), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, 
V|D },
+{"fallback_forbid", "forbid the fallback to software path when hardware init 
fails", OFFSET(fallback_forbid), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, V|D },
 {NULL},
 };
 
-- 
2.17.1

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

Re: [FFmpeg-devel] [PATCH 4/4] lavfi/opencl: remove peak option of tonemap_opencl

2018-10-30 Thread myp...@gmail.com

On Mon, Oct 29, 2018 at 1:21 PM Ruiling Song  wrote:
>
> Since the filter use auto-calculate the peak value,
> the option does not work as expected. So, remove it.
>
> Signed-off-by: Ruiling Song 
> ---
>  libavfilter/vf_tonemap_opencl.c | 7 ++-
>  1 file changed, 2 insertions(+), 5 deletions(-)
>
> diff --git a/libavfilter/vf_tonemap_opencl.c
b/libavfilter/vf_tonemap_opencl.c
> index cd293c2..88b3107 100644
> --- a/libavfilter/vf_tonemap_opencl.c
> +++ b/libavfilter/vf_tonemap_opencl.c
> @@ -62,7 +62,6 @@ typedef struct TonemapOpenCLContext {
>
>  enum TonemapAlgorithm tonemap;
>  enum AVPixelFormatformat;
> -doublepeak;
>  doubleparam;
>  doubledesat_param;
>  doubletarget_peak;
> @@ -349,7 +348,7 @@ static int tonemap_opencl_filter_frame(AVFilterLink
*inlink, AVFrame *input)
>  AVFrame *output = NULL;
>  cl_int cle;
>  int err;
> -double peak = ctx->peak;
> +double peak;
>
>  AVHWFramesContext *input_frames_ctx =
>  (AVHWFramesContext*)input->hw_frames_ctx->data;
> @@ -371,8 +370,7 @@ static int tonemap_opencl_filter_frame(AVFilterLink
*inlink, AVFrame *input)
>  if (err < 0)
>  goto fail;
>
> -if (!peak)
> -peak = ff_determine_signal_peak(input);
> +peak = ff_determine_signal_peak(input);
>
>  if (ctx->trc != -1)
>  output->color_trc = ctx->trc;
> @@ -518,7 +516,6 @@ static const AVOption tonemap_opencl_options[] = {
>  { "limited",   0,   0,
AV_OPT_TYPE_CONST, {.i64 = AVCOL_RANGE_MPEG}, 0, 0, FLAGS, "range"
},
>  { "full",  0,   0,
AV_OPT_TYPE_CONST, {.i64 = AVCOL_RANGE_JPEG}, 0, 0, FLAGS, "range"
},
>  { "format","output pixel format", OFFSET(format),
AV_OPT_TYPE_PIXEL_FMT, {.i64 = AV_PIX_FMT_NONE}, AV_PIX_FMT_NONE, INT_MAX,
FLAGS, "fmt" },
> -{ "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 },
> --
LGTM if we don't use this option
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH 3/4] doc/filters: add tonemap_opencl document.

2018-10-30 Thread Gyan

On Mon, Oct 29, 2018 at 11:29 AM Ruiling Song 
wrote:

> Signed-off-by: Ruiling Song 
> ---
>  doc/filters.texi | 96
> 
>  1 file changed, 96 insertions(+)
>
> diff --git a/doc/filters.texi b/doc/filters.texi
> index 83df460..f884ba4 100644
> --- a/doc/filters.texi
> +++ b/doc/filters.texi
> @@ -16387,6 +16387,7 @@ tmix=frames=3:weights="-1 2 -1":scale=1
>  @end example
>  @end itemize
>
> +@anchor{tonemap}
>  @section tonemap
>  Tone map colors from different dynamic ranges.
>
> @@ -18440,6 +18441,101 @@ Apply sobel operator with scale set to 2 and
> delta set to 10
>  @end example
>  @end itemize
>
> +@section tonemap_opencl
> +
> +Perform HDR(PQ/HLG) to SDR conversion with tone-mapping.
> +
> +It accepts the following parameters:
> +
> +@table @option
> +@item tonemap
> +Specify the tone-mapping operator to be used. Same as tonemap option in
> @ref{tonemap}.
> +
> +@item param
> +Tune the tone mapping algorithm. same as param option in @ref{tonemap}.
> +
> +@item desat
> +Apply desaturation for highlights that exceed this level of brightness.
> The
> +higher the parameter, the more color information will be preserved. This
> +setting helps prevent unnaturally blown-out colors for super-highlights,
> by
> +(smoothly) turning into white instead. This makes images feel more
> natural,
> +at the cost of reducing information about out-of-range colors.
> +
> +The default value is 0.5, and the algorithm here is a little different
> from
> +the cpu version tonemap currently. A setting of 0.0 disables this option.
> +
> +@item threshold
> +The tonemapping algorithm parameters is fine-tuned per each scene. And a
> threshold
> +is used to detect whether the scene has changed or not. If the distance
> beween
> +the current frame average brightness and the current running average
> exceeds
> +a threshold value, we would re-calculate scene average and peak
> brightness.
> +The default value is 0.2.
> +
> +@item format
> +Specify the output pixel format.
> +
> +Currently supported formats are:
> +@table @var
> +@item p010
> +@item nv12
> +@end table
> +
> +@item range, r
> +Set the output color range.
> +
> +Possible values are:
> +@table @var
> +@item tv/mpeg
> +@item pc/jpeg
> +@end table
> +
> +Default is same as input.
> +
> +@item primaries, p
> +Set the output color primaries.
> +
> +Possible values are:
> +@table @var
> +@item bt709
> +@item bt2020
> +@end table
> +
> +Default is same as input.
> +
> +@item transfer, t
> +Set the output transfer characteristics.
> +
> +Possible values are:
> +@table @var
> +@item bt709
> +@item bt2020
> +@end table
> +
> +Default is bt709.
> +
> +@item matrix, m
> +Set the output colorspace matrix.
> +
> +Possible value are:
> +@table @var
> +@item bt709
> +@item bt2020
> +@end table
> +
> +Default is same as input.
> +
> +@end table
> +
> +@subsection Example
> +
> +@itemize
> +@item
> +Convert HDR(PQ/HLG) video to bt2020-transfer-characteristic p010 format
> using linear operator.
> +@example
> +-i INPUT -vf
> "format=p010,hwupload,tonemap_opencl=t=bt2020:tonemap=linear:format=p010,hwdownload,format=p010"
> OUTPUT
> +@end example
> +@end itemize
> +
>  @section unsharp_opencl
>

LGTM.

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

Re: [FFmpeg-devel] [PATCH 1/4] doc/filters: add document for opencl filters

2018-10-30 Thread Gyan

On Mon, Oct 29, 2018 at 10:50 AM Ruiling Song 
wrote:

> Signed-off-by: Danil Iashchenko 
> Signed-off-by: Ruiling Song 
> ---
> Seems like Danil is not working on this recently.
> So I re-submit this patch to address the comment over overlay_opencl.
>

Looks good for a push now. I'll make any changes later on.

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

Re: [FFmpeg-devel] [PATCH] avcodec/libvpxdec: fix setting auto threads

2018-10-30 Thread James Zern

On Mon, Oct 29, 2018 at 10:58 PM James Zern  wrote:
>
> On Sat, Oct 27, 2018 at 1:54 PM James Almer  wrote:
> >
> > On 10/27/2018 5:52 PM, James Zern wrote:
> > > a thread count of 0 is treated the same as 1, use av_cpu_count() to get
> > > the correct thread count when auto threads is requested.
> > >
> > > this matches the fix in libvpxenc:
> > > 27df34bf1f avcodec/libvpxenc: fix setting amount of threads used for 
> > > encoding
> > >
> > > Signed-off-by: James Zern 
> > > ---
> > >  libavcodec/libvpxdec.c | 3 +--
> > >  1 file changed, 1 insertion(+), 2 deletions(-)
> > >

Applied with the limit of 16 restored. Thanks for the review.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avcodec/libvpxdec: fix setting auto threads

2018-10-30 Thread James Zern

On Sat, Oct 27, 2018 at 1:54 PM James Almer  wrote:
>
> On 10/27/2018 5:52 PM, James Zern wrote:
> > a thread count of 0 is treated the same as 1, use av_cpu_count() to get
> > the correct thread count when auto threads is requested.
> >
> > this matches the fix in libvpxenc:
> > 27df34bf1f avcodec/libvpxenc: fix setting amount of threads used for 
> > encoding
> >
> > Signed-off-by: James Zern 
> > ---
> >  libavcodec/libvpxdec.c | 3 +--
> >  1 file changed, 1 insertion(+), 2 deletions(-)
> >
> > diff --git a/libavcodec/libvpxdec.c b/libavcodec/libvpxdec.c
> > index 04f27d3396..5193ac51c3 100644
> > --- a/libavcodec/libvpxdec.c
> > +++ b/libavcodec/libvpxdec.c
> > @@ -47,8 +47,7 @@ static av_cold int vpx_init(AVCodecContext *avctx,
> >  {
> >  VPxContext *ctx = avctx->priv_data;
> >  struct vpx_codec_dec_cfg deccfg = {
> > -/* token partitions+1 would be a decent choice */
> > -.threads = FFMIN(avctx->thread_count, 16)
> > +.threads = avctx->thread_count ? avctx->thread_count : 
> > av_cpu_count()
>
> If the limit of 16 threads is correct and unless libvpx clips it on its
> own, then you shouldn't remove the check.
>

If anything it was a limit put in at one stage of vp8 development, the
comment is certainly vp8 specific. vp8/9 both clip in their own ways
[1][2], but looking at it there may have been some changes to the loop
filter that aren't limiting this number. I'll restore it to be safe.
On the libvpxenc side I don't think there was ever a check in the
wrapper, but the library does put a hard cap on the requested threads
(64) and will fail otherwise.

[1] 
https://chromium.googlesource.com/webm/libvpx/+/v1.7.0/vp8/decoder/threading.c#596
[2] 
https://chromium.googlesource.com/webm/libvpx/+/v1.7.0/vp9/decoder/vp9_decodeframe.c#1542

> LGTM otherwise.
>
> >  };
> >
> >  av_log(avctx, AV_LOG_INFO, "%s\n", vpx_codec_version_str());
> >
>
> ___
> 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

46 matches

Mail list logo