From: Kaustubh Raste
Replace generic with block size specific function.
Signed-off-by: Kaustubh Raste
---
libavcodec/mips/h264chroma_msa.c | 237 ++
1 file changed, 112 insertions(+), 125 deletions(-)
diff --git a/libavcodec/mips/h264chroma_msa.c b/libavco
From: Kaustubh Raste
Replace generic with block size specific function.
Signed-off-by: Kaustubh Raste
---
libavcodec/mips/h264dsp_msa.c | 423 ++-
libavutil/mips/generic_macros_msa.h | 36 +++
2 files changed, 306 insertions(+), 153 deletions(-)
diff -
From: Kaustubh Raste
Signed-off-by: Kaustubh Raste
---
libavcodec/mips/h264pred_msa.c | 215 +---
1 file changed, 92 insertions(+), 123 deletions(-)
diff --git a/libavcodec/mips/h264pred_msa.c b/libavcodec/mips/h264pred_msa.c
index c297aec..b9990c1 100644
-
> -Original Message-
> From: Li, Zhong
> Sent: Thursday, September 21, 2017 2:13 PM
> To: ffmpeg-devel@ffmpeg.org
> Cc: s...@jkqxz.net; Zhao, Jun ; nfx...@googlemail.com;
> Li, Zhong
> Subject: [FFmpeg-devel][PATCH V2] mpeg2_hwaccel: always submit twice
> for a frame with two fileds.
>
>
This allows the use of the `ignore_chapters` option to avoid performing
extra seeks at startup without producing "subtitle" chapter streams.
---
libavformat/mov.c | 16
1 file changed, 12 insertions(+), 4 deletions(-)
diff --git a/libavformat/mov.c b/libavformat/mov.c
index 25197
---
libavformat/dashenc.c | 13 +
1 file changed, 13 insertions(+)
diff --git a/libavformat/dashenc.c b/libavformat/dashenc.c
index 6471649eb7..5e5bea54c4 100644
--- a/libavformat/dashenc.c
+++ b/libavformat/dashenc.c
@@ -90,6 +90,7 @@ typedef struct DASHContext {
AVRational min_
---
libavdevice/fbdev_enc.c | 1 +
libavformat/os_support.h | 6 ++
libavutil/file.c | 1 +
3 files changed, 8 insertions(+)
diff --git a/libavdevice/fbdev_enc.c b/libavdevice/fbdev_enc.c
index b4e5f84975..a9be608383 100644
--- a/libavdevice/fbdev_enc.c
+++ b/libavdevice/fbdev_enc.c
From: Aman Gupta
If the VideoToolbox session needs to be restarted, and
videotoolbox_start() fails for some reason (for instance, if the video
is interlaced and the decoder is running on iOS), avcodec will return
AVERROR_EXTERNAL. This can be used by the API user to switch to another
decoder.
---
From: Aman Gupta
This callback will be used by the VideoToolbox H264 hwaccel so that it
can receive SPS and PPS NALUs. VideoToolbox requires PPS changes to be
fed into the decoder session, and for the session to be recreated when
the SPS changes.
---
libavcodec/avcodec.h | 12
libav
From: Aman Gupta
This allows decode_slice to be invoked multiple times before end_frame,
causing slices to accumulate before being fed into the VT decoder.
An upcoming commit will re-use decode_slice for SPS and PPS nalus, so
they can be propagated into the VT decoder session along with slide
da
From: Aman Gupta
---
libavcodec/videotoolboxenc.c | 2 --
1 file changed, 2 deletions(-)
diff --git a/libavcodec/videotoolboxenc.c b/libavcodec/videotoolboxenc.c
index eba6cc672f..a7bef72da8 100644
--- a/libavcodec/videotoolboxenc.c
+++ b/libavcodec/videotoolboxenc.c
@@ -1853,8 +1853,6 @@ stati
From: Aman Gupta
The only reason videotoolbox wouldn't produce frames is if the data fed
to it was invalid, so returning AVERROR_INVALIDDATA makes sense here.
Further, it means AVERROR_EXTERNAL can be used in further commits to signal
fatal VideoToolbox errors, letting the user know that they ne
From: Aman Gupta
Removes the avctx->extradata_size requirement when creating avcC, since
avctx->extradata is only used in the esds code path.
This fixes an issue where the VideoToolbox decoder would not work unless
avformat_find_stream_info() was called.
---
libavcodec/videotoolbox.c | 65 +
From: Aman Gupta
These helpers will be used in later commits to automatically restart
the decoder session when SPS changes are encountered.
---
libavcodec/videotoolbox.c | 36 ++--
1 file changed, 22 insertions(+), 14 deletions(-)
diff --git a/libavcodec/videotoo
From: Aman Gupta
VideoToolbox's support for interlaced H264 is quite poor.
On macOS, VTSessionCopySupportedPropertyDictionary() will show that
kVTDecompressionPropertyKey_FieldMode is supported. Possible values for
this option include DeinterlaceFields and BothFields.
However, files that use MB
From: Aman Gupta
---
libavcodec/videotoolbox.c | 17 -
1 file changed, 16 insertions(+), 1 deletion(-)
diff --git a/libavcodec/videotoolbox.c b/libavcodec/videotoolbox.c
index 1de556f3e8..f56ab1f8c9 100644
--- a/libavcodec/videotoolbox.c
+++ b/libavcodec/videotoolbox.c
@@ -425,7
On 9/25/2017 3:08 PM, James Almer wrote:
> On 9/23/2017 8:35 PM, James Almer wrote:
>> It does the same thing as av_packet_ref().
>>
>> Signed-off-by: James Almer
>> ---
>> libavcodec/avcodec.h | 3 +++
>> libavcodec/avpacket.c | 4 ++--
>> 2 files changed, 5 insertions(+), 2 deletions(-)
>>
>>
On 2017/9/21 15:21, Jun Zhao wrote:
Ping? and can you help to review the small patch ?
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
From: Martin Storsjö
This reduces the need for an edit list; streams that start with
e.g. dts=-1, pts=0 can be encoded as dts=0, pts=0 (which is valid
in mov/mp4) by shifting the dts values of all packets forward.
This avoids the need for edit lists for such streams (while they
still are needed f
From: Martin Storsjö
This reduces the need for an edit list; streams that start with
e.g. dts=-1, pts=0 can be encoded as dts=0, pts=0 (which is valid
in mov/mp4) by shifting the dts values of all packets forward.
This avoids the need for edit lists for such streams (while they
still are needed f
W dniu 2017-09-25 o 22:53, Carl Eugen Hoyos pisze:
> 2017-09-23 19:18 GMT+02:00 Mateusz :
>> W dniu 2017-09-23 o 17:01, Michael Niedermayer pisze:
>>> On Fri, Sep 22, 2017 at 02:10:01AM +0200, Mateusz wrote:
To reduce bit depth in planar YUV or gray pixel formats
ffmpeg uses DITHER_COPY m
On Wed, May 24, 2017 at 6:47 AM, wm4 wrote:
> With the new decode API, you can't handle errors directly in the API
> user - you only know that the hwaccel did not initialize at all.
>
> Add some approximate logging.
> ---
> libavcodec/videotoolbox.c | 6 ++
> 1 file changed, 6 insertions(+)
On 09/25/2017 02:27 PM, Carl Eugen Hoyos wrote:
> 2017-09-25 23:12 GMT+02:00 John Stebbins :
>> FYI, such a facility for marking frames to drop would also
>> be useful for frame accurate playback of MP4 edit lists.
> Do you have samples with edit lists that do not play
> correctly with current FFmp
On Sat, Sep 23, 2017 at 5:44 PM Helmut K. C. Tessarek
wrote:
> On 2017-09-22 16:34, Joseph Van Riper wrote:
> > I would prefer not to sprinkle changes throughout something
> > if I can avoid it.
>
> Have you thought about a third party library then?
>
> This library can be maintained by the Direc
2017-09-25 23:12 GMT+02:00 John Stebbins :
> FYI, such a facility for marking frames to drop would also
> be useful for frame accurate playback of MP4 edit lists.
Do you have samples with edit lists that do not play
correctly with current FFmpeg?
(Others will hopefully comment on this issue, I ce
On 09/25/2017 01:12 PM, Carl Eugen Hoyos wrote:
> 2017-09-25 19:10 GMT+02:00 John Stebbins :
>> When keyframe intervals of dash segments are not perfectly aligned,
>> fragments in the stream can overlap in time. Append new "trun" index
>> entries to the end of the index instead of sorting by time
2017-09-23 19:18 GMT+02:00 Mateusz :
> W dniu 2017-09-23 o 17:01, Michael Niedermayer pisze:
>> On Fri, Sep 22, 2017 at 02:10:01AM +0200, Mateusz wrote:
>>> To reduce bit depth in planar YUV or gray pixel formats
>>> ffmpeg uses DITHER_COPY macro.
>>> Now it makes images greener and with visible di
On Sun, Sep 24, 2017 at 09:14:28PM -0400, Kevin Mark wrote:
> Hi Ronald,
>
> On Sat, Sep 23, 2017 at 11:54 AM, Ronald S. Bultje wrote:
> >
> > Yes, -E works on Mac.
>
> Thanks! Are there any remaining blockers for a potential merge?
There is no gurantee that 3 av_log() calls will be executed wi
It leaks memory and destroys the dst packet in case of failure, and it
ultimately duplicates functionality already existing in the saner
av_packet_copy_props().
Signed-off-by: James Almer
---
libavcodec/avcodec.h | 3 +++
1 file changed, 3 insertions(+)
diff --git a/libavcodec/avcodec.h b/libav
2017-09-19 22:54 GMT+02:00 Thomas Mundt :
> The attached patch needs to be applied on top of
> "avfilter/interlace: add support for 10 and 12 bit".
Pushed.
Thank you, Carl Eugen
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mai
2017-09-25 19:10 GMT+02:00 John Stebbins :
> When keyframe intervals of dash segments are not perfectly aligned,
> fragments in the stream can overlap in time. Append new "trun" index
> entries to the end of the index instead of sorting by timestamp.
> Sorting by timestamp causes packets to be read
2017-09-25 21:25 GMT+02:00 Martin Vignali :
> 2017-09-25 20:19 GMT+02:00 Carl Eugen Hoyos :
>
>> 2017-09-25 9:48 GMT+02:00 Martin Vignali :
>> >>
>> >> Sorry if I misunderstand:
>> >> If the format only contains one layer, why is the pix_fmt
>> >> rgb0 and not gray8?
>>
>> > I follow the existing t
2017-09-25 20:49 GMT+02:00 Bjorn Roche :
> In the meantime, I've attached the patch if anyone wants to look at it.
Can you explain why you had to patch the gif encoder?
I thought it is able to encode transparency, no?
Carl Eugen
___
ffmpeg-devel mailin
On 9/25/2017 4:28 PM, Rostislav Pehlivanov wrote:
> Signed-off-by: Rostislav Pehlivanov
> ---
> doc/APIchanges | 3 +++
> libavutil/frame.h | 6 ++
> libavutil/version.h | 2 +-
> 3 files changed, 10 insertions(+), 1 deletion(-)
>
> diff --git a/doc/APIchanges b/doc/APIchanges
> index
On 9/25/2017 4:28 PM, Rostislav Pehlivanov wrote:
> Signed-off-by: Rostislav Pehlivanov
> ---
> libavcodec/pngdec.c | 48
> 1 file changed, 48 insertions(+)
>
> diff --git a/libavcodec/pngdec.c b/libavcodec/pngdec.c
> index 0d6612ccca..2d005f90fc
On 9/25/2017 4:28 PM, Rostislav Pehlivanov wrote:
> Signed-off-by: Rostislav Pehlivanov
> ---
> ffprobe.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/ffprobe.c b/ffprobe.c
> index b2e8949d9f..171f856c2d 100644
> --- a/ffprobe.c
> +++ b/ffprobe.c
> @@ -2227,6 +2227,9 @@ static voi
Signed-off-by: Rostislav Pehlivanov
---
doc/APIchanges | 3 +++
libavutil/frame.h | 6 ++
libavutil/version.h | 2 +-
3 files changed, 10 insertions(+), 1 deletion(-)
diff --git a/doc/APIchanges b/doc/APIchanges
index d06144f1e9..427d7e5613 100644
--- a/doc/APIchanges
+++ b/doc/APIcha
Signed-off-by: Rostislav Pehlivanov
---
ffprobe.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/ffprobe.c b/ffprobe.c
index b2e8949d9f..171f856c2d 100644
--- a/ffprobe.c
+++ b/ffprobe.c
@@ -2227,6 +2227,9 @@ static void show_frame(WriterContext *w, AVFrame *frame,
AVStream *stream,
Signed-off-by: Rostislav Pehlivanov
---
libavcodec/pngdec.c | 48
1 file changed, 48 insertions(+)
diff --git a/libavcodec/pngdec.c b/libavcodec/pngdec.c
index 0d6612ccca..2d005f90fc 100644
--- a/libavcodec/pngdec.c
+++ b/libavcodec/pngdec.c
@@ -2
2017-09-25 20:19 GMT+02:00 Carl Eugen Hoyos :
> 2017-09-25 9:48 GMT+02:00 Martin Vignali :
> >>
> >> Sorry if I misunderstand:
> >> If the format only contains one layer, why is the pix_fmt
> >> rgb0 and not gray8?
>
> > I follow the existing texture dsp design, who output RGB0
> > or RGBA dependi
On 9/25/2017 3:49 PM, Bjorn Roche wrote:
> Hi there,
>
> I have a patch for this issue:
>
> https://trac.ffmpeg.org/ticket/4443
>
> However, when running the automated ("fate") tests, I get the error pasted
> below. I'm not sure what's going on since the output formats look different
> when runn
On Mon, Sep 25, 2017 at 3:06 AM, wm4 wrote:
> On Mon, 25 Sep 2017 09:02:36 +0200
> Hendrik Leppkes wrote:
>
> > On Mon, Sep 25, 2017 at 3:31 AM, Aman Gupta wrote:
> > >
> > > How do the other hwaccels handle mid-stream SPS changes?
> > >
> >
> > Real HWAccels (ie. VAAPI, VDPAU or DXVA) communic
Hi there,
I have a patch for this issue:
https://trac.ffmpeg.org/ticket/4443
However, when running the automated ("fate") tests, I get the error pasted
below. I'm not sure what's going on since the output formats look different
when running a the test vs when I run the same command with my built
On Wed, 20 Sep 2017, Jeyapal, Karthick wrote:
Then please add the relevant copyright line to the file you changed.
(Probably "Rafaël Carré" but please verify yourself)
Carl Eugen
Thanks for the clarification. I have added the copyright line as you
have advised. (Yes, I have verified that th
2017-09-25 9:48 GMT+02:00 Martin Vignali :
>>
>> Sorry if I misunderstand:
>> If the format only contains one layer, why is the pix_fmt
>> rgb0 and not gray8?
> I follow the existing texture dsp design, who output RGB0
> or RGBA depending of the texture.
> But you're right, the format is gray only
2017-09-25 9:55 GMT+02:00 Martin Vignali :
>>
>> Same question as before:
>> How can 24bit input be encoded into alpha?
>
> For now, the texturedspenc, only support rgba as input.
> This encoder, use the alpha part of the rgba input and
> ignore the rgb part.
>
> But maybe it's more convenient, to
On 9/23/2017 8:35 PM, James Almer wrote:
> It does the same thing as av_packet_ref().
>
> Signed-off-by: James Almer
> ---
> libavcodec/avcodec.h | 3 +++
> libavcodec/avpacket.c | 4 ++--
> 2 files changed, 5 insertions(+), 2 deletions(-)
>
> diff --git a/libavcodec/avcodec.h b/libavcodec/avc
On 09/25/2017 10:31 AM, wm4 wrote:
> On Mon, 25 Sep 2017 10:10:58 -0700
> John Stebbins wrote:
>
>> When keyframe intervals of dash segments are not perfectly aligned,
>> fragments in the stream can overlap in time. Append new "trun" index
>> entries to the end of the index instead of sorting by t
On Thu, 21 Sep 2017, Gildas Fargeas wrote:
Hey, do you need more stuff changed on this patch ?
No, just waiting for other pending decklink patches to push (and test)
them at once.
Regards,
Marton
2017-09-07 14:46 GMT+02:00 Gildas Fargeas :
Ok, I changed the texi, added the bitrate an
On 9/25/2017 2:45 PM, wm4 wrote:
> On Mon, 25 Sep 2017 14:37:52 -0300
> James Almer wrote:
>
>> On 9/25/2017 2:29 PM, wm4 wrote:
>>> On Mon, 25 Sep 2017 14:07:54 -0300
>>> James Almer wrote:
>>>
On 9/25/2017 1:43 PM, wm4 wrote:
> On Mon, 25 Sep 2017 10:58:31 -0300
> James Almer
On 09/25/2017 10:12 AM, Carl Eugen Hoyos wrote:
> 2017-09-25 19:10 GMT+02:00 John Stebbins :
>> When keyframe intervals of dash segments are not perfectly aligned,
>> fragments in the stream can overlap in time. Append new "trun" index
>> entries to the end of the index instead of sorting by timest
On Mon, 25 Sep 2017 14:37:52 -0300
James Almer wrote:
> On 9/25/2017 2:29 PM, wm4 wrote:
> > On Mon, 25 Sep 2017 14:07:54 -0300
> > James Almer wrote:
> >
> >> On 9/25/2017 1:43 PM, wm4 wrote:
> >>> On Mon, 25 Sep 2017 10:58:31 -0300
> >>> James Almer wrote:
> >>>
> > Using av_pac
On 9/25/2017 2:29 PM, wm4 wrote:
> On Mon, 25 Sep 2017 14:07:54 -0300
> James Almer wrote:
>
>> On 9/25/2017 1:43 PM, wm4 wrote:
>>> On Mon, 25 Sep 2017 10:58:31 -0300
>>> James Almer wrote:
>>>
> Using av_packet_copy_props() instead of introducing yet another weird
> function would b
On Mon, 25 Sep 2017 10:10:58 -0700
John Stebbins wrote:
> When keyframe intervals of dash segments are not perfectly aligned,
> fragments in the stream can overlap in time. Append new "trun" index
> entries to the end of the index instead of sorting by timestamp.
> Sorting by timestamp causes pac
On Mon, 25 Sep 2017 14:07:54 -0300
James Almer wrote:
> On 9/25/2017 1:43 PM, wm4 wrote:
> > On Mon, 25 Sep 2017 10:58:31 -0300
> > James Almer wrote:
> >
> >>> Using av_packet_copy_props() instead of introducing yet another weird
> >>> function would be better.
> >>
> >> It can't be used
2017-09-25 19:10 GMT+02:00 John Stebbins :
> When keyframe intervals of dash segments are not perfectly aligned,
> fragments in the stream can overlap in time. Append new "trun" index
> entries to the end of the index instead of sorting by timestamp.
> Sorting by timestamp causes packets to be read
When keyframe intervals of dash segments are not perfectly aligned,
fragments in the stream can overlap in time. Append new "trun" index
entries to the end of the index instead of sorting by timestamp.
Sorting by timestamp causes packets to be read out of decode order and
results in decode errors.
On 9/25/2017 1:43 PM, wm4 wrote:
> On Mon, 25 Sep 2017 10:58:31 -0300
> James Almer wrote:
>
>>> Using av_packet_copy_props() instead of introducing yet another weird
>>> function would be better.
>>
>> It can't be used in some cases. Look at the last two patches in the set.
>> copy_props can't
2017-09-19 22:54 GMT+02:00 Thomas Mundt :
> The attached patch needs to be applied on top of
> "avfilter/interlace: add support for 10 and 12 bit".
> Thanks.
>
Ping
This patch is needful cosmetic after last commit.
___
ffmpeg-devel mailing list
ffmpeg-d
On Mon, 25 Sep 2017 10:58:31 -0300
James Almer wrote:
> > Using av_packet_copy_props() instead of introducing yet another weird
> > function would be better.
>
> It can't be used in some cases. Look at the last two patches in the set.
> copy_props can't be used there without some extra pointle
On 9/25/2017 6:50 AM, wm4 wrote:
> On Sun, 24 Sep 2017 22:06:09 -0300
> James Almer wrote:
>
>> It's added as a replacemet of av_copy_packet_side_data() using the proper
>> av_packet_* namespace, and differs from it in the following ways:
>>
>> - Side data already allocated and copied is properly
From: Kaustubh Raste
Signed-off-by: Kaustubh Raste
---
libavcodec/mips/hevc_lpf_sao_msa.c | 197
1 file changed, 135 insertions(+), 62 deletions(-)
diff --git a/libavcodec/mips/hevc_lpf_sao_msa.c
b/libavcodec/mips/hevc_lpf_sao_msa.c
index 39c647e..c192265
Align order of "start_time" option within filter documentation to actual
implementation and add missing AV_OPT_FLAG_FILTERING_PARAM flag. Fix
indent of "round" named constants and clear unused field values.
Also fix some documentation cosmetics.
Signed-off-by: Tobias Rapp
---
doc/filters.texi
Allows to specify the action to be performed when reading the last frame
from the internal FIFO buffer. By default the last frame is written to
filter output depending on the timestamp rounding method. When using
"pass" action the last frame is passed through if input duration
has not been reached
On Mon, 25 Sep 2017 09:02:36 +0200
Hendrik Leppkes wrote:
> On Mon, Sep 25, 2017 at 3:31 AM, Aman Gupta wrote:
> >
> > How do the other hwaccels handle mid-stream SPS changes?
> >
>
> Real HWAccels (ie. VAAPI, VDPAU or DXVA) communicate the SPS/PPS
> content for every frame, they don't keep a
On Sun, 24 Sep 2017 22:06:09 -0300
James Almer wrote:
> It's added as a replacemet of av_copy_packet_side_data() using the proper
> av_packet_* namespace, and differs from it in the following ways:
>
> - Side data already allocated and copied is properly freed in case of failure.
> - The dst pac
On 23.09.2017 17:05, Michael Niedermayer wrote:
On Fri, Sep 22, 2017 at 08:52:26AM +0200, Tobias Rapp wrote:
On 22.09.2017 01:58, Michael Niedermayer wrote:
On Thu, Sep 21, 2017 at 04:55:51PM +0200, Tobias Rapp wrote:
Align order of "start_time" option to documentation and add missing
AV_OPT_F
>
>
> Yes.
>
> Or we could switch now to a file that we know we will have
> to add.
> No strong opinion here: I just wanted to point out that we
> will need another sample - that we already have. I wonder
> if keeping the size of fate small(er) is also a relevant
> argument.
>
Ok,
feel free to rep
2017-09-25 9:52 GMT+02:00 Martin Vignali :
>> >
>> > But we can of course have more than one sample
>> > for cineform...
>
> I don't know the exact guidelines for fate test, but i think
> we can have several sample, who test several things.
>
> For now, there is no fate test for this decoder
> we c
>
> Same question as before:
> How can 24bit input be encoded into alpha?
>
>
>
> For now, the texturedspenc, only support rgba as input.
This encoder, use the alpha part of the rgba input and ignore the rgb part.
But maybe it's more convenient, to use Gray8 as input for HapAlphaOnly
and not suppo
> >
> > But we can of course have more than one sample
> > for cineform...
> >
I don't know the exact guidelines for fate test, but i think
we can have several sample, who test several things.
For now, there is no fate test for this decoder
we can use the sample of this patch, for a first fate t
W dniu 2017-09-25 o 01:42, Carl Eugen Hoyos pisze:
> 2017-09-23 19:18 GMT+02:00 Mateusz :
>
>> In reality -- current version of DITHER_COPY mess
>> up the average color tone.
>
> You could explain how we can reproduce this.
Please take any video with colors that you can see change to green color
>
> Sorry if I misunderstand:
> If the format only contains one layer, why is the pix_fmt
> rgb0 and not gray8?
>
> Carl Eugen
>
>
> Hello,
I follow the existing texture dsp design, who output RGB0 or RGBA depending
of the texture.
But you're right, the format is gray only,
Martin
___
On Mon, Sep 25, 2017 at 3:31 AM, Aman Gupta wrote:
>
> How do the other hwaccels handle mid-stream SPS changes?
>
Real HWAccels (ie. VAAPI, VDPAU or DXVA) communicate the SPS/PPS
content for every frame, they don't keep a persistent state internally
- that way the only "state" is the frame size a
74 matches
Mail list logo
