Development list for x264, a free library for encoding H264/AVC video streams.

headers

Davide Bergamini | 6 Mar 2011 15:39

About --sar option

Hi, thanks for your work!

I have a couple of question/suggestion about the --sar option.

I think the real meaning of this option is not sar (storage aspect ratio), but its par (pixel aspect ratio)

and it should be renamed "--input-par". (like --input-res)

Example: a pal dvd mpeg2 widescreen

sar = 5:4 (its the pixel resolution ratio 720/576)

dar = 16:9 (its the output ratio)

par = 64:45 (its the single pixel width/height ratio)

_______________________________________________
x264-devel mailing list
x264-devel <at> videolan.org
http://mailman.videolan.org/listinfo/x264-devel

headers

BugMaster | 7 Mar 2011 09:08

Re: About --sar option

On Sun, 6 Mar 2011 15:39:33 +0100, Davide Bergamini wrote:
> Hi, thanks for your work!
> I have a couple of question/suggestion about the --sar option.
> I think the real meaning of this option is not sar (storage aspect
> ratio), but its par (pixel aspect ratio)
> and it should be renamed "--input-par". (like --input-res)
> Example: a pal dvd mpeg2 widescreen
> sar = 5:4 (its the pixel resolution ratio 720/576)
> dar = 16:9 (its the output ratio)
> par = 64:45 (its the single pixel width/height ratio)
>  
>  

Hi. SAR acronym is not for storage aspect ratio, but for sample aspect
ratio. And yes sample aspect ratio has almost the same meaning as
pixel aspect ratio, but we use sample aspect ratio because this is the
name used in official H.264 specs.

_______________________________________________
x264-devel mailing list
x264-devel <at> videolan.org
http://mailman.videolan.org/listinfo/x264-devel

headers

雾卡 | 7 Mar 2011 11:31

Wrong reply

I think you have sent the wrong reply to me，that's not my questions。My original letter is:

I am doing study on the Rate Control of Macroblock level based on X264, I have read some documents of this aspect, but there exists some questions, I need your help to make these clear:

1、Dose X264 supports the Macroblock level Rate Control?

2、If it supports, Is it right I choose the VBV(the same as CBR?)mode? And only VBV mode support the Macroblock layer rate control ,What about other modes?
3、If VBV allows macroblock layer RC, how should I set the command line parameter about the Rate Control, must I set the following parameters, and how to?

-B/--bitrate ,--vbv-maxrate ,--vbv-bufsize ,--vbv-init

Look forward to your early reply. So appreciate!

sincerely yours

Jane

------------------ Original ------------------

From: "BugMaster"<BugMaster <at> narod.ru>;

Date: Mon, Mar 7, 2011 04:08 PM

To: "Mailing list for x264 developers"<x264-devel <at> videolan.org>;

Subject: Re: [x264-devel] About --sar option

On Sun, 6 Mar 2011 15:39:33 +0100, Davide Bergamini wrote:
> Hi, thanks for your work!
> I have a couple of question/suggestion about the --sar option.
> I think the real meaning of this option is not sar (storage aspect
> ratio), but its par (pixel aspect ratio)
> and it should be renamed "--input-par". (like --input-res)
> Example: a pal dvd mpeg2 widescreen
> sar = 5:4 (its the pixel resolution ratio 720/576)
> dar = 16:9 (its the output ratio)
> par = 64:45 (its the single pixel width/height ratio)
>
>

Hi. SAR acronym is not for storage aspect ratio, but for sample aspect
ratio. And yes sample aspect ratio has almost the same meaning as
pixel aspect ratio, but we use sample aspect ratio because this is the
name used in official H.264 specs.

_______________________________________________
x264-devel mailing list
x264-devel <at> videolan.org
http://mailman.videolan.org/listinfo/x264-devel

_______________________________________________
x264-devel mailing list
x264-devel <at> videolan.org
http://mailman.videolan.org/listinfo/x264-devel

headers

Davide Bergamini | 7 Mar 2011 13:22

Re: About --sar option

Thanks you for the explanation! I think the wikipedia article about pixel aspect raio should be updated, I suggested in its discussion page. Bye!

2011/3/7 BugMaster <BugMaster <at> narod.ru>

On Sun, 6 Mar 2011 15:39:33 +0100, Davide Bergamini wrote:
> Hi, thanks for your work!
> I have a couple of question/suggestion about the --sar option.
> I think the real meaning of this option is not sar (storage aspect
> ratio), but its par (pixel aspect ratio)
> and it should be renamed "--input-par". (like --input-res)
> Example: a pal dvd mpeg2 widescreen
> sar = 5:4 (its the pixel resolution ratio 720/576)
> dar = 16:9 (its the output ratio)
> par = 64:45 (its the single pixel width/height ratio)
>
>

Hi. SAR acronym is not for storage aspect ratio, but for sample aspect
ratio. And yes sample aspect ratio has almost the same meaning as
pixel aspect ratio, but we use sample aspect ratio because this is the
name used in official H.264 specs.

_______________________________________________
x264-devel mailing list
x264-devel <at> videolan.org
http://mailman.videolan.org/listinfo/x264-devel

_______________________________________________
x264-devel mailing list
x264-devel <at> videolan.org
http://mailman.videolan.org/listinfo/x264-devel

headers

sandeep mk | 7 Mar 2011 18:53

- Help in calculation of PSNRs -

Hi,

I am a newbie to x264. I wanted to know how PSNR is calculated between an 8 bit .yuv input file encoded using 10 bit-x264 (in High10 profile) and the 10 bit .yuv output file decoded from a JM decoder.

Thanks,
Sandeep

_______________________________________________
x264-devel mailing list
x264-devel <at> videolan.org
http://mailman.videolan.org/listinfo/x264-devel

headers

Nick Lewycky | 8 Mar 2011 01:03

[PATCH] Remove "8+%1" syntax which isn't supported by LLVM's assembler. FFmpeg has removed theirs already for the same reason.

The ".o | objdump -d" files built before and after this change are the same,
tested by building with and without -DHIGH_BIT_DEPTH.
---
 common/x86/predict-c.c |   22 ++++++++++++----------
 1 files changed, 12 insertions(+), 10 deletions(-)

diff --git a/common/x86/predict-c.c b/common/x86/predict-c.c
index 610740e..fa307bd 100644
--- a/common/x86/predict-c.c
+++ b/common/x86/predict-c.c
 <at>  <at>  -191,10 +191,10  <at>  <at>  static void x264_predict_16x16_p_ssse3( uint8_t *src )
     int H, V;
 #if HIGH_BIT_DEPTH
     asm (
-        "movdqu        -2+%1, %%xmm1 \n"
-        "movdqa        16+%1, %%xmm0 \n"
-        "pmaddwd          %2, %%xmm0 \n"
-        "pmaddwd          %3, %%xmm1 \n"
+        "movdqu           %1, %%xmm1 \n"
+        "movdqa           %2, %%xmm0 \n"
+        "pmaddwd          %3, %%xmm0 \n"
+        "pmaddwd          %4, %%xmm1 \n"
         "paddd        %%xmm1, %%xmm0 \n"
         "movhlps      %%xmm0, %%xmm1 \n"
         "paddd        %%xmm1, %%xmm0 \n"
 <at>  <at>  -202,15 +202,16  <at>  <at>  static void x264_predict_16x16_p_ssse3( uint8_t *src )
         "paddd        %%xmm1, %%xmm0 \n"
         "movd         %%xmm0, %0     \n"
         :"=r"(H)
-        :"m"(src[-FDEC_STRIDE]), "m"(*pw_12345678), "m"(*pw_m87654321)
+        :"m"(src[-FDEC_STRIDE-1]), "m"(src[-FDEC_STRIDE+8]),
+         "m"(*pw_12345678), "m"(*pw_m87654321)
     );
 #else
     asm (
         "movq           %1, %%mm1 \n"
-        "movq         8+%1, %%mm0 \n"
-        "palignr $7, -8+%1, %%mm1 \n"
-        "pmaddubsw      %2, %%mm0 \n"
-        "pmaddubsw      %3, %%mm1 \n"
+        "movq           %2, %%mm0 \n"
+        "palignr $7,    %3, %%mm1 \n"
+        "pmaddubsw      %4, %%mm0 \n"
+        "pmaddubsw      %5, %%mm1 \n"
         "paddw       %%mm1, %%mm0 \n"
         "pshufw $14, %%mm0, %%mm1 \n"
         "paddw       %%mm1, %%mm0 \n"
 <at>  <at>  -219,7 +220,8  <at>  <at>  static void x264_predict_16x16_p_ssse3( uint8_t *src )
         "movd        %%mm0, %0    \n"
         "movsx         %w0, %0    \n"
         :"=r"(H)
-        :"m"(src[-FDEC_STRIDE]), "m"(*pb_12345678), "m"(*pb_m87654321)
+        :"m"(src[-FDEC_STRIDE]), "m"(src[-FDEC_STRIDE+8]),
+         "m"(src[-FDEC_STRIDE-8]), "m"(*pb_12345678), "m"(*pb_m87654321)
     );
 #endif
     V = 8 * ( src[15*FDEC_STRIDE-1] - src[-1*FDEC_STRIDE-1] )

headers

Jason Garrett-Glaser | 8 Mar 2011 03:11

Re: [PATCH] Remove "8+%1" syntax which isn't supported by LLVM's assembler. FFmpeg has removed theirs already for the same reason.

On Mon, Mar 7, 2011 at 4:03 PM, Nick Lewycky <nlewycky <at> google.com> wrote:
> The ".o | objdump -d" files built before and after this change are the same,
> tested by building with and without -DHIGH_BIT_DEPTH.
> ---
>  common/x86/predict-c.c |   22 ++++++++++++----------
>  1 files changed, 12 insertions(+), 10 deletions(-)
>
> diff --git a/common/x86/predict-c.c b/common/x86/predict-c.c
> index 610740e..fa307bd 100644
> --- a/common/x86/predict-c.c
> +++ b/common/x86/predict-c.c
>  <at>  <at>  -191,10 +191,10  <at>  <at>  static void x264_predict_16x16_p_ssse3( uint8_t *src )
>     int H, V;
>  #if HIGH_BIT_DEPTH
>     asm (
> -        "movdqu        -2+%1, %%xmm1 \n"
> -        "movdqa        16+%1, %%xmm0 \n"
> -        "pmaddwd          %2, %%xmm0 \n"
> -        "pmaddwd          %3, %%xmm1 \n"
> +        "movdqu           %1, %%xmm1 \n"
> +        "movdqa           %2, %%xmm0 \n"
> +        "pmaddwd          %3, %%xmm0 \n"
> +        "pmaddwd          %4, %%xmm1 \n"
>         "paddd        %%xmm1, %%xmm0 \n"
>         "movhlps      %%xmm0, %%xmm1 \n"
>         "paddd        %%xmm1, %%xmm0 \n"
>  <at>  <at>  -202,15 +202,16  <at>  <at>  static void x264_predict_16x16_p_ssse3( uint8_t *src )
>         "paddd        %%xmm1, %%xmm0 \n"
>         "movd         %%xmm0, %0     \n"
>         :"=r"(H)
> -        :"m"(src[-FDEC_STRIDE]), "m"(*pw_12345678), "m"(*pw_m87654321)
> +        :"m"(src[-FDEC_STRIDE-1]), "m"(src[-FDEC_STRIDE+8]),
> +         "m"(*pw_12345678), "m"(*pw_m87654321)
>     );
>  #else
>     asm (
>         "movq           %1, %%mm1 \n"
> -        "movq         8+%1, %%mm0 \n"
> -        "palignr $7, -8+%1, %%mm1 \n"
> -        "pmaddubsw      %2, %%mm0 \n"
> -        "pmaddubsw      %3, %%mm1 \n"
> +        "movq           %2, %%mm0 \n"
> +        "palignr $7,    %3, %%mm1 \n"
> +        "pmaddubsw      %4, %%mm0 \n"
> +        "pmaddubsw      %5, %%mm1 \n"
>         "paddw       %%mm1, %%mm0 \n"
>         "pshufw $14, %%mm0, %%mm1 \n"
>         "paddw       %%mm1, %%mm0 \n"
>  <at>  <at>  -219,7 +220,8  <at>  <at>  static void x264_predict_16x16_p_ssse3( uint8_t *src )
>         "movd        %%mm0, %0    \n"
>         "movsx         %w0, %0    \n"
>         :"=r"(H)
> -        :"m"(src[-FDEC_STRIDE]), "m"(*pb_12345678), "m"(*pb_m87654321)
> +        :"m"(src[-FDEC_STRIDE]), "m"(src[-FDEC_STRIDE+8]),
> +         "m"(src[-FDEC_STRIDE-8]), "m"(*pb_12345678), "m"(*pb_m87654321)
>     );
>  #endif
>     V = 8 * ( src[15*FDEC_STRIDE-1] - src[-1*FDEC_STRIDE-1] )
> --
> 1.7.3.1
>

Applied locally, will be in the next push.

Jason
_______________________________________________
x264-devel mailing list
x264-devel <at> videolan.org
http://mailman.videolan.org/listinfo/x264-devel

headers

sandeep mk | 8 Mar 2011 19:19

Re: - Help in calculation of PSNRs -

Hi,

I am a newbie to x264. I had two questions regarding the PSNRs in H.264.

1. I have an 8 bit .yuv input file input to the x264 encoder which gives me an 8 bit encoded stream. This encoded stream is decoded by a 8 bit decoder and lets call the output as 'Stream 1'. Next, I have a 10 bit .yuv input file input to the x264 encoder which gives me an 10 bit encoded stream and is further decoded by a 10 bit decoder. Lets call this stream as 'Stream 2'. My question is how do we compare the PSNRs of 'Stream 1' and 'Stream 2' or what is the best was of comparing the quality of the streams 1 and 2.

2. My second question is how is PSNR calculated between an 8 bit .yuv input file encoded using 10 bit-x264 (in High10 profile) and the 10 bit .yuv output file decoded from 10 bit h264 decoder.

Thanks,
Sandeep

_______________________________________________
x264-devel mailing list
x264-devel <at> videolan.org
http://mailman.videolan.org/listinfo/x264-devel

headers

Max Lapshin | 12 Mar 2011 19:26

Encode motion in selected rectangle

I'm creating now encoder using libx264: it takes video from
surveillance camera, unpacks it,
process and encode only small rectangle.

I have hardware tool that can tell me where movement is, so in each
moment I have map of blocks, where movement is and
let libx264 work only in this small region. It is required to lower
bitrate, because there is noise on matrix and I want
to use this trick to block any useless calculations on FullHD image
and encode only real movement.

I've saw topic:
http://mailman.videolan.org/pipermail/x264-devel/2010-November/007874.html
(about infrared cameras),
but I haven't understood: is it what I need?

I thought about some API, that will require from me list of changed
macroblocks.  Google didn't help me, because I haven't found
anything about my problem there.
_______________________________________________
x264-devel mailing list
x264-devel <at> videolan.org
http://mailman.videolan.org/listinfo/x264-devel

headers

Shalom Levytam | 15 Mar 2011 05:27

x264 Install: Cannot Follow Symbolic Link

Hi,

I'm getting numerous "cannot follow symbolic link" errors when trying to install x264 on Solaris.

Here's what I'm seeing: http://pastebin.com/bGnGuuMh

Can anyone be of help?  

Thanks,

slevytam
_______________________________________________
x264-devel mailing list
x264-devel <at> videolan.org
http://mailman.videolan.org/listinfo/x264-devel

41	May 2013
98	April 2013
45	March 2013
72	February 2013
62	January 2013
44	December 2012
64	November 2012
67	October 2012
41	September 2012
74	August 2012
77	July 2012
17	June 2012
45	May 2012
40	April 2012
96	March 2012
117	February 2012
120	January 2012
42	December 2011
19	November 2011
53	October 2011
55	September 2011
96	August 2011
83	July 2011
80	June 2011
72	May 2011
76	April 2011
62	March 2011
103	February 2011
112	January 2011
82	December 2010
135	November 2010
60	October 2010
107	September 2010
80	August 2010
140	July 2010
163	June 2010
128	May 2010
104	April 2010
96	March 2010
172	February 2010
159	January 2010
62	December 2009
93	November 2009
102	October 2009
154	September 2009
137	August 2009
122	July 2009
48	June 2009
79	May 2009
80	April 2009
98	March 2009
110	February 2009
225	January 2009
111	December 2008
110	November 2008
109	October 2008
132	September 2008
80	August 2008
111	July 2008
109	June 2008
86	May 2008
154	April 2008
242	March 2008
45	February 2008
123	January 2008
59	December 2007
148	November 2007
96	October 2007
37	September 2007
71	August 2007
100	July 2007
103	June 2007
105	May 2007
113	April 2007
145	March 2007
77	February 2007
169	January 2007
101	December 2006
109	November 2006
84	October 2006
94	September 2006
125	August 2006
79	July 2006
66	June 2006
107	May 2006
102	April 2006
136	March 2006
108	February 2006
60	January 2006
65	December 2005
76	November 2005
141	October 2005
76	September 2005
77	August 2005
115	July 2005
73	June 2005
140	May 2005
131	April 2005
83	March 2005
19	February 2005

Mon	Tue	Wed	Thu	Fri	Sat	Sun

	1	2	3	4	5	6
7	8	9	10	11	12	13
14	15	16	17	18	19	20
21	22	23	24	25	26	27
28	29	30	31