[FFmpeg-devel] [PATCH] lavfi: add xbr filter
arwa arif
arwaarif1994 at gmail.com
Sun Oct 26 10:21:48 CET 2014
I think the segmentation fault is maybe because of wrong input format. The
code runs for rgb24 only. I have attached the updated patch along with the
tested input file. I was not able to attach a video file. So, for now I am
attaching an image instead.
On Sun, Oct 26, 2014 at 4:13 AM, Michael Niedermayer <michaelni at gmx.at>
wrote:
> On Sat, Oct 25, 2014 at 11:32:43PM +0530, arwa arif wrote:
> > Can you please specify what is meant by stand-alone changes? Do I need to
> > add the non-default functions in different commit? I am not very sure if
> I
> > understood it right. Apart from that, I have updated the patch.
> >
> > On Sat, Oct 25, 2014 at 10:16 PM, Nicolas George <george at nsup.org>
> wrote:
> >
> > > Le quartidi 4 brumaire, an CCXXIII, arwa arif a écrit :
> > > > > please post a new patch instead of a patch on top of a previous
> > > > > patch
> > > > libavfilter/vf_xbr.c | 303
> > > ++++++++++++++++++++++++++++++++++++++++++++++++++
> > > > 1 file changed, 303 insertions(+)
> > > > create mode 100644 libavfilter/vf_xbr.c
> > >
> > > This patch does not contain the changes to Makefile and allfilters.c,
> so I
> > > believe you still have a bit of tweaking to do with Git.
> > >
> > > If you used a branch (that is widely advisable), you should be able to
> type
> > > this:
> > >
> > > git log --stat master..
> > >
> > > Then you should see a single commit, yours, with changes to Makefile,
> > > allfilters.c and all other common files you needed to changes, and of
> > > course
> > > the new file(s).
> > >
> > > (If you have stand-alone changes, such as moving code into a shared
> > > function
> > > to use in your actual patch, then it should go in a separate commit.
> But it
> > > does not seem to apply here.)
> > >
> > > Also, I suspect you forgot to add the documentation for the filter in
> > > doc/filters.texi. A few words are enough, but at the very least let
> people
> > > know what it does, because xbr looks like just three random letters --
> not
> > > your fault of course.
> > >
> > > Regards,
> > >
> > > --
> > > Nicolas George
> > >
> > > _______________________________________________
> > > ffmpeg-devel mailing list
> > > ffmpeg-devel at ffmpeg.org
> > > http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
> > >
> > >
>
> > doc/filters.texi | 7 +
> > libavfilter/Makefile | 1
> > libavfilter/allfilters.c | 1
> > libavfilter/vf_xbr.c | 303
> +++++++++++++++++++++++++++++++++++++++++++++++
> > 4 files changed, 312 insertions(+)
> > a92794344ddf0d5d4ea3205091ee4a19699c5c06
> 0001-PATCH-lavfi-add-xbr-filter.patch
> > From cf3a7a6bcb9735c6f6157f80da208c1c191b3e02 Mon Sep 17 00:00:00 2001
> > From: Arwa Arif <arwaarif1994 at gmail.com>
> > Date: Sat, 25 Oct 2014 22:04:51 +0530
> > Subject: [PATCH] [PATCH]lavfi: add xbr filter
> >
> > Makefile
> >
> > allfilters.c
> >
> > filters.texi
> > ---
> > doc/filters.texi | 7 ++
> > libavfilter/Makefile | 1 +
> > libavfilter/allfilters.c | 1 +
> > libavfilter/vf_xbr.c | 303
> ++++++++++++++++++++++++++++++++++++++++++++++
> > 4 files changed, 312 insertions(+)
> > create mode 100644 libavfilter/vf_xbr.c
>
> I tried the filter but it segfaults
> ./ffplay matrixbench_mpeg2.mpg -vf xbr
>
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 0x7fffcfff7700 (LWP 14880)]
> 0x00000000004b40c5 in d (in=0x7fffc85eb000, x1=-1, y1=576, x2=0, y2=577)
> at libavfilter/vf_xbr.c:49
> 49 int r2 = *(in->data[0] + y2 * in->linesize[0] + x2*3);
> (gdb) bt
> #0 0x00000000004b40c5 in d (in=0x7fffc85eb000, x1=-1, y1=576, x2=0,
> y2=577) at libavfilter/vf_xbr.c:49
> #1 0x00000000004b49db in apply_edge_detection_rules (in=0x7fffc85eb000,
> out=0x7fffc85ead20, x=0, y=575) at libavfilter/vf_xbr.c:159
> #2 0x00000000004b5173 in filter_frame (inlink=0x7fffc85a7800,
> in=0x7fffc85eb000) at libavfilter/vf_xbr.c:270
> #3 0x0000000000444270 in ff_filter_frame_framed (link=0x7fffc85a7800,
> frame=0x7fffc85eb000) at libavfilter/avfilter.c:1103
> #4 0x00000000004447c4 in ff_filter_frame (link=0x7fffc85a7800,
> frame=0x7fffc85eb000) at libavfilter/avfilter.c:1183
> #5 0x00000000004a4aff in filter_frame (link=0x7fffc85a70a0, in=0x0) at
> libavfilter/vf_scale.c:530
> #6 0x0000000000444270 in ff_filter_frame_framed (link=0x7fffc85a70a0,
> frame=0x7fffc85ead20) at libavfilter/avfilter.c:1103
> #7 0x00000000004447c4 in ff_filter_frame (link=0x7fffc85a70a0,
> frame=0x7fffc85ead20) at libavfilter/avfilter.c:1183
> #8 0x000000000044b1e3 in request_frame (link=0x7fffc85a70a0) at
> libavfilter/buffersrc.c:499
> #9 0x000000000044212b in ff_request_frame (link=0x7fffc85a70a0) at
> libavfilter/avfilter.c:351
> #10 0x0000000000442159 in ff_request_frame (link=0x7fffc85a7800) at
> libavfilter/avfilter.c:353
> #11 0x0000000000442159 in ff_request_frame (link=0x7fffc85a71e0) at
> libavfilter/avfilter.c:353
> #12 0x0000000000442159 in ff_request_frame (link=0x7fffc85a7680) at
> libavfilter/avfilter.c:353
> #13 0x0000000000442159 in ff_request_frame (link=0x7fffc85a6c00) at
> libavfilter/avfilter.c:353
> #14 0x0000000000448f1e in av_buffersink_get_frame_flags
> (ctx=0x7fffc85a6220, frame=0x7fffc80008c0, flags=0) at
> libavfilter/buffersink.c:137
> #15 0x000000000041f526 in video_thread (arg=0x7fffeb1a7040) at
> ffplay.c:2168
> #16 0x00007ffff64a6fd5 in ?? () from
> /usr/lib/x86_64-linux-gnu/libSDL-1.2.so.0
> #17 0x00007ffff64ea999 in ?? () from
> /usr/lib/x86_64-linux-gnu/libSDL-1.2.so.0
> #18 0x00007ffff006ae9a in start_thread (arg=0x7fffcfff7700) at
> pthread_create.c:308
> #19 0x00007fffefd9831d in clone () at
> ../sysdeps/unix/sysv/linux/x86_64/clone.S:112
> #20 0x0000000000000000 in ?? ()
>
>
> [...]
>
> > +const int THRESHHOLD_Y = 48;
> > +const int THRESHHOLD_U = 7;
> > +const int THRESHHOLD_V = 6;
>
> these should be #define
>
>
> > +
> > +/**
> > +* Calculates the weight of difference of the pixels, by transforming
> these
> > +* pixels into their Y'UV parts. It then uses the threshold used by HQx
> filters:
> > +* 48*Y + 7*U + 6*V, to give it those smooth looking edges.
> > +**/
> > +static int d(AVFrame *in,int x1,int y1,int x2,int y2){
> > +
> > + int r1 = *(in->data[0] + y1 * in->linesize[0] + x1*3);
> > + int g1 = *(in->data[0] + y1 * in->linesize[0] + x1*3 + 1);
> > + int b1 = *(in->data[0] + y1 * in->linesize[0] + x1*3 + 2);
> > +
> > + int r2 = *(in->data[0] + y2 * in->linesize[0] + x2*3);
> > + int g2 = *(in->data[0] + y2 * in->linesize[0] + x2*3 + 1);
> > + int b2 = *(in->data[0] + y2 * in->linesize[0] + x2*3 + 2);
> > +
> > + int r = abs(r1 - r2);
> > + int g = abs(g1 - g2);
> > + int b = abs(b1 - b2);
> > +
> > + /*Convert RGB to Y'UV*/
> > + int y = ( ( 66 * r + 129 * g + 25 * b + 128) >> 8) + 16;
> > + int u = ( ( -38 * r - 74 * g + 112 * b + 128) >> 8) + 128;
> > + int v = ( ( 112 * r - 94 * g - 18 * b + 128) >> 8) + 128;
> > +
> > + /*Add HQx filters threshold & return*/
> > + return (y * THRESHHOLD_Y) + (u* THRESHHOLD_U) + (v* THRESHHOLD_V);
>
> the position of the abs() looks strange
> if this is supposed to be a vector distance in yuv space then the
> abs are needed on the yuv terms
>
>
> [...]
>
> > + av_frame_copy_props(out, in);
> > + for(i=0;i<inlink->w;i++)
> > + for(j=0;j<inlink->h;j++)
> > + apply_edge_detection_rules(in,out,i,j);
>
> tabs and trailing whitespace isnt allowed in ffmpeg git
>
> [...]
>
> --
> Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
>
> Everything should be made as simple as possible, but not simpler.
> -- Albert Einstein
>
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-PATCH-lavfi-add-xbr-filter.patch
Type: text/x-patch
Size: 12215 bytes
Desc: not available
URL: <https://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20141026/2c6d942a/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: images.jpeg
Type: image/jpeg
Size: 5333 bytes
Desc: not available
URL: <https://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20141026/2c6d942a/attachment.jpeg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: output.jpeg
Type: image/jpeg
Size: 15919 bytes
Desc: not available
URL: <https://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20141026/2c6d942a/attachment-0001.jpeg>
More information about the ffmpeg-devel
mailing list