[FFmpeg-devel] [PATCH] avfilter: port pullup filter from libmpcodecs

[Paul B Mahol]

On 9/17/13, Stefano Sabatini <stefasab at gmail.com> wrote:
> On date Monday 2013-09-16 14:14:33 +0000, Paul B Mahol encoded:
>> Signed-off-by: Paul B Mahol <onemda at gmail.com>
>> ---
>>  LICENSE                          |   1 +
>>  configure                        |   1 +
>>  doc/filters.texi                 |  53 +++
>>  libavfilter/Makefile             |   1 +
>>  libavfilter/allfilters.c         |   1 +
>>  libavfilter/vf_pullup.c          | 746
>> +++++++++++++++++++++++++++++++++++++++
>>  libavfilter/vf_pullup.h          |  71 ++++
>>  libavfilter/x86/Makefile         |   1 +
>>  libavfilter/x86/vf_pullup_init.c | 221 ++++++++++++
>>  9 files changed, 1096 insertions(+)
>>  create mode 100644 libavfilter/vf_pullup.c
>>  create mode 100644 libavfilter/vf_pullup.h
>>  create mode 100644 libavfilter/x86/vf_pullup_init.c
> [...]
>> +#define OFFSET(x) offsetof(PullupContext, x)
>> +#define FLAGS AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
>> +
>> +static const AVOption pullup_options[] = {
>
>> +    { "jl", "junk left",  OFFSET(junk_left),  AV_OPT_TYPE_INT, {.i64=1},
>> 0, INT_MAX, FLAGS },
>> +    { "jr", "junk right", OFFSET(junk_right), AV_OPT_TYPE_INT, {.i64=1},
>> 0, INT_MAX, FLAGS },
>> +    { "jt", "junk top",   OFFSET(junk_top),   AV_OPT_TYPE_INT, {.i64=4},
>> 0, INT_MAX, FLAGS },
>> +    { "jd", "junk down",  OFFSET(junk_down),  AV_OPT_TYPE_INT, {.i64=4},
>> 0, INT_MAX, FLAGS },
>
> set junk * size ?
>
>> +    { "sb", "strict breaks", OFFSET(strict_breaks), AV_OPT_TYPE_INT,
>> {.i64=0},-1, 1, FLAGS },
>
> set strict breaks
>
>> +    { "mp", "metric plane",  OFFSET(metric_plane),  AV_OPT_TYPE_INT,
>> {.i64=0}, 0, 2, FLAGS, "mp" },
>
> set metric plane
>
>> +    { "y", "luma",        0, AV_OPT_TYPE_CONST, {.i64=0}, 0, 0, FLAGS,
>> "mp" },
>> +    { "u", "chroma blue", 0, AV_OPT_TYPE_CONST, {.i64=1}, 0, 0, FLAGS,
>> "mp" },
>> +    { "v", "chroma red",  0, AV_OPT_TYPE_CONST, {.i64=2}, 0, 0, FLAGS,
>> "mp" },
>> +    { NULL }
>> +};
>> +
>> +AVFILTER_DEFINE_CLASS(pullup);
>> +
>> +static int query_formats(AVFilterContext *ctx)
>> +{
>> +    static const enum AVPixelFormat pix_fmts[] = {
>> +        AV_PIX_FMT_YUVJ444P, AV_PIX_FMT_YUVJ440P,
>> +        AV_PIX_FMT_YUVJ422P, AV_PIX_FMT_YUVJ420P,
>> +        AV_PIX_FMT_YUV444P,  AV_PIX_FMT_YUV440P,
>> +        AV_PIX_FMT_YUV422P,  AV_PIX_FMT_YUV420P,
>> +        AV_PIX_FMT_YUV411P,  AV_PIX_FMT_YUV410P,
>> +        AV_PIX_FMT_YUVJ411P, AV_PIX_FMT_GRAY8,
>> +        AV_PIX_FMT_NONE
>> +    };
>> +    ff_set_common_formats(ctx, ff_make_format_list(pix_fmts));
>> +    return 0;
>> +}
>> +
>
>> +#define ABS(a) (((a) ^ ((a) >> 31)) - ((a) >> 31))
>
> Note: is this different/faster than FFABS?

It produces different output.

>
> [...]
>
> No objections from me if it is bit-identical with the ported filter
> (and I'm not going to give an indepth review).
> --
> FFmpeg = Fascinating and Fancy Mastering Political Entertaining Gigant
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>