support for ffmpeg's drc_scale option [patch]

[peter@…]

I added support for drc_scale ([e]ac3 decoder dynamic range compression factor) in lavdopts. All the other lavdopts apply to video, so I had to make the variable extern instead of static like all the others.

​http://forum.doom9.org/showthread.php?p=1348332#post1348332
(that's ​http://85.230.118.136/showthread.php?p=1348332#post1348332 until forum.doom9.org's DNS record updates...)

Index: libmpcodecs/ad_ffmpeg.c
===================================================================
--- libmpcodecs/ad_ffmpeg.c (revision 29972)
+++ libmpcodecs/ad_ffmpeg.c (working copy)
@@ -71,6 +71,11 @@

lavc_context->codec_type = CODEC_TYPE_AUDIO;
lavc_context->codec_id = lavc_codec->id; not sure if required, imho not --A'rpi

+ {
+ extern float lavc_param_drc_scale; parsed from lavdopts in vd_ffmpeg.c
+ lavc_context->drc_scale = lavc_param_drc_scale;
+ }
+

/* alloc extra data */
if (sh_audio->wf && sh_audio->wf->cbSize > 0) {

lavc_context->extradata = av_mallocz(sh_audio->wf->cbSize + FF_INPUT_BUFFER_PADDING_SIZE);

Index: libmpcodecs/vd_ffmpeg.c
===================================================================
--- libmpcodecs/vd_ffmpeg.c (revision 29972)
+++ libmpcodecs/vd_ffmpeg.c (working copy)
@@ -79,6 +79,7 @@

static char *lavc_param_skip_frame_str = NULL;
static int lavc_param_threads=1;
static int lavc_param_bitexact=0;

+ float lavc_param_drc_scale=0.0; audio param, for ad_ffmpeg.c

static char *lavc_avopt = NULL;

const m_option_t lavc_decode_opts_conf[]={

@@ -99,6 +100,7 @@

{"skipframe", &lavc_param_skip_frame_str, CONF_TYPE_STRING, 0, 0, 0, NULL},
{"threads", &lavc_param_threads, CONF_TYPE_INT, CONF_RANGE, 1, 8, NULL},
{"bitexact", &lavc_param_bitexact, CONF_TYPE_FLAG, 0, 0, CODEC_FLAG_BITEXACT, NULL},

+ {"drc_scale", &lavc_param_drc_scale, CONF_TYPE_FLOAT, CONF_RANGE, 0, 2.0, NULL},

{"o", &lavc_avopt, CONF_TYPE_STRING, 0, 0, 0, NULL},
{NULL, NULL, 0, 0, 0, 0, NULL}

};

I made the allowed range for drc_scale 0 - 2.0, so curious people could hear what it sounds like with more than the full amount of dynamic range compression scaling applied. The code in libavcodec/ac3dec.c that uses drc_scale doesn't have a problem with that.

[peter@…]

the tiny patch

[compn]

i'm pretty sure patch rejected because of putting audio option in video code.

[reimar]

The -a52drc option was extended to apply to both decoders, so you do not have to care about which one is actually used.
It is not well tested, so report if you have issues with it.

[reimar]

(Fixed)