lavfi/aconvert: use libswresample.
authorClément Bœsch <ubitux@gmail.com>
Thu, 26 Jan 2012 07:49:15 +0000 (08:49 +0100)
committerClément Bœsch <ubitux@gmail.com>
Thu, 2 Feb 2012 13:31:18 +0000 (14:31 +0100)
This commit also drops the planar parameter; you now need to use the 'p'
suffix in order to request a planar sample format.

configure
doc/filters.texi
libavfilter/Makefile
libavfilter/af_aconvert.c
libavfilter/af_aconvert_rematrix.c [deleted file]
libavfilter/version.h

index b4e433c..aa45fd7 100755 (executable)
--- a/configure
+++ b/configure
@@ -1648,6 +1648,7 @@ tls_protocol_select="tcp_protocol"
 udp_protocol_deps="network"
 
 # filters
+aconvert_filter_deps="swresample"
 amovie_filter_deps="avcodec avformat"
 aresample_filter_deps="swresample"
 ass_filter_deps="libass"
index 1f3522e..c64f899 100644 (file)
@@ -104,17 +104,15 @@ Below is a description of the currently available audio filters.
 Convert the input audio format to the specified formats.
 
 The filter accepts a string of the form:
-"@var{sample_format}:@var{channel_layout}:@var{packing_format}".
+"@var{sample_format}:@var{channel_layout}".
 
-@var{sample_format} specifies the sample format, and can be a string or
-the corresponding numeric value defined in @file{libavutil/samplefmt.h}.
+@var{sample_format} specifies the sample format, and can be a string or the
+corresponding numeric value defined in @file{libavutil/samplefmt.h}. Use 'p'
+suffix for a planar sample format.
 
 @var{channel_layout} specifies the channel layout, and can be a string
 or the corresponding number value defined in @file{libavutil/audioconvert.h}.
 
-@var{packing_format} specifies the type of packing in output, can be one
-of "planar" or "packed", or the corresponding numeric values "0" or "1".
-
 The special parameter "auto", signifies that the filter will
 automatically select the output format depending on the output filter.
 
@@ -122,16 +120,15 @@ Some examples follow.
 
 @itemize
 @item
-Convert input to unsigned 8-bit, stereo, packed:
+Convert input to float, planar, stereo:
 @example
-aconvert=u8:stereo:packed
+aconvert=fltp:stereo
 @end example
 
 @item
-Convert input to unsigned 8-bit, automatically select out channel layout
-and packing format:
+Convert input to unsigned 8-bit, automatically select out channel layout:
 @example
-aconvert=u8:auto:auto
+aconvert=u8:auto
 @end example
 @end itemize
 
index ff0ba75..9fbb59b 100644 (file)
@@ -3,7 +3,7 @@ include $(SUBDIR)../config.mak
 NAME = avfilter
 FFLIBS = avutil
 
-FFLIBS-$(CONFIG_ACONVERT_FILTER)             += avcodec
+FFLIBS-$(CONFIG_ACONVERT_FILTER)             += swresample
 FFLIBS-$(CONFIG_AMOVIE_FILTER)               += avformat avcodec
 FFLIBS-$(CONFIG_ARESAMPLE_FILTER)            += swresample
 FFLIBS-$(CONFIG_MOVIE_FILTER)                += avformat avcodec
index e3c7f8c..8c1b5dc 100644 (file)
 /**
  * @file
  * sample format and channel layout conversion audio filter
- * based on code in libavcodec/resample.c by Fabrice Bellard and
- * libavcodec/audioconvert.c by Michael Niedermayer
  */
 
-#include "libavutil/audioconvert.h"
 #include "libavutil/avstring.h"
-#include "libavcodec/audioconvert.h"
+#include "libswresample/swresample.h"
 #include "avfilter.h"
 #include "internal.h"
 
 typedef struct {
-    enum AVSampleFormat  out_sample_fmt,  in_sample_fmt;   ///< in/out sample formats
-    int64_t              out_chlayout,    in_chlayout;     ///< in/out channel layout
-    int                  out_nb_channels, in_nb_channels;  ///< number of in/output channels
-    enum AVFilterPacking out_packing_fmt, in_packing_fmt;  ///< output packing format
-
-    int max_nb_samples;                     ///< maximum number of buffered samples
-    AVFilterBufferRef *mix_samplesref;      ///< rematrixed buffer
-    AVFilterBufferRef *out_samplesref;      ///< output buffer after required conversions
-
-    uint8_t *in_mix[8], *out_mix[8];        ///< input/output for rematrixing functions
-    uint8_t *packed_data[8];                ///< pointers for packing conversion
-    int out_strides[8], in_strides[8];      ///< input/output strides for av_audio_convert
-    uint8_t **in_conv, **out_conv;          ///< input/output for av_audio_convert
-
-    AVAudioConvert *audioconvert_ctx;       ///< context for conversion to output sample format
-
-    void (*convert_chlayout)();             ///< function to do the requested rematrixing
+    enum AVSampleFormat  out_sample_fmt;
+    int64_t              out_chlayout;
+    struct SwrContext *swr;
 } AConvertContext;
 
-#define REMATRIX_FUNC_SIG(NAME) static void REMATRIX_FUNC_NAME(NAME) \
-    (FMT_TYPE *outp[], FMT_TYPE *inp[], int nb_samples, AConvertContext *aconvert)
-
-#define FMT_TYPE uint8_t
-#define REMATRIX_FUNC_NAME(NAME) NAME ## _u8
-#include "af_aconvert_rematrix.c"
-
-#define FMT_TYPE int16_t
-#define REMATRIX_FUNC_NAME(NAME) NAME ## _s16
-#include "af_aconvert_rematrix.c"
-
-#define FMT_TYPE int32_t
-#define REMATRIX_FUNC_NAME(NAME) NAME ## _s32
-#include "af_aconvert_rematrix.c"
-
-#define FLOATING
-
-#define FMT_TYPE float
-#define REMATRIX_FUNC_NAME(NAME) NAME ## _flt
-#include "af_aconvert_rematrix.c"
-
-#define FMT_TYPE double
-#define REMATRIX_FUNC_NAME(NAME) NAME ## _dbl
-#include "af_aconvert_rematrix.c"
-
-#define FMT_TYPE uint8_t
-#define REMATRIX_FUNC_NAME(NAME) NAME
-REMATRIX_FUNC_SIG(stereo_remix_planar)
-{
-    int size = av_get_bytes_per_sample(aconvert->in_sample_fmt) * nb_samples;
-
-    memcpy(outp[0], inp[0], size);
-    memcpy(outp[1], inp[aconvert->in_nb_channels == 1 ? 0 : 1], size);
-}
-
-#define REGISTER_FUNC_PACKING(INCHLAYOUT, OUTCHLAYOUT, FUNC, PACKING)   \
-    {INCHLAYOUT, OUTCHLAYOUT, PACKING, AV_SAMPLE_FMT_U8,  FUNC##_u8},   \
-    {INCHLAYOUT, OUTCHLAYOUT, PACKING, AV_SAMPLE_FMT_S16, FUNC##_s16},  \
-    {INCHLAYOUT, OUTCHLAYOUT, PACKING, AV_SAMPLE_FMT_S32, FUNC##_s32},  \
-    {INCHLAYOUT, OUTCHLAYOUT, PACKING, AV_SAMPLE_FMT_FLT, FUNC##_flt},  \
-    {INCHLAYOUT, OUTCHLAYOUT, PACKING, AV_SAMPLE_FMT_DBL, FUNC##_dbl},
-
-#define REGISTER_FUNC(INCHLAYOUT, OUTCHLAYOUT, FUNC)                                \
-    REGISTER_FUNC_PACKING(INCHLAYOUT, OUTCHLAYOUT, FUNC##_packed, AVFILTER_PACKED)  \
-    REGISTER_FUNC_PACKING(INCHLAYOUT, OUTCHLAYOUT, FUNC##_planar, AVFILTER_PLANAR)
-
-static const struct RematrixFunctionInfo {
-    int64_t in_chlayout, out_chlayout;
-    int planar, sfmt;
-    void (*func)();
-} rematrix_funcs[] = {
-    REGISTER_FUNC        (AV_CH_LAYOUT_STEREO,  AV_CH_LAYOUT_5POINT1, stereo_to_surround_5p1)
-    REGISTER_FUNC        (AV_CH_LAYOUT_5POINT1, AV_CH_LAYOUT_STEREO,  surround_5p1_to_stereo)
-    REGISTER_FUNC_PACKING(AV_CH_LAYOUT_STEREO,  AV_CH_LAYOUT_MONO,    stereo_to_mono_packed, AVFILTER_PACKED)
-    REGISTER_FUNC_PACKING(AV_CH_LAYOUT_MONO,    AV_CH_LAYOUT_STEREO,  mono_to_stereo_packed, AVFILTER_PACKED)
-    REGISTER_FUNC        (0,                    AV_CH_LAYOUT_MONO,    mono_downmix)
-    REGISTER_FUNC_PACKING(0,                    AV_CH_LAYOUT_STEREO,  stereo_downmix_packed, AVFILTER_PACKED)
-
-    // This function works for all sample formats
-    {0, AV_CH_LAYOUT_STEREO, AVFILTER_PLANAR, -1, stereo_remix_planar}
-};
-
 static av_cold int init(AVFilterContext *ctx, const char *args0, void *opaque)
 {
     AConvertContext *aconvert = ctx->priv;
@@ -124,7 +45,6 @@ static av_cold int init(AVFilterContext *ctx, const char *args0, void *opaque)
 
     aconvert->out_sample_fmt  = AV_SAMPLE_FMT_NONE;
     aconvert->out_chlayout    = 0;
-    aconvert->out_packing_fmt = -1;
 
     if ((arg = av_strtok(args, ":", &ptr)) && strcmp(arg, "auto")) {
         if ((ret = ff_parse_sample_format(&aconvert->out_sample_fmt, arg, ctx)) < 0)
@@ -134,10 +54,6 @@ static av_cold int init(AVFilterContext *ctx, const char *args0, void *opaque)
         if ((ret = ff_parse_channel_layout(&aconvert->out_chlayout, arg, ctx)) < 0)
             goto end;
     }
-    if ((arg = av_strtok(NULL, ":", &ptr)) && strcmp(arg, "auto")) {
-        if ((ret = ff_parse_packing_format((int *)&aconvert->out_packing_fmt, arg, ctx)) < 0)
-            goto end;
-    }
 
 end:
     av_freep(&args);
@@ -147,10 +63,7 @@ end:
 static av_cold void uninit(AVFilterContext *ctx)
 {
     AConvertContext *aconvert = ctx->priv;
-    avfilter_unref_buffer(aconvert->mix_samplesref);
-    avfilter_unref_buffer(aconvert->out_samplesref);
-    if (aconvert->audioconvert_ctx)
-        av_audio_convert_free(aconvert->audioconvert_ctx);
+    swr_free(&aconvert->swr);
 }
 
 static int query_formats(AVFilterContext *ctx)
@@ -159,6 +72,7 @@ static int query_formats(AVFilterContext *ctx)
     AConvertContext *aconvert = ctx->priv;
     AVFilterLink *inlink  = ctx->inputs[0];
     AVFilterLink *outlink = ctx->outputs[0];
+    int out_packing = av_sample_fmt_is_planar(aconvert->out_sample_fmt);
 
     avfilter_formats_ref(avfilter_make_all_formats(AVMEDIA_TYPE_AUDIO),
                          &inlink->out_formats);
@@ -182,219 +96,64 @@ static int query_formats(AVFilterContext *ctx)
 
     avfilter_formats_ref(avfilter_make_all_packing_formats(),
                          &inlink->out_packing);
-    if (aconvert->out_packing_fmt != -1) {
-        formats = NULL;
-        avfilter_add_format(&formats, aconvert->out_packing_fmt);
-        avfilter_formats_ref(formats, &outlink->in_packing);
-    } else
-        avfilter_formats_ref(avfilter_make_all_packing_formats(),
-                             &outlink->in_packing);
+    formats = NULL;
+    avfilter_add_format(&formats, out_packing);
+    avfilter_formats_ref(formats, &outlink->in_packing);
 
     return 0;
 }
 
 static int config_output(AVFilterLink *outlink)
 {
-    AVFilterLink *inlink = outlink->src->inputs[0];
-    AConvertContext *aconvert = outlink->src->priv;
+    int ret;
+    AVFilterContext *ctx = outlink->src;
+    AVFilterLink *inlink = ctx->inputs[0];
+    AConvertContext *aconvert = ctx->priv;
     char buf1[64], buf2[64];
 
-    aconvert->in_sample_fmt  = inlink->format;
-    aconvert->in_packing_fmt = inlink->planar;
-    if (aconvert->out_packing_fmt == -1)
-        aconvert->out_packing_fmt = outlink->planar;
-    aconvert->in_chlayout    = inlink->channel_layout;
-    aconvert->in_nb_channels =
-        av_get_channel_layout_nb_channels(inlink->channel_layout);
-
     /* if not specified in args, use the format and layout of the output */
     if (aconvert->out_sample_fmt == AV_SAMPLE_FMT_NONE)
         aconvert->out_sample_fmt = outlink->format;
     if (aconvert->out_chlayout   == 0)
         aconvert->out_chlayout   = outlink->channel_layout;
-    aconvert->out_nb_channels  =
-        av_get_channel_layout_nb_channels(outlink->channel_layout);
+
+    aconvert->swr = swr_alloc_set_opts(aconvert->swr,
+                                       aconvert->out_chlayout, aconvert->out_sample_fmt, inlink->sample_rate,
+                                       inlink->channel_layout, inlink->format,           inlink->sample_rate,
+                                       0, ctx);
+    if (!aconvert->swr)
+        return AVERROR(ENOMEM);
+    ret = swr_init(aconvert->swr);
+    if (ret < 0)
+        return ret;
 
     av_get_channel_layout_string(buf1, sizeof(buf1),
                                  -1, inlink ->channel_layout);
     av_get_channel_layout_string(buf2, sizeof(buf2),
                                  -1, outlink->channel_layout);
-    av_log(outlink->src, AV_LOG_INFO,
+    av_log(ctx, AV_LOG_INFO,
            "fmt:%s cl:%s planar:%i -> fmt:%s cl:%s planar:%i\n",
            av_get_sample_fmt_name(inlink ->format), buf1, inlink ->planar,
            av_get_sample_fmt_name(outlink->format), buf2, outlink->planar);
 
-    /* compute which channel layout conversion to use */
-    if (inlink->channel_layout != outlink->channel_layout) {
-        int i;
-        for (i = 0; i < sizeof(rematrix_funcs); i++) {
-            const struct RematrixFunctionInfo *f = &rematrix_funcs[i];
-            if ((f->in_chlayout  == 0 || f->in_chlayout  == inlink ->channel_layout) &&
-                (f->out_chlayout == 0 || f->out_chlayout == outlink->channel_layout) &&
-                (f->planar == -1 || f->planar == inlink->planar) &&
-                (f->sfmt   == -1 || f->sfmt   == inlink->format)
-               ) {
-                aconvert->convert_chlayout = f->func;
-                break;
-            }
-        }
-        if (!aconvert->convert_chlayout) {
-            av_log(outlink->src, AV_LOG_ERROR,
-                   "Unsupported channel layout conversion '%s -> %s' requested!\n",
-                   buf1, buf2);
-            return AVERROR(EINVAL);
-        }
-    }
-
     return 0;
 }
 
-static int init_buffers(AVFilterLink *inlink, int nb_samples)
-{
-    AConvertContext *aconvert = inlink->dst->priv;
-    AVFilterLink * const outlink = inlink->dst->outputs[0];
-    int i, packed_stride = 0;
-    const unsigned
-        packing_conv = inlink->planar != outlink->planar &&
-                       aconvert->out_nb_channels != 1,
-        format_conv  = inlink->format != outlink->format;
-    int nb_channels  = aconvert->out_nb_channels;
-
-    uninit(inlink->dst);
-    aconvert->max_nb_samples = nb_samples;
-
-    if (aconvert->convert_chlayout) {
-        /* allocate buffer for storing intermediary mixing samplesref */
-        uint8_t *data[8];
-        int linesize[8];
-        int nb_channels = av_get_channel_layout_nb_channels(outlink->channel_layout);
-
-        if (av_samples_alloc(data, linesize, nb_channels, nb_samples,
-                             inlink->format, 16) < 0)
-            goto fail_no_mem;
-        aconvert->mix_samplesref =
-            avfilter_get_audio_buffer_ref_from_arrays(data, linesize, AV_PERM_WRITE,
-                                                      nb_samples, inlink->format,
-                                                      outlink->channel_layout,
-                                                      inlink->planar);
-        if (!aconvert->mix_samplesref)
-            goto fail_no_mem;
-    }
-
-    // if there's a format/packing conversion we need an audio_convert context
-    if (format_conv || packing_conv) {
-        aconvert->out_samplesref =
-            avfilter_get_audio_buffer(outlink, AV_PERM_WRITE, nb_samples);
-        if (!aconvert->out_samplesref)
-            goto fail_no_mem;
-
-        aconvert->in_strides [0] = av_get_bytes_per_sample(inlink ->format);
-        aconvert->out_strides[0] = av_get_bytes_per_sample(outlink->format);
-
-        aconvert->out_conv = aconvert->out_samplesref->data;
-        if (aconvert->mix_samplesref)
-            aconvert->in_conv = aconvert->mix_samplesref->data;
-
-        if (packing_conv) {
-            // packed -> planar
-            if (outlink->planar == AVFILTER_PLANAR) {
-                if (aconvert->mix_samplesref)
-                    aconvert->packed_data[0] = aconvert->mix_samplesref->data[0];
-                aconvert->in_conv         = aconvert->packed_data;
-                packed_stride             = aconvert->in_strides[0];
-                aconvert->in_strides[0]  *= nb_channels;
-            // planar -> packed
-            } else {
-                aconvert->packed_data[0]  = aconvert->out_samplesref->data[0];
-                aconvert->out_conv        = aconvert->packed_data;
-                packed_stride             = aconvert->out_strides[0];
-                aconvert->out_strides[0] *= nb_channels;
-            }
-        } else if (outlink->planar == AVFILTER_PACKED) {
-            /* If there's no packing conversion, and the stream is packed
-             * then we treat the entire stream as one big channel
-             */
-            nb_channels = 1;
-        }
-
-        for (i = 1; i < nb_channels; i++) {
-            aconvert->packed_data[i] = aconvert->packed_data[i-1] + packed_stride;
-            aconvert->in_strides[i]  = aconvert->in_strides[0];
-            aconvert->out_strides[i] = aconvert->out_strides[0];
-        }
-
-        aconvert->audioconvert_ctx =
-                av_audio_convert_alloc(outlink->format, nb_channels,
-                                       inlink->format,  nb_channels, NULL, 0);
-        if (!aconvert->audioconvert_ctx)
-            goto fail_no_mem;
-    }
-
-    return 0;
-
-fail_no_mem:
-    av_log(inlink->dst, AV_LOG_ERROR, "Could not allocate memory.\n");
-    return AVERROR(ENOMEM);
-}
-
 static void filter_samples(AVFilterLink *inlink, AVFilterBufferRef *insamplesref)
 {
     AConvertContext *aconvert = inlink->dst->priv;
-    AVFilterBufferRef *curbuf = insamplesref;
-    AVFilterLink * const outlink = inlink->dst->outputs[0];
-    int chan_mult;
-
-    /* in/reinint the internal buffers if this is the first buffer
-     * provided or it is needed to use a bigger one */
-    if (!aconvert->max_nb_samples ||
-        (curbuf->audio->nb_samples > aconvert->max_nb_samples))
-        if (init_buffers(inlink, curbuf->audio->nb_samples) < 0) {
-            av_log(inlink->dst, AV_LOG_ERROR, "Could not initialize buffers.\n");
-            return;
-        }
+    const int n = insamplesref->audio->nb_samples;
+    AVFilterLink *const outlink = inlink->dst->outputs[0];
+    AVFilterBufferRef *outsamplesref = avfilter_get_audio_buffer(outlink, AV_PERM_WRITE, n);
 
-    /* if channel mixing is required */
-    if (aconvert->mix_samplesref) {
-        memcpy(aconvert->in_mix,  curbuf->data, sizeof(aconvert->in_mix));
-        memcpy(aconvert->out_mix, aconvert->mix_samplesref->data, sizeof(aconvert->out_mix));
-        aconvert->convert_chlayout(aconvert->out_mix,
-                                   aconvert->in_mix,
-                                   curbuf->audio->nb_samples,
-                                   aconvert);
-        curbuf = aconvert->mix_samplesref;
-    }
-
-    if (aconvert->audioconvert_ctx) {
-        if (!aconvert->mix_samplesref) {
-            if (aconvert->in_conv == aconvert->packed_data) {
-                int i, packed_stride = av_get_bytes_per_sample(inlink->format);
-                aconvert->packed_data[0] = curbuf->data[0];
-                for (i = 1; i < aconvert->out_nb_channels; i++)
-                    aconvert->packed_data[i] = aconvert->packed_data[i-1] + packed_stride;
-            } else {
-                aconvert->in_conv = curbuf->data;
-            }
-        }
-
-        chan_mult = inlink->planar == outlink->planar && inlink->planar == 0 ?
-            aconvert->out_nb_channels : 1;
-
-        av_audio_convert(aconvert->audioconvert_ctx,
-                         (void * const *) aconvert->out_conv,
-                         aconvert->out_strides,
-                         (const void * const *) aconvert->in_conv,
-                         aconvert->in_strides,
-                         curbuf->audio->nb_samples * chan_mult);
-
-        curbuf = aconvert->out_samplesref;
-    }
+    swr_convert(aconvert->swr, outsamplesref->data, n,
+                        (void *)insamplesref->data, n);
 
-    avfilter_copy_buffer_ref_props(curbuf, insamplesref);
-    curbuf->audio->channel_layout = outlink->channel_layout;
-    curbuf->audio->planar         = outlink->planar;
+    avfilter_copy_buffer_ref_props(outsamplesref, insamplesref);
+    outsamplesref->audio->channel_layout = outlink->channel_layout;
+    outsamplesref->audio->planar         = outlink->planar;
 
-    avfilter_filter_samples(inlink->dst->outputs[0],
-                            avfilter_ref_buffer(curbuf, ~0));
+    avfilter_filter_samples(outlink, outsamplesref);
     avfilter_unref_buffer(insamplesref);
 }
 
diff --git a/libavfilter/af_aconvert_rematrix.c b/libavfilter/af_aconvert_rematrix.c
deleted file mode 100644 (file)
index d75ca5a..0000000
+++ /dev/null
@@ -1,172 +0,0 @@
-/*
- * Copyright (c) 2011 Mina Nagy Zaki
- *
- * This file is part of FFmpeg.
- *
- * FFmpeg is free software; you can redistribute it and/or
- * modify it under the terms of the GNU Lesser General Public
- * License as published by the Free Software Foundation; either
- * version 2.1 of the License, or (at your option) any later version.
- *
- * FFmpeg is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
- * Lesser General Public License for more details.
- *
- * You should have received a copy of the GNU Lesser General Public
- * License along with FFmpeg; if not, write to the Free Software
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-/**
- * @file
- * audio rematrixing functions, based on functions from libavcodec/resample.c
- */
-
-#if defined(FLOATING)
-# define DIV2 /2
-#else
-# define DIV2 >>1
-#endif
-
-REMATRIX_FUNC_SIG(stereo_to_mono_packed)
-{
-    while (nb_samples >= 4) {
-        outp[0][0] = (inp[0][0] + inp[0][1]) DIV2;
-        outp[0][1] = (inp[0][2] + inp[0][3]) DIV2;
-        outp[0][2] = (inp[0][4] + inp[0][5]) DIV2;
-        outp[0][3] = (inp[0][6] + inp[0][7]) DIV2;
-        outp[0] += 4;
-        inp[0]  += 8;
-        nb_samples -= 4;
-    }
-    while (nb_samples--) {
-        outp[0][0] = (inp[0][0] + inp[0][1]) DIV2;
-        outp[0]++;
-        inp[0] += 2;
-    }
-}
-
-REMATRIX_FUNC_SIG(stereo_downmix_packed)
-{
-    while (nb_samples--) {
-        *outp[0]++ = inp[0][0];
-        *outp[0]++ = inp[0][1];
-        inp[0] += aconvert->in_nb_channels;
-    }
-}
-
-REMATRIX_FUNC_SIG(mono_to_stereo_packed)
-{
-    while (nb_samples >= 4) {
-        outp[0][0] = outp[0][1] = inp[0][0];
-        outp[0][2] = outp[0][3] = inp[0][1];
-        outp[0][4] = outp[0][5] = inp[0][2];
-        outp[0][6] = outp[0][7] = inp[0][3];
-        outp[0] += 8;
-        inp[0]  += 4;
-        nb_samples -= 4;
-    }
-    while (nb_samples--) {
-        outp[0][0] = outp[0][1] = inp[0][0];
-        outp[0] += 2;
-        inp[0]  += 1;
-    }
-}
-
-/**
- * This is for when we have more than 2 input channels, need to downmix to mono
- * and do not have a conversion formula available.  We just use first two input
- * channels - left and right. This is a placeholder until more conversion
- * functions are written.
- */
-REMATRIX_FUNC_SIG(mono_downmix_packed)
-{
-    while (nb_samples--) {
-        outp[0][0] = (inp[0][0] + inp[0][1]) DIV2;
-        inp[0] += aconvert->in_nb_channels;
-        outp[0]++;
-    }
-}
-
-REMATRIX_FUNC_SIG(mono_downmix_planar)
-{
-    FMT_TYPE *out = outp[0];
-
-    while (nb_samples >= 4) {
-        out[0] = (inp[0][0] + inp[1][0]) DIV2;
-        out[1] = (inp[0][1] + inp[1][1]) DIV2;
-        out[2] = (inp[0][2] + inp[1][2]) DIV2;
-        out[3] = (inp[0][3] + inp[1][3]) DIV2;
-        out    += 4;
-        inp[0] += 4;
-        inp[1] += 4;
-        nb_samples -= 4;
-    }
-    while (nb_samples--) {
-        out[0] = (inp[0][0] + inp[1][0]) DIV2;
-        out++;
-        inp[0]++;
-        inp[1]++;
-    }
-}
-
-/* Stereo to 5.1 output */
-REMATRIX_FUNC_SIG(stereo_to_surround_5p1_packed)
-{
-    while (nb_samples--) {
-      outp[0][0] = inp[0][0];  /* left */
-      outp[0][1] = inp[0][1];  /* right */
-      outp[0][2] = (inp[0][0] + inp[0][1]) DIV2; /* center */
-      outp[0][3] = 0;          /* low freq */
-      outp[0][4] = 0;          /* FIXME: left surround: -3dB or -6dB or -9dB of stereo left  */
-      outp[0][5] = 0;          /* FIXME: right surroud: -3dB or -6dB or -9dB of stereo right */
-      inp[0]  += 2;
-      outp[0] += 6;
-    }
-}
-
-REMATRIX_FUNC_SIG(stereo_to_surround_5p1_planar)
-{
-    while (nb_samples--) {
-      *outp[0]++ = *inp[0];    /* left */
-      *outp[1]++ = *inp[1];    /* right */
-      *outp[2]++ = (*inp[0] + *inp[1]) DIV2; /* center */
-      *outp[3]++ = 0;          /* low freq */
-      *outp[4]++ = 0;          /* FIXME: left surround: -3dB or -6dB or -9dB of stereo left  */
-      *outp[5]++ = 0;          /* FIXME: right surroud: -3dB or -6dB or -9dB of stereo right */
-      inp[0]++; inp[1]++;
-    }
-}
-
-
-/*
-5.1 to stereo input: [fl, fr, c, lfe, rl, rr]
-- Left = front_left + rear_gain * rear_left + center_gain * center
-- Right = front_right + rear_gain * rear_right + center_gain * center
-Where rear_gain is usually around 0.5-1.0 and
-      center_gain is almost always 0.7 (-3 dB)
-*/
-REMATRIX_FUNC_SIG(surround_5p1_to_stereo_packed)
-{
-    while (nb_samples--) {
-        *outp[0]++ = inp[0][0] + (0.5 * inp[0][4]) + (0.7 * inp[0][2]); //FIXME CLIPPING!
-        *outp[0]++ = inp[0][1] + (0.5 * inp[0][5]) + (0.7 * inp[0][2]); //FIXME CLIPPING!
-
-        inp[0] += 6;
-    }
-}
-
-REMATRIX_FUNC_SIG(surround_5p1_to_stereo_planar)
-{
-    while (nb_samples--) {
-        *outp[0]++ = *inp[0] + (0.5 * *inp[4]) + (0.7 * *inp[2]); //FIXME CLIPPING!
-        *outp[1]++ = *inp[1] + (0.5 * *inp[5]) + (0.7 * *inp[2]); //FIXME CLIPPING!
-
-        inp[0]++; inp[1]++; inp[2]++; inp[3]++; inp[4]++; inp[5]++;
-    }
-}
-
-#undef DIV2
-#undef REMATRIX_FUNC_NAME
-#undef FMT_TYPE
index 393fb91..11e038d 100644 (file)
@@ -29,7 +29,7 @@
 #include "libavutil/avutil.h"
 
 #define LIBAVFILTER_VERSION_MAJOR  2
-#define LIBAVFILTER_VERSION_MINOR 60
+#define LIBAVFILTER_VERSION_MINOR 61
 #define LIBAVFILTER_VERSION_MICRO 100
 
 #define LIBAVFILTER_VERSION_INT AV_VERSION_INT(LIBAVFILTER_VERSION_MAJOR, \