[FFmpeg-cvslog] avfilter: add mix filter

Paul B Mahol git at videolan.org
Fri Nov 24 17:51:53 EET 2017


ffmpeg | branch: master | Paul B Mahol <onemda at gmail.com> | Mon Nov 20 19:58:44 2017 +0100| [ec5328aa6f3c1650155dd2a257167f4130abb800] | committer: Paul B Mahol

avfilter: add mix filter

Signed-off-by: Paul B Mahol <onemda at gmail.com>

> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=ec5328aa6f3c1650155dd2a257167f4130abb800
---

 Changelog                |   1 +
 doc/filters.texi         |  28 +++++
 libavfilter/Makefile     |   1 +
 libavfilter/allfilters.c |   1 +
 libavfilter/version.h    |   4 +-
 libavfilter/vf_mix.c     | 283 +++++++++++++++++++++++++++++++++++++++++++++++
 6 files changed, 316 insertions(+), 2 deletions(-)

diff --git a/Changelog b/Changelog
index f6b70c4a34..6ef144c315 100644
--- a/Changelog
+++ b/Changelog
@@ -18,6 +18,7 @@ version <next>:
 - mcompand audio filter
 - acontrast audio filter
 - OpenCL overlay filter
+- video mix filter
 
 
 version 3.4:
diff --git a/doc/filters.texi b/doc/filters.texi
index bee626e775..76929e4db5 100644
--- a/doc/filters.texi
+++ b/doc/filters.texi
@@ -10569,6 +10569,34 @@ Default method is @samp{fdiff}.
 Scene change detection threshold. Default is @code{5.0}.
 @end table
 
+ at section mix
+
+Mix several video input streams into one video stream.
+
+A description of the accepted options follows.
+
+ at table @option
+ at item nb_inputs
+The number of inputs. If unspecified, it defaults to 2.
+
+ at item weights
+Specify weight of each input video stream as sequence.
+Each weight is separated by space.
+
+ at item duration
+Specify how end of stream is determined.
+ at table @samp
+ at item longest
+The duration of the longest input. (default)
+
+ at item shortest
+The duration of the shortest input.
+
+ at item first
+The duration of the first input.
+ at end table
+ at end table
+
 @section mpdecimate
 
 Drop frames that do not differ greatly from the previous frame in
diff --git a/libavfilter/Makefile b/libavfilter/Makefile
index ec0deff86f..c47373a65b 100644
--- a/libavfilter/Makefile
+++ b/libavfilter/Makefile
@@ -240,6 +240,7 @@ OBJS-$(CONFIG_MESTIMATE_FILTER)              += vf_mestimate.o motion_estimation
 OBJS-$(CONFIG_METADATA_FILTER)               += f_metadata.o
 OBJS-$(CONFIG_MIDEQUALIZER_FILTER)           += vf_midequalizer.o framesync.o
 OBJS-$(CONFIG_MINTERPOLATE_FILTER)           += vf_minterpolate.o motion_estimation.o
+OBJS-$(CONFIG_MIX_FILTER)                    += vf_mix.o
 OBJS-$(CONFIG_MPDECIMATE_FILTER)             += vf_mpdecimate.o
 OBJS-$(CONFIG_NEGATE_FILTER)                 += vf_lut.o
 OBJS-$(CONFIG_NLMEANS_FILTER)                += vf_nlmeans.o
diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
index 50e9d9a0cb..3a9cd00fdf 100644
--- a/libavfilter/allfilters.c
+++ b/libavfilter/allfilters.c
@@ -250,6 +250,7 @@ static void register_all(void)
     REGISTER_FILTER(METADATA,       metadata,       vf);
     REGISTER_FILTER(MIDEQUALIZER,   midequalizer,   vf);
     REGISTER_FILTER(MINTERPOLATE,   minterpolate,   vf);
+    REGISTER_FILTER(MIX,            mix,            vf);
     REGISTER_FILTER(MPDECIMATE,     mpdecimate,     vf);
     REGISTER_FILTER(NEGATE,         negate,         vf);
     REGISTER_FILTER(NLMEANS,        nlmeans,        vf);
diff --git a/libavfilter/version.h b/libavfilter/version.h
index b449fd821d..3242c44618 100644
--- a/libavfilter/version.h
+++ b/libavfilter/version.h
@@ -30,8 +30,8 @@
 #include "libavutil/version.h"
 
 #define LIBAVFILTER_VERSION_MAJOR   7
-#define LIBAVFILTER_VERSION_MINOR   2
-#define LIBAVFILTER_VERSION_MICRO 102
+#define LIBAVFILTER_VERSION_MINOR   3
+#define LIBAVFILTER_VERSION_MICRO 100
 
 #define LIBAVFILTER_VERSION_INT AV_VERSION_INT(LIBAVFILTER_VERSION_MAJOR, \
                                                LIBAVFILTER_VERSION_MINOR, \
diff --git a/libavfilter/vf_mix.c b/libavfilter/vf_mix.c
new file mode 100644
index 0000000000..261ab066ef
--- /dev/null
+++ b/libavfilter/vf_mix.c
@@ -0,0 +1,283 @@
+/*
+ * Copyright (c) 2017 Paul B Mahol
+ *
+ * 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
+ */
+
+#include "libavutil/avstring.h"
+#include "libavutil/imgutils.h"
+#include "libavutil/intreadwrite.h"
+#include "libavutil/opt.h"
+#include "libavutil/pixdesc.h"
+
+#include "avfilter.h"
+#include "formats.h"
+#include "internal.h"
+#include "framesync.h"
+#include "video.h"
+
+typedef struct MixContext {
+    const AVClass *class;
+    const AVPixFmtDescriptor *desc;
+    char *weights_str;
+    int nb_inputs;
+    int duration;
+    float *weights;
+    float wfactor;
+
+    int depth;
+    int nb_planes;
+    int linesize[4];
+    int height[4];
+
+    AVFrame **frames;
+    FFFrameSync fs;
+} MixContext;
+
+static int query_formats(AVFilterContext *ctx)
+{
+    AVFilterFormats *pix_fmts = NULL;
+    int fmt, ret;
+
+    for (fmt = 0; av_pix_fmt_desc_get(fmt); fmt++) {
+        const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(fmt);
+        if (!(desc->flags & AV_PIX_FMT_FLAG_PAL ||
+              desc->flags & AV_PIX_FMT_FLAG_HWACCEL ||
+              desc->flags & AV_PIX_FMT_FLAG_BITSTREAM) &&
+            (ret = ff_add_format(&pix_fmts, fmt)) < 0)
+            return ret;
+    }
+
+    return ff_set_common_formats(ctx, pix_fmts);
+}
+
+static av_cold int init(AVFilterContext *ctx)
+{
+    MixContext *s = ctx->priv;
+    char *p, *arg, *saveptr = NULL;
+    int i, ret;
+
+    s->frames = av_calloc(s->nb_inputs, sizeof(*s->frames));
+    if (!s->frames)
+        return AVERROR(ENOMEM);
+
+    s->weights = av_calloc(s->nb_inputs, sizeof(*s->weights));
+    if (!s->weights)
+        return AVERROR(ENOMEM);
+
+    for (i = 0; i < s->nb_inputs; i++) {
+        AVFilterPad pad = { 0 };
+
+        pad.type = AVMEDIA_TYPE_VIDEO;
+        pad.name = av_asprintf("input%d", i);
+        if (!pad.name)
+            return AVERROR(ENOMEM);
+
+        if ((ret = ff_insert_inpad(ctx, i, &pad)) < 0) {
+            av_freep(&pad.name);
+            return ret;
+        }
+    }
+
+    p = s->weights_str;
+    for (i = 0; i < s->nb_inputs; i++) {
+        if (!(arg = av_strtok(p, " ", &saveptr)))
+            break;
+
+        p = NULL;
+        sscanf(arg, "%f", &s->weights[i]);
+        s->wfactor += s->weights[i];
+    }
+    s->wfactor = 1 / s->wfactor;
+
+    return 0;
+}
+
+static int process_frame(FFFrameSync *fs)
+{
+    AVFilterContext *ctx = fs->parent;
+    AVFilterLink *outlink = ctx->outputs[0];
+    MixContext *s = fs->opaque;
+    AVFrame **in = s->frames;
+    AVFrame *out;
+    int i, p, ret, x, y;
+
+    for (i = 0; i < s->nb_inputs; i++) {
+        if ((ret = ff_framesync_get_frame(&s->fs, i, &in[i], 0)) < 0)
+            return ret;
+    }
+
+    out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
+    if (!out)
+        return AVERROR(ENOMEM);
+    out->pts = av_rescale_q(s->fs.pts, s->fs.time_base, outlink->time_base);
+
+    if (s->depth <= 8) {
+        for (p = 0; p < s->nb_planes; p++) {
+            uint8_t *dst = out->data[p];
+
+            for (y = 0; y < s->height[p]; y++) {
+                for (x = 0; x < s->linesize[p]; x++) {
+                    int val = 0;
+
+                    for (i = 0; i < s->nb_inputs; i++) {
+                        uint8_t src = in[i]->data[p][y * s->linesize[p] + x];
+
+                        val += src * s->weights[i];
+                    }
+
+                    dst[x] = val * s->wfactor;
+                }
+
+                dst += out->linesize[p];
+            }
+        }
+    } else {
+        for (p = 0; p < s->nb_planes; p++) {
+            uint16_t *dst = (uint16_t *)out->data[p];
+
+            for (y = 0; y < s->height[p]; y++) {
+                for (x = 0; x < s->linesize[p]; x++) {
+                    int val = 0;
+
+                    for (i = 0; i < s->nb_inputs; i++) {
+                        uint16_t src = AV_RN16(in[i]->data[p] + y * s->linesize[p] + x * 2);
+
+                        val += src * s->weights[i];
+                    }
+
+                    dst[x] = val * s->wfactor;
+                }
+
+                dst += out->linesize[p] / 2;
+            }
+        }
+    }
+
+    return ff_filter_frame(outlink, out);
+}
+
+static int config_output(AVFilterLink *outlink)
+{
+    AVFilterContext *ctx = outlink->src;
+    MixContext *s = ctx->priv;
+    AVRational time_base = ctx->inputs[0]->time_base;
+    AVRational frame_rate = ctx->inputs[0]->frame_rate;
+    AVFilterLink *inlink = ctx->inputs[0];
+    int height = ctx->inputs[0]->h;
+    int width = ctx->inputs[0]->w;
+    FFFrameSyncIn *in;
+    int i, ret;
+
+    for (i = 1; i < s->nb_inputs; i++) {
+        if (ctx->inputs[i]->h != height || ctx->inputs[i]->w != width) {
+            av_log(ctx, AV_LOG_ERROR, "Input %d size (%dx%d) does not match input %d size (%dx%d).\n", i, ctx->inputs[i]->w, ctx->inputs[i]->h, 0, width, height);
+            return AVERROR(EINVAL);
+        }
+    }
+
+    s->desc = av_pix_fmt_desc_get(outlink->format);
+    if (!s->desc)
+        return AVERROR_BUG;
+    s->nb_planes = av_pix_fmt_count_planes(outlink->format);
+    s->depth = s->desc->comp[0].depth;
+
+    outlink->w          = width;
+    outlink->h          = height;
+    outlink->time_base  = time_base;
+    outlink->frame_rate = frame_rate;
+
+    if ((ret = ff_framesync_init(&s->fs, ctx, s->nb_inputs)) < 0)
+        return ret;
+
+    in = s->fs.in;
+    s->fs.opaque = s;
+    s->fs.on_event = process_frame;
+
+    if ((ret = av_image_fill_linesizes(s->linesize, inlink->format, inlink->w)) < 0)
+        return ret;
+
+    s->height[1] = s->height[2] = AV_CEIL_RSHIFT(inlink->h, s->desc->log2_chroma_h);
+    s->height[0] = s->height[3] = inlink->h;
+
+    for (i = 0; i < s->nb_inputs; i++) {
+        AVFilterLink *inlink = ctx->inputs[i];
+
+        in[i].time_base = inlink->time_base;
+        in[i].sync   = 1;
+        in[i].before = EXT_STOP;
+        in[i].after  = (s->duration == 1 || (s->duration == 2 && i == 0)) ? EXT_STOP : EXT_INFINITY;
+    }
+
+    return ff_framesync_configure(&s->fs);
+}
+
+static av_cold void uninit(AVFilterContext *ctx)
+{
+    MixContext *s = ctx->priv;
+    int i;
+
+    ff_framesync_uninit(&s->fs);
+    av_freep(&s->frames);
+    av_freep(&s->weights);
+
+    for (i = 0; i < ctx->nb_inputs; i++)
+        av_freep(&ctx->input_pads[i].name);
+}
+
+static int activate(AVFilterContext *ctx)
+{
+    MixContext *s = ctx->priv;
+    return ff_framesync_activate(&s->fs);
+}
+
+#define OFFSET(x) offsetof(MixContext, x)
+#define FLAGS AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_FILTERING_PARAM
+
+static const AVOption mix_options[] = {
+    { "inputs", "set number of inputs", OFFSET(nb_inputs), AV_OPT_TYPE_INT, {.i64=2}, 2, INT_MAX, .flags = FLAGS },
+    { "weights", "set weight for each input", OFFSET(weights_str), AV_OPT_TYPE_STRING, {.str="1 1"}, 0, 0, .flags = FLAGS },
+    { "duration", "how to determine end of stream", OFFSET(duration), AV_OPT_TYPE_INT, {.i64=0}, 0, 2, .flags = FLAGS, "duration" },
+        { "longest",  "Duration of longest input",  0, AV_OPT_TYPE_CONST, {.i64=0}, 0, 0, FLAGS, "duration" },
+        { "shortest", "Duration of shortest input", 0, AV_OPT_TYPE_CONST, {.i64=1}, 0, 0, FLAGS, "duration" },
+        { "first",    "Duration of first input",    0, AV_OPT_TYPE_CONST, {.i64=2}, 0, 0, FLAGS, "duration" },
+    { NULL },
+};
+
+static const AVFilterPad outputs[] = {
+    {
+        .name          = "default",
+        .type          = AVMEDIA_TYPE_VIDEO,
+        .config_props  = config_output,
+    },
+    { NULL }
+};
+
+AVFILTER_DEFINE_CLASS(mix);
+
+AVFilter ff_vf_mix = {
+    .name          = "mix",
+    .description   = NULL_IF_CONFIG_SMALL("Mix video inputs."),
+    .priv_size     = sizeof(MixContext),
+    .priv_class    = &mix_class,
+    .query_formats = query_formats,
+    .outputs       = outputs,
+    .init          = init,
+    .uninit        = uninit,
+    .activate      = activate,
+    .flags         = AVFILTER_FLAG_DYNAMIC_INPUTS,
+};



More information about the ffmpeg-cvslog mailing list