[FFmpeg-devel] [PATCH] avfilter/vf_alphamerge: use framesync

Paul B Mahol onemda at gmail.com
Mon Aug 31 00:10:16 EEST 2020


Signed-off-by: Paul B Mahol <onemda at gmail.com>
---
 doc/filters.texi            |   6 --
 libavfilter/vf_alphamerge.c | 136 +++++++++++++++++-------------------
 2 files changed, 63 insertions(+), 79 deletions(-)

diff --git a/doc/filters.texi b/doc/filters.texi
index b0cdf92845..6d2c234b37 100644
--- a/doc/filters.texi
+++ b/doc/filters.texi
@@ -6470,12 +6470,6 @@ and a separate video created with @var{alphaextract}, you might use:
 movie=in_alpha.mkv [alpha]; [in][alpha] alphamerge [out]
 @end example
 
-Since this filter is designed for reconstruction, it operates on frame
-sequences without considering timestamps, and terminates when either
-input reaches end of stream. This will cause problems if your encoding
-pipeline drops frames. If you're trying to apply an image as an
-overlay to a video stream, consider the @var{overlay} filter instead.
-
 @section amplify
 
 Amplify differences between current pixel and pixels of adjacent frames in
diff --git a/libavfilter/vf_alphamerge.c b/libavfilter/vf_alphamerge.c
index a509f10103..866d91f9e8 100644
--- a/libavfilter/vf_alphamerge.c
+++ b/libavfilter/vf_alphamerge.c
@@ -32,6 +32,7 @@
 #include "drawutils.h"
 #include "formats.h"
 #include "filters.h"
+#include "framesync.h"
 #include "internal.h"
 #include "video.h"
 
@@ -42,10 +43,54 @@ typedef struct AlphaMergeContext {
 
     int is_packed_rgb;
     uint8_t rgba_map[4];
-    AVFrame *main_frame;
-    AVFrame *alpha_frame;
+
+    FFFrameSync fs;
 } AlphaMergeContext;
 
+static int do_alphamerge(FFFrameSync *fs)
+{
+    AVFilterContext *ctx = fs->parent;
+    AlphaMergeContext *s = ctx->priv;
+    AVFrame *main_buf, *alpha_buf;
+    int ret;
+
+    ret = ff_framesync_dualinput_get_writable(fs, &main_buf, &alpha_buf);
+    if (ret < 0)
+        return ret;
+    if (!alpha_buf)
+        return ff_filter_frame(ctx->outputs[0], main_buf);
+
+    if (s->is_packed_rgb) {
+        int x, y;
+        uint8_t *pin, *pout;
+        for (y = 0; y < main_buf->height; y++) {
+            pin = alpha_buf->data[0] + y * alpha_buf->linesize[0];
+            pout = main_buf->data[0] + y * main_buf->linesize[0] + s->rgba_map[A];
+            for (x = 0; x < main_buf->width; x++) {
+                *pout = *pin;
+                pin += 1;
+                pout += 4;
+            }
+        }
+    } else {
+        const int main_linesize = main_buf->linesize[A];
+        const int alpha_linesize = alpha_buf->linesize[Y];
+        av_image_copy_plane(main_buf->data[A], main_linesize,
+                            alpha_buf->data[Y], alpha_linesize,
+                            FFMIN(main_linesize, alpha_linesize), alpha_buf->height);
+    }
+
+    return ff_filter_frame(ctx->outputs[0], main_buf);
+}
+
+static av_cold int init(AVFilterContext *ctx)
+{
+    AlphaMergeContext *s = ctx->priv;
+
+    s->fs.on_event = do_alphamerge;
+    return 0;
+}
+
 static int query_formats(AVFilterContext *ctx)
 {
     static const enum AVPixelFormat main_fmts[] = {
@@ -78,8 +123,11 @@ static int config_input_main(AVFilterLink *inlink)
 static int config_output(AVFilterLink *outlink)
 {
     AVFilterContext *ctx = outlink->src;
+    AlphaMergeContext *s = ctx->priv;
     AVFilterLink *mainlink = ctx->inputs[0];
     AVFilterLink *alphalink = ctx->inputs[1];
+    int ret;
+
     if (mainlink->w != alphalink->w || mainlink->h != alphalink->h) {
         av_log(ctx, AV_LOG_ERROR,
                "Input frame sizes do not match (%dx%d vs %dx%d).\n",
@@ -88,89 +136,29 @@ static int config_output(AVFilterLink *outlink)
         return AVERROR(EINVAL);
     }
 
+    if ((ret = ff_framesync_init_dualinput(&s->fs, ctx)) < 0)
+        return ret;
+
     outlink->w = mainlink->w;
     outlink->h = mainlink->h;
     outlink->time_base = mainlink->time_base;
     outlink->sample_aspect_ratio = mainlink->sample_aspect_ratio;
     outlink->frame_rate = mainlink->frame_rate;
-    return 0;
+
+    return ff_framesync_configure(&s->fs);
 }
 
-static void draw_frame(AVFilterContext *ctx,
-                       AVFrame *main_buf,
-                       AVFrame *alpha_buf)
+static int activate(AVFilterContext *ctx)
 {
     AlphaMergeContext *s = ctx->priv;
-    int h = main_buf->height;
-
-    if (s->is_packed_rgb) {
-        int x, y;
-        uint8_t *pin, *pout;
-        for (y = 0; y < h; y++) {
-            pin = alpha_buf->data[0] + y * alpha_buf->linesize[0];
-            pout = main_buf->data[0] + y * main_buf->linesize[0] + s->rgba_map[A];
-            for (x = 0; x < main_buf->width; x++) {
-                *pout = *pin;
-                pin += 1;
-                pout += 4;
-            }
-        }
-    } else {
-        const int main_linesize = main_buf->linesize[A];
-        const int alpha_linesize = alpha_buf->linesize[Y];
-        av_image_copy_plane(main_buf->data[A], main_linesize,
-                            alpha_buf->data[Y], alpha_linesize,
-                            FFMIN(main_linesize, alpha_linesize), alpha_buf->height);
-    }
+    return ff_framesync_activate(&s->fs);
 }
 
-static int activate(AVFilterContext *ctx)
+static av_cold void uninit(AVFilterContext *ctx)
 {
-    AVFilterLink *outlink = ctx->outputs[0];
     AlphaMergeContext *s = ctx->priv;
-    int ret;
-
-    FF_FILTER_FORWARD_STATUS_BACK_ALL(outlink, ctx);
-
-    if (!s->main_frame) {
-        ret = ff_inlink_consume_frame(ctx->inputs[0], &s->main_frame);
-        if (ret < 0)
-            return ret;
-    }
-
-    if (!s->alpha_frame) {
-        ret = ff_inlink_consume_frame(ctx->inputs[1], &s->alpha_frame);
-        if (ret < 0)
-            return ret;
-    }
-
-    if (s->main_frame && s->alpha_frame) {
-        if (!ctx->is_disabled)
-            draw_frame(ctx, s->main_frame, s->alpha_frame);
-        ret = ff_filter_frame(outlink, s->main_frame);
-        av_frame_free(&s->alpha_frame);
-        s->main_frame = NULL;
-        return ret;
-    }
-
-    FF_FILTER_FORWARD_STATUS(ctx->inputs[0], outlink);
-    FF_FILTER_FORWARD_STATUS(ctx->inputs[1], outlink);
-
-    if (ff_outlink_frame_wanted(ctx->outputs[0]) &&
-        !ff_outlink_get_status(ctx->inputs[0]) &&
-        !s->main_frame) {
-        ff_inlink_request_frame(ctx->inputs[0]);
-        return 0;
-    }
-
-    if (ff_outlink_frame_wanted(ctx->outputs[0]) &&
-        !ff_outlink_get_status(ctx->inputs[1]) &&
-        !s->alpha_frame) {
-        ff_inlink_request_frame(ctx->inputs[1]);
-        return 0;
-    }
 
-    return FFERROR_NOT_READY;
+    ff_framesync_uninit(&s->fs);
 }
 
 static const AVFilterPad alphamerge_inputs[] = {
@@ -178,7 +166,6 @@ static const AVFilterPad alphamerge_inputs[] = {
         .name             = "main",
         .type             = AVMEDIA_TYPE_VIDEO,
         .config_props     = config_input_main,
-        .needs_writable   = 1,
     },{
         .name             = "alpha",
         .type             = AVMEDIA_TYPE_VIDEO,
@@ -199,17 +186,20 @@ static const AVOption alphamerge_options[] = {
     { NULL }
 };
 
-AVFILTER_DEFINE_CLASS(alphamerge);
+FRAMESYNC_DEFINE_CLASS(alphamerge, AlphaMergeContext, fs);
 
 AVFilter ff_vf_alphamerge = {
     .name           = "alphamerge",
     .description    = NULL_IF_CONFIG_SMALL("Copy the luma value of the second "
                       "input into the alpha channel of the first input."),
+    .preinit        = alphamerge_framesync_preinit,
     .priv_size      = sizeof(AlphaMergeContext),
     .priv_class     = &alphamerge_class,
+    .init           = init,
     .query_formats  = query_formats,
     .inputs         = alphamerge_inputs,
     .outputs        = alphamerge_outputs,
+    .uninit         = uninit,
     .activate       = activate,
     .flags          = AVFILTER_FLAG_SUPPORT_TIMELINE_INTERNAL,
 };
-- 
2.17.1



More information about the ffmpeg-devel mailing list