Merge commit '4ce701b4e640d4723a4005d664f31f8342fac40e'

[ffmpeg] / libavfilter / vf_mix.c

diff --git a/libavfilter/vf_mix.c b/libavfilter/vf_mix.c
index 55307a4bb5965732e6fb174f9ca91416917ae3f8..d0cc7cb83071fa139477a8023dd6bba707dfb214 100644 (file)
--- a/libavfilter/vf_mix.c
+++ b/libavfilter/vf_mix.c
@@ -37,12 +37,14 @@ typedef struct MixContext {
     int nb_inputs;
     int duration;
     float *weights;
+    float scale;
     float wfactor;
 
     int tmix;
     int nb_frames;
 
     int depth;
+    int max;
     int nb_planes;
     int linesize[4];
     int height[4];
@@ -72,7 +74,7 @@ static av_cold int init(AVFilterContext *ctx)
 {
     MixContext *s = ctx->priv;
     char *p, *arg, *saveptr = NULL;
-    int i, ret;
+    int i, ret, last = 0;
 
     s->tmix = !strcmp(ctx->filter->name, "tmix");
 
@@ -108,31 +110,50 @@ static av_cold int init(AVFilterContext *ctx)
         p = NULL;
         sscanf(arg, "%f", &s->weights[i]);
         s->wfactor += s->weights[i];
+        last = i;
+    }
+    for (; i < s->nb_inputs; i++) {
+        s->weights[i] = s->weights[last];
+        s->wfactor += s->weights[i];
+    }
+    if (s->scale == 0) {
+        s->wfactor = 1 / s->wfactor;
+    } else {
+        s->wfactor = s->scale;
     }
-    s->wfactor = 1 / s->wfactor;
 
     return 0;
 }
 
-static void mix_frames(MixContext *s, AVFrame **in, AVFrame *out)
+typedef struct ThreadData {
+    AVFrame **in, *out;
+} ThreadData;
+
+static int mix_frames(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
 {
+    MixContext *s = ctx->priv;
+    ThreadData *td = arg;
+    AVFrame **in = td->in;
+    AVFrame *out = td->out;
     int i, p, x, y;
 
     if (s->depth <= 8) {
         for (p = 0; p < s->nb_planes; p++) {
-            uint8_t *dst = out->data[p];
+            const int slice_start = (s->height[p] * jobnr) / nb_jobs;
+            const int slice_end = (s->height[p] * (jobnr+1)) / nb_jobs;
+            uint8_t *dst = out->data[p] + slice_start * out->linesize[p];
 
-            for (y = 0; y < s->height[p]; y++) {
+            for (y = slice_start; y < slice_end; 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];
+                        uint8_t src = in[i]->data[p][y * in[i]->linesize[p] + x];
 
                         val += src * s->weights[i];
                     }
 
-                    dst[x] = val * s->wfactor;
+                    dst[x] = av_clip_uint8(val * s->wfactor);
                 }
 
                 dst += out->linesize[p];
@@ -140,25 +161,29 @@ static void mix_frames(MixContext *s, AVFrame **in, AVFrame *out)
         }
     } else {
         for (p = 0; p < s->nb_planes; p++) {
-            uint16_t *dst = (uint16_t *)out->data[p];
+            const int slice_start = (s->height[p] * jobnr) / nb_jobs;
+            const int slice_end = (s->height[p] * (jobnr+1)) / nb_jobs;
+            uint16_t *dst = (uint16_t *)(out->data[p] + slice_start * out->linesize[p]);
 
-            for (y = 0; y < s->height[p]; y++) {
+            for (y = slice_start; y < slice_end; y++) {
                 for (x = 0; x < s->linesize[p] / 2; 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);
+                        uint16_t src = AV_RN16(in[i]->data[p] + y * in[i]->linesize[p] + x * 2);
 
                         val += src * s->weights[i];
                     }
 
-                    dst[x] = val * s->wfactor;
+                    dst[x] = av_clip(val * s->wfactor, 0, s->max);
                 }
 
                 dst += out->linesize[p] / 2;
             }
         }
     }
+
+    return 0;
 }
 
 static int process_frame(FFFrameSync *fs)
@@ -168,6 +193,7 @@ static int process_frame(FFFrameSync *fs)
     MixContext *s = fs->opaque;
     AVFrame **in = s->frames;
     AVFrame *out;
+    ThreadData td;
     int i, ret;
 
     for (i = 0; i < s->nb_inputs; i++) {
@@ -180,7 +206,9 @@ static int process_frame(FFFrameSync *fs)
         return AVERROR(ENOMEM);
     out->pts = av_rescale_q(s->fs.pts, s->fs.time_base, outlink->time_base);
 
-    mix_frames(s, in, out);
+    td.in = in;
+    td.out = out;
+    ctx->internal->execute(ctx, mix_frames, &td, NULL, FFMIN(s->height[0], ff_filter_get_nb_threads(ctx)));
 
     return ff_filter_frame(outlink, out);
 }
@@ -211,6 +239,7 @@ static int config_output(AVFilterLink *outlink)
         return AVERROR_BUG;
     s->nb_planes = av_pix_fmt_count_planes(outlink->format);
     s->depth = s->desc->comp[0].depth;
+    s->max = (1 << s->depth) - 1;
 
     if ((ret = av_image_fill_linesizes(s->linesize, inlink->format, inlink->w)) < 0)
         return ret;
@@ -275,6 +304,7 @@ static int activate(AVFilterContext *ctx)
 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 },
+    { "scale", "set scale", OFFSET(scale), AV_OPT_TYPE_FLOAT, {.dbl=0}, 0, INT16_MAX, .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" },
@@ -304,7 +334,7 @@ AVFilter ff_vf_mix = {
     .init          = init,
     .uninit        = uninit,
     .activate      = activate,
-    .flags         = AVFILTER_FLAG_DYNAMIC_INPUTS,
+    .flags         = AVFILTER_FLAG_DYNAMIC_INPUTS | AVFILTER_FLAG_SLICE_THREADS,
 };
 
 #endif /* CONFIG_MIX_FILTER */
@@ -315,6 +345,7 @@ static int tmix_filter_frame(AVFilterLink *inlink, AVFrame *in)
     AVFilterContext *ctx = inlink->dst;
     AVFilterLink *outlink = ctx->outputs[0];
     MixContext *s = ctx->priv;
+    ThreadData td;
     AVFrame *out;
 
     if (s->nb_frames < s->nb_inputs) {
@@ -332,7 +363,9 @@ static int tmix_filter_frame(AVFilterLink *inlink, AVFrame *in)
         return AVERROR(ENOMEM);
     out->pts = s->frames[0]->pts;
 
-    mix_frames(s, s->frames, out);
+    td.out = out;
+    td.in = s->frames;
+    ctx->internal->execute(ctx, mix_frames, &td, NULL, FFMIN(s->height[0], ff_filter_get_nb_threads(ctx)));
 
     return ff_filter_frame(outlink, out);
 }
@@ -340,6 +373,7 @@ static int tmix_filter_frame(AVFilterLink *inlink, AVFrame *in)
 static const AVOption tmix_options[] = {
     { "frames", "set number of successive frames to mix", OFFSET(nb_inputs), AV_OPT_TYPE_INT, {.i64=3}, 2, 128, .flags = FLAGS },
     { "weights", "set weight for each frame", OFFSET(weights_str), AV_OPT_TYPE_STRING, {.str="1 1 1"}, 0, 0, .flags = FLAGS },
+    { "scale", "set scale", OFFSET(scale), AV_OPT_TYPE_FLOAT, {.dbl=0}, 0, INT16_MAX, .flags = FLAGS },
     { NULL },
 };
 
@@ -364,6 +398,7 @@ AVFilter ff_vf_tmix = {
     .inputs        = inputs,
     .init          = init,
     .uninit        = uninit,
+    .flags         = AVFILTER_FLAG_SLICE_THREADS,
 };
 
 #endif /* CONFIG_TMIX_FILTER */