[FFmpeg-devel] avfilter: add xstack filter

Submitted by Paul B Mahol on Oct. 24, 2018, 7:04 p.m.

Details

Message ID 20181024190440.3707-1-onemda@gmail.com
State New
Headers show

Commit Message

Paul B Mahol Oct. 24, 2018, 7:04 p.m.
Signed-off-by: Paul B Mahol <onemda@gmail.com>
---
 doc/filters.texi         |  45 ++++++++++++
 libavfilter/Makefile     |   1 +
 libavfilter/allfilters.c |   1 +
 libavfilter/vf_stack.c   | 151 ++++++++++++++++++++++++++++++++++++---
 4 files changed, 187 insertions(+), 11 deletions(-)

Comments

Lou Logan Oct. 25, 2018, 10:34 p.m.
On Wed, Oct 24, 2018, at 11:04 AM, Paul B Mahol wrote:
> Signed-off-by: Paul B Mahol <onemda@gmail.com>
> ---
>  doc/filters.texi         |  45 ++++++++++++
>  libavfilter/Makefile     |   1 +
>  libavfilter/allfilters.c |   1 +
>  libavfilter/vf_stack.c   | 151 ++++++++++++++++++++++++++++++++++++---
>  4 files changed, 187 insertions(+), 11 deletions(-)
> 
> diff --git a/doc/filters.texi b/doc/filters.texi
> index 740eec670c..51e6a0fc1c 100644
> --- a/doc/filters.texi
> +++ b/doc/filters.texi
> @@ -17798,6 +17798,51 @@ Set the scaling dimension: @code{2} for 
> @code{2xBR}, @code{3} for
>  Default is @code{3}.
>  @end table
>  
> +@section xstack
> +Stack video inputs into custom layout.
> +
> +All streams must be of same pixel format.
> +
> +The filter accept the following option:
> +
> +@table @option
> +@item inputs
> +Set number of input streams. Default is 2.
> +
> +@item layout
> +Specify layout of inputs. This option needs to be always set.

This option requires the desired layout configuration to be explicitly set by the user.

> +This sets position of each video input in output. Each input
> +is separated by '|'. First number represents column and second row,
> +numbers are separated by '_'.

This sentence is confusing and can be made clearer by breaking it apart:

  The first number represents the column, and the second number represents the row.
  Numbers start at 0 and are separated by '_'.

Everything else LGTM.
Moritz Barsnick Oct. 26, 2018, 11:44 a.m.
On Thu, Oct 25, 2018 at 14:34:49 -0800, Lou Logan wrote:
> On Wed, Oct 24, 2018, at 11:04 AM, Paul B Mahol wrote:

Additional nit:

> > +The filter accept the following option:
                ^ accepts

Otherwise I agree with Lou's corrections.

Practical filter btw., should save users from a lot of fiddling with
padding and math.

Moritz
Gyan Nov. 1, 2018, 7:48 a.m.
On Thu, Oct 25, 2018 at 12:35 AM Paul B Mahol <onemda@gmail.com> wrote:

>
> +        if (s->is_horizontal || s->is_vertical) {
> +            if ((ret = av_image_fill_linesizes(linesize, inlink->format,
> inlink->w)) < 0) {
> +                av_frame_free(&out);
> +                return ret;
> +            }
>

 Nice filter. One unfortunate side-effect is that unused parts of the
canvas are zero-ed which in YUV formats is a shade of dark green. Users
will generally want black. Would be nice to add a color option which fills
the canvas before av_image_copy_plane.

Gyan

Patch hide | download patch | download mbox

diff --git a/doc/filters.texi b/doc/filters.texi
index 740eec670c..51e6a0fc1c 100644
--- a/doc/filters.texi
+++ b/doc/filters.texi
@@ -17798,6 +17798,51 @@  Set the scaling dimension: @code{2} for @code{2xBR}, @code{3} for
 Default is @code{3}.
 @end table
 
+@section xstack
+Stack video inputs into custom layout.
+
+All streams must be of same pixel format.
+
+The filter accept the following option:
+
+@table @option
+@item inputs
+Set number of input streams. Default is 2.
+
+@item layout
+Specify layout of inputs. This option needs to be always set.
+This sets position of each video input in output. Each input
+is separated by '|'. First number represents column and second row,
+numbers are separated by '_'. Optionally one can use wX and hX,
+where X is video input from which to take width or height.
+Multiple values can be used when separated by '+'. In such
+case values are summed together.
+
+@item shortest
+If set to 1, force the output to terminate when the shortest input
+terminates. Default value is 0.
+@end table
+
+@subsection Examples
+
+@itemize
+@item
+Display 4 inputs into 2x2 grid,
+note that if inputs are of different sizes unused gaps might appear,
+as not all of output video is used.
+@example
+xstack=inputs=4:layout=0_0|0_h0|w0_0|w0_h0
+@end example
+
+@item
+Display 4 inputs into 1x4 grid,
+note that if inputs are of different sizes unused gaps might appear,
+as not all of output video is used.
+@example
+xstack=inputs=4:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2
+@end example
+@end itemize
+
 @anchor{yadif}
 @section yadif
 
diff --git a/libavfilter/Makefile b/libavfilter/Makefile
index 7beec310f8..a98c64b7ce 100644
--- a/libavfilter/Makefile
+++ b/libavfilter/Makefile
@@ -405,6 +405,7 @@  OBJS-$(CONFIG_W3FDIF_FILTER)                 += vf_w3fdif.o
 OBJS-$(CONFIG_WAVEFORM_FILTER)               += vf_waveform.o
 OBJS-$(CONFIG_WEAVE_FILTER)                  += vf_weave.o
 OBJS-$(CONFIG_XBR_FILTER)                    += vf_xbr.o
+OBJS-$(CONFIG_XSTACK_FILTER)                 += vf_stack.o framesync.o
 OBJS-$(CONFIG_YADIF_FILTER)                  += vf_yadif.o
 OBJS-$(CONFIG_ZMQ_FILTER)                    += f_zmq.o
 OBJS-$(CONFIG_ZOOMPAN_FILTER)                += vf_zoompan.o
diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
index 6f1e7cfb7d..b2cb58fc38 100644
--- a/libavfilter/allfilters.c
+++ b/libavfilter/allfilters.c
@@ -386,6 +386,7 @@  extern AVFilter ff_vf_w3fdif;
 extern AVFilter ff_vf_waveform;
 extern AVFilter ff_vf_weave;
 extern AVFilter ff_vf_xbr;
+extern AVFilter ff_vf_xstack;
 extern AVFilter ff_vf_yadif;
 extern AVFilter ff_vf_zmq;
 extern AVFilter ff_vf_zoompan;
diff --git a/libavfilter/vf_stack.c b/libavfilter/vf_stack.c
index b2b8c68041..8731674aa7 100644
--- a/libavfilter/vf_stack.c
+++ b/libavfilter/vf_stack.c
@@ -29,14 +29,23 @@ 
 #include "framesync.h"
 #include "video.h"
 
+typedef struct StackItem {
+    int x[4], y[4];
+    int linesize[4];
+    int height[4];
+} StackItem;
+
 typedef struct StackContext {
     const AVClass *class;
     const AVPixFmtDescriptor *desc;
     int nb_inputs;
+    char *layout;
     int shortest;
     int is_vertical;
+    int is_horizontal;
     int nb_planes;
 
+    StackItem *items;
     AVFrame **frames;
     FFFrameSync fs;
 } StackContext;
@@ -66,10 +75,19 @@  static av_cold int init(AVFilterContext *ctx)
     if (!strcmp(ctx->filter->name, "vstack"))
         s->is_vertical = 1;
 
+    if (!strcmp(ctx->filter->name, "hstack"))
+        s->is_horizontal = 1;
+
     s->frames = av_calloc(s->nb_inputs, sizeof(*s->frames));
     if (!s->frames)
         return AVERROR(ENOMEM);
 
+    if (!strcmp(ctx->filter->name, "xstack")) {
+        s->items = av_calloc(s->nb_inputs, sizeof(*s->items));
+        if (!s->items)
+            return AVERROR(ENOMEM);
+    }
+
     for (i = 0; i < s->nb_inputs; i++) {
         AVFilterPad pad = { 0 };
 
@@ -112,13 +130,15 @@  static int process_frame(FFFrameSync *fs)
         int linesize[4];
         int height[4];
 
-        if ((ret = av_image_fill_linesizes(linesize, inlink->format, inlink->w)) < 0) {
-            av_frame_free(&out);
-            return ret;
-        }
+        if (s->is_horizontal || s->is_vertical) {
+            if ((ret = av_image_fill_linesizes(linesize, inlink->format, inlink->w)) < 0) {
+                av_frame_free(&out);
+                return ret;
+            }
 
-        height[1] = height[2] = AV_CEIL_RSHIFT(inlink->h, s->desc->log2_chroma_h);
-        height[0] = height[3] = inlink->h;
+            height[1] = height[2] = AV_CEIL_RSHIFT(inlink->h, s->desc->log2_chroma_h);
+            height[0] = height[3] = inlink->h;
+        }
 
         for (p = 0; p < s->nb_planes; p++) {
             if (s->is_vertical) {
@@ -128,13 +148,21 @@  static int process_frame(FFFrameSync *fs)
                                     in[i]->linesize[p],
                                     linesize[p], height[p]);
                 offset[p] += height[p];
-            } else {
+            } else if (s->is_horizontal) {
                 av_image_copy_plane(out->data[p] + offset[p],
                                     out->linesize[p],
                                     in[i]->data[p],
                                     in[i]->linesize[p],
                                     linesize[p], height[p]);
                 offset[p] += linesize[p];
+            } else {
+                StackItem *item = &s->items[i];
+
+                av_image_copy_plane(out->data[p] + out->linesize[p] * item->y[p] + item->x[p],
+                                    out->linesize[p],
+                                    in[i]->data[p],
+                                    in[i]->linesize[p],
+                                    item->linesize[p], item->height[p]);
             }
         }
     }
@@ -154,6 +182,10 @@  static int config_output(AVFilterLink *outlink)
     FFFrameSyncIn *in;
     int i, ret;
 
+    s->desc = av_pix_fmt_desc_get(outlink->format);
+    if (!s->desc)
+        return AVERROR_BUG;
+
     if (s->is_vertical) {
         for (i = 1; i < s->nb_inputs; i++) {
             if (ctx->inputs[i]->w != width) {
@@ -162,7 +194,7 @@  static int config_output(AVFilterLink *outlink)
             }
             height += ctx->inputs[i]->h;
         }
-    } else {
+    } else if (s->is_horizontal) {
         for (i = 1; i < s->nb_inputs; i++) {
             if (ctx->inputs[i]->h != height) {
                 av_log(ctx, AV_LOG_ERROR, "Input %d height %d does not match input %d height %d.\n", i, ctx->inputs[i]->h, 0, height);
@@ -170,11 +202,81 @@  static int config_output(AVFilterLink *outlink)
             }
             width += ctx->inputs[i]->w;
         }
+    } else {
+        char *arg, *p = s->layout, *saveptr = NULL;
+        char *arg2, *p2, *saveptr2 = NULL;
+        char *arg3, *p3, *saveptr3 = NULL;
+        int inw, inh, size;
+
+        for (i = 0; i < s->nb_inputs; i++) {
+            AVFilterLink *inlink = ctx->inputs[i];
+            StackItem *item = &s->items[i];
+
+            if (!(arg = av_strtok(p, "|", &saveptr)))
+                return AVERROR(EINVAL);
+
+            p = NULL;
+
+            if ((ret = av_image_fill_linesizes(item->linesize, inlink->format, inlink->w)) < 0) {
+                return ret;
+            }
+
+            item->height[1] = item->height[2] = AV_CEIL_RSHIFT(inlink->h, s->desc->log2_chroma_h);
+            item->height[0] = item->height[3] = inlink->h;
+
+            p2 = arg;
+            inw = inh = 0;
+
+            for (int j = 0; j < 2; j++) {
+                if (!(arg2 = av_strtok(p2, "_", &saveptr2)))
+                    return AVERROR(EINVAL);
+
+                p2 = NULL;
+                p3 = arg2;
+                while ((arg3 = av_strtok(p3, "+", &saveptr3))) {
+                    p3 = NULL;
+                    if (sscanf(arg3, "w%d", &size) == 1) {
+                        if (size == i || size < 0 || size >= s->nb_inputs)
+                            return AVERROR(EINVAL);
+
+                        if (!j)
+                            inw += ctx->inputs[size]->w;
+                        else
+                            inh += ctx->inputs[size]->w;
+                    } else if (sscanf(arg3, "h%d", &size) == 1) {
+                        if (size == i || size < 0 || size >= s->nb_inputs)
+                            return AVERROR(EINVAL);
+
+                        if (!j)
+                            inw += ctx->inputs[size]->h;
+                        else
+                            inh += ctx->inputs[size]->h;
+                    } else if (sscanf(arg3, "%d", &size) == 1) {
+                        if (size < 0)
+                            return AVERROR(EINVAL);
+
+                        if (!j)
+                            inw += size;
+                        else
+                            inh += size;
+                    } else {
+                        return AVERROR(EINVAL);
+                    }
+                }
+            }
+
+            if ((ret = av_image_fill_linesizes(item->x, inlink->format, inw)) < 0) {
+                return ret;
+            }
+
+            item->y[1] = item->y[2] = AV_CEIL_RSHIFT(inh, s->desc->log2_chroma_h);
+            item->y[0] = item->y[3] = inh;
+
+            width  = FFMAX(width,  inlink->w + inw);
+            height = FFMAX(height, inlink->h + inh);
+        }
     }
 
-    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);
 
     outlink->w          = width;
@@ -209,6 +311,7 @@  static av_cold void uninit(AVFilterContext *ctx)
 
     ff_framesync_uninit(&s->fs);
     av_freep(&s->frames);
+    av_freep(&s->items);
 
     for (i = 0; i < ctx->nb_inputs; i++)
         av_freep(&ctx->input_pads[i].name);
@@ -276,3 +379,29 @@  AVFilter ff_vf_vstack = {
 };
 
 #endif /* CONFIG_VSTACK_FILTER */
+
+#if CONFIG_XSTACK_FILTER
+
+static const AVOption xstack_options[] = {
+    { "inputs", "set number of inputs", OFFSET(nb_inputs), AV_OPT_TYPE_INT, {.i64=2}, 2, INT_MAX, .flags = FLAGS },
+    { "layout", "set custom layout", OFFSET(layout), AV_OPT_TYPE_STRING, {.str="0_0|w0_0"}, 0, 0, .flags = FLAGS },
+    { "shortest", "force termination when the shortest input terminates", OFFSET(shortest), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, .flags = FLAGS },
+    { NULL },
+};
+
+AVFILTER_DEFINE_CLASS(xstack);
+
+AVFilter ff_vf_xstack = {
+    .name          = "xstack",
+    .description   = NULL_IF_CONFIG_SMALL("Stack video inputs into custom layout."),
+    .priv_size     = sizeof(StackContext),
+    .priv_class    = &xstack_class,
+    .query_formats = query_formats,
+    .outputs       = outputs,
+    .init          = init,
+    .uninit        = uninit,
+    .activate      = activate,
+    .flags         = AVFILTER_FLAG_DYNAMIC_INPUTS,
+};
+
+#endif /* CONFIG_XSTACK_FILTER */