[FFmpeg-cvslog] doc/filters: Document OpenCL program filters

Mark Thompson git at videolan.org
Mon Jan 8 01:51:22 EET 2018


ffmpeg | branch: master | Mark Thompson <sw at jkqxz.net> | Wed Jan  3 22:43:53 2018 +0000| [03c08d59fb2b998106829063b3840e63c2837923] | committer: Mark Thompson

doc/filters: Document OpenCL program filters

Include some example programs.

> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=03c08d59fb2b998106829063b3840e63c2837923
---

 doc/filters.texi | 202 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 202 insertions(+)

diff --git a/doc/filters.texi b/doc/filters.texi
index f6954c947c..a13aef2196 100644
--- a/doc/filters.texi
+++ b/doc/filters.texi
@@ -12524,6 +12524,136 @@ Set value which will be multiplied with filtered result.
 Set value which will be added to filtered result.
 @end table
 
+ at anchor{program_opencl}
+ at section program_opencl
+
+Filter video using an OpenCL program.
+
+ at table @option
+
+ at item source
+OpenCL program source file.
+
+ at item kernel
+Kernel name in program.
+
+ at item inputs
+Number of inputs to the filter.  Defaults to 1.
+
+ at item size, s
+Size of output frames.  Defaults to the same as the first input.
+
+ at end table
+
+The program source file must contain a kernel function with the given name,
+which will be run once for each plane of the output.  Each run on a plane
+gets enqueued as a separate 2D global NDRange with one work-item for each
+pixel to be generated.  The global ID offset for each work-item is therefore
+the coordinates of a pixel in the destination image.
+
+The kernel function needs to take the following arguments:
+ at itemize
+ at item
+Destination image, @var{__write_only image2d_t}.
+
+This image will become the output; the kernel should write all of it.
+ at item
+Frame index, @var{unsigned int}.
+
+This is a counter starting from zero and increasing by one for each frame.
+ at item
+Source images, @var{__read_only image2d_t}.
+
+These are the most recent images on each input.  The kernel may read from
+them to generate the output, but they can't be written to.
+ at end itemize
+
+Example programs:
+
+ at itemize
+ at item
+Copy the input to the output (output must be the same size as the input).
+ at verbatim
+__kernel void copy(__write_only image2d_t destination,
+                   unsigned int index,
+                   __read_only  image2d_t source)
+{
+    const sampler_t sampler = CLK_NORMALIZED_COORDS_FALSE;
+
+    int2 location = (int2)(get_global_id(0), get_global_id(1));
+
+    float4 value = read_imagef(source, sampler, location);
+
+    write_imagef(destination, location, value);
+}
+ at end verbatim
+
+ at item
+Apply a simple transformation, rotating the input by an amount increasing
+with the index counter.  Pixel values are linearly interpolated by the
+sampler, and the output need not have the same dimensions as the input.
+ at verbatim
+__kernel void rotate_image(__write_only image2d_t dst,
+                           unsigned int index,
+                           __read_only  image2d_t src)
+{
+    const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
+                               CLK_FILTER_LINEAR);
+
+    float angle = (float)index / 100.0f;
+
+    float2 dst_dim = convert_float2(get_image_dim(dst));
+    float2 src_dim = convert_float2(get_image_dim(src));
+
+    float2 dst_cen = dst_dim / 2.0f;
+    float2 src_cen = src_dim / 2.0f;
+
+    int2   dst_loc = (int2)(get_global_id(0), get_global_id(1));
+
+    float2 dst_pos = convert_float2(dst_loc) - dst_cen;
+    float2 src_pos = {
+        cos(angle) * dst_pos.x - sin(angle) * dst_pos.y,
+        sin(angle) * dst_pos.x + cos(angle) * dst_pos.y
+    };
+    src_pos = src_pos * src_dim / dst_dim;
+
+    float2 src_loc = src_pos + src_cen;
+
+    if (src_loc.x < 0.0f      || src_loc.y < 0.0f ||
+        src_loc.x > src_dim.x || src_loc.y > src_dim.y)
+        write_imagef(dst, dst_loc, 0.5f);
+    else
+        write_imagef(dst, dst_loc, read_imagef(src, sampler, src_loc));
+}
+ at end verbatim
+
+ at item
+Blend two inputs together, with the amount of each input used varying
+with the index counter.
+ at verbatim
+__kernel void blend_images(__write_only image2d_t dst,
+                           unsigned int index,
+                           __read_only  image2d_t src1,
+                           __read_only  image2d_t src2)
+{
+    const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
+                               CLK_FILTER_LINEAR);
+
+    float blend = (cos((float)index / 50.0f) + 1.0f) / 2.0f;
+
+    int2  dst_loc = (int2)(get_global_id(0), get_global_id(1));
+    int2 src1_loc = dst_loc * get_image_dim(src1) / get_image_dim(dst);
+    int2 src2_loc = dst_loc * get_image_dim(src2) / get_image_dim(dst);
+
+    float4 val1 = read_imagef(src1, sampler, src1_loc);
+    float4 val2 = read_imagef(src2, sampler, src2_loc);
+
+    write_imagef(dst, dst_loc, val1 * blend + val2 * (1.0f - blend));
+}
+ at end verbatim
+
+ at end itemize
+
 @section pseudocolor
 
 Alter frame colors in video with pseudocolors.
@@ -17498,6 +17628,78 @@ Set the color of the created image. Accepts the same syntax of the
 corresponding @option{color} option.
 @end table
 
+ at section openclsrc
+
+Generate video using an OpenCL program.
+
+ at table @option
+
+ at item source
+OpenCL program source file.
+
+ at item kernel
+Kernel name in program.
+
+ at item size, s
+Size of frames to generate.  This must be set.
+
+ at item format
+Pixel format to use for the generated frames.  This must be set.
+
+ at item rate, r
+Number of frames generated every second.  Default value is '25'.
+
+ at end table
+
+For details of how the program loading works, see the @ref{program_opencl}
+filter.
+
+Example programs:
+
+ at itemize
+ at item
+Generate a colour ramp by setting pixel values from the position of the pixel
+in the output image.  (Note that this will work with all pixel formats, but
+the generated output will not be the same.)
+ at verbatim
+__kernel void ramp(__write_only image2d_t dst,
+                   unsigned int index)
+{
+    int2 loc = (int2)(get_global_id(0), get_global_id(1));
+
+    float4 val;
+    val.xy = val.zw = convert_float2(loc) / convert_float2(get_image_dim(dst));
+
+    write_imagef(dst, loc, val);
+}
+ at end verbatim
+
+ at item
+Generate a Sierpinski carpet pattern, panning by a single pixel each frame.
+ at verbatim
+__kernel void sierpinski_carpet(__write_only image2d_t dst,
+                                unsigned int index)
+{
+    int2 loc = (int2)(get_global_id(0), get_global_id(1));
+
+    float4 value = 0.0f;
+    int x = loc.x + index;
+    int y = loc.y + index;
+    while (x > 0 || y > 0) {
+        if (x % 3 == 1 && y % 3 == 1) {
+            value = 1.0f;
+            break;
+        }
+        x /= 3;
+        y /= 3;
+    }
+
+    write_imagef(dst, loc, value);
+}
+ at end verbatim
+
+ at end itemize
+
 @c man end VIDEO SOURCES
 
 @chapter Video Sinks



More information about the ffmpeg-cvslog mailing list