lavfi: add haldclut filter.

Fixes Ticket #2517.

6 files changed, 309 insertions, 16 deletions

diff --git a/Changelog b/Changelog
index f42e5e3cdc..e3d6ff6ed9 100644
--- a/Changelog
+++ b/Changelog
@@ -55,6 +55,7 @@ version <next>:
 - 3D LUT filter (lut3d)
 - SMPTE 302M audio encoder
 - support for slice multithreading in libavfilter
+- Hald CLUT support (generation and filtering)
 
 
 version 1.2:
diff --git a/doc/filters.texi b/doc/filters.texi
index ff4fa1a8bc..75a2d6a88d 100644
--- a/doc/filters.texi
+++ b/doc/filters.texi
@@ -4177,6 +4177,79 @@ gradfun=radius=8
 
 @end itemize
 
+@anchor{haldclut}
+@section haldclut
+
+Apply a Hald CLUT to a video stream.
+
+First input is the video stream to process, and second one is the Hald CLUT.
+The Hald CLUT input can be a simple picture or a complete video stream.
+
+The filter accepts the following options:
+
+@table @option
+@item shortest
+Force termination when the shortest input terminates. Default is @code{0}.
+@item repeatlast
+Continue applying the last CLUT after the end of the stream. A value of
+@code{0} disable the filter after the last frame of the CLUT is reached.
+Default is @code{1}.
+@end table
+
+@code{haldclut} also has the same interpolation options as @ref{lut3d} (both
+filters share the same internals).
+
+More information about the Hald CLUT can be found on Eskil Steenberg's website
+(Hald CLUT author) at @url{http://www.quelsolaar.com/technology/clut.html}.
+
+@subsection Workflow examples
+
+@subsubsection Hald CLUT video stream
+
+Generate an identity Hald CLUT stream altered with various effects:
+@example
+ffmpeg -f lavfi -i @ref{haldclutsrc}=8 -vf "hue=H=2*PI*t:s=sin(2*PI*t)+1, curves=cross_process" -t 10 -c:v ffv1 clut.nut
+@end example
+
+Note: make sure you use a lossless codec.
+
+Then use it with @code{haldclut} to apply it on some random stream:
+@example
+ffmpeg -f lavfi -i mandelbrot -i clut.nut -filter_complex '[0][1] haldclut' -t 20 mandelclut.mkv
+@end example
+
+The Hald CLUT will be applied to the 10 first seconds (duration of
+@file{clut.nut}), then the latest picture of that CLUT stream will be applied
+to the remaining frames of the @code{mandelbrot} stream.
+
+@subsubsection Hald CLUT with preview
+
+A Hald CLUT is supposed to be a squared image of @code{Level*Level*Level} by
+@code{Level*Level*Level} pixels. For a given Hald CLUT, FFmpeg will select the
+biggest possible square starting at the top left of the picture. The remaining
+padding pixels (bottom or right) will be ignored. This area can be used to add
+a preview of the Hald CLUT.
+
+Typically, the following generated Hald CLUT will be supported by the
+@code{haldclut} filter:
+
+@example
+ffmpeg -f lavfi -i @ref{haldclutsrc}=8 -vf "
+   pad=iw+320 [padded_clut];
+   smptebars=s=320x256, split [a][b];
+   [padded_clut][a] overlay=W-320:h, curves=color_negative [main];
+   [main][b] overlay=W-320" -frames:v 1 clut.png
+@end example
+
+It contains the original and a preview of the effect of the CLUT: SMPTE color
+bars are displayed on the right-top, and below the same color bars processed by
+the color changes.
+
+Then, the effect of this Hald CLUT can be visualized with:
+@example
+ffplay input.mkv -vf "movie=clut.png, [in] haldclut"
+@end example
+
 @section hflip
 
 Flip the input video horizontally.
@@ -4603,6 +4676,7 @@ kerndeint=map=1
 @end example
 @end itemize
 
+@anchor{lut3d}
 @section lut3d
 
 Apply a 3D LUT to an input video.
@@ -7401,11 +7475,19 @@ ffplay -f lavfi life=s=300x200:mold=10:r=60:ratio=0.1:death_color=#C83232:life_c
 @end example
 @end itemize
 
+@anchor{color}
+@anchor{haldclutsrc}
+@anchor{nullsrc}
+@anchor{rgbtestsrc}
+@anchor{smptebars}
+@anchor{smptehdbars}
+@anchor{testsrc}
 @section color, haldclutsrc, nullsrc, rgbtestsrc, smptebars, smptehdbars, testsrc
 
 The @code{color} source provides an uniformly colored input.
 
-The @code{haldclutsrc} source provides an identity Hald CLUT.
+The @code{haldclutsrc} source provides an identity Hald CLUT. See also
+@ref{haldclut} filter.
 
 The @code{nullsrc} source returns unprocessed video frames. It is
 mainly useful to be employed in analysis / debugging tools, or as the
diff --git a/libavfilter/Makefile b/libavfilter/Makefile
index 9682cba216..4ee28bfb63 100644
--- a/libavfilter/Makefile
+++ b/libavfilter/Makefile
@@ -135,6 +135,7 @@ OBJS-$(CONFIG_FPS_FILTER)                    += vf_fps.o
 OBJS-$(CONFIG_FREI0R_FILTER)                 += vf_frei0r.o
 OBJS-$(CONFIG_GEQ_FILTER)                    += vf_geq.o
 OBJS-$(CONFIG_GRADFUN_FILTER)                += vf_gradfun.o
+OBJS-$(CONFIG_HALDCLUT_FILTER)               += vf_lut3d.o dualinput.o
 OBJS-$(CONFIG_HFLIP_FILTER)                  += vf_hflip.o
 OBJS-$(CONFIG_HISTEQ_FILTER)                 += vf_histeq.o
 OBJS-$(CONFIG_HISTOGRAM_FILTER)              += vf_histogram.o
diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
index 075eccade4..58bffc7fa3 100644
--- a/libavfilter/allfilters.c
+++ b/libavfilter/allfilters.c
@@ -133,6 +133,7 @@ void avfilter_register_all(void)
     REGISTER_FILTER(FREI0R,         frei0r,         vf);
     REGISTER_FILTER(GEQ,            geq,            vf);
     REGISTER_FILTER(GRADFUN,        gradfun,        vf);
+    REGISTER_FILTER(HALDCLUT,       haldclut,       vf);
     REGISTER_FILTER(HFLIP,          hflip,          vf);
     REGISTER_FILTER(HISTEQ,         histeq,         vf);
     REGISTER_FILTER(HISTOGRAM,      histogram,      vf);
diff --git a/libavfilter/version.h b/libavfilter/version.h
index fa4c1ddfab..51e13b99e4 100644
--- a/libavfilter/version.h
+++ b/libavfilter/version.h
@@ -30,7 +30,7 @@
 #include "libavutil/avutil.h"
 
 #define LIBAVFILTER_VERSION_MAJOR  3
-#define LIBAVFILTER_VERSION_MINOR  71
+#define LIBAVFILTER_VERSION_MINOR  72
 #define LIBAVFILTER_VERSION_MICRO 100
 
 #define LIBAVFILTER_VERSION_INT AV_VERSION_INT(LIBAVFILTER_VERSION_MAJOR, \
diff --git a/libavfilter/vf_lut3d.c b/libavfilter/vf_lut3d.c
index df674d7f26..9a92043a60 100644
--- a/libavfilter/vf_lut3d.c
+++ b/libavfilter/vf_lut3d.c
@@ -31,6 +31,7 @@
 #include "libavutil/avstring.h"
 #include "avfilter.h"
 #include "drawutils.h"
+#include "dualinput.h"
 #include "formats.h"
 #include "internal.h"
 #include "video.h"
@@ -51,7 +52,9 @@ struct rgbvec {
     float r, g, b;
 };
 
-#define MAX_LEVEL 36
+/* 3D LUT don't often go up to level 32, but it is common to have a Hald CLUT
+ * of 512x512 (64x64x64) */
+#define MAX_LEVEL 64
 
 typedef struct LUT3DContext {
     const AVClass *class;
@@ -64,20 +67,23 @@ typedef struct LUT3DContext {
     struct rgbvec (*interp_16)(const struct LUT3DContext*, uint16_t, uint16_t, uint16_t);
     struct rgbvec lut[MAX_LEVEL][MAX_LEVEL][MAX_LEVEL];
     int lutsize;
+#if CONFIG_HALDCLUT_FILTER
+    uint8_t clut_rgba_map[4];
+    int clut_step;
+    int clut_is16bit;
+    int clut_width;
+    FFDualInputContext dinput;
+#endif
 } LUT3DContext;
 
 #define OFFSET(x) offsetof(LUT3DContext, x)
 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
-static const AVOption lut3d_options[] = {
-    { "file",   "set 3D LUT file name", OFFSET(file), AV_OPT_TYPE_STRING, {.str=NULL}, .flags = FLAGS },
-    { "interp", "select interpolation mode", OFFSET(interpolation), AV_OPT_TYPE_INT, {.i64=INTERPOLATE_TETRAHEDRAL}, 0, NB_INTERP_MODE-1, FLAGS, "interp_mode" },
-        { "nearest",     "use values from the nearest defined points",            0, AV_OPT_TYPE_CONST, {.i64=INTERPOLATE_NEAREST},     INT_MIN, INT_MAX, FLAGS, "interp_mode" },
-        { "trilinear",   "interpolate values using the 8 points defining a cube", 0, AV_OPT_TYPE_CONST, {.i64=INTERPOLATE_TRILINEAR},   INT_MIN, INT_MAX, FLAGS, "interp_mode" },
-        { "tetrahedral", "interpolate values using a tetrahedron",                0, AV_OPT_TYPE_CONST, {.i64=INTERPOLATE_TETRAHEDRAL}, INT_MIN, INT_MAX, FLAGS, "interp_mode" },
+#define COMMON_OPTIONS \
+    { "interp", "select interpolation mode", OFFSET(interpolation), AV_OPT_TYPE_INT, {.i64=INTERPOLATE_TETRAHEDRAL}, 0, NB_INTERP_MODE-1, FLAGS, "interp_mode" }, \
+        { "nearest",     "use values from the nearest defined points",            0, AV_OPT_TYPE_CONST, {.i64=INTERPOLATE_NEAREST},     INT_MIN, INT_MAX, FLAGS, "interp_mode" }, \
+        { "trilinear",   "interpolate values using the 8 points defining a cube", 0, AV_OPT_TYPE_CONST, {.i64=INTERPOLATE_TRILINEAR},   INT_MIN, INT_MAX, FLAGS, "interp_mode" }, \
+        { "tetrahedral", "interpolate values using a tetrahedron",                0, AV_OPT_TYPE_CONST, {.i64=INTERPOLATE_TETRAHEDRAL}, INT_MIN, INT_MAX, FLAGS, "interp_mode" }, \
     { NULL }
-};
-
-AVFILTER_DEFINE_CLASS(lut3d);
 
 static inline float lerpf(float v0, float v1, float f)
 {
@@ -404,7 +410,9 @@ static void set_identity_matrix(LUT3DContext *lut3d, int size)
     }
 }
 
-static av_cold int init(AVFilterContext *ctx)
+#if CONFIG_LUT3D_FILTER
+/* TODO: move to the CONFIG_LUT3D_FILTER definition scope at the bottom */
+static av_cold int lut3d_init(AVFilterContext *ctx)
 {
     int ret;
     FILE *f;
@@ -454,6 +462,7 @@ end:
     fclose(f);
     return ret;
 }
+#endif
 
 static int query_formats(AVFilterContext *ctx)
 {
@@ -523,7 +532,7 @@ static int config_input(AVFilterLink *inlink)
     }                                                                                               \
 } while (0)
 
-static int filter_frame(AVFilterLink *inlink, AVFrame *in)
+static AVFrame *apply_lut(AVFilterLink *inlink, AVFrame *in)
 {
     int x, y, direct = 0;
     AVFilterContext *ctx = inlink->dst;
@@ -543,7 +552,7 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *in)
         out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
         if (!out) {
             av_frame_free(&in);
-            return AVERROR(ENOMEM);
+            return NULL;
         }
         av_frame_copy_props(out, in);
     }
@@ -554,9 +563,26 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *in)
     if (!direct)
         av_frame_free(&in);
 
+    return out;
+}
+
+static int filter_frame(AVFilterLink *inlink, AVFrame *in)
+{
+    AVFilterLink *outlink = inlink->dst->outputs[0];
+    AVFrame *out = apply_lut(inlink, in);
+    if (!out)
+        return AVERROR(ENOMEM);
     return ff_filter_frame(outlink, out);
 }
 
+#if CONFIG_LUT3D_FILTER
+static const AVOption lut3d_options[] = {
+    { "file", "set 3D LUT file name", OFFSET(file), AV_OPT_TYPE_STRING, {.str=NULL}, .flags = FLAGS },
+    COMMON_OPTIONS
+};
+
+AVFILTER_DEFINE_CLASS(lut3d);
+
 static const AVFilterPad lut3d_inputs[] = {
     {
         .name         = "default",
@@ -579,10 +605,192 @@ AVFilter avfilter_vf_lut3d = {
     .name          = "lut3d",
     .description   = NULL_IF_CONFIG_SMALL("Adjust colors using a 3D LUT."),
     .priv_size     = sizeof(LUT3DContext),
-    .init          = init,
+    .init          = lut3d_init,
     .query_formats = query_formats,
     .inputs        = lut3d_inputs,
     .outputs       = lut3d_outputs,
     .priv_class    = &lut3d_class,
     .flags         = AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC,
 };
+#endif
+
+#if CONFIG_HALDCLUT_FILTER
+
+static void update_clut(LUT3DContext *lut3d, const AVFrame *frame)
+{
+    const uint8_t *data = frame->data[0];
+    const int linesize  = frame->linesize[0];
+    const int w = lut3d->clut_width;
+    const int step = lut3d->clut_step;
+    const uint8_t *rgba_map = lut3d->clut_rgba_map;
+    const int level = lut3d->lutsize;
+
+#define LOAD_CLUT(nbits) do {                                           \
+    int i, j, k, x = 0, y = 0;                                          \
+                                                                        \
+    for (k = 0; k < level; k++) {                                       \
+        for (j = 0; j < level; j++) {                                   \
+            for (i = 0; i < level; i++) {                               \
+                const uint##nbits##_t *src = (const uint##nbits##_t *)  \
+                    (data + y*linesize + x*step);                       \
+                struct rgbvec *vec = &lut3d->lut[k][j][i];              \
+                vec->r = src[rgba_map[0]] / (float)((1<<(nbits)) - 1);  \
+                vec->g = src[rgba_map[1]] / (float)((1<<(nbits)) - 1);  \
+                vec->b = src[rgba_map[2]] / (float)((1<<(nbits)) - 1);  \
+                if (++x == w) {                                         \
+                    x = 0;                                              \
+                    y++;                                                \
+                }                                                       \
+            }                                                           \
+        }                                                               \
+    }                                                                   \
+} while (0)
+
+    if (!lut3d->clut_is16bit) LOAD_CLUT(8);
+    else                      LOAD_CLUT(16);
+}
+
+
+static int config_output(AVFilterLink *outlink)
+{
+    AVFilterContext *ctx = outlink->src;
+
+    outlink->w = ctx->inputs[0]->w;
+    outlink->h = ctx->inputs[0]->h;
+    outlink->time_base = ctx->inputs[0]->time_base;
+    return 0;
+}
+
+static int filter_frame_main(AVFilterLink *inlink, AVFrame *inpicref)
+{
+    LUT3DContext *s = inlink->dst->priv;
+    return ff_dualinput_filter_frame_main(&s->dinput, inlink, inpicref);
+}
+
+static int filter_frame_clut(AVFilterLink *inlink, AVFrame *inpicref)
+{
+    LUT3DContext *s = inlink->dst->priv;
+    return ff_dualinput_filter_frame_second(&s->dinput, inlink, inpicref);
+}
+
+static int request_frame(AVFilterLink *outlink)
+{
+    LUT3DContext *s = outlink->src->priv;
+    return ff_dualinput_request_frame(&s->dinput, outlink);
+}
+
+static int config_clut(AVFilterLink *inlink)
+{
+    int size, level, w, h;
+    AVFilterContext *ctx = inlink->dst;
+    LUT3DContext *lut3d = ctx->priv;
+    const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(inlink->format);
+
+    lut3d->clut_is16bit = 0;
+    switch (inlink->format) {
+    case AV_PIX_FMT_RGB48:
+    case AV_PIX_FMT_BGR48:
+    case AV_PIX_FMT_RGBA64:
+    case AV_PIX_FMT_BGRA64:
+        lut3d->clut_is16bit = 1;
+    }
+
+    lut3d->clut_step = av_get_padded_bits_per_pixel(desc) >> 3;
+    ff_fill_rgba_map(lut3d->clut_rgba_map, inlink->format);
+
+    if (inlink->w > inlink->h)
+        av_log(ctx, AV_LOG_INFO, "Padding on the right (%dpx) of the "
+               "Hald CLUT will be ignored\n", inlink->w - inlink->h);
+    else if (inlink->w < inlink->h)
+        av_log(ctx, AV_LOG_INFO, "Padding at the bottom (%dpx) of the "
+               "Hald CLUT will be ignored\n", inlink->h - inlink->w);
+    lut3d->clut_width = w = h = FFMIN(inlink->w, inlink->h);
+
+    for (level = 1; level*level*level < w; level++);
+    size = level*level*level;
+    if (size != w) {
+        av_log(ctx, AV_LOG_WARNING, "The Hald CLUT width does not match the level\n");
+        return AVERROR_INVALIDDATA;
+    }
+    av_assert0(w == h && w == size);
+    level *= level;
+    if (level > MAX_LEVEL) {
+        const int max_clut_level = sqrt(MAX_LEVEL);
+        const int max_clut_size  = max_clut_level*max_clut_level*max_clut_level;
+        av_log(ctx, AV_LOG_ERROR, "Too large Hald CLUT "
+               "(maximum level is %d, or %dx%d CLUT)\n",
+               max_clut_level, max_clut_size, max_clut_size);
+        return AVERROR(EINVAL);
+    }
+    lut3d->lutsize = level;
+
+    return 0;
+}
+
+static AVFrame *update_apply_clut(AVFilterContext *ctx, AVFrame *main,
+                                  const AVFrame *second)
+{
+    AVFilterLink *inlink = ctx->inputs[0];
+    update_clut(ctx->priv, second);
+    return apply_lut(inlink, main);
+}
+
+static av_cold int haldclut_init(AVFilterContext *ctx)
+{
+    LUT3DContext *lut3d = ctx->priv;
+    lut3d->dinput.process = update_apply_clut;
+    return 0;
+}
+
+static av_cold void haldclut_uninit(AVFilterContext *ctx)
+{
+    LUT3DContext *lut3d = ctx->priv;
+    ff_dualinput_uninit(&lut3d->dinput);
+}
+
+static const AVOption haldclut_options[] = {
+    { "shortest",   "force termination when the shortest input terminates", OFFSET(dinput.shortest),   AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, FLAGS },
+    { "repeatlast", "continue applying the last clut after eos",            OFFSET(dinput.repeatlast), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, FLAGS },
+    COMMON_OPTIONS
+};
+
+AVFILTER_DEFINE_CLASS(haldclut);
+
+static const AVFilterPad haldclut_inputs[] = {
+    {
+        .name         = "main",
+        .type         = AVMEDIA_TYPE_VIDEO,
+        .filter_frame = filter_frame_main,
+        .config_props = config_input,
+    },{
+        .name         = "clut",
+        .type         = AVMEDIA_TYPE_VIDEO,
+        .filter_frame = filter_frame_clut,
+        .config_props = config_clut,
+    },
+    { NULL }
+};
+
+static const AVFilterPad haldclut_outputs[] = {
+    {
+        .name = "default",
+        .type = AVMEDIA_TYPE_VIDEO,
+        .request_frame = request_frame,
+        .config_props = config_output,
+    },
+    { NULL }
+};
+
+AVFilter avfilter_vf_haldclut = {
+    .name          = "haldclut",
+    .description   = NULL_IF_CONFIG_SMALL("Adjust colors using a Hald CLUT."),
+    .priv_size     = sizeof(LUT3DContext),
+    .init          = haldclut_init,
+    .uninit        = haldclut_uninit,
+    .query_formats = query_formats,
+    .inputs        = haldclut_inputs,
+    .outputs       = haldclut_outputs,
+    .priv_class    = &haldclut_class,
+    .flags         = AVFILTER_FLAG_SUPPORT_TIMELINE_INTERNAL,
+};
+#endif