[FFmpeg-cvslog] avfilter: added motion estimation and interpolation filters

Davinder Singh git at videolan.org
Fri Aug 26 12:21:26 EEST 2016


ffmpeg | branch: master | Davinder Singh <ds.mudhar at gmail.com> | Tue Aug 23 18:08:35 2016 +0530| [b07d4a0fb20db387e0aa516f1465517b083c4e49] | committer: Michael Niedermayer

avfilter: added motion estimation and interpolation filters

Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>

> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=b07d4a0fb20db387e0aa516f1465517b083c4e49
---

 doc/filters.texi                |  128 ++++
 libavfilter/Makefile            |    2 +
 libavfilter/allfilters.c        |    2 +
 libavfilter/motion_estimation.c |  432 ++++++++++++++
 libavfilter/motion_estimation.h |   87 +++
 libavfilter/version.h           |    2 +-
 libavfilter/vf_mestimate.c      |  377 ++++++++++++
 libavfilter/vf_minterpolate.c   | 1247 +++++++++++++++++++++++++++++++++++++++
 8 files changed, 2276 insertions(+), 1 deletion(-)

diff --git a/doc/filters.texi b/doc/filters.texi
index 4d1f5a2..a9b6ad1 100644
--- a/doc/filters.texi
+++ b/doc/filters.texi
@@ -9434,6 +9434,134 @@ format=rgb24,mergeplanes=0x000102:yuv444p
 @end example
 @end itemize
 
+ at section mestimate
+
+Estimate and export motion vectors using block matching algorithms.
+Motion vectors are stored in frame side data to be used by other filters.
+
+This filter accepts the following options:
+ at table @option
+ at item method
+Specify the motion estimation method. Accepts one of the following values:
+
+ at table @samp
+ at item esa
+Exhaustive search algorithm.
+ at item tss
+Three step search algorithm.
+ at item tdls
+Two dimensional logarithmic search algorithm.
+ at item ntss
+New three step search algorithm.
+ at item fss
+Four step search algorithm.
+ at item ds
+Diamond search algorithm.
+ at item hexbs
+Hexagon-based search algorithm.
+ at item epzs
+Enhanced predictive zonal search algorithm.
+ at item umh
+Uneven multi-hexagon search algorithm.
+ at end table
+Default value is @samp{esa}.
+
+ at item mb_size
+Macroblock size. Default @code{16}.
+
+ at item search_param
+Search parameter. Default @code{7}.
+ at end table
+
+ at section minterpolate
+
+Convert the video to specified frame rate using motion interpolation.
+
+This filter accepts the following options:
+ at table @option
+ at item fps
+Specify the output frame rate. This can be rational e.g. @code{60000/1001}. Frames are dropped if @var{fps} is lower than source fps. Default @code{60}.
+
+ at item mi_mode
+Motion interpolation mode. Following values are accepted:
+ at table @samp
+ at item dup
+Duplicate previous or next frame for interpolating new ones.
+ at item blend
+Blend source frames. Interpolated frame is mean of previous and next frames.
+ at item mci
+Motion compensated interpolation. Following options are effective when this mode is selected:
+
+ at table @samp
+ at item mc_mode
+Motion compensation mode. Following values are accepted:
+ at table @samp
+ at item obmc
+Overlapped block motion compensation.
+ at item aobmc
+Adaptive overlapped block motion compensation. Window weighting coefficients are controlled adaptively according to the reliabilities of the neighboring motion vectors to reduce oversmoothing.
+ at end table
+Default mode is @samp{obmc}.
+
+ at item me_mode
+Motion estimation mode. Following values are accepted:
+ at table @samp
+ at item bidir
+Bidirectional motion estimation. Motion vectors are estimated for each source frame in both forward and backward directions.
+ at item bilat
+Bilateral motion estimation. Motion vectors are estimated directly for interpolated frame.
+ at end table
+Default mode is @samp{bilat}.
+
+ at item me
+The algorithm to be used for motion estimation. Following values are accepted:
+ at table @samp
+ at item esa
+Exhaustive search algorithm.
+ at item tss
+Three step search algorithm.
+ at item tdls
+Two dimensional logarithmic search algorithm.
+ at item ntss
+New three step search algorithm.
+ at item fss
+Four step search algorithm.
+ at item ds
+Diamond search algorithm.
+ at item hexbs
+Hexagon-based search algorithm.
+ at item epzs
+Enhanced predictive zonal search algorithm.
+ at item umh
+Uneven multi-hexagon search algorithm.
+ at end table
+Default algorithm is @samp{epzs}.
+
+ at item mb_size
+Macroblock size. Default @code{16}.
+
+ at item search_param
+Motion estimation search parameter. Default @code{32}.
+
+ at item vsmbc
+Enable variable-size block motion compensation. Motion estimation is applied with smaller block sizes at object boundaries in order to make the them less blur. Default is @code{0} (disabled).
+ at end table
+ at end table
+
+ at item scd
+Scene change detection method. Scene change leads motion vectors to be in random direction. Scene change detection replace interpolated frames by duplicate ones. May not be needed for other modes. Following values are accepted:
+ at table @samp
+ at item none
+Disable scene change detection.
+ at item fdiff
+Frame difference. Corresponding pixel values are compared and if it statisfies @var{scd_threshold} scene change is detected.
+ at end table
+Default method is @samp{fdiff}.
+
+ at item scd_threshold
+Scene change detection threshold. Default is @code{5.0}.
+ at end table
+
 @section mpdecimate
 
 Drop frames that do not differ greatly from the previous frame in
diff --git a/libavfilter/Makefile b/libavfilter/Makefile
index d00b40c..9eb823e 100644
--- a/libavfilter/Makefile
+++ b/libavfilter/Makefile
@@ -213,7 +213,9 @@ OBJS-$(CONFIG_MASKEDCLAMP_FILTER)            += vf_maskedclamp.o framesync.o
 OBJS-$(CONFIG_MASKEDMERGE_FILTER)            += vf_maskedmerge.o framesync.o
 OBJS-$(CONFIG_MCDEINT_FILTER)                += vf_mcdeint.o
 OBJS-$(CONFIG_MERGEPLANES_FILTER)            += vf_mergeplanes.o framesync.o
+OBJS-$(CONFIG_MESTIMATE_FILTER)              += vf_mestimate.o motion_estimation.o
 OBJS-$(CONFIG_METADATA_FILTER)               += f_metadata.o
+OBJS-$(CONFIG_MINTERPOLATE_FILTER)           += vf_minterpolate.o motion_estimation.o
 OBJS-$(CONFIG_MPDECIMATE_FILTER)             += vf_mpdecimate.o
 OBJS-$(CONFIG_NEGATE_FILTER)                 += vf_lut.o
 OBJS-$(CONFIG_NNEDI_FILTER)                  += vf_nnedi.o
diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
index ca1bfb7..32afeb0 100644
--- a/libavfilter/allfilters.c
+++ b/libavfilter/allfilters.c
@@ -230,7 +230,9 @@ void avfilter_register_all(void)
     REGISTER_FILTER(MASKEDMERGE,    maskedmerge,    vf);
     REGISTER_FILTER(MCDEINT,        mcdeint,        vf);
     REGISTER_FILTER(MERGEPLANES,    mergeplanes,    vf);
+    REGISTER_FILTER(MESTIMATE,      mestimate,      vf);
     REGISTER_FILTER(METADATA,       metadata,       vf);
+    REGISTER_FILTER(MINTERPOLATE,   minterpolate,   vf);
     REGISTER_FILTER(MPDECIMATE,     mpdecimate,     vf);
     REGISTER_FILTER(NEGATE,         negate,         vf);
     REGISTER_FILTER(NNEDI,          nnedi,          vf);
diff --git a/libavfilter/motion_estimation.c b/libavfilter/motion_estimation.c
new file mode 100644
index 0000000..91cc003
--- /dev/null
+++ b/libavfilter/motion_estimation.c
@@ -0,0 +1,432 @@
+/**
+ * Copyright (c) 2016 Davinder Singh (DSM_) <ds.mudhar<@gmail.com>
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "motion_estimation.h"
+
+static const int8_t sqr1[8][2]  = {{ 0,-1}, { 0, 1}, {-1, 0}, { 1, 0}, {-1,-1}, {-1, 1}, { 1,-1}, { 1, 1}};
+static const int8_t dia1[4][2]  = {{-1, 0}, { 0,-1}, { 1, 0}, { 0, 1}};
+static const int8_t dia2[8][2]  = {{-2, 0}, {-1,-1}, { 0,-2}, { 1,-1}, { 2, 0}, { 1, 1}, { 0, 2}, {-1, 1}};
+static const int8_t hex2[6][2]  = {{-2, 0}, {-1,-2}, {-1, 2}, { 1,-2}, { 1, 2}, { 2, 0}};
+static const int8_t hex4[16][2] = {{-4,-2}, {-4,-1}, {-4, 0}, {-4, 1}, {-4, 2},
+                                   { 4,-2}, { 4,-1}, { 4, 0}, { 4, 1}, { 4, 2},
+                                   {-2, 3}, { 0, 4}, { 2, 3}, {-2,-3}, { 0,-4}, { 2,-3}};
+
+#define COST_MV(x, y)\
+do {\
+    cost = me_ctx->get_cost(me_ctx, x_mb, y_mb, x, y);\
+    if (cost < cost_min) {\
+        cost_min = cost;\
+        mv[0] = x;\
+        mv[1] = y;\
+    }\
+} while(0)
+
+#define COST_P_MV(x, y)\
+if (x >= x_min && x <= x_max && y >= y_min && y <= y_max)\
+    COST_MV(x, y);
+
+void ff_me_init_context(AVMotionEstContext *me_ctx, int mb_size, int search_param,
+                        int width, int height, int x_min, int x_max, int y_min, int y_max)
+{
+    me_ctx->width = width;
+    me_ctx->height = height;
+    me_ctx->mb_size = mb_size;
+    me_ctx->search_param = search_param;
+    me_ctx->get_cost = &ff_me_cmp_sad;
+    me_ctx->x_min = x_min;
+    me_ctx->x_max = x_max;
+    me_ctx->y_min = y_min;
+    me_ctx->y_max = y_max;
+}
+
+uint64_t ff_me_cmp_sad(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int x_mv, int y_mv)
+{
+    const int linesize = me_ctx->linesize;
+    uint8_t *data_ref = me_ctx->data_ref;
+    uint8_t *data_cur = me_ctx->data_cur;
+    uint64_t sad = 0;
+    int i, j;
+
+    data_ref += y_mv * linesize;
+    data_cur += y_mb * linesize;
+
+    for (j = 0; j < me_ctx->mb_size; j++)
+        for (i = 0; i < me_ctx->mb_size; i++)
+            sad += FFABS(data_ref[x_mv + i + j * linesize] - data_cur[x_mb + i + j * linesize]);
+
+    return sad;
+}
+
+uint64_t ff_me_search_esa(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv)
+{
+    int x, y;
+    int x_min = FFMAX(me_ctx->x_min, x_mb - me_ctx->search_param);
+    int y_min = FFMAX(me_ctx->y_min, y_mb - me_ctx->search_param);
+    int x_max = FFMIN(x_mb + me_ctx->search_param, me_ctx->x_max);
+    int y_max = FFMIN(y_mb + me_ctx->search_param, me_ctx->y_max);
+    uint64_t cost, cost_min;
+
+    if (!(cost_min = me_ctx->get_cost(me_ctx, x_mb, y_mb, x_mb, y_mb)))
+        return cost_min;
+
+    for (y = y_min; y <= y_max; y++)
+        for (x = x_min; x <= x_max; x++)
+            COST_MV(x, y);
+
+    return cost_min;
+}
+
+uint64_t ff_me_search_tss(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv)
+{
+    int x, y;
+    int x_min = FFMAX(me_ctx->x_min, x_mb - me_ctx->search_param);
+    int y_min = FFMAX(me_ctx->y_min, y_mb - me_ctx->search_param);
+    int x_max = FFMIN(x_mb + me_ctx->search_param, me_ctx->x_max);
+    int y_max = FFMIN(y_mb + me_ctx->search_param, me_ctx->y_max);
+    uint64_t cost, cost_min;
+    int step = ROUNDED_DIV(me_ctx->search_param, 2);
+    int i;
+
+    mv[0] = x_mb;
+    mv[1] = y_mb;
+
+    if (!(cost_min = me_ctx->get_cost(me_ctx, x_mb, y_mb, x_mb, y_mb)))
+        return cost_min;
+
+    do {
+        x = mv[0];
+        y = mv[1];
+
+        for (i = 0; i < 8; i++)
+            COST_P_MV(x + sqr1[i][0] * step, y + sqr1[i][1] * step);
+
+        step = step >> 1;
+
+    } while (step > 0);
+
+    return cost_min;
+}
+
+uint64_t ff_me_search_tdls(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv)
+{
+    int x, y;
+    int x_min = FFMAX(me_ctx->x_min, x_mb - me_ctx->search_param);
+    int y_min = FFMAX(me_ctx->y_min, y_mb - me_ctx->search_param);
+    int x_max = FFMIN(x_mb + me_ctx->search_param, me_ctx->x_max);
+    int y_max = FFMIN(y_mb + me_ctx->search_param, me_ctx->y_max);
+    uint64_t cost, cost_min;
+    int step = ROUNDED_DIV(me_ctx->search_param, 2);
+    int i;
+
+    mv[0] = x_mb;
+    mv[1] = y_mb;
+
+    if (!(cost_min = me_ctx->get_cost(me_ctx, x_mb, y_mb, x_mb, y_mb)))
+        return cost_min;
+
+    do {
+        x = mv[0];
+        y = mv[1];
+
+        for (i = 0; i < 4; i++)
+            COST_P_MV(x + dia1[i][0] * step, y + dia1[i][1] * step);
+
+        if (x == mv[0] && y == mv[1])
+            step = step >> 1;
+
+    } while (step > 0);
+
+    return cost_min;
+}
+
+uint64_t ff_me_search_ntss(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv)
+{
+    int x, y;
+    int x_min = FFMAX(me_ctx->x_min, x_mb - me_ctx->search_param);
+    int y_min = FFMAX(me_ctx->y_min, y_mb - me_ctx->search_param);
+    int x_max = FFMIN(x_mb + me_ctx->search_param, me_ctx->x_max);
+    int y_max = FFMIN(y_mb + me_ctx->search_param, me_ctx->y_max);
+    uint64_t cost, cost_min;
+    int step = ROUNDED_DIV(me_ctx->search_param, 2);
+    int first_step = 1;
+    int i;
+
+    mv[0] = x_mb;
+    mv[1] = y_mb;
+
+    if (!(cost_min = me_ctx->get_cost(me_ctx, x_mb, y_mb, x_mb, y_mb)))
+        return cost_min;
+
+    do {
+        x = mv[0];
+        y = mv[1];
+
+        for (i = 0; i < 8; i++)
+            COST_P_MV(x + sqr1[i][0] * step, y + sqr1[i][1] * step);
+
+        /* addition to TSS in NTSS */
+        if (first_step) {
+
+            for (i = 0; i < 8; i++)
+                COST_P_MV(x + sqr1[i][0], y + sqr1[i][1]);
+
+            if (x == mv[0] && y == mv[1])
+                return cost_min;
+
+            if (FFABS(x - mv[0]) <= 1 && FFABS(y - mv[1]) <= 1) {
+                x = mv[0];
+                y = mv[1];
+
+                for (i = 0; i < 8; i++)
+                    COST_P_MV(x + sqr1[i][0], y + sqr1[i][1]);
+                return cost_min;
+            }
+
+            first_step = 0;
+        }
+
+        step = step >> 1;
+
+    } while (step > 0);
+
+    return cost_min;
+}
+
+uint64_t ff_me_search_fss(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv)
+{
+    int x, y;
+    int x_min = FFMAX(me_ctx->x_min, x_mb - me_ctx->search_param);
+    int y_min = FFMAX(me_ctx->y_min, y_mb - me_ctx->search_param);
+    int x_max = FFMIN(x_mb + me_ctx->search_param, me_ctx->x_max);
+    int y_max = FFMIN(y_mb + me_ctx->search_param, me_ctx->y_max);
+    uint64_t cost, cost_min;
+    int step = 2;
+    int i;
+
+    mv[0] = x_mb;
+    mv[1] = y_mb;
+
+    if (!(cost_min = me_ctx->get_cost(me_ctx, x_mb, y_mb, x_mb, y_mb)))
+        return cost_min;
+
+    do {
+        x = mv[0];
+        y = mv[1];
+
+        for (i = 0; i < 8; i++)
+            COST_P_MV(x + sqr1[i][0] * step, y + sqr1[i][1] * step);
+
+        if (x == mv[0] && y == mv[1])
+            step = step >> 1;
+
+    } while (step > 0);
+
+    return cost_min;
+}
+
+uint64_t ff_me_search_ds(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv)
+{
+    int x, y;
+    int x_min = FFMAX(me_ctx->x_min, x_mb - me_ctx->search_param);
+    int y_min = FFMAX(me_ctx->y_min, y_mb - me_ctx->search_param);
+    int x_max = FFMIN(x_mb + me_ctx->search_param, me_ctx->x_max);
+    int y_max = FFMIN(y_mb + me_ctx->search_param, me_ctx->y_max);
+    uint64_t cost, cost_min;
+    int i;
+    int dir_x, dir_y;
+
+    if (!(cost_min = me_ctx->get_cost(me_ctx, x_mb, y_mb, x_mb, y_mb)))
+        return cost_min;
+
+    x = x_mb; y = y_mb;
+    dir_x = dir_y = 0;
+
+    do {
+        x = mv[0];
+        y = mv[1];
+
+    #if 1
+        for (i = 0; i < 8; i++)
+            COST_P_MV(x + dia2[i][0], y + dia2[i][1]);
+    #else
+        /* this version skips previously examined 3 or 5 locations based on prev origin */
+        if (dir_x <= 0)
+            COST_P_MV(x - 2, y);
+        if (dir_x <= 0 && dir_y <= 0)
+            COST_P_MV(x - 1, y - 1);
+        if (dir_y <= 0)
+            COST_P_MV(x, y - 2);
+        if (dir_x >= 0 && dir_y <= 0)
+            COST_P_MV(x + 1, y - 1);
+        if (dir_x >= 0)
+            COST_P_MV(x + 2, y);
+        if (dir_x >= 0 && dir_y >= 0)
+            COST_P_MV(x + 1, y + 1);
+        if (dir_y >= 0)
+            COST_P_MV(x, y + 2);
+        if (dir_x <= 0 && dir_y >= 0)
+            COST_P_MV(x - 1, y + 1);
+
+        dir_x = mv[0] - x;
+        dir_y = mv[1] - y;
+    #endif
+
+    } while (x != mv[0] || y != mv[1]);
+
+    for (i = 0; i < 4; i++)
+        COST_P_MV(x + dia1[i][0], y + dia1[i][1]);
+
+    return cost_min;
+}
+
+uint64_t ff_me_search_hexbs(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv)
+{
+    int x, y;
+    int x_min = FFMAX(me_ctx->x_min, x_mb - me_ctx->search_param);
+    int y_min = FFMAX(me_ctx->y_min, y_mb - me_ctx->search_param);
+    int x_max = FFMIN(x_mb + me_ctx->search_param, me_ctx->x_max);
+    int y_max = FFMIN(y_mb + me_ctx->search_param, me_ctx->y_max);
+    uint64_t cost, cost_min;
+    int i;
+
+    if (!(cost_min = me_ctx->get_cost(me_ctx, x_mb, y_mb, x_mb, y_mb)))
+        return cost_min;
+
+    do {
+        x = mv[0];
+        y = mv[1];
+
+        for (i = 0; i < 6; i++)
+            COST_P_MV(x + hex2[i][0], y + hex2[i][1]);
+
+    } while (x != mv[0] || y != mv[1]);
+
+    for (i = 0; i < 4; i++)
+        COST_P_MV(x + dia1[i][0], y + dia1[i][1]);
+
+    return cost_min;
+}
+
+/* two subsets of predictors are used
+   me->pred_x|y is set to median of current frame's left, top, top-right
+   set 1: me->preds[0] has: (0, 0), left, top, top-right, collocated block in prev frame
+   set 2: me->preds[1] has: accelerator mv, top, left, right, bottom adj mb of prev frame
+*/
+uint64_t ff_me_search_epzs(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv)
+{
+    int x, y;
+    int x_min = FFMAX(me_ctx->x_min, x_mb - me_ctx->search_param);
+    int y_min = FFMAX(me_ctx->y_min, y_mb - me_ctx->search_param);
+    int x_max = FFMIN(x_mb + me_ctx->search_param, me_ctx->x_max);
+    int y_max = FFMIN(y_mb + me_ctx->search_param, me_ctx->y_max);
+    uint64_t cost, cost_min;
+    int i;
+
+    AVMotionEstPredictor *preds = me_ctx->preds;
+
+    cost_min = UINT64_MAX;
+
+    COST_P_MV(x_mb + me_ctx->pred_x, y_mb + me_ctx->pred_y);
+
+    for (i = 0; i < preds[0].nb; i++)
+        COST_P_MV(x_mb + preds[0].mvs[i][0], y_mb + preds[0].mvs[i][1]);
+
+    for (i = 0; i < preds[1].nb; i++)
+        COST_P_MV(x_mb + preds[1].mvs[i][0], y_mb + preds[1].mvs[i][1]);
+
+    do {
+        x = mv[0];
+        y = mv[1];
+
+        for (i = 0; i < 4; i++)
+            COST_P_MV(x + dia1[i][0], y + dia1[i][1]);
+
+    } while (x != mv[0] || y != mv[1]);
+
+    return cost_min;
+}
+
+/* required predictor order: median, (0,0), left, top, top-right
+   rules when mb not available:
+   replace left with (0, 0)
+   replace top-right with top-left
+   replace top two with left
+   repeated can be skipped, if no predictors are used, set me_ctx->pred to (0,0)
+*/
+uint64_t ff_me_search_umh(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv)
+{
+    int x, y;
+    int x_min = FFMAX(me_ctx->x_min, x_mb - me_ctx->search_param);
+    int y_min = FFMAX(me_ctx->y_min, y_mb - me_ctx->search_param);
+    int x_max = FFMIN(x_mb + me_ctx->search_param, me_ctx->x_max);
+    int y_max = FFMIN(y_mb + me_ctx->search_param, me_ctx->y_max);
+    uint64_t cost, cost_min;
+    int d, i;
+    int end_x, end_y;
+
+    AVMotionEstPredictor *pred = &me_ctx->preds[0];
+
+    cost_min = UINT64_MAX;
+
+    COST_P_MV(x_mb + me_ctx->pred_x, y_mb + me_ctx->pred_y);
+
+    for (i = 0; i < pred->nb; i++)
+        COST_P_MV(x_mb + pred->mvs[i][0], y_mb + pred->mvs[i][1]);
+
+    // Unsymmetrical-cross Search
+    x = mv[0];
+    y = mv[1];
+    for (d = 1; d <= me_ctx->search_param; d += 2) {
+        COST_P_MV(x - d, y);
+        COST_P_MV(x + d, y);
+        if (d <= me_ctx->search_param / 2) {
+            COST_P_MV(x, y - d);
+            COST_P_MV(x, y + d);
+        }
+    }
+
+    // Uneven Multi-Hexagon-Grid Search
+    end_x = FFMIN(mv[0] + 2, x_max);
+    end_y = FFMIN(mv[1] + 2, y_max);
+    for (y = FFMAX(y_min, mv[1] - 2); y <= end_y; y++)
+        for (x = FFMAX(x_min, mv[0] - 2); x <= end_x; x++)
+            COST_P_MV(x, y);
+
+    x = mv[0];
+    y = mv[1];
+    for (d = 1; d <= me_ctx->search_param / 4; d++)
+        for (i = 1; i < 16; i++)
+            COST_P_MV(x + hex4[i][0] * d, y + hex4[i][1] * d);
+
+    // Extended Hexagon-based Search
+    do {
+        x = mv[0];
+        y = mv[1];
+
+        for (i = 0; i < 6; i++)
+            COST_P_MV(x + hex2[i][0], y + hex2[i][1]);
+
+    } while (x != mv[0] || y != mv[1]);
+
+    for (i = 0; i < 4; i++)
+        COST_P_MV(x + dia1[i][0], y + dia1[i][1]);
+
+    return cost_min;
+}
diff --git a/libavfilter/motion_estimation.h b/libavfilter/motion_estimation.h
new file mode 100644
index 0000000..6ae29dd
--- /dev/null
+++ b/libavfilter/motion_estimation.h
@@ -0,0 +1,87 @@
+/**
+ * Copyright (c) 2016 Davinder Singh (DSM_) <ds.mudhar<@gmail.com>
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#ifndef AVFILTER_MOTION_ESTIMATION_H
+#define AVFILTER_MOTION_ESTIMATION_H
+
+#include "libavutil/avutil.h"
+
+#define AV_ME_METHOD_ESA        1
+#define AV_ME_METHOD_TSS        2
+#define AV_ME_METHOD_TDLS       3
+#define AV_ME_METHOD_NTSS       4
+#define AV_ME_METHOD_FSS        5
+#define AV_ME_METHOD_DS         6
+#define AV_ME_METHOD_HEXBS      7
+#define AV_ME_METHOD_EPZS       8
+#define AV_ME_METHOD_UMH        9
+
+typedef struct AVMotionEstPredictor {
+    int mvs[10][2];
+    int nb;
+} AVMotionEstPredictor;
+
+typedef struct AVMotionEstContext {
+    uint8_t *data_cur, *data_ref;
+    int linesize;
+
+    int mb_size;
+    int search_param;
+
+    int width;
+    int height;
+
+    int x_min;
+    int x_max;
+    int y_min;
+    int y_max;
+
+    int pred_x;     ///< median predictor x
+    int pred_y;     ///< median predictor y
+    AVMotionEstPredictor preds[2];
+
+    uint64_t (*get_cost)(struct AVMotionEstContext *me_ctx, int x_mb, int y_mb,
+                         int mv_x, int mv_y);
+} AVMotionEstContext;
+
+void ff_me_init_context(AVMotionEstContext *me_ctx, int mb_size, int search_param,
+                        int width, int height, int x_min, int x_max, int y_min, int y_max);
+
+uint64_t ff_me_cmp_sad(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int x_mv, int y_mv);
+
+uint64_t ff_me_search_esa(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv);
+
+uint64_t ff_me_search_tss(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv);
+
+uint64_t ff_me_search_tdls(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv);
+
+uint64_t ff_me_search_ntss(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv);
+
+uint64_t ff_me_search_fss(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv);
+
+uint64_t ff_me_search_ds(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv);
+
+uint64_t ff_me_search_hexbs(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv);
+
+uint64_t ff_me_search_epzs(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv);
+
+uint64_t ff_me_search_umh(AVMotionEstContext *me_ctx, int x_mb, int y_mb, int *mv);
+
+#endif /* AVFILTER_MOTION_ESTIMATION_H */
diff --git a/libavfilter/version.h b/libavfilter/version.h
index 05b9fdf..6f6e516 100644
--- a/libavfilter/version.h
+++ b/libavfilter/version.h
@@ -30,7 +30,7 @@
 #include "libavutil/version.h"
 
 #define LIBAVFILTER_VERSION_MAJOR   6
-#define LIBAVFILTER_VERSION_MINOR  55
+#define LIBAVFILTER_VERSION_MINOR  56
 #define LIBAVFILTER_VERSION_MICRO 100
 
 #define LIBAVFILTER_VERSION_INT AV_VERSION_INT(LIBAVFILTER_VERSION_MAJOR, \
diff --git a/libavfilter/vf_mestimate.c b/libavfilter/vf_mestimate.c
new file mode 100644
index 0000000..7ecfe7d
--- /dev/null
+++ b/libavfilter/vf_mestimate.c
@@ -0,0 +1,377 @@
+/**
+ * Copyright (c) 2016 Davinder Singh (DSM_) <ds.mudhar<@gmail.com>
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "motion_estimation.h"
+#include "libavcodec/mathops.h"
+#include "libavutil/avassert.h"
+#include "libavutil/common.h"
+#include "libavutil/imgutils.h"
+#include "libavutil/opt.h"
+#include "libavutil/pixdesc.h"
+#include "libavutil/motion_vector.h"
+#include "avfilter.h"
+#include "formats.h"
+#include "internal.h"
+#include "video.h"
+
+typedef struct MEContext {
+    const AVClass *class;
+    AVMotionEstContext me_ctx;
+    int method;                         ///< motion estimation method
+
+    int mb_size;                        ///< macroblock size
+    int search_param;                   ///< search parameter
+    int b_width, b_height, b_count;
+    int log2_mb_size;
+
+    AVFrame *prev, *cur, *next;
+
+    int (*mv_table[3])[2][2];           ///< motion vectors of current & prev 2 frames
+} MEContext;
+
+#define OFFSET(x) offsetof(MEContext, x)
+#define FLAGS AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
+#define CONST(name, help, val, unit) { name, help, 0, AV_OPT_TYPE_CONST, {.i64=val}, 0, 0, FLAGS, unit }
+
+static const AVOption mestimate_options[] = {
+    { "method", "motion estimation method", OFFSET(method), AV_OPT_TYPE_INT, {.i64 = AV_ME_METHOD_ESA}, AV_ME_METHOD_ESA, AV_ME_METHOD_UMH, FLAGS, "method" },
+        CONST("esa",   "exhaustive search",                  AV_ME_METHOD_ESA,      "method"),
+        CONST("tss",   "three step search",                  AV_ME_METHOD_TSS,      "method"),
+        CONST("tdls",  "two dimensional logarithmic search", AV_ME_METHOD_TDLS,     "method"),
+        CONST("ntss",  "new three step search",              AV_ME_METHOD_NTSS,     "method"),
+        CONST("fss",   "four step search",                   AV_ME_METHOD_FSS,      "method"),
+        CONST("ds",    "diamond search",                     AV_ME_METHOD_DS,       "method"),
+        CONST("hexbs", "hexagon-based search",               AV_ME_METHOD_HEXBS,    "method"),
+        CONST("epzs",  "enhanced predictive zonal search",   AV_ME_METHOD_EPZS,     "method"),
+        CONST("umh",   "uneven multi-hexagon search",        AV_ME_METHOD_UMH,      "method"),
+    { "mb_size", "macroblock size", OFFSET(mb_size), AV_OPT_TYPE_INT, {.i64 = 16}, 8, INT_MAX, FLAGS },
+    { "search_param", "search parameter", OFFSET(search_param), AV_OPT_TYPE_INT, {.i64 = 7}, 4, INT_MAX, FLAGS },
+    { NULL }
+};
+
+AVFILTER_DEFINE_CLASS(mestimate);
+
+static int query_formats(AVFilterContext *ctx)
+{
+    static const enum AVPixelFormat pix_fmts[] = {
+        AV_PIX_FMT_YUV410P, AV_PIX_FMT_YUV411P,
+        AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUV422P,
+        AV_PIX_FMT_YUV440P, AV_PIX_FMT_YUV444P,
+        AV_PIX_FMT_YUVJ444P, AV_PIX_FMT_YUVJ440P,
+        AV_PIX_FMT_YUVJ422P, AV_PIX_FMT_YUVJ420P,
+        AV_PIX_FMT_YUVJ411P,
+        AV_PIX_FMT_YUVA420P, AV_PIX_FMT_YUVA422P, AV_PIX_FMT_YUVA444P,
+        AV_PIX_FMT_GRAY8,
+        AV_PIX_FMT_NONE
+    };
+
+    AVFilterFormats *fmts_list = ff_make_format_list(pix_fmts);
+    if (!fmts_list)
+        return AVERROR(ENOMEM);
+    return ff_set_common_formats(ctx, fmts_list);
+}
+
+static int config_input(AVFilterLink *inlink)
+{
+    MEContext *s = inlink->dst->priv;
+    int i;
+
+    s->log2_mb_size = av_ceil_log2_c(s->mb_size);
+    s->mb_size = 1 << s->log2_mb_size;
+
+    s->b_width  = inlink->w >> s->log2_mb_size;
+    s->b_height = inlink->h >> s->log2_mb_size;
+    s->b_count = s->b_width * s->b_height;
+
+    for (i = 0; i < 3; i++) {
+        s->mv_table[i] = av_mallocz_array(s->b_count, sizeof(*s->mv_table[0]));
+        if (!s->mv_table[i])
+            return AVERROR(ENOMEM);
+    }
+
+    ff_me_init_context(&s->me_ctx, s->mb_size, s->search_param, inlink->w, inlink->h, 0, (s->b_width - 1) << s->log2_mb_size, 0, (s->b_height - 1) << s->log2_mb_size);
+
+    return 0;
+}
+
+static void add_mv_data(AVMotionVector *mv, int mb_size,
+                        int x, int y, int x_mv, int y_mv, int dir)
+{
+    mv->w = mb_size;
+    mv->h = mb_size;
+    mv->dst_x = x + (mb_size >> 1);
+    mv->dst_y = y + (mb_size >> 1);
+    mv->src_x = x_mv + (mb_size >> 1);
+    mv->src_y = y_mv + (mb_size >> 1);
+    mv->source = dir ? 1 : -1;
+    mv->flags = 0;
+}
+
+#define SEARCH_MV(method)\
+    do {\
+        for (mb_y = 0; mb_y < s->b_height; mb_y++)\
+            for (mb_x = 0; mb_x < s->b_width; mb_x++) {\
+                const int x_mb = mb_x << s->log2_mb_size;\
+                const int y_mb = mb_y << s->log2_mb_size;\
+                int mv[2] = {x_mb, y_mb};\
+                ff_me_search_##method(me_ctx, x_mb, y_mb, mv);\
+                add_mv_data(((AVMotionVector *) sd->data) + mv_count++, me_ctx->mb_size, x_mb, y_mb, mv[0], mv[1], dir);\
+            }\
+    } while (0)
+
+#define ADD_PRED(preds, px, py)\
+    do {\
+        preds.mvs[preds.nb][0] = px;\
+        preds.mvs[preds.nb][1] = py;\
+        preds.nb++;\
+    } while(0)
+
+static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
+{
+    AVFilterContext *ctx = inlink->dst;
+    MEContext *s = ctx->priv;
+    AVMotionEstContext *me_ctx = &s->me_ctx;
+    AVFrameSideData *sd;
+    AVFrame *out;
+    int mb_x, mb_y, dir;
+    int32_t mv_count = 0;
+    int ret;
+
+    if (frame->pts == AV_NOPTS_VALUE) {
+        ret = ff_filter_frame(ctx->outputs[0], frame);
+        return ret;
+    }
+
+    av_frame_free(&s->prev);
+    s->prev = s->cur;
+    s->cur  = s->next;
+    s->next = frame;
+
+    s->mv_table[2] = memcpy(s->mv_table[2], s->mv_table[1], sizeof(*s->mv_table[1]) * s->b_count);
+    s->mv_table[1] = memcpy(s->mv_table[1], s->mv_table[0], sizeof(*s->mv_table[0]) * s->b_count);
+
+    if (!s->cur) {
+        s->cur = av_frame_clone(frame);
+        if (!s->cur)
+            return AVERROR(ENOMEM);
+    }
+
+    if (!s->prev)
+        return 0;
+
+    out = av_frame_clone(s->cur);
+    if (!out)
+        return AVERROR(ENOMEM);
+
+    sd = av_frame_new_side_data(out, AV_FRAME_DATA_MOTION_VECTORS, 2 * s->b_count * sizeof(AVMotionVector));
+    if (!sd) {
+        av_frame_free(&out);
+        return AVERROR(ENOMEM);
+    }
+
+    me_ctx->data_cur = s->cur->data[0];
+    me_ctx->linesize = s->cur->linesize[0];
+
+    for (dir = 0; dir < 2; dir++) {
+        me_ctx->data_ref = (dir ? s->next : s->prev)->data[0];
+
+        if (s->method == AV_ME_METHOD_DS)
+            SEARCH_MV(ds);
+        else if (s->method == AV_ME_METHOD_ESA)
+            SEARCH_MV(esa);
+        else if (s->method == AV_ME_METHOD_FSS)
+            SEARCH_MV(fss);
+        else if (s->method == AV_ME_METHOD_NTSS)
+            SEARCH_MV(ntss);
+        else if (s->method == AV_ME_METHOD_TDLS)
+            SEARCH_MV(tdls);
+        else if (s->method == AV_ME_METHOD_TSS)
+            SEARCH_MV(tss);
+        else if (s->method == AV_ME_METHOD_HEXBS)
+            SEARCH_MV(hexbs);
+        else if (s->method == AV_ME_METHOD_UMH) {
+            for (mb_y = 0; mb_y < s->b_height; mb_y++)
+                for (mb_x = 0; mb_x < s->b_width; mb_x++) {
+                    const int mb_i = mb_x + mb_y * s->b_width;
+                    const int x_mb = mb_x << s->log2_mb_size;
+                    const int y_mb = mb_y << s->log2_mb_size;
+                    int mv[2] = {x_mb, y_mb};
+
+                    AVMotionEstPredictor *preds = me_ctx->preds;
+                    preds[0].nb = 0;
+
+                    ADD_PRED(preds[0], 0, 0);
+
+                    //left mb in current frame
+                    if (mb_x > 0)
+                        ADD_PRED(preds[0], s->mv_table[0][mb_i - 1][dir][0], s->mv_table[0][mb_i - 1][dir][1]);
+
+                    if (mb_y > 0) {
+                        //top mb in current frame
+                        ADD_PRED(preds[0], s->mv_table[0][mb_i - s->b_width][dir][0], s->mv_table[0][mb_i - s->b_width][dir][1]);
+
+                        //top-right mb in current frame
+                        if (mb_x + 1 < s->b_width)
+                            ADD_PRED(preds[0], s->mv_table[0][mb_i - s->b_width + 1][dir][0], s->mv_table[0][mb_i - s->b_width + 1][dir][1]);
+                        //top-left mb in current frame
+                        else if (mb_x > 0)
+                            ADD_PRED(preds[0], s->mv_table[0][mb_i - s->b_width - 1][dir][0], s->mv_table[0][mb_i - s->b_width - 1][dir][1]);
+                    }
+
+                    //median predictor
+                    if (preds[0].nb == 4) {
+                        me_ctx->pred_x = mid_pred(preds[0].mvs[1][0], preds[0].mvs[2][0], preds[0].mvs[3][0]);
+                        me_ctx->pred_y = mid_pred(preds[0].mvs[1][1], preds[0].mvs[2][1], preds[0].mvs[3][1]);
+                    } else if (preds[0].nb == 3) {
+                        me_ctx->pred_x = mid_pred(0, preds[0].mvs[1][0], preds[0].mvs[2][0]);
+                        me_ctx->pred_y = mid_pred(0, preds[0].mvs[1][1], preds[0].mvs[2][1]);
+                    } else if (preds[0].nb == 2) {
+                        me_ctx->pred_x = preds[0].mvs[1][0];
+                        me_ctx->pred_y = preds[0].mvs[1][1];
+                    } else {
+                        me_ctx->pred_x = 0;
+                        me_ctx->pred_y = 0;
+                    }
+
+                    ff_me_search_umh(me_ctx, x_mb, y_mb, mv);
+
+                    s->mv_table[0][mb_i][dir][0] = mv[0] - x_mb;
+                    s->mv_table[0][mb_i][dir][1] = mv[1] - y_mb;
+                    add_mv_data(((AVMotionVector *) sd->data) + mv_count++, me_ctx->mb_size, x_mb, y_mb, mv[0], mv[1], dir);
+                }
+
+        } else if (s->method == AV_ME_METHOD_EPZS) {
+
+            for (mb_y = 0; mb_y < s->b_height; mb_y++)
+                for (mb_x = 0; mb_x < s->b_width; mb_x++) {
+                    const int mb_i = mb_x + mb_y * s->b_width;
+                    const int x_mb = mb_x << s->log2_mb_size;
+                    const int y_mb = mb_y << s->log2_mb_size;
+                    int mv[2] = {x_mb, y_mb};
+
+                    AVMotionEstPredictor *preds = me_ctx->preds;
+                    preds[0].nb = 0;
+                    preds[1].nb = 0;
+
+                    ADD_PRED(preds[0], 0, 0);
+
+                    //left mb in current frame
+                    if (mb_x > 0)
+                        ADD_PRED(preds[0], s->mv_table[0][mb_i - 1][dir][0], s->mv_table[0][mb_i - 1][dir][1]);
+
+                    //top mb in current frame
+                    if (mb_y > 0)
+                        ADD_PRED(preds[0], s->mv_table[0][mb_i - s->b_width][dir][0], s->mv_table[0][mb_i - s->b_width][dir][1]);
+
+                    //top-right mb in current frame
+                    if (mb_y > 0 && mb_x + 1 < s->b_width)
+                        ADD_PRED(preds[0], s->mv_table[0][mb_i - s->b_width + 1][dir][0], s->mv_table[0][mb_i - s->b_width + 1][dir][1]);
+
+                    //median predictor
+                    if (preds[0].nb == 4) {
+                        me_ctx->pred_x = mid_pred(preds[0].mvs[1][0], preds[0].mvs[2][0], preds[0].mvs[3][0]);
+                        me_ctx->pred_y = mid_pred(preds[0].mvs[1][1], preds[0].mvs[2][1], preds[0].mvs[3][1]);
+                    } else if (preds[0].nb == 3) {
+                        me_ctx->pred_x = mid_pred(0, preds[0].mvs[1][0], preds[0].mvs[2][0]);
+                        me_ctx->pred_y = mid_pred(0, preds[0].mvs[1][1], preds[0].mvs[2][1]);
+                    } else if (preds[0].nb == 2) {
+                        me_ctx->pred_x = preds[0].mvs[1][0];
+                        me_ctx->pred_y = preds[0].mvs[1][1];
+                    } else {
+                        me_ctx->pred_x = 0;
+                        me_ctx->pred_y = 0;
+                    }
+
+                    //collocated mb in prev frame
+                    ADD_PRED(preds[0], s->mv_table[1][mb_i][dir][0], s->mv_table[1][mb_i][dir][1]);
+
+                    //accelerator motion vector of collocated block in prev frame
+                    ADD_PRED(preds[1], s->mv_table[1][mb_i][dir][0] + (s->mv_table[1][mb_i][dir][0] - s->mv_table[2][mb_i][dir][0]),
+                                       s->mv_table[1][mb_i][dir][1] + (s->mv_table[1][mb_i][dir][1] - s->mv_table[2][mb_i][dir][1]));
+
+                    //left mb in prev frame
+                    if (mb_x > 0)
+                        ADD_PRED(preds[1], s->mv_table[1][mb_i - 1][dir][0], s->mv_table[1][mb_i - 1][dir][1]);
+
+                    //top mb in prev frame
+                    if (mb_y > 0)
+                        ADD_PRED(preds[1], s->mv_table[1][mb_i - s->b_width][dir][0], s->mv_table[1][mb_i - s->b_width][dir][1]);
+
+                    //right mb in prev frame
+                    if (mb_x + 1 < s->b_width)
+                        ADD_PRED(preds[1], s->mv_table[1][mb_i + 1][dir][0], s->mv_table[1][mb_i + 1][dir][1]);
+
+                    //bottom mb in prev frame
+                    if (mb_y + 1 < s->b_height)
+                        ADD_PRED(preds[1], s->mv_table[1][mb_i + s->b_width][dir][0], s->mv_table[1][mb_i + s->b_width][dir][1]);
+
+                    ff_me_search_epzs(me_ctx, x_mb, y_mb, mv);
+
+                    s->mv_table[0][mb_i][dir][0] = mv[0] - x_mb;
+                    s->mv_table[0][mb_i][dir][1] = mv[1] - y_mb;
+                    add_mv_data(((AVMotionVector *) sd->data) + mv_count++, s->mb_size, x_mb, y_mb, mv[0], mv[1], dir);
+                }
+        }
+    }
+
+    return ff_filter_frame(ctx->outputs[0], out);
+}
+
+static av_cold void uninit(AVFilterContext *ctx)
+{
+    MEContext *s = ctx->priv;
+    int i;
+
+    av_frame_free(&s->prev);
+    av_frame_free(&s->cur);
+    av_frame_free(&s->next);
+
+    for (i = 0; i < 3; i++)
+        av_freep(&s->mv_table[i]);
+}
+
+static const AVFilterPad mestimate_inputs[] = {
+    {
+        .name          = "default",
+        .type          = AVMEDIA_TYPE_VIDEO,
+        .filter_frame  = filter_frame,
+        .config_props  = config_input,
+    },
+    { NULL }
+};
+
+static const AVFilterPad mestimate_outputs[] = {
+    {
+        .name          = "default",
+        .type          = AVMEDIA_TYPE_VIDEO,
+    },
+    { NULL }
+};
+
+AVFilter ff_vf_mestimate = {
+    .name          = "mestimate",
+    .description   = NULL_IF_CONFIG_SMALL("Generate motion vectors."),
+    .priv_size     = sizeof(MEContext),
+    .priv_class    = &mestimate_class,
+    .uninit        = uninit,
+    .query_formats = query_formats,
+    .inputs        = mestimate_inputs,
+    .outputs       = mestimate_outputs,
+};
diff --git a/libavfilter/vf_minterpolate.c b/libavfilter/vf_minterpolate.c
new file mode 100644
index 0000000..5b41cd8
--- /dev/null
+++ b/libavfilter/vf_minterpolate.c
@@ -0,0 +1,1247 @@
+/**
+ * Copyright (c) 2014-2015 Michael Niedermayer <michaelni at gmx.at>
+ * Copyright (c) 2016 Davinder Singh (DSM_) <ds.mudhar<@gmail.com>
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "motion_estimation.h"
+#include "libavcodec/mathops.h"
+#include "libavutil/avassert.h"
+#include "libavutil/common.h"
+#include "libavutil/motion_vector.h"
+#include "libavutil/opt.h"
+#include "libavutil/pixdesc.h"
+#include "libavutil/pixelutils.h"
+#include "avfilter.h"
+#include "formats.h"
+#include "internal.h"
+#include "video.h"
+
+#define ME_MODE_BIDIR 0
+#define ME_MODE_BILAT 1
+
+#define MC_MODE_OBMC 0
+#define MC_MODE_AOBMC 1
+
+#define SCD_METHOD_NONE 0
+#define SCD_METHOD_FDIFF 1
+
+#define NB_FRAMES 4
+#define NB_PIXEL_MVS 32
+#define NB_CLUSTERS 128
+
+#define ALPHA_MAX 1024
+#define CLUSTER_THRESHOLD 4
+#define PX_WEIGHT_MAX 255
+#define COST_PRED_SCALE 64
+
+static const uint8_t obmc_linear32[1024] = {
+  0,  0,  0,  0,  4,  4,  4,  4,  4,  4,  4,  4,  8,  8,  8,  8,  8,  8,  8,  8,  4,  4,  4,  4,  4,  4,  4,  4,  0,  0,  0,  0,
+  0,  4,  4,  4,  8,  8,  8, 12, 12, 16, 16, 16, 20, 20, 20, 24, 24, 20, 20, 20, 16, 16, 16, 12, 12,  8,  8,  8,  4,  4,  4,  0,
+  0,  4,  8,  8, 12, 12, 16, 20, 20, 24, 28, 28, 32, 32, 36, 40, 40, 36, 32, 32, 28, 28, 24, 20, 20, 16, 12, 12,  8,  8,  4,  0,
+  0,  4,  8, 12, 16, 20, 24, 28, 28, 32, 36, 40, 44, 48, 52, 56, 56, 52, 48, 44, 40, 36, 32, 28, 28, 24, 20, 16, 12,  8,  4,  0,
+  4,  8, 12, 16, 20, 24, 28, 32, 40, 44, 48, 52, 56, 60, 64, 68, 68, 64, 60, 56, 52, 48, 44, 40, 32, 28, 24, 20, 16, 12,  8,  4,
+  4,  8, 12, 20, 24, 32, 36, 40, 48, 52, 56, 64, 68, 76, 80, 84, 84, 80, 76, 68, 64, 56, 52, 48, 40, 36, 32, 24, 20, 12,  8,  4,
+  4,  8, 16, 24, 28, 36, 44, 48, 56, 60, 68, 76, 80, 88, 96,100,100, 96, 88, 80, 76, 68, 60, 56, 48, 44, 36, 28, 24, 16,  8,  4,
+  4, 12, 20, 28, 32, 40, 48, 56, 64, 72, 80, 88, 92,100,108,116,116,108,100, 92, 88, 80, 72, 64, 56, 48, 40, 32, 28, 20, 12,  4,
+  4, 12, 20, 28, 40, 48, 56, 64, 72, 80, 88, 96,108,116,124,132,132,124,116,108, 96, 88, 80, 72, 64, 56, 48, 40, 28, 20, 12,  4,
+  4, 16, 24, 32, 44, 52, 60, 72, 80, 92,100,108,120,128,136,148,148,136,128,120,108,100, 92, 80, 72, 60, 52, 44, 32, 24, 16,  4,
+  4, 16, 28, 36, 48, 56, 68, 80, 88,100,112,120,132,140,152,164,164,152,140,132,120,112,100, 88, 80, 68, 56, 48, 36, 28, 16,  4,
+  4, 16, 28, 40, 52, 64, 76, 88, 96,108,120,132,144,156,168,180,180,168,156,144,132,120,108, 96, 88, 76, 64, 52, 40, 28, 16,  4,
+  8, 20, 32, 44, 56, 68, 80, 92,108,120,132,144,156,168,180,192,192,180,168,156,144,132,120,108, 92, 80, 68, 56, 44, 32, 20,  8,
+  8, 20, 32, 48, 60, 76, 88,100,116,128,140,156,168,184,196,208,208,196,184,168,156,140,128,116,100, 88, 76, 60, 48, 32, 20,  8,
+  8, 20, 36, 52, 64, 80, 96,108,124,136,152,168,180,196,212,224,224,212,196,180,168,152,136,124,108, 96, 80, 64, 52, 36, 20,  8,
+  8, 24, 40, 56, 68, 84,100,116,132,148,164,180,192,208,224,240,240,224,208,192,180,164,148,132,116,100, 84, 68, 56, 40, 24,  8,
+  8, 24, 40, 56, 68, 84,100,116,132,148,164,180,192,208,224,240,240,224,208,192,180,164,148,132,116,100, 84, 68, 56, 40, 24,  8,
+  8, 20, 36, 52, 64, 80, 96,108,124,136,152,168,180,196,212,224,224,212,196,180,168,152,136,124,108, 96, 80, 64, 52, 36, 20,  8,
+  8, 20, 32, 48, 60, 76, 88,100,116,128,140,156,168,184,196,208,208,196,184,168,156,140,128,116,100, 88, 76, 60, 48, 32, 20,  8,
+  8, 20, 32, 44, 56, 68, 80, 92,108,120,132,144,156,168,180,192,192,180,168,156,144,132,120,108, 92, 80, 68, 56, 44, 32, 20,  8,
+  4, 16, 28, 40, 52, 64, 76, 88, 96,108,120,132,144,156,168,180,180,168,156,144,132,120,108, 96, 88, 76, 64, 52, 40, 28, 16,  4,
+  4, 16, 28, 36, 48, 56, 68, 80, 88,100,112,120,132,140,152,164,164,152,140,132,120,112,100, 88, 80, 68, 56, 48, 36, 28, 16,  4,
+  4, 16, 24, 32, 44, 52, 60, 72, 80, 92,100,108,120,128,136,148,148,136,128,120,108,100, 92, 80, 72, 60, 52, 44, 32, 24, 16,  4,
+  4, 12, 20, 28, 40, 48, 56, 64, 72, 80, 88, 96,108,116,124,132,132,124,116,108, 96, 88, 80, 72, 64, 56, 48, 40, 28, 20, 12,  4,
+  4, 12, 20, 28, 32, 40, 48, 56, 64, 72, 80, 88, 92,100,108,116,116,108,100, 92, 88, 80, 72, 64, 56, 48, 40, 32, 28, 20, 12,  4,
+  4,  8, 16, 24, 28, 36, 44, 48, 56, 60, 68, 76, 80, 88, 96,100,100, 96, 88, 80, 76, 68, 60, 56, 48, 44, 36, 28, 24, 16,  8,  4,
+  4,  8, 12, 20, 24, 32, 36, 40, 48, 52, 56, 64, 68, 76, 80, 84, 84, 80, 76, 68, 64, 56, 52, 48, 40, 36, 32, 24, 20, 12,  8,  4,
+  4,  8, 12, 16, 20, 24, 28, 32, 40, 44, 48, 52, 56, 60, 64, 68, 68, 64, 60, 56, 52, 48, 44, 40, 32, 28, 24, 20, 16, 12,  8,  4,
+  0,  4,  8, 12, 16, 20, 24, 28, 28, 32, 36, 40, 44, 48, 52, 56, 56, 52, 48, 44, 40, 36, 32, 28, 28, 24, 20, 16, 12,  8,  4,  0,
+  0,  4,  8,  8, 12, 12, 16, 20, 20, 24, 28, 28, 32, 32, 36, 40, 40, 36, 32, 32, 28, 28, 24, 20, 20, 16, 12, 12,  8,  8,  4,  0,
+  0,  4,  4,  4,  8,  8,  8, 12, 12, 16, 16, 16, 20, 20, 20, 24, 24, 20, 20, 20, 16, 16, 16, 12, 12,  8,  8,  8,  4,  4,  4,  0,
+  0,  0,  0,  0,  4,  4,  4,  4,  4,  4,  4,  4,  8,  8,  8,  8,  8,  8,  8,  8,  4,  4,  4,  4,  4,  4,  4,  4,  0,  0,  0,  0,
+};
+
+static const uint8_t obmc_linear16[256] = {
+  0,  4,  4,  8,  8, 12, 12, 16, 16, 12, 12,  8,  8,  4,  4,  0,
+  4,  8, 16, 20, 28, 32, 40, 44, 44, 40, 32, 28, 20, 16,  8,  4,
+  4, 16, 24, 36, 44, 56, 64, 76, 76, 64, 56, 44, 36, 24, 16,  4,
+  8, 20, 36, 48, 64, 76, 92,104,104, 92, 76, 64, 48, 36, 20,  8,
+  8, 28, 44, 64, 80,100,116,136,136,116,100, 80, 64, 44, 28,  8,
+ 12, 32, 56, 76,100,120,144,164,164,144,120,100, 76, 56, 32, 12,
+ 12, 40, 64, 92,116,144,168,196,196,168,144,116, 92, 64, 40, 12,
+ 16, 44, 76,104,136,164,196,224,224,196,164,136,104, 76, 44, 16,
+ 16, 44, 76,104,136,164,196,224,224,196,164,136,104, 76, 44, 16,
+ 12, 40, 64, 92,116,144,168,196,196,168,144,116, 92, 64, 40, 12,
+ 12, 32, 56, 76,100,120,144,164,164,144,120,100, 76, 56, 32, 12,
+  8, 28, 44, 64, 80,100,116,136,136,116,100, 80, 64, 44, 28,  8,
+  8, 20, 36, 48, 64, 76, 92,104,104, 92, 76, 64, 48, 36, 20,  8,
+  4, 16, 24, 36, 44, 56, 64, 76, 76, 64, 56, 44, 36, 24, 16,  4,
+  4,  8, 16, 20, 28, 32, 40, 44, 44, 40, 32, 28, 20, 16,  8,  4,
+  0,  4,  4,  8,  8, 12, 12, 16, 16, 12, 12,  8,  8,  4,  4,  0,
+};
+
+static const uint8_t obmc_linear8[64] = {
+  4, 12, 20, 28, 28, 20, 12,  4,
+ 12, 36, 60, 84, 84, 60, 36, 12,
+ 20, 60,100,140,140,100, 60, 20,
+ 28, 84,140,196,196,140, 84, 28,
+ 28, 84,140,196,196,140, 84, 28,
+ 20, 60,100,140,140,100, 60, 20,
+ 12, 36, 60, 84, 84, 60, 36, 12,
+  4, 12, 20, 28, 28, 20, 12,  4,
+};
+
+static const uint8_t obmc_linear4[16] = {
+ 16, 48, 48, 16,
+ 48,144,144, 48,
+ 48,144,144, 48,
+ 16, 48, 48, 16,
+};
+
+static const uint8_t * const obmc_tab_linear[4]= {
+    obmc_linear32, obmc_linear16, obmc_linear8, obmc_linear4
+};
+
+enum MIMode {
+    MI_MODE_DUP         = 0,
+    MI_MODE_BLEND       = 1,
+    MI_MODE_MCI         = 2,
+};
+
+typedef struct Cluster {
+    int64_t sum[2];
+    int nb;
+} Cluster;
+
+typedef struct Block {
+    int16_t mvs[2][2];
+    int cid;
+    uint64_t sbad;
+    int sb;
+    struct Block *subs;
+} Block;
+
+typedef struct Pixel {
+    int16_t mvs[NB_PIXEL_MVS][2];
+    uint32_t weights[NB_PIXEL_MVS];
+    int8_t refs[NB_PIXEL_MVS];
+    int nb;
+} Pixel;
+
+typedef struct Frame {
+    AVFrame *avf;
+    Block *blocks;
+} Frame;
+
+typedef struct MIContext {
+    const AVClass *class;
+    AVMotionEstContext me_ctx;
+    AVRational frame_rate;
+    enum MIMode mi_mode;
+    int mc_mode;
+    int me_mode;
+    int me_method;
+    int mb_size;
+    int search_param;
+    int vsbmc;
+
+    Frame frames[NB_FRAMES];
+    Cluster clusters[NB_CLUSTERS];
+    Block *int_blocks;
+    Pixel *pixels;
+    int (*mv_table[3])[2][2];
+    int64_t out_pts;
+    int b_width, b_height, b_count;
+    int log2_mb_size;
+
+    int scd_method;
+    int scene_changed;
+    av_pixelutils_sad_fn sad;
+    double prev_mafd;
+    double scd_threshold;
+
+    int chroma_height;
+    int chroma_width;
+    int chroma_h_shift;
+    int chroma_v_shift;
+    int nb_planes;
+} MIContext;
+
+#define OFFSET(x) offsetof(MIContext, x)
+#define FLAGS AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
+#define CONST(name, help, val, unit) { name, help, 0, AV_OPT_TYPE_CONST, {.i64=val}, 0, 0, FLAGS, unit }
+
+static const AVOption minterpolate_options[] = {
+    { "fps", "output's frame rate", OFFSET(frame_rate), AV_OPT_TYPE_VIDEO_RATE, {.str = "60"}, 0, INT_MAX, FLAGS },
+    { "mi_mode", "motion interpolation mode", OFFSET(mi_mode), AV_OPT_TYPE_INT, {.i64 = MI_MODE_MCI}, MI_MODE_DUP, MI_MODE_MCI, FLAGS, "mi_mode" },
+        CONST("dup",    "duplicate frames",                     MI_MODE_DUP,            "mi_mode"),
+        CONST("blend",  "blend frames",                         MI_MODE_BLEND,          "mi_mode"),
+        CONST("mci",    "motion compensated interpolation",     MI_MODE_MCI,            "mi_mode"),
+    { "mc_mode", "motion compensation mode", OFFSET(mc_mode), AV_OPT_TYPE_INT, {.i64 = MC_MODE_OBMC}, MC_MODE_OBMC, MC_MODE_AOBMC, FLAGS, "mc_mode" },
+        CONST("obmc",   "overlapped block motion compensation", MC_MODE_OBMC,           "mc_mode"),
+        CONST("aobmc",  "adaptive overlapped block motion compensation", MC_MODE_AOBMC, "mc_mode"),
+    { "me_mode", "motion estimation mode", OFFSET(me_mode), AV_OPT_TYPE_INT, {.i64 = ME_MODE_BILAT}, ME_MODE_BIDIR, ME_MODE_BILAT, FLAGS, "me_mode" },
+        CONST("bidir",  "bidirectional motion estimation",      ME_MODE_BIDIR,          "me_mode"),
+        CONST("bilat",  "bilateral motion estimation",          ME_MODE_BILAT,          "me_mode"),
+    { "me", "motion estimation method", OFFSET(me_method), AV_OPT_TYPE_INT, {.i64 = AV_ME_METHOD_EPZS}, AV_ME_METHOD_ESA, AV_ME_METHOD_UMH, FLAGS, "me" },
+        CONST("esa",    "exhaustive search",                    AV_ME_METHOD_ESA,       "me"),
+        CONST("tss",    "three step search",                    AV_ME_METHOD_TSS,       "me"),
+        CONST("tdls",   "two dimensional logarithmic search",   AV_ME_METHOD_TDLS,      "me"),
+        CONST("ntss",   "new three step search",                AV_ME_METHOD_NTSS,      "me"),
+        CONST("fss",    "four step search",                     AV_ME_METHOD_FSS,       "me"),
+        CONST("ds",     "diamond search",                       AV_ME_METHOD_DS,        "me"),
+        CONST("hexbs",  "hexagon-based search",                 AV_ME_METHOD_HEXBS,     "me"),
+        CONST("epzs",   "enhanced predictive zonal search",     AV_ME_METHOD_EPZS,      "me"),
+        CONST("umh",    "uneven multi-hexagon search",          AV_ME_METHOD_UMH,       "me"),
+    { "mb_size", "macroblock size", OFFSET(mb_size), AV_OPT_TYPE_INT, {.i64 = 16}, 4, 16, FLAGS },
+    { "search_param", "search parameter", OFFSET(search_param), AV_OPT_TYPE_INT, {.i64 = 32}, 4, INT_MAX, FLAGS },
+    { "vsbmc", "variable-size block motion compensation", OFFSET(vsbmc), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1, FLAGS },
+    { "scd", "scene change detection method", OFFSET(scd_method), AV_OPT_TYPE_INT, {.i64 = SCD_METHOD_FDIFF}, SCD_METHOD_NONE, SCD_METHOD_FDIFF, FLAGS, "scene" },
+        CONST("none",   "disable detection",                    SCD_METHOD_NONE,        "scene"),
+        CONST("fdiff",  "frame difference",                     SCD_METHOD_FDIFF,       "scene"),
+    { "scd_threshold", "scene change threshold", OFFSET(scd_threshold), AV_OPT_TYPE_DOUBLE, {.dbl = 5.0}, 0, 100.0, FLAGS },
+    { NULL }
+};
+
+AVFILTER_DEFINE_CLASS(minterpolate);
+
+static int query_formats(AVFilterContext *ctx)
+{
+    static const enum AVPixelFormat pix_fmts[] = {
+        AV_PIX_FMT_YUV410P, AV_PIX_FMT_YUV411P,
+        AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUV422P,
+        AV_PIX_FMT_YUV440P, AV_PIX_FMT_YUV444P,
+        AV_PIX_FMT_YUVJ444P, AV_PIX_FMT_YUVJ440P,
+        AV_PIX_FMT_YUVJ422P, AV_PIX_FMT_YUVJ420P,
+        AV_PIX_FMT_YUVJ411P,
+        AV_PIX_FMT_YUVA420P, AV_PIX_FMT_YUVA422P, AV_PIX_FMT_YUVA444P,
+        AV_PIX_FMT_GRAY8,
+        AV_PIX_FMT_NONE
+    };
+
+    AVFilterFormats *fmts_list = ff_make_format_list(pix_fmts);
+    if (!fmts_list)
+        return AVERROR(ENOMEM);
+    return ff_set_common_formats(ctx, fmts_list);
+}
+
+static uint64_t get_sbad(AVMotionEstContext *me_ctx, int x, int y, int x_mv, int y_mv)
+{
+    uint8_t *data_cur = me_ctx->data_cur;
+    uint8_t *data_next = me_ctx->data_ref;
+    int linesize = me_ctx->linesize;
+    int mv_x1 = x_mv - x;
+    int mv_y1 = y_mv - y;
+    int mv_x, mv_y, i, j;
+    uint64_t sbad = 0;
+
+    x = av_clip(x, me_ctx->x_min, me_ctx->x_max);
+    y = av_clip(y, me_ctx->y_min, me_ctx->y_max);
+    mv_x = av_clip(x_mv - x, -FFMIN(x - me_ctx->x_min, me_ctx->x_max - x), FFMIN(x - me_ctx->x_min, me_ctx->x_max - x));
+    mv_y = av_clip(y_mv - y, -FFMIN(y - me_ctx->y_min, me_ctx->y_max - y), FFMIN(y - me_ctx->y_min, me_ctx->y_max - y));
+
+    data_cur += (y + mv_y) * linesize;
+    data_next += (y - mv_y) * linesize;
+
+    for (j = 0; j < me_ctx->mb_size; j++)
+        for (i = 0; i < me_ctx->mb_size; i++)
+            sbad += FFABS(data_cur[x + mv_x + i + j * linesize] - data_next[x - mv_x + i + j * linesize]);
+
+    return sbad + (FFABS(mv_x1 - me_ctx->pred_x) + FFABS(mv_y1 - me_ctx->pred_y)) * COST_PRED_SCALE;
+}
+
+static uint64_t get_sbad_ob(AVMotionEstContext *me_ctx, int x, int y, int x_mv, int y_mv)
+{
+    uint8_t *data_cur = me_ctx->data_cur;
+    uint8_t *data_next = me_ctx->data_ref;
+    int linesize = me_ctx->linesize;
+    int x_min = me_ctx->x_min + me_ctx->mb_size / 2;
+    int x_max = me_ctx->x_max - me_ctx->mb_size / 2;
+    int y_min = me_ctx->y_min + me_ctx->mb_size / 2;
+    int y_max = me_ctx->y_max - me_ctx->mb_size / 2;
+    int mv_x1 = x_mv - x;
+    int mv_y1 = y_mv - y;
+    int mv_x, mv_y, i, j;
+    uint64_t sbad = 0;
+
+    x = av_clip(x, x_min, x_max);
+    y = av_clip(y, y_min, y_max);
+    mv_x = av_clip(x_mv - x, -FFMIN(x - x_min, x_max - x), FFMIN(x - x_min, x_max - x));
+    mv_y = av_clip(y_mv - y, -FFMIN(y - y_min, y_max - y), FFMIN(y - y_min, y_max - y));
+
+    for (j = -me_ctx->mb_size / 2; j < me_ctx->mb_size * 3 / 2; j++)
+        for (i = -me_ctx->mb_size / 2; i < me_ctx->mb_size * 3 / 2; i++)
+            sbad += FFABS(data_cur[x + mv_x + i + (y + mv_y + j) * linesize] - data_next[x - mv_x + i + (y - mv_y + j) * linesize]);
+
+    return sbad + (FFABS(mv_x1 - me_ctx->pred_x) + FFABS(mv_y1 - me_ctx->pred_y)) * COST_PRED_SCALE;
+}
+
+static uint64_t get_sad_ob(AVMotionEstContext *me_ctx, int x, int y, int x_mv, int y_mv)
+{
+    uint8_t *data_ref = me_ctx->data_ref;
+    uint8_t *data_cur = me_ctx->data_cur;
+    int linesize = me_ctx->linesize;
+    int x_min = me_ctx->x_min + me_ctx->mb_size / 2;
+    int x_max = me_ctx->x_max - me_ctx->mb_size / 2;
+    int y_min = me_ctx->y_min + me_ctx->mb_size / 2;
+    int y_max = me_ctx->y_max - me_ctx->mb_size / 2;
+    int mv_x = x_mv - x;
+    int mv_y = y_mv - y;
+    int i, j;
+    uint64_t sad = 0;
+
+    x = av_clip(x, x_min, x_max);
+    y = av_clip(y, y_min, y_max);
+    x_mv = av_clip(x_mv, x_min, x_max);
+    y_mv = av_clip(y_mv, y_min, y_max);
+
+    for (j = -me_ctx->mb_size / 2; j < me_ctx->mb_size * 3 / 2; j++)
+        for (i = -me_ctx->mb_size / 2; i < me_ctx->mb_size * 3 / 2; i++)
+            sad += FFABS(data_ref[x_mv + i + (y_mv + j) * linesize] - data_cur[x + i + (y + j) * linesize]);
+
+    return sad + (FFABS(mv_x - me_ctx->pred_x) + FFABS(mv_y - me_ctx->pred_y)) * COST_PRED_SCALE;
+}
+
+static int config_input(AVFilterLink *inlink)
+{
+    MIContext *mi_ctx = inlink->dst->priv;
+    AVMotionEstContext *me_ctx = &mi_ctx->me_ctx;
+    const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(inlink->format);
+    const int height = inlink->h;
+    const int width  = inlink->w;
+    int i;
+
+    mi_ctx->chroma_height = AV_CEIL_RSHIFT(inlink->h, desc->log2_chroma_h);
+    mi_ctx->chroma_width = AV_CEIL_RSHIFT(inlink->w, desc->log2_chroma_w);
+
+    avcodec_get_chroma_sub_sample(inlink->format, &mi_ctx->chroma_h_shift, &mi_ctx->chroma_v_shift);
+
+    mi_ctx->nb_planes = av_pix_fmt_count_planes(inlink->format);
+
+    mi_ctx->log2_mb_size = av_ceil_log2_c(mi_ctx->mb_size);
+    mi_ctx->mb_size = 1 << mi_ctx->log2_mb_size;
+
+    mi_ctx->b_width  = width >> mi_ctx->log2_mb_size;
+    mi_ctx->b_height = height >> mi_ctx->log2_mb_size;
+    mi_ctx->b_count = mi_ctx->b_width * mi_ctx->b_height;
+
+    for (i = 0; i < NB_FRAMES; i++) {
+        Frame *frame = &mi_ctx->frames[i];
+        frame->blocks = av_mallocz_array(mi_ctx->b_count, sizeof(Block));
+        if (!frame->blocks)
+            return AVERROR(ENOMEM);
+    }
+
+    if (mi_ctx->mi_mode == MI_MODE_MCI) {
+        if (!(mi_ctx->pixels = av_mallocz_array(width * height, sizeof(Pixel))))
+            return AVERROR(ENOMEM);
+
+        if (mi_ctx->me_mode == ME_MODE_BILAT)
+            if (!(mi_ctx->int_blocks = av_mallocz_array(mi_ctx->b_count, sizeof(Block))))
+                return AVERROR(ENOMEM);
+
+        if (mi_ctx->me_method == AV_ME_METHOD_EPZS) {
+            for (i = 0; i < 3; i++) {
+                mi_ctx->mv_table[i] = av_mallocz_array(mi_ctx->b_count, sizeof(*mi_ctx->mv_table[0]));
+                if (!mi_ctx->mv_table[i])
+                    return AVERROR(ENOMEM);
+            }
+        }
+    }
+
+    if (mi_ctx->scd_method == SCD_METHOD_FDIFF) {
+        mi_ctx->sad = av_pixelutils_get_sad_fn(3, 3, 2, mi_ctx);
+        if (!mi_ctx->sad)
+            return AVERROR(EINVAL);
+    }
+
+    ff_me_init_context(me_ctx, mi_ctx->mb_size, mi_ctx->search_param, width, height, 0, (mi_ctx->b_width - 1) << mi_ctx->log2_mb_size, 0, (mi_ctx->b_height - 1) << mi_ctx->log2_mb_size);
+
+    if (mi_ctx->me_mode == ME_MODE_BIDIR)
+        me_ctx->get_cost = &get_sad_ob;
+    else if (mi_ctx->me_mode == ME_MODE_BILAT)
+        me_ctx->get_cost = &get_sbad_ob;
+
+    return 0;
+}
+
+static int config_output(AVFilterLink *outlink)
+{
+    MIContext *mi_ctx = outlink->src->priv;
+
+    outlink->frame_rate = mi_ctx->frame_rate;
+    outlink->time_base  = av_inv_q(mi_ctx->frame_rate);
+
+    return 0;
+}
+
+#define ADD_PRED(preds, px, py)\
+    do {\
+        preds.mvs[preds.nb][0] = px;\
+        preds.mvs[preds.nb][1] = py;\
+        preds.nb++;\
+    } while(0)
+
+static void search_mv(MIContext *mi_ctx, Block *blocks, int mb_x, int mb_y, int dir)
+{
+    AVMotionEstContext *me_ctx = &mi_ctx->me_ctx;
+    AVMotionEstPredictor *preds = me_ctx->preds;
+    Block *block = &blocks[mb_x + mb_y * mi_ctx->b_width];
+
+    const int x_mb = mb_x << mi_ctx->log2_mb_size;
+    const int y_mb = mb_y << mi_ctx->log2_mb_size;
+    const int mb_i = mb_x + mb_y * mi_ctx->b_width;
+    int mv[2] = {x_mb, y_mb};
+
+    switch (mi_ctx->me_method) {
+        case AV_ME_METHOD_ESA:
+            ff_me_search_esa(me_ctx, x_mb, y_mb, mv);
+            break;
+        case AV_ME_METHOD_TSS:
+            ff_me_search_tss(me_ctx, x_mb, y_mb, mv);
+            break;
+        case AV_ME_METHOD_TDLS:
+            ff_me_search_tdls(me_ctx, x_mb, y_mb, mv);
+            break;
+        case AV_ME_METHOD_NTSS:
+            ff_me_search_ntss(me_ctx, x_mb, y_mb, mv);
+            break;
+        case AV_ME_METHOD_FSS:
+            ff_me_search_fss(me_ctx, x_mb, y_mb, mv);
+            break;
+        case AV_ME_METHOD_DS:
+            ff_me_search_ds(me_ctx, x_mb, y_mb, mv);
+            break;
+        case AV_ME_METHOD_HEXBS:
+            ff_me_search_hexbs(me_ctx, x_mb, y_mb, mv);
+            break;
+        case AV_ME_METHOD_EPZS:
+
+            preds[0].nb = 0;
+            preds[1].nb = 0;
+
+            ADD_PRED(preds[0], 0, 0);
+
+            //left mb in current frame
+            if (mb_x > 0)
+                ADD_PRED(preds[0], mi_ctx->mv_table[0][mb_i - 1][dir][0], mi_ctx->mv_table[0][mb_i - 1][dir][1]);
+
+            //top mb in current frame
+            if (mb_y > 0)
+                ADD_PRED(preds[0], mi_ctx->mv_table[0][mb_i - mi_ctx->b_width][dir][0], mi_ctx->mv_table[0][mb_i - mi_ctx->b_width][dir][1]);
+
+            //top-right mb in current frame
+            if (mb_y > 0 && mb_x + 1 < mi_ctx->b_width)
+                ADD_PRED(preds[0], mi_ctx->mv_table[0][mb_i - mi_ctx->b_width + 1][dir][0], mi_ctx->mv_table[0][mb_i - mi_ctx->b_width + 1][dir][1]);
+
+            //median predictor
+            if (preds[0].nb == 4) {
+                me_ctx->pred_x = mid_pred(preds[0].mvs[1][0], preds[0].mvs[2][0], preds[0].mvs[3][0]);
+                me_ctx->pred_y = mid_pred(preds[0].mvs[1][1], preds[0].mvs[2][1], preds[0].mvs[3][1]);
+            } else if (preds[0].nb == 3) {
+                me_ctx->pred_x = mid_pred(0, preds[0].mvs[1][0], preds[0].mvs[2][0]);
+                me_ctx->pred_y = mid_pred(0, preds[0].mvs[1][1], preds[0].mvs[2][1]);
+            } else if (preds[0].nb == 2) {
+                me_ctx->pred_x = preds[0].mvs[1][0];
+                me_ctx->pred_y = preds[0].mvs[1][1];
+            } else {
+                me_ctx->pred_x = 0;
+                me_ctx->pred_y = 0;
+            }
+
+            //collocated mb in prev frame
+            ADD_PRED(preds[0], mi_ctx->mv_table[1][mb_i][dir][0], mi_ctx->mv_table[1][mb_i][dir][1]);
+
+            //accelerator motion vector of collocated block in prev frame
+            ADD_PRED(preds[1], mi_ctx->mv_table[1][mb_i][dir][0] + (mi_ctx->mv_table[1][mb_i][dir][0] - mi_ctx->mv_table[2][mb_i][dir][0]),
+                               mi_ctx->mv_table[1][mb_i][dir][1] + (mi_ctx->mv_table[1][mb_i][dir][1] - mi_ctx->mv_table[2][mb_i][dir][1]));
+
+            //left mb in prev frame
+            if (mb_x > 0)
+                ADD_PRED(preds[1], mi_ctx->mv_table[1][mb_i - 1][dir][0], mi_ctx->mv_table[1][mb_i - 1][dir][1]);
+
+            //top mb in prev frame
+            if (mb_y > 0)
+                ADD_PRED(preds[1], mi_ctx->mv_table[1][mb_i - mi_ctx->b_width][dir][0], mi_ctx->mv_table[1][mb_i - mi_ctx->b_width][dir][1]);
+
+            //right mb in prev frame
+            if (mb_x + 1 < mi_ctx->b_width)
+                ADD_PRED(preds[1], mi_ctx->mv_table[1][mb_i + 1][dir][0], mi_ctx->mv_table[1][mb_i + 1][dir][1]);
+
+            //bottom mb in prev frame
+            if (mb_y + 1 < mi_ctx->b_height)
+                ADD_PRED(preds[1], mi_ctx->mv_table[1][mb_i + mi_ctx->b_width][dir][0], mi_ctx->mv_table[1][mb_i + mi_ctx->b_width][dir][1]);
+
+            ff_me_search_epzs(me_ctx, x_mb, y_mb, mv);
+
+            mi_ctx->mv_table[0][mb_i][dir][0] = mv[0] - x_mb;
+            mi_ctx->mv_table[0][mb_i][dir][1] = mv[1] - y_mb;
+
+            break;
+        case AV_ME_METHOD_UMH:
+
+            preds[0].nb = 0;
+
+            ADD_PRED(preds[0], 0, 0);
+
+            //left mb in current frame
+            if (mb_x > 0)
+                ADD_PRED(preds[0], blocks[mb_i - 1].mvs[dir][0], blocks[mb_i - 1].mvs[dir][1]);
+
+            if (mb_y > 0) {
+                //top mb in current frame
+                ADD_PRED(preds[0], blocks[mb_i - mi_ctx->b_width].mvs[dir][0], blocks[mb_i - mi_ctx->b_width].mvs[dir][1]);
+
+                //top-right mb in current frame
+                if (mb_x + 1 < mi_ctx->b_width)
+                    ADD_PRED(preds[0], blocks[mb_i - mi_ctx->b_width + 1].mvs[dir][0], blocks[mb_i - mi_ctx->b_width + 1].mvs[dir][1]);
+                //top-left mb in current frame
+                else if (mb_x > 0)
+                    ADD_PRED(preds[0], blocks[mb_i - mi_ctx->b_width - 1].mvs[dir][0], blocks[mb_i - mi_ctx->b_width - 1].mvs[dir][1]);
+            }
+
+            //median predictor
+            if (preds[0].nb == 4) {
+                me_ctx->pred_x = mid_pred(preds[0].mvs[1][0], preds[0].mvs[2][0], preds[0].mvs[3][0]);
+                me_ctx->pred_y = mid_pred(preds[0].mvs[1][1], preds[0].mvs[2][1], preds[0].mvs[3][1]);
+            } else if (preds[0].nb == 3) {
+                me_ctx->pred_x = mid_pred(0, preds[0].mvs[1][0], preds[0].mvs[2][0]);
+                me_ctx->pred_y = mid_pred(0, preds[0].mvs[1][1], preds[0].mvs[2][1]);
+            } else if (preds[0].nb == 2) {
+                me_ctx->pred_x = preds[0].mvs[1][0];
+                me_ctx->pred_y = preds[0].mvs[1][1];
+            } else {
+                me_ctx->pred_x = 0;
+                me_ctx->pred_y = 0;
+            }
+
+            ff_me_search_umh(me_ctx, x_mb, y_mb, mv);
+
+            break;
+    }
+
+    block->mvs[dir][0] = mv[0] - x_mb;
+    block->mvs[dir][1] = mv[1] - y_mb;
+}
+
+static void bilateral_me(MIContext *mi_ctx)
+{
+    Block *block;
+    int mb_x, mb_y;
+
+    for (mb_y = 0; mb_y < mi_ctx->b_height; mb_y++)
+        for (mb_x = 0; mb_x < mi_ctx->b_width; mb_x++) {
+            block = &mi_ctx->int_blocks[mb_x + mb_y * mi_ctx->b_width];
+
+            block->cid = 0;
+            block->sb = 0;
+
+            block->mvs[0][0] = 0;
+            block->mvs[0][1] = 0;
+        }
+
+    for (mb_y = 0; mb_y < mi_ctx->b_height; mb_y++)
+        for (mb_x = 0; mb_x < mi_ctx->b_width; mb_x++)
+            search_mv(mi_ctx, mi_ctx->int_blocks, mb_x, mb_y, 0);
+}
+
+static int var_size_bme(MIContext *mi_ctx, Block *block, int x_mb, int y_mb, int n)
+{
+    AVMotionEstContext *me_ctx = &mi_ctx->me_ctx;
+    uint64_t cost_sb, cost_old;
+    int mb_size = me_ctx->mb_size;
+    int search_param = me_ctx->search_param;
+    int mv_x, mv_y;
+    int x, y;
+    int ret;
+
+    me_ctx->mb_size = 1 << n;
+    cost_old = me_ctx->get_cost(me_ctx, x_mb, y_mb, x_mb + block->mvs[0][0], y_mb + block->mvs[0][1]);
+    me_ctx->mb_size = mb_size;
+
+    if (!cost_old) {
+        block->sb = 0;
+        return 0;
+    }
+
+    if (!block->subs) {
+        block->subs = av_mallocz_array(4, sizeof(Block));
+        if (!block->subs)
+            return AVERROR(ENOMEM);
+    }
+
+    block->sb = 1;
+
+    for (y = 0; y < 2; y++)
+        for (x = 0; x < 2; x++) {
+            Block *sb = &block->subs[x + y * 2];
+            int mv[2] = {x_mb + block->mvs[0][0], y_mb + block->mvs[0][1]};
+
+            me_ctx->mb_size = 1 << (n - 1);
+            me_ctx->search_param = 2;
+            me_ctx->pred_x = block->mvs[0][0];
+            me_ctx->pred_y = block->mvs[0][1];
+
+            cost_sb = ff_me_search_ds(&mi_ctx->me_ctx, x_mb + block->mvs[0][0], y_mb + block->mvs[0][1], mv);
+            mv_x = mv[0] - x_mb;
+            mv_y = mv[1] - y_mb;
+
+            me_ctx->mb_size = mb_size;
+            me_ctx->search_param = search_param;
+
+            if (cost_sb < cost_old / 4) {
+                sb->mvs[0][0] = mv_x;
+                sb->mvs[0][1] = mv_y;
+
+                if (n > 1) {
+                    if (ret = var_size_bme(mi_ctx, sb, x_mb + (x << (n - 1)), y_mb + (y << (n - 1)), n - 1))
+                        return ret;
+                } else
+                    sb->sb = 0;
+            } else {
+                block->sb = 0;
+                return 0;
+            }
+        }
+
+    return 0;
+}
+
+static int cluster_mvs(MIContext *mi_ctx)
+{
+    int changed, c, c_max = 0;
+    int mb_x, mb_y, x, y;
+    int mv_x, mv_y, avg_x, avg_y, dx, dy;
+    int d, ret;
+    Block *block;
+    Cluster *cluster, *cluster_new;
+
+    do {
+        changed = 0;
+        for (mb_y = 0; mb_y < mi_ctx->b_height; mb_y++)
+            for (mb_x = 0; mb_x < mi_ctx->b_width; mb_x++) {
+                block = &mi_ctx->int_blocks[mb_x + mb_y * mi_ctx->b_width];
+                c = block->cid;
+                cluster = &mi_ctx->clusters[c];
+                mv_x = block->mvs[0][0];
+                mv_y = block->mvs[0][1];
+
+                if (cluster->nb < 2)
+                    continue;
+
+                avg_x = cluster->sum[0] / cluster->nb;
+                avg_y = cluster->sum[1] / cluster->nb;
+                dx = avg_x - mv_x;
+                dy = avg_y - mv_y;
+
+                if (FFABS(avg_x - mv_x) > CLUSTER_THRESHOLD || FFABS(avg_y - mv_y) > CLUSTER_THRESHOLD) {
+
+                    for (d = 1; d < 5; d++)
+                        for (y = FFMAX(mb_y - d, 0); y < FFMIN(mb_y + d + 1, mi_ctx->b_height); y++)
+                            for (x = FFMAX(mb_x - d, 0); x < FFMIN(mb_x + d + 1, mi_ctx->b_width); x++) {
+                                Block *nb = &mi_ctx->int_blocks[x + y * mi_ctx->b_width];
+                                if (nb->cid > block->cid) {
+                                    if (nb->cid < c || c == block->cid)
+                                        c = nb->cid;
+                                }
+                            }
+
+                    if (c == block->cid)
+                        c = c_max + 1;
+
+                    if (c >= NB_CLUSTERS) {
+                        continue;
+                    }
+
+                    cluster_new = &mi_ctx->clusters[c];
+                    cluster_new->sum[0] += mv_x;
+                    cluster_new->sum[1] += mv_y;
+                    cluster->sum[0] -= mv_x;
+                    cluster->sum[1] -= mv_y;
+                    cluster_new->nb++;
+                    cluster->nb--;
+
+                    c_max = FFMAX(c_max, c);
+                    block->cid = c;
+
+                    changed = 1;
+                }
+            }
+    } while (changed);
+
+    /* find boundaries */
+    for (mb_y = 0; mb_y < mi_ctx->b_height; mb_y++)
+        for (mb_x = 0; mb_x < mi_ctx->b_width; mb_x++) {
+            block = &mi_ctx->int_blocks[mb_x + mb_y * mi_ctx->b_width];
+            for (y = FFMAX(mb_y - 1, 0); y < FFMIN(mb_y + 2, mi_ctx->b_height); y++)
+                for (x = FFMAX(mb_x - 1, 0); x < FFMIN(mb_x + 2, mi_ctx->b_width); x++) {
+                    dx = x - mb_x;
+                    dy = y - mb_y;
+
+                    if ((x - mb_x) && (y - mb_y) || !dx && !dy)
+                        continue;
+
+                    if (!mb_x || !mb_y || mb_x == mi_ctx->b_width - 1 || mb_y == mi_ctx->b_height - 1)
+                        continue;
+
+                    if (block->cid != mi_ctx->int_blocks[x + y * mi_ctx->b_width].cid) {
+                        if (!dx && block->cid == mi_ctx->int_blocks[x + (mb_y - dy) * mi_ctx->b_width].cid ||
+                            !dy && block->cid == mi_ctx->int_blocks[(mb_x - dx) + y * mi_ctx->b_width].cid) {
+                            if (ret = var_size_bme(mi_ctx, block, mb_x << mi_ctx->log2_mb_size, mb_y << mi_ctx->log2_mb_size, mi_ctx->log2_mb_size))
+                                return ret;
+                        }
+                    }
+                }
+        }
+
+    return 0;
+}
+
+static int inject_frame(AVFilterLink *inlink, AVFrame *avf_in)
+{
+    AVFilterContext *ctx = inlink->dst;
+    MIContext *mi_ctx = ctx->priv;
+    Frame frame_tmp, *frame;
+    int mb_x, mb_y, dir;
+
+    av_frame_free(&mi_ctx->frames[0].avf);
+    frame_tmp = mi_ctx->frames[0];
+    memmove(&mi_ctx->frames[0], &mi_ctx->frames[1], sizeof(mi_ctx->frames[0]) * (NB_FRAMES - 1));
+    mi_ctx->frames[NB_FRAMES - 1] = frame_tmp;
+    mi_ctx->frames[NB_FRAMES - 1].avf = avf_in;
+    frame = &mi_ctx->frames[NB_FRAMES - 1];
+
+    if (mi_ctx->mi_mode == MI_MODE_MCI) {
+
+        if (mi_ctx->me_method == AV_ME_METHOD_EPZS) {
+            mi_ctx->mv_table[2] = memcpy(mi_ctx->mv_table[2], mi_ctx->mv_table[1], sizeof(*mi_ctx->mv_table[1]) * mi_ctx->b_count);
+            mi_ctx->mv_table[1] = memcpy(mi_ctx->mv_table[1], mi_ctx->mv_table[0], sizeof(*mi_ctx->mv_table[0]) * mi_ctx->b_count);
+        }
+
+        if (mi_ctx->me_mode == ME_MODE_BIDIR) {
+
+            if (mi_ctx->frames[1].avf) {
+                for (dir = 0; dir < 2; dir++) {
+                    mi_ctx->me_ctx.linesize = mi_ctx->frames[2].avf->linesize[0];
+                    mi_ctx->me_ctx.data_cur = mi_ctx->frames[2].avf->data[0];
+                    mi_ctx->me_ctx.data_ref = mi_ctx->frames[dir ? 3 : 1].avf->data[0];
+
+                    for (mb_y = 0; mb_y < mi_ctx->b_height; mb_y++)
+                        for (mb_x = 0; mb_x < mi_ctx->b_width; mb_x++)
+                            search_mv(mi_ctx, mi_ctx->frames[2].blocks, mb_x, mb_y, dir);
+                }
+            }
+
+        } else if (mi_ctx->me_mode == ME_MODE_BILAT) {
+            Block *block;
+            int i, ret;
+
+            if (!mi_ctx->frames[0].avf)
+                return 0;
+
+            mi_ctx->me_ctx.linesize = mi_ctx->frames[0].avf->linesize[0];
+            mi_ctx->me_ctx.data_cur = mi_ctx->frames[1].avf->data[0];
+            mi_ctx->me_ctx.data_ref = mi_ctx->frames[2].avf->data[0];
+
+            bilateral_me(mi_ctx);
+
+            if (mi_ctx->mc_mode == MC_MODE_AOBMC) {
+
+                for (mb_y = 0; mb_y < mi_ctx->b_height; mb_y++)
+                    for (mb_x = 0; mb_x < mi_ctx->b_width; mb_x++) {
+                        int x_mb = mb_x << mi_ctx->log2_mb_size;
+                        int y_mb = mb_y << mi_ctx->log2_mb_size;
+                        block = &mi_ctx->int_blocks[mb_x + mb_y * mi_ctx->b_width];
+
+                        block->sbad = get_sbad(&mi_ctx->me_ctx, x_mb, y_mb, x_mb + block->mvs[0][0], y_mb + block->mvs[0][1]);
+                    }
+            }
+
+            if (mi_ctx->vsbmc) {
+
+                for (i = 0; i < NB_CLUSTERS; i++) {
+                    mi_ctx->clusters[i].sum[0] = 0;
+                    mi_ctx->clusters[i].sum[1] = 0;
+                    mi_ctx->clusters[i].nb = 0;
+                }
+
+                for (mb_y = 0; mb_y < mi_ctx->b_height; mb_y++)
+                    for (mb_x = 0; mb_x < mi_ctx->b_width; mb_x++) {
+                        block = &mi_ctx->int_blocks[mb_x + mb_y * mi_ctx->b_width];
+
+                        mi_ctx->clusters[0].sum[0] += block->mvs[0][0];
+                        mi_ctx->clusters[0].sum[1] += block->mvs[0][1];
+                    }
+
+                mi_ctx->clusters[0].nb = mi_ctx->b_count;
+
+                if (ret = cluster_mvs(mi_ctx))
+                    return ret;
+            }
+        }
+    }
+
+    return 0;
+}
+
+static int detect_scene_change(MIContext *mi_ctx)
+{
+    AVMotionEstContext *me_ctx = &mi_ctx->me_ctx;
+    int x, y;
+    int linesize = me_ctx->linesize;
+    uint8_t *p1 = mi_ctx->frames[1].avf->data[0];
+    uint8_t *p2 = mi_ctx->frames[2].avf->data[0];
+
+    if (mi_ctx->scd_method == SCD_METHOD_FDIFF) {
+        double ret = 0, mafd, diff;
+        int64_t sad;
+
+        for (sad = y = 0; y < me_ctx->height; y += 8)
+            for (x = 0; x < linesize; x += 8)
+                sad += mi_ctx->sad(p1 + x + y * linesize, linesize, p2 + x + y * linesize, linesize);
+
+        emms_c();
+        mafd = (double) sad / (me_ctx->height * me_ctx->width * 3);
+        diff = fabs(mafd - mi_ctx->prev_mafd);
+        ret  = av_clipf(FFMIN(mafd, diff), 0, 100.0);
+        mi_ctx->prev_mafd = mafd;
+
+        return ret >= mi_ctx->scd_threshold;
+    }
+
+    return 0;
+}
+
+#define ADD_PIXELS(b_weight, mv_x, mv_y)\
+    do {\
+        if (!b_weight || pixel->nb + 1 >= NB_PIXEL_MVS)\
+            continue;\
+        pixel->refs[pixel->nb] = 1;\
+        pixel->weights[pixel->nb] = b_weight * (ALPHA_MAX - alpha);\
+        pixel->mvs[pixel->nb][0] = av_clip((mv_x * alpha) / ALPHA_MAX, x_min, x_max);\
+        pixel->mvs[pixel->nb][1] = av_clip((mv_y * alpha) / ALPHA_MAX, y_min, y_max);\
+        pixel->nb++;\
+        pixel->refs[pixel->nb] = 2;\
+        pixel->weights[pixel->nb] = b_weight * alpha;\
+        pixel->mvs[pixel->nb][0] = av_clip(-mv_x * (ALPHA_MAX - alpha) / ALPHA_MAX, x_min, x_max);\
+        pixel->mvs[pixel->nb][1] = av_clip(-mv_y * (ALPHA_MAX - alpha) / ALPHA_MAX, y_min, y_max);\
+        pixel->nb++;\
+    } while(0)
+
+static void bidirectional_obmc(MIContext *mi_ctx, int alpha)
+{
+    int x, y;
+    int width = mi_ctx->frames[0].avf->width;
+    int height = mi_ctx->frames[0].avf->height;
+    int mb_y, mb_x, dir;
+
+    for (y = 0; y < height; y++)
+        for (x = 0; x < width; x++)
+            mi_ctx->pixels[x + y * width].nb = 0;
+
+    for (dir = 0; dir < 2; dir++)
+        for (mb_y = 0; mb_y < mi_ctx->b_height; mb_y++)
+            for (mb_x = 0; mb_x < mi_ctx->b_width; mb_x++) {
+                int a = dir ? alpha : (ALPHA_MAX - alpha);
+                int mv_x = mi_ctx->frames[2 - dir].blocks[mb_x + mb_y * mi_ctx->b_width].mvs[dir][0];
+                int mv_y = mi_ctx->frames[2 - dir].blocks[mb_x + mb_y * mi_ctx->b_width].mvs[dir][1];
+                int start_x, start_y;
+                int startc_x, startc_y, endc_x, endc_y;
+
+                start_x = (mb_x << mi_ctx->log2_mb_size) - mi_ctx->mb_size / 2 + mv_x * a / ALPHA_MAX;
+                start_y = (mb_y << mi_ctx->log2_mb_size) - mi_ctx->mb_size / 2 + mv_y * a / ALPHA_MAX;
+
+                startc_x = av_clip(start_x, 0, width - 1);
+                startc_y = av_clip(start_y, 0, height - 1);
+                endc_x = av_clip(start_x + (2 << mi_ctx->log2_mb_size), 0, width - 1);
+                endc_y = av_clip(start_y + (2 << mi_ctx->log2_mb_size), 0, height - 1);
+
+                if (dir) {
+                    mv_x = -mv_x;
+                    mv_y = -mv_y;
+                }
+
+                for (y = startc_y; y < endc_y; y++) {
+                    int y_min = -y;
+                    int y_max = height - y - 1;
+                    for (x = startc_x; x < endc_x; x++) {
+                        int x_min = -x;
+                        int x_max = width - x - 1;
+                        int obmc_weight = obmc_tab_linear[4 - mi_ctx->log2_mb_size][(x - start_x) + ((y - start_y) << (mi_ctx->log2_mb_size + 1))];
+                        Pixel *pixel = &mi_ctx->pixels[x + y * width];
+
+                        ADD_PIXELS(obmc_weight, mv_x, mv_y);
+                    }
+                }
+            }
+}
+
+static void set_frame_data(MIContext *mi_ctx, int alpha, AVFrame *avf_out)
+{
+    int x, y, plane;
+
+    for (plane = 0; plane < mi_ctx->nb_planes; plane++) {
+        int width = avf_out->width;
+        int height = avf_out->height;
+        int chroma = plane == 1 || plane == 2;
+
+        for (y = 0; y < height; y++)
+            for (x = 0; x < width; x++) {
+                int x_mv, y_mv;
+                int weight_sum = 0;
+                int i, val = 0;
+                Pixel *pixel = &mi_ctx->pixels[x + y * avf_out->width];
+
+                for (i = 0; i < pixel->nb; i++)
+                    weight_sum += pixel->weights[i];
+
+                if (!weight_sum || !pixel->nb) {
+                    pixel->weights[0] = ALPHA_MAX - alpha;
+                    pixel->refs[0] = 1;
+                    pixel->mvs[0][0] = 0;
+                    pixel->mvs[0][1] = 0;
+                    pixel->weights[1] = alpha;
+                    pixel->refs[1] = 2;
+                    pixel->mvs[1][0] = 0;
+                    pixel->mvs[1][1] = 0;
+                    pixel->nb = 2;
+
+                    weight_sum = ALPHA_MAX;
+                }
+
+                for (i = 0; i < pixel->nb; i++) {
+                    Frame *frame = &mi_ctx->frames[pixel->refs[i]];
+                    if (chroma) {
+                        x_mv = (x >> mi_ctx->chroma_h_shift) + (pixel->mvs[i][0] >> mi_ctx->chroma_h_shift);
+                        y_mv = (y >> mi_ctx->chroma_v_shift) + (pixel->mvs[i][1] >> mi_ctx->chroma_v_shift);
+                    } else {
+                        x_mv = x + pixel->mvs[i][0];
+                        y_mv = y + pixel->mvs[i][1];
+                    }
+
+                    val += pixel->weights[i] * frame->avf->data[plane][x_mv + y_mv * frame->avf->linesize[plane]];
+                }
+
+                val = ROUNDED_DIV(val, weight_sum);
+
+                if (chroma)
+                    avf_out->data[plane][(x >> mi_ctx->chroma_h_shift) + (y >> mi_ctx->chroma_v_shift) * avf_out->linesize[plane]] = val;
+                else
+                    avf_out->data[plane][x + y * avf_out->linesize[plane]] = val;
+            }
+    }
+}
+
+static void var_size_bmc(MIContext *mi_ctx, Block *block, int x_mb, int y_mb, int n, int alpha)
+{
+    int sb_x, sb_y;
+    int width = mi_ctx->frames[0].avf->width;
+    int height = mi_ctx->frames[0].avf->height;
+
+    for (sb_y = 0; sb_y < 2; sb_y++)
+        for (sb_x = 0; sb_x < 2; sb_x++) {
+            Block *sb = &block->subs[sb_x + sb_y * 2];
+
+            if (sb->sb)
+                var_size_bmc(mi_ctx, sb, x_mb + (sb_x << (n - 1)), y_mb + (sb_y << (n - 1)), n - 1, alpha);
+            else {
+                int x, y;
+                int mv_x = sb->mvs[0][0] * 2;
+                int mv_y = sb->mvs[0][1] * 2;
+
+                int start_x = x_mb + (sb_x << (n - 1));
+                int start_y = y_mb + (sb_y << (n - 1));
+                int end_x = start_x + (1 << (n - 1));
+                int end_y = start_y + (1 << (n - 1));
+
+                for (y = start_y; y < end_y; y++)  {
+                    int y_min = -y;
+                    int y_max = height - y - 1;
+                    for (x = start_x; x < end_x; x++) {
+                        int x_min = -x;
+                        int x_max = width - x - 1;
+                        Pixel *pixel = &mi_ctx->pixels[x + y * width];
+
+                        ADD_PIXELS(PX_WEIGHT_MAX, mv_x, mv_y);
+                    }
+                }
+            }
+        }
+}
+
+static void bilateral_obmc(MIContext *mi_ctx, Block *block, int mb_x, int mb_y, int alpha)
+{
+    int x, y;
+    int width = mi_ctx->frames[0].avf->width;
+    int height = mi_ctx->frames[0].avf->height;
+
+    Block *nb;
+    int nb_x, nb_y;
+    uint64_t sbads[9];
+
+    int mv_x = block->mvs[0][0] * 2;
+    int mv_y = block->mvs[0][1] * 2;
+    int start_x, start_y;
+    int startc_x, startc_y, endc_x, endc_y;
+
+    if (mi_ctx->mc_mode == MC_MODE_AOBMC)
+        for (nb_y = FFMAX(0, mb_y - 1); nb_y < FFMIN(mb_y + 2, mi_ctx->b_height); nb_y++)
+            for (nb_x = FFMAX(0, mb_x - 1); nb_x < FFMIN(mb_x + 2, mi_ctx->b_width); nb_x++) {
+                int x_nb = nb_x << mi_ctx->log2_mb_size;
+                int y_nb = nb_y << mi_ctx->log2_mb_size;
+
+                if (nb_x - mb_x || nb_y - mb_y)
+                    sbads[nb_x - mb_x + 1 + (nb_y - mb_y + 1) * 3] = get_sbad(&mi_ctx->me_ctx, x_nb, y_nb, x_nb + block->mvs[0][0], y_nb + block->mvs[0][1]);
+            }
+
+    start_x = (mb_x << mi_ctx->log2_mb_size) - mi_ctx->mb_size / 2;
+    start_y = (mb_y << mi_ctx->log2_mb_size) - mi_ctx->mb_size / 2;
+
+    startc_x = av_clip(start_x, 0, width - 1);
+    startc_y = av_clip(start_y, 0, height - 1);
+    endc_x = av_clip(start_x + (2 << mi_ctx->log2_mb_size), 0, width - 1);
+    endc_y = av_clip(start_y + (2 << mi_ctx->log2_mb_size), 0, height - 1);
+
+    for (y = startc_y; y < endc_y; y++) {
+        int y_min = -y;
+        int y_max = height - y - 1;
+        for (x = startc_x; x < endc_x; x++) {
+            int x_min = -x;
+            int x_max = width - x - 1;
+            int obmc_weight = obmc_tab_linear[4 - mi_ctx->log2_mb_size][(x - start_x) + ((y - start_y) << (mi_ctx->log2_mb_size + 1))];
+            Pixel *pixel = &mi_ctx->pixels[x + y * width];
+
+            if (mi_ctx->mc_mode == MC_MODE_AOBMC) {
+                nb_x = (((x - start_x) >> (mi_ctx->log2_mb_size - 1)) * 2 - 3) / 2;
+                nb_y = (((y - start_y) >> (mi_ctx->log2_mb_size - 1)) * 2 - 3) / 2;
+
+                if (nb_x || nb_y) {
+                    uint64_t sbad = sbads[nb_x + 1 + (nb_y + 1) * 3];
+                    nb = &mi_ctx->int_blocks[mb_x + nb_x + (mb_y + nb_y) * mi_ctx->b_width];
+
+                    if (sbad && sbad != UINT64_MAX && nb->sbad != UINT64_MAX) {
+                        int phi = av_clip(ALPHA_MAX * nb->sbad / sbad, 0, ALPHA_MAX);
+                        obmc_weight = obmc_weight * phi / ALPHA_MAX;
+                    }
+                }
+            }
+
+            ADD_PIXELS(obmc_weight, mv_x, mv_y);
+        }
+    }
+}
+
+static void interpolate(AVFilterLink *inlink, AVFrame *avf_out)
+{
+    AVFilterContext *ctx = inlink->dst;
+    AVFilterLink *outlink = ctx->outputs[0];
+    MIContext *mi_ctx = ctx->priv;
+    int x, y;
+    int plane, alpha;
+    int64_t pts;
+
+    pts = av_rescale(avf_out->pts, (int64_t) ALPHA_MAX * outlink->time_base.num * inlink->time_base.den,
+                                   (int64_t)             outlink->time_base.den * inlink->time_base.num);
+
+    alpha = (pts - mi_ctx->frames[1].avf->pts * ALPHA_MAX) / (mi_ctx->frames[2].avf->pts - mi_ctx->frames[1].avf->pts);
+    alpha = av_clip(alpha, 0, ALPHA_MAX);
+
+    if (alpha == 0 || alpha == ALPHA_MAX) {
+        av_frame_copy(avf_out, alpha ? mi_ctx->frames[2].avf : mi_ctx->frames[1].avf);
+        return;
+    }
+
+    if (mi_ctx->scene_changed) {
+        /* duplicate frame */
+        av_frame_copy(avf_out, alpha > ALPHA_MAX / 2 ? mi_ctx->frames[2].avf : mi_ctx->frames[1].avf);
+        return;
+    }
+
+    switch(mi_ctx->mi_mode) {
+        case MI_MODE_DUP:
+            av_frame_copy(avf_out, alpha > ALPHA_MAX / 2 ? mi_ctx->frames[2].avf : mi_ctx->frames[1].avf);
+
+            break;
+        case MI_MODE_BLEND:
+            for (plane = 0; plane < mi_ctx->nb_planes; plane++) {
+                int width = avf_out->width;
+                int height = avf_out->height;
+
+                if (plane == 1 || plane == 2) {
+                    width = mi_ctx->chroma_width;
+                    height = mi_ctx->chroma_height;
+                }
+
+                for (y = 0; y < height; y++) {
+                    for (x = 0; x < width; x++) {
+                        avf_out->data[plane][x + y * avf_out->linesize[plane]] =
+                                          alpha  * mi_ctx->frames[2].avf->data[plane][x + y * mi_ctx->frames[2].avf->linesize[plane]] +
+                            ((ALPHA_MAX - alpha) * mi_ctx->frames[1].avf->data[plane][x + y * mi_ctx->frames[1].avf->linesize[plane]] + 512) >> 10;
+                    }
+                }
+            }
+
+            break;
+        case MI_MODE_MCI:
+            if (mi_ctx->me_mode == ME_MODE_BIDIR) {
+                bidirectional_obmc(mi_ctx, alpha);
+                set_frame_data(mi_ctx, alpha, avf_out);
+
+            } else if (mi_ctx->me_mode == ME_MODE_BILAT) {
+                int mb_x, mb_y;
+                Block *block;
+
+                for (y = 0; y < mi_ctx->frames[0].avf->height; y++)
+                    for (x = 0; x < mi_ctx->frames[0].avf->width; x++)
+                        mi_ctx->pixels[x + y * mi_ctx->frames[0].avf->width].nb = 0;
+
+                for (mb_y = 0; mb_y < mi_ctx->b_height; mb_y++)
+                    for (mb_x = 0; mb_x < mi_ctx->b_width; mb_x++) {
+                        block = &mi_ctx->int_blocks[mb_x + mb_y * mi_ctx->b_width];
+
+                        if (block->sb)
+                            var_size_bmc(mi_ctx, block, mb_x << mi_ctx->log2_mb_size, mb_y << mi_ctx->log2_mb_size, mi_ctx->log2_mb_size, alpha);
+
+                        bilateral_obmc(mi_ctx, block, mb_x, mb_y, alpha);
+
+                    }
+
+                set_frame_data(mi_ctx, alpha, avf_out);
+            }
+
+            break;
+    }
+}
+
+static int filter_frame(AVFilterLink *inlink, AVFrame *avf_in)
+{
+    AVFilterContext *ctx = inlink->dst;
+    AVFilterLink *outlink = ctx->outputs[0];
+    MIContext *mi_ctx = ctx->priv;
+    int ret;
+
+    if (avf_in->pts == AV_NOPTS_VALUE) {
+        ret = ff_filter_frame(ctx->outputs[0], avf_in);
+        return ret;
+    }
+
+    if (!mi_ctx->frames[NB_FRAMES - 1].avf || avf_in->pts < mi_ctx->frames[NB_FRAMES - 1].avf->pts) {
+        av_log(ctx, AV_LOG_VERBOSE, "Initializing out pts from input pts %"PRId64"\n", avf_in->pts);
+        mi_ctx->out_pts = av_rescale_q(avf_in->pts, inlink->time_base, outlink->time_base);
+    }
+
+    if (!mi_ctx->frames[NB_FRAMES - 1].avf)
+        if (ret = inject_frame(inlink, av_frame_clone(avf_in)))
+            return ret;
+
+    if (ret = inject_frame(inlink, avf_in))
+        return ret;
+
+    if (!mi_ctx->frames[0].avf)
+        return 0;
+
+    mi_ctx->scene_changed = detect_scene_change(mi_ctx);
+
+    for (;;) {
+        AVFrame *avf_out;
+
+        if (av_compare_ts(mi_ctx->out_pts, outlink->time_base, mi_ctx->frames[2].avf->pts, inlink->time_base) > 0)
+            break;
+
+        if (!(avf_out = ff_get_video_buffer(ctx->outputs[0], inlink->w, inlink->h)))
+            return AVERROR(ENOMEM);
+
+        av_frame_copy_props(avf_out, mi_ctx->frames[NB_FRAMES - 1].avf);
+        avf_out->pts = mi_ctx->out_pts++;
+
+        interpolate(inlink, avf_out);
+
+        if ((ret = ff_filter_frame(ctx->outputs[0], avf_out)) < 0)
+            return ret;
+    }
+
+    return 0;
+}
+
+static av_cold void free_blocks(Block *block, int sb)
+{
+    if (block->subs)
+        free_blocks(block->subs, 1);
+    if (sb)
+        av_freep(&block);
+}
+
+static av_cold void uninit(AVFilterContext *ctx)
+{
+    MIContext *mi_ctx = ctx->priv;
+    int i, m;
+
+    av_freep(&mi_ctx->pixels);
+    if (mi_ctx->int_blocks)
+        for (m = 0; m < mi_ctx->b_count; m++)
+            free_blocks(&mi_ctx->int_blocks[m], 0);
+    av_freep(&mi_ctx->int_blocks);
+
+    for (i = 0; i < NB_FRAMES; i++) {
+        Frame *frame = &mi_ctx->frames[i];
+        av_freep(&frame->blocks);
+        av_frame_free(&frame->avf);
+    }
+
+    for (i = 0; i < 3; i++)
+        av_freep(&mi_ctx->mv_table[i]);
+}
+
+static const AVFilterPad minterpolate_inputs[] = {
+    {
+        .name          = "default",
+        .type          = AVMEDIA_TYPE_VIDEO,
+        .filter_frame  = filter_frame,
+        .config_props  = config_input,
+    },
+    { NULL }
+};
+
+static const AVFilterPad minterpolate_outputs[] = {
+    {
+        .name          = "default",
+        .type          = AVMEDIA_TYPE_VIDEO,
+        .config_props  = config_output,
+    },
+    { NULL }
+};
+
+AVFilter ff_vf_minterpolate = {
+    .name          = "minterpolate",
+    .description   = NULL_IF_CONFIG_SMALL("Frame rate conversion using Motion Interpolation."),
+    .priv_size     = sizeof(MIContext),
+    .priv_class    = &minterpolate_class,
+    .uninit        = uninit,
+    .query_formats = query_formats,
+    .inputs        = minterpolate_inputs,
+    .outputs       = minterpolate_outputs,
+};




More information about the ffmpeg-cvslog mailing list