mpegvideo_motion.c

Go to the documentation of this file.

/*
 * Copyright (c) 2000,2001 Fabrice Bellard
 * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
 *
 * 4MV & hq & B-frame encoding stuff by Michael Niedermayer <michaelni@gmx.at>
 *
 * 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 <string.h>

#include "libavutil/avassert.h"
#include "libavutil/internal.h"
#include "avcodec.h"
#include "h261.h"
#include "mpegutils.h"
#include "mpegvideo.h"
#include "mjpegenc.h"
#include "msmpeg4.h"
#include "qpeldsp.h"
#include "wmv2.h"
#include <limits.h>

static void gmc1_motion(MpegEncContext *s,
                        uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
                        uint8_t **ref_picture)
{
    uint8_t *ptr;
    int src_x, src_y, motion_x, motion_y;
    ptrdiff_t offset, linesize, uvlinesize;
    int emu = 0;

    motion_x   = s->sprite_offset[0][0];
    motion_y   = s->sprite_offset[0][1];
    src_x      = s->mb_x * 16 + (motion_x >> (s->sprite_warping_accuracy + 1));
    src_y      = s->mb_y * 16 + (motion_y >> (s->sprite_warping_accuracy + 1));
    motion_x <<= (3 - s->sprite_warping_accuracy);
    motion_y <<= (3 - s->sprite_warping_accuracy);
    src_x      = av_clip(src_x, -16, s->width);
    if (src_x == s->width)
        motion_x = 0;
    src_y = av_clip(src_y, -16, s->height);
    if (src_y == s->height)
        motion_y = 0;

    linesize   = s->linesize;
    uvlinesize = s->uvlinesize;

    ptr = ref_picture[0] + src_y * linesize + src_x;

    if ((unsigned)src_x >= FFMAX(s->h_edge_pos - 17, 0) ||
        (unsigned)src_y >= FFMAX(s->v_edge_pos - 17, 0)) {
        s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer, ptr,
                                 linesize, linesize,
                                 17, 17,
                                 src_x, src_y,
                                 s->h_edge_pos, s->v_edge_pos);
        ptr = s->sc.edge_emu_buffer;
    }

    if ((motion_x | motion_y) & 7) {
        s->mdsp.gmc1(dest_y, ptr, linesize, 16,
                     motion_x & 15, motion_y & 15, 128 - s->no_rounding);
        s->mdsp.gmc1(dest_y + 8, ptr + 8, linesize, 16,
                     motion_x & 15, motion_y & 15, 128 - s->no_rounding);
    } else {
        int dxy;

        dxy = ((motion_x >> 3) & 1) | ((motion_y >> 2) & 2);
        if (s->no_rounding) {
            s->hdsp.put_no_rnd_pixels_tab[0][dxy](dest_y, ptr, linesize, 16);
        } else {
            s->hdsp.put_pixels_tab[0][dxy](dest_y, ptr, linesize, 16);
        }
    }

    if (CONFIG_GRAY && s->avctx->flags & AV_CODEC_FLAG_GRAY)
        return;

    motion_x   = s->sprite_offset[1][0];
    motion_y   = s->sprite_offset[1][1];
    src_x      = s->mb_x * 8 + (motion_x >> (s->sprite_warping_accuracy + 1));
    src_y      = s->mb_y * 8 + (motion_y >> (s->sprite_warping_accuracy + 1));
    motion_x <<= (3 - s->sprite_warping_accuracy);
    motion_y <<= (3 - s->sprite_warping_accuracy);
    src_x      = av_clip(src_x, -8, s->width >> 1);
    if (src_x == s->width >> 1)
        motion_x = 0;
    src_y = av_clip(src_y, -8, s->height >> 1);
    if (src_y == s->height >> 1)
        motion_y = 0;

    offset = (src_y * uvlinesize) + src_x;
    ptr    = ref_picture[1] + offset;
    if ((unsigned)src_x >= FFMAX((s->h_edge_pos >> 1) - 9, 0) ||
        (unsigned)src_y >= FFMAX((s->v_edge_pos >> 1) - 9, 0)) {
        s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer, ptr,
                                 uvlinesize, uvlinesize,
                                 9, 9,
                                 src_x, src_y,
                                 s->h_edge_pos >> 1, s->v_edge_pos >> 1);
        ptr = s->sc.edge_emu_buffer;
        emu = 1;
    }
    s->mdsp.gmc1(dest_cb, ptr, uvlinesize, 8,
                 motion_x & 15, motion_y & 15, 128 - s->no_rounding);

    ptr = ref_picture[2] + offset;
    if (emu) {
        s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer, ptr,
                                 uvlinesize, uvlinesize,
                                 9, 9,
                                 src_x, src_y,
                                 s->h_edge_pos >> 1, s->v_edge_pos >> 1);
        ptr = s->sc.edge_emu_buffer;
    }
    s->mdsp.gmc1(dest_cr, ptr, uvlinesize, 8,
                 motion_x & 15, motion_y & 15, 128 - s->no_rounding);
}

static void gmc_motion(MpegEncContext *s,
                       uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
                       uint8_t **ref_picture)
{
    uint8_t *ptr;
    int linesize, uvlinesize;
    const int a = s->sprite_warping_accuracy;
    int ox, oy;

    linesize   = s->linesize;
    uvlinesize = s->uvlinesize;

    ptr = ref_picture[0];

    ox = s->sprite_offset[0][0] + s->sprite_delta[0][0] * s->mb_x * 16 +
         s->sprite_delta[0][1] * s->mb_y * 16;
    oy = s->sprite_offset[0][1] + s->sprite_delta[1][0] * s->mb_x * 16 +
         s->sprite_delta[1][1] * s->mb_y * 16;

    s->mdsp.gmc(dest_y, ptr, linesize, 16,
                ox, oy,
                s->sprite_delta[0][0], s->sprite_delta[0][1],
                s->sprite_delta[1][0], s->sprite_delta[1][1],
                a + 1, (1 << (2 * a + 1)) - s->no_rounding,
                s->h_edge_pos, s->v_edge_pos);
    s->mdsp.gmc(dest_y + 8, ptr, linesize, 16,
                ox + s->sprite_delta[0][0] * 8,
                oy + s->sprite_delta[1][0] * 8,
                s->sprite_delta[0][0], s->sprite_delta[0][1],
                s->sprite_delta[1][0], s->sprite_delta[1][1],
                a + 1, (1 << (2 * a + 1)) - s->no_rounding,
                s->h_edge_pos, s->v_edge_pos);

    if (CONFIG_GRAY && s->avctx->flags & AV_CODEC_FLAG_GRAY)
        return;

    ox = s->sprite_offset[1][0] + s->sprite_delta[0][0] * s->mb_x * 8 +
         s->sprite_delta[0][1] * s->mb_y * 8;
    oy = s->sprite_offset[1][1] + s->sprite_delta[1][0] * s->mb_x * 8 +
         s->sprite_delta[1][1] * s->mb_y * 8;

    ptr = ref_picture[1];
    s->mdsp.gmc(dest_cb, ptr, uvlinesize, 8,
                ox, oy,
                s->sprite_delta[0][0], s->sprite_delta[0][1],
                s->sprite_delta[1][0], s->sprite_delta[1][1],
                a + 1, (1 << (2 * a + 1)) - s->no_rounding,
                (s->h_edge_pos + 1) >> 1, (s->v_edge_pos + 1) >> 1);

    ptr = ref_picture[2];
    s->mdsp.gmc(dest_cr, ptr, uvlinesize, 8,
                ox, oy,
                s->sprite_delta[0][0], s->sprite_delta[0][1],
                s->sprite_delta[1][0], s->sprite_delta[1][1],
                a + 1, (1 << (2 * a + 1)) - s->no_rounding,
                (s->h_edge_pos + 1) >> 1, (s->v_edge_pos + 1) >> 1);
}

static inline int hpel_motion(MpegEncContext *s,
                              uint8_t *dest, uint8_t *src,
                              int src_x, int src_y,
                              op_pixels_func *pix_op,
                              int motion_x, int motion_y)
{
    int dxy = 0;
    int emu = 0;

    src_x += motion_x >> 1;
    src_y += motion_y >> 1;

    /* WARNING: do no forget half pels */
    src_x = av_clip(src_x, -16, s->width); // FIXME unneeded for emu?
    if (src_x != s->width)
        dxy |= motion_x & 1;
    src_y = av_clip(src_y, -16, s->height);
    if (src_y != s->height)
        dxy |= (motion_y & 1) << 1;
    src += src_y * s->linesize + src_x;

        if ((unsigned)src_x >= FFMAX(s->h_edge_pos - (motion_x & 1) - 7, 0) ||
            (unsigned)src_y >= FFMAX(s->v_edge_pos - (motion_y & 1) - 7, 0)) {
            s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer, src,
                                     s->linesize, s->linesize,
                                     9, 9,
                                     src_x, src_y,
                                     s->h_edge_pos, s->v_edge_pos);
            src = s->sc.edge_emu_buffer;
            emu = 1;
        }
    pix_op[dxy](dest, src, s->linesize, 8);
    return emu;
}

static av_always_inline
void mpeg_motion_internal(MpegEncContext *s,
                          uint8_t *dest_y,
                          uint8_t *dest_cb,
                          uint8_t *dest_cr,
                          int field_based,
                          int bottom_field,
                          int field_select,
                          uint8_t **ref_picture,
                          op_pixels_func (*pix_op)[4],
                          int motion_x,
                          int motion_y,
                          int h,
                          int is_mpeg12,
                          int mb_y)
{
    uint8_t *ptr_y, *ptr_cb, *ptr_cr;
    int dxy, uvdxy, mx, my, src_x, src_y,
        uvsrc_x, uvsrc_y, v_edge_pos;
    ptrdiff_t uvlinesize, linesize;

#if 0
    if (s->quarter_sample) {
        motion_x >>= 1;
        motion_y >>= 1;
    }
#endif

    v_edge_pos = s->v_edge_pos >> field_based;
    linesize   = s->current_picture.f->linesize[0] << field_based;
    uvlinesize = s->current_picture.f->linesize[1] << field_based;

    dxy   = ((motion_y & 1) << 1) | (motion_x & 1);
    src_x = s->mb_x * 16 + (motion_x >> 1);
    src_y = (mb_y << (4 - field_based)) + (motion_y >> 1);

    if (!is_mpeg12 && s->out_format == FMT_H263) {
        if ((s->workaround_bugs & FF_BUG_HPEL_CHROMA) && field_based) {
            mx      = (motion_x >> 1) | (motion_x & 1);
            my      = motion_y >> 1;
            uvdxy   = ((my & 1) << 1) | (mx & 1);
            uvsrc_x = s->mb_x * 8 + (mx >> 1);
            uvsrc_y = (mb_y << (3 - field_based)) + (my >> 1);
        } else {
            uvdxy   = dxy | (motion_y & 2) | ((motion_x & 2) >> 1);
            uvsrc_x = src_x >> 1;
            uvsrc_y = src_y >> 1;
        }
    // Even chroma mv's are full pel in H261
    } else if (!is_mpeg12 && s->out_format == FMT_H261) {
        mx      = motion_x / 4;
        my      = motion_y / 4;
        uvdxy   = 0;
        uvsrc_x = s->mb_x * 8 + mx;
        uvsrc_y = mb_y * 8 + my;
    } else {
        if (s->chroma_y_shift) {
            mx      = motion_x / 2;
            my      = motion_y / 2;
            uvdxy   = ((my & 1) << 1) | (mx & 1);
            uvsrc_x = s->mb_x * 8 + (mx >> 1);
            uvsrc_y = (mb_y << (3 - field_based)) + (my >> 1);
        } else {
            if (s->chroma_x_shift) {
                // Chroma422
                mx      = motion_x / 2;
                uvdxy   = ((motion_y & 1) << 1) | (mx & 1);
                uvsrc_x = s->mb_x * 8 + (mx >> 1);
                uvsrc_y = src_y;
            } else {
                // Chroma444
                uvdxy   = dxy;
                uvsrc_x = src_x;
                uvsrc_y = src_y;
            }
        }
    }

    ptr_y  = ref_picture[0] + src_y * linesize + src_x;
    ptr_cb = ref_picture[1] + uvsrc_y * uvlinesize + uvsrc_x;
    ptr_cr = ref_picture[2] + uvsrc_y * uvlinesize + uvsrc_x;

    if ((unsigned)src_x >= FFMAX(s->h_edge_pos - (motion_x & 1) - 15   , 0) ||
        (unsigned)src_y >= FFMAX(   v_edge_pos - (motion_y & 1) - h + 1, 0)) {
        if (is_mpeg12 ||
            s->codec_id == AV_CODEC_ID_MPEG2VIDEO ||
            s->codec_id == AV_CODEC_ID_MPEG1VIDEO) {
            av_log(s->avctx, AV_LOG_DEBUG,
                   "MPEG motion vector out of boundary (%d %d)\n", src_x,
                   src_y);
            return;
        }
        src_y = (unsigned)src_y << field_based;
        s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer, ptr_y,
                                 s->linesize, s->linesize,
                                 17, 17 + field_based,
                                 src_x, src_y,
                                 s->h_edge_pos, s->v_edge_pos);
        ptr_y = s->sc.edge_emu_buffer;
        if (!CONFIG_GRAY || !(s->avctx->flags & AV_CODEC_FLAG_GRAY)) {
            uint8_t *ubuf = s->sc.edge_emu_buffer + 18 * s->linesize;
            uint8_t *vbuf = ubuf + 9 * s->uvlinesize;
            uvsrc_y = (unsigned)uvsrc_y << field_based;
            s->vdsp.emulated_edge_mc(ubuf, ptr_cb,
                                     s->uvlinesize, s->uvlinesize,
                                     9, 9 + field_based,
                                     uvsrc_x, uvsrc_y,
                                     s->h_edge_pos >> 1, s->v_edge_pos >> 1);
            s->vdsp.emulated_edge_mc(vbuf, ptr_cr,
                                     s->uvlinesize, s->uvlinesize,
                                     9, 9 + field_based,
                                     uvsrc_x, uvsrc_y,
                                     s->h_edge_pos >> 1, s->v_edge_pos >> 1);
            ptr_cb = ubuf;
            ptr_cr = vbuf;
        }
    }

    /* FIXME use this for field pix too instead of the obnoxious hack which
     * changes picture.data */
    if (bottom_field) {
        dest_y  += s->linesize;
        dest_cb += s->uvlinesize;
        dest_cr += s->uvlinesize;
    }

    if (field_select) {
        ptr_y  += s->linesize;
        ptr_cb += s->uvlinesize;
        ptr_cr += s->uvlinesize;
    }

    pix_op[0][dxy](dest_y, ptr_y, linesize, h);

    if (!CONFIG_GRAY || !(s->avctx->flags & AV_CODEC_FLAG_GRAY)) {
        pix_op[s->chroma_x_shift][uvdxy]
            (dest_cb, ptr_cb, uvlinesize, h >> s->chroma_y_shift);
        pix_op[s->chroma_x_shift][uvdxy]
            (dest_cr, ptr_cr, uvlinesize, h >> s->chroma_y_shift);
    }
    if (!is_mpeg12 && (CONFIG_H261_ENCODER || CONFIG_H261_DECODER) &&
        s->out_format == FMT_H261) {
        ff_h261_loop_filter(s);
    }
}
/* apply one mpeg motion vector to the three components */
static void mpeg_motion(MpegEncContext *s,
                        uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
                        int field_select, uint8_t **ref_picture,
                        op_pixels_func (*pix_op)[4],
                        int motion_x, int motion_y, int h, int mb_y)
{
#if !CONFIG_SMALL
    if (s->out_format == FMT_MPEG1)
        mpeg_motion_internal(s, dest_y, dest_cb, dest_cr, 0, 0,
                             field_select, ref_picture, pix_op,
                             motion_x, motion_y, h, 1, mb_y);
    else
#endif
        mpeg_motion_internal(s, dest_y, dest_cb, dest_cr, 0, 0,
                             field_select, ref_picture, pix_op,
                             motion_x, motion_y, h, 0, mb_y);
}

static void mpeg_motion_field(MpegEncContext *s, uint8_t *dest_y,
                              uint8_t *dest_cb, uint8_t *dest_cr,
                              int bottom_field, int field_select,
                              uint8_t **ref_picture,
                              op_pixels_func (*pix_op)[4],
                              int motion_x, int motion_y, int h, int mb_y)
{
#if !CONFIG_SMALL
    if (s->out_format == FMT_MPEG1)
        mpeg_motion_internal(s, dest_y, dest_cb, dest_cr, 1,
                             bottom_field, field_select, ref_picture, pix_op,
                             motion_x, motion_y, h, 1, mb_y);
    else
#endif
        mpeg_motion_internal(s, dest_y, dest_cb, dest_cr, 1,
                             bottom_field, field_select, ref_picture, pix_op,
                             motion_x, motion_y, h, 0, mb_y);
}

// FIXME: SIMDify, avg variant, 16x16 version
static inline void put_obmc(uint8_t *dst, uint8_t *src[5], int stride)
{
    int x;
    uint8_t *const top    = src[1];
    uint8_t *const left   = src[2];
    uint8_t *const mid    = src[0];
    uint8_t *const right  = src[3];
    uint8_t *const bottom = src[4];
#define OBMC_FILTER(x, t, l, m, r, b)\
    dst[x]= (t*top[x] + l*left[x] + m*mid[x] + r*right[x] + b*bottom[x] + 4)>>3
#define OBMC_FILTER4(x, t, l, m, r, b)\
    OBMC_FILTER(x         , t, l, m, r, b);\
    OBMC_FILTER(x+1       , t, l, m, r, b);\
    OBMC_FILTER(x  +stride, t, l, m, r, b);\
    OBMC_FILTER(x+1+stride, t, l, m, r, b);

    x = 0;
    OBMC_FILTER (x    , 2, 2, 4, 0, 0);
    OBMC_FILTER (x + 1, 2, 1, 5, 0, 0);
    OBMC_FILTER4(x + 2, 2, 1, 5, 0, 0);
    OBMC_FILTER4(x + 4, 2, 0, 5, 1, 0);
    OBMC_FILTER (x + 6, 2, 0, 5, 1, 0);
    OBMC_FILTER (x + 7, 2, 0, 4, 2, 0);
    x += stride;
    OBMC_FILTER (x    , 1, 2, 5, 0, 0);
    OBMC_FILTER (x + 1, 1, 2, 5, 0, 0);
    OBMC_FILTER (x + 6, 1, 0, 5, 2, 0);
    OBMC_FILTER (x + 7, 1, 0, 5, 2, 0);
    x += stride;
    OBMC_FILTER4(x    , 1, 2, 5, 0, 0);
    OBMC_FILTER4(x + 2, 1, 1, 6, 0, 0);
    OBMC_FILTER4(x + 4, 1, 0, 6, 1, 0);
    OBMC_FILTER4(x + 6, 1, 0, 5, 2, 0);
    x += 2 * stride;
    OBMC_FILTER4(x    , 0, 2, 5, 0, 1);
    OBMC_FILTER4(x + 2, 0, 1, 6, 0, 1);
    OBMC_FILTER4(x + 4, 0, 0, 6, 1, 1);
    OBMC_FILTER4(x + 6, 0, 0, 5, 2, 1);
    x += 2*stride;
    OBMC_FILTER (x    , 0, 2, 5, 0, 1);
    OBMC_FILTER (x + 1, 0, 2, 5, 0, 1);
    OBMC_FILTER4(x + 2, 0, 1, 5, 0, 2);
    OBMC_FILTER4(x + 4, 0, 0, 5, 1, 2);
    OBMC_FILTER (x + 6, 0, 0, 5, 2, 1);
    OBMC_FILTER (x + 7, 0, 0, 5, 2, 1);
    x += stride;
    OBMC_FILTER (x    , 0, 2, 4, 0, 2);
    OBMC_FILTER (x + 1, 0, 1, 5, 0, 2);
    OBMC_FILTER (x + 6, 0, 0, 5, 1, 2);
    OBMC_FILTER (x + 7, 0, 0, 4, 2, 2);
}

/* obmc for 1 8x8 luma block */
static inline void obmc_motion(MpegEncContext *s,
                               uint8_t *dest, uint8_t *src,
                               int src_x, int src_y,
                               op_pixels_func *pix_op,
                               int16_t mv[5][2] /* mid top left right bottom */)
#define MID    0
{
    int i;
    uint8_t *ptr[5];

    av_assert2(s->quarter_sample == 0);

    for (i = 0; i < 5; i++) {
        if (i && mv[i][0] == mv[MID][0] && mv[i][1] == mv[MID][1]) {
            ptr[i] = ptr[MID];
        } else {
            ptr[i] = s->sc.obmc_scratchpad + 8 * (i & 1) +
                     s->linesize * 8 * (i >> 1);
            hpel_motion(s, ptr[i], src, src_x, src_y, pix_op,
                        mv[i][0], mv[i][1]);
        }
    }

    put_obmc(dest, ptr, s->linesize);
}

static inline void qpel_motion(MpegEncContext *s,
                               uint8_t *dest_y,
                               uint8_t *dest_cb,
                               uint8_t *dest_cr,
                               int field_based, int bottom_field,
                               int field_select, uint8_t **ref_picture,
                               op_pixels_func (*pix_op)[4],
                               qpel_mc_func (*qpix_op)[16],
                               int motion_x, int motion_y, int h)
{
    uint8_t *ptr_y, *ptr_cb, *ptr_cr;
    int dxy, uvdxy, mx, my, src_x, src_y, uvsrc_x, uvsrc_y, v_edge_pos;
    ptrdiff_t linesize, uvlinesize;

    dxy   = ((motion_y & 3) << 2) | (motion_x & 3);

    src_x = s->mb_x *  16                 + (motion_x >> 2);
    src_y = s->mb_y * (16 >> field_based) + (motion_y >> 2);

    v_edge_pos = s->v_edge_pos >> field_based;
    linesize   = s->linesize   << field_based;
    uvlinesize = s->uvlinesize << field_based;

    if (field_based) {
        mx = motion_x / 2;
        my = motion_y >> 1;
    } else if (s->workaround_bugs & FF_BUG_QPEL_CHROMA2) {
        static const int rtab[8] = { 0, 0, 1, 1, 0, 0, 0, 1 };
        mx = (motion_x >> 1) + rtab[motion_x & 7];
        my = (motion_y >> 1) + rtab[motion_y & 7];
    } else if (s->workaround_bugs & FF_BUG_QPEL_CHROMA) {
        mx = (motion_x >> 1) | (motion_x & 1);
        my = (motion_y >> 1) | (motion_y & 1);
    } else {
        mx = motion_x / 2;
        my = motion_y / 2;
    }
    mx = (mx >> 1) | (mx & 1);
    my = (my >> 1) | (my & 1);

    uvdxy = (mx & 1) | ((my & 1) << 1);
    mx  >>= 1;
    my  >>= 1;

    uvsrc_x = s->mb_x *  8                 + mx;
    uvsrc_y = s->mb_y * (8 >> field_based) + my;

    ptr_y  = ref_picture[0] + src_y   * linesize   + src_x;
    ptr_cb = ref_picture[1] + uvsrc_y * uvlinesize + uvsrc_x;
    ptr_cr = ref_picture[2] + uvsrc_y * uvlinesize + uvsrc_x;

    if ((unsigned)src_x >= FFMAX(s->h_edge_pos - (motion_x & 3) - 15   , 0) ||
        (unsigned)src_y >= FFMAX(   v_edge_pos - (motion_y & 3) - h + 1, 0)) {
        s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer, ptr_y,
                                 s->linesize, s->linesize,
                                 17, 17 + field_based,
                                 src_x, src_y * (1 << field_based),
                                 s->h_edge_pos, s->v_edge_pos);
        ptr_y = s->sc.edge_emu_buffer;
        if (!CONFIG_GRAY || !(s->avctx->flags & AV_CODEC_FLAG_GRAY)) {
            uint8_t *ubuf = s->sc.edge_emu_buffer + 18 * s->linesize;
            uint8_t *vbuf = ubuf + 9 * s->uvlinesize;
            s->vdsp.emulated_edge_mc(ubuf, ptr_cb,
                                     s->uvlinesize, s->uvlinesize,
                                     9, 9 + field_based,
                                     uvsrc_x, uvsrc_y * (1 << field_based),
                                     s->h_edge_pos >> 1, s->v_edge_pos >> 1);
            s->vdsp.emulated_edge_mc(vbuf, ptr_cr,
                                     s->uvlinesize, s->uvlinesize,
                                     9, 9 + field_based,
                                     uvsrc_x, uvsrc_y * (1 << field_based),
                                     s->h_edge_pos >> 1, s->v_edge_pos >> 1);
            ptr_cb = ubuf;
            ptr_cr = vbuf;
        }
    }

    if (!field_based)
        qpix_op[0][dxy](dest_y, ptr_y, linesize);
    else {
        if (bottom_field) {
            dest_y  += s->linesize;
            dest_cb += s->uvlinesize;
            dest_cr += s->uvlinesize;
        }

        if (field_select) {
            ptr_y  += s->linesize;
            ptr_cb += s->uvlinesize;
            ptr_cr += s->uvlinesize;
        }
        // damn interlaced mode
        // FIXME boundary mirroring is not exactly correct here
        qpix_op[1][dxy](dest_y, ptr_y, linesize);
        qpix_op[1][dxy](dest_y + 8, ptr_y + 8, linesize);
    }
    if (!CONFIG_GRAY || !(s->avctx->flags & AV_CODEC_FLAG_GRAY)) {
        pix_op[1][uvdxy](dest_cr, ptr_cr, uvlinesize, h >> 1);
        pix_op[1][uvdxy](dest_cb, ptr_cb, uvlinesize, h >> 1);
    }
}

/**
 * H.263 chroma 4mv motion compensation.
 */
static void chroma_4mv_motion(MpegEncContext *s,
                              uint8_t *dest_cb, uint8_t *dest_cr,
                              uint8_t **ref_picture,
                              op_pixels_func *pix_op,
                              int mx, int my)
{
    uint8_t *ptr;
    int src_x, src_y, dxy, emu = 0;
    ptrdiff_t offset;

    /* In case of 8X8, we construct a single chroma motion vector
     * with a special rounding */
    mx = ff_h263_round_chroma(mx);
    my = ff_h263_round_chroma(my);

    dxy  = ((my & 1) << 1) | (mx & 1);
    mx >>= 1;
    my >>= 1;

    src_x = s->mb_x * 8 + mx;
    src_y = s->mb_y * 8 + my;
    src_x = av_clip(src_x, -8, (s->width >> 1));
    if (src_x == (s->width >> 1))
        dxy &= ~1;
    src_y = av_clip(src_y, -8, (s->height >> 1));
    if (src_y == (s->height >> 1))
        dxy &= ~2;

    offset = src_y * s->uvlinesize + src_x;
    ptr    = ref_picture[1] + offset;
    if ((unsigned)src_x >= FFMAX((s->h_edge_pos >> 1) - (dxy  & 1) - 7, 0) ||
        (unsigned)src_y >= FFMAX((s->v_edge_pos >> 1) - (dxy >> 1) - 7, 0)) {
        s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer, ptr,
                                 s->uvlinesize, s->uvlinesize,
                                 9, 9, src_x, src_y,
                                 s->h_edge_pos >> 1, s->v_edge_pos >> 1);
        ptr = s->sc.edge_emu_buffer;
        emu = 1;
    }
    pix_op[dxy](dest_cb, ptr, s->uvlinesize, 8);

    ptr = ref_picture[2] + offset;
    if (emu) {
        s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer, ptr,
                                 s->uvlinesize, s->uvlinesize,
                                 9, 9, src_x, src_y,
                                 s->h_edge_pos >> 1, s->v_edge_pos >> 1);
        ptr = s->sc.edge_emu_buffer;
    }
    pix_op[dxy](dest_cr, ptr, s->uvlinesize, 8);
}

static inline void prefetch_motion(MpegEncContext *s, uint8_t **pix, int dir)
{
    /* fetch pixels for estimated mv 4 macroblocks ahead
     * optimized for 64byte cache lines */
    const int shift = s->quarter_sample ? 2 : 1;
    const int mx    = (s->mv[dir][0][0] >> shift) + 16 * s->mb_x + 8;
    const int my    = (s->mv[dir][0][1] >> shift) + 16 * s->mb_y;
    int off         = mx + (my + (s->mb_x & 3) * 4) * s->linesize + 64;

    s->vdsp.prefetch(pix[0] + off, s->linesize, 4);
    off = (mx >> 1) + ((my >> 1) + (s->mb_x & 7)) * s->uvlinesize + 64;
    s->vdsp.prefetch(pix[1] + off, pix[2] - pix[1], 2);
}

static inline void apply_obmc(MpegEncContext *s,
                              uint8_t *dest_y,
                              uint8_t *dest_cb,
                              uint8_t *dest_cr,
                              uint8_t **ref_picture,
                              op_pixels_func (*pix_op)[4])
{
    LOCAL_ALIGNED_8(int16_t, mv_cache, [4], [4][2]);
    Picture *cur_frame   = &s->current_picture;
    int mb_x = s->mb_x;
    int mb_y = s->mb_y;
    const int xy         = mb_x + mb_y * s->mb_stride;
    const int mot_stride = s->b8_stride;
    const int mot_xy     = mb_x * 2 + mb_y * 2 * mot_stride;
    int mx, my, i;

    av_assert2(!s->mb_skipped);

    AV_COPY32(mv_cache[1][1], cur_frame->motion_val[0][mot_xy]);
    AV_COPY32(mv_cache[1][2], cur_frame->motion_val[0][mot_xy + 1]);

    AV_COPY32(mv_cache[2][1],
              cur_frame->motion_val[0][mot_xy + mot_stride]);
    AV_COPY32(mv_cache[2][2],
              cur_frame->motion_val[0][mot_xy + mot_stride + 1]);

    AV_COPY32(mv_cache[3][1],
              cur_frame->motion_val[0][mot_xy + mot_stride]);
    AV_COPY32(mv_cache[3][2],
              cur_frame->motion_val[0][mot_xy + mot_stride + 1]);

    if (mb_y == 0 || IS_INTRA(cur_frame->mb_type[xy - s->mb_stride])) {
        AV_COPY32(mv_cache[0][1], mv_cache[1][1]);
        AV_COPY32(mv_cache[0][2], mv_cache[1][2]);
    } else {
        AV_COPY32(mv_cache[0][1],
                  cur_frame->motion_val[0][mot_xy - mot_stride]);
        AV_COPY32(mv_cache[0][2],
                  cur_frame->motion_val[0][mot_xy - mot_stride + 1]);
    }

    if (mb_x == 0 || IS_INTRA(cur_frame->mb_type[xy - 1])) {
        AV_COPY32(mv_cache[1][0], mv_cache[1][1]);
        AV_COPY32(mv_cache[2][0], mv_cache[2][1]);
    } else {
        AV_COPY32(mv_cache[1][0], cur_frame->motion_val[0][mot_xy - 1]);
        AV_COPY32(mv_cache[2][0],
                  cur_frame->motion_val[0][mot_xy - 1 + mot_stride]);
    }

    if (mb_x + 1 >= s->mb_width || IS_INTRA(cur_frame->mb_type[xy + 1])) {
        AV_COPY32(mv_cache[1][3], mv_cache[1][2]);
        AV_COPY32(mv_cache[2][3], mv_cache[2][2]);
    } else {
        AV_COPY32(mv_cache[1][3], cur_frame->motion_val[0][mot_xy + 2]);
        AV_COPY32(mv_cache[2][3],
                  cur_frame->motion_val[0][mot_xy + 2 + mot_stride]);
    }

    mx = 0;
    my = 0;
    for (i = 0; i < 4; i++) {
        const int x      = (i & 1) + 1;
        const int y      = (i >> 1) + 1;
        int16_t mv[5][2] = {
            { mv_cache[y][x][0],     mv_cache[y][x][1]         },
            { mv_cache[y - 1][x][0], mv_cache[y - 1][x][1]     },
            { mv_cache[y][x - 1][0], mv_cache[y][x - 1][1]     },
            { mv_cache[y][x + 1][0], mv_cache[y][x + 1][1]     },
            { mv_cache[y + 1][x][0], mv_cache[y + 1][x][1]     }
        };
        // FIXME cleanup
        obmc_motion(s, dest_y + ((i & 1) * 8) + (i >> 1) * 8 * s->linesize,
                    ref_picture[0],
                    mb_x * 16 + (i & 1) * 8, mb_y * 16 + (i >> 1) * 8,
                    pix_op[1],
                    mv);

        mx += mv[0][0];
        my += mv[0][1];
    }
    if (!CONFIG_GRAY || !(s->avctx->flags & AV_CODEC_FLAG_GRAY))
        chroma_4mv_motion(s, dest_cb, dest_cr,
                          ref_picture, pix_op[1],
                          mx, my);
}

static inline void apply_8x8(MpegEncContext *s,
                             uint8_t *dest_y,
                             uint8_t *dest_cb,
                             uint8_t *dest_cr,
                             int dir,
                             uint8_t **ref_picture,
                             qpel_mc_func (*qpix_op)[16],
                             op_pixels_func (*pix_op)[4])
{
    int dxy, mx, my, src_x, src_y;
    int i;
    int mb_x = s->mb_x;
    int mb_y = s->mb_y;
    uint8_t *ptr, *dest;

    mx = 0;
    my = 0;
    if (s->quarter_sample) {
        for (i = 0; i < 4; i++) {
            int motion_x = s->mv[dir][i][0];
            int motion_y = s->mv[dir][i][1];

            dxy   = ((motion_y & 3) << 2) | (motion_x & 3);
            src_x = mb_x * 16 + (motion_x >> 2) + (i & 1) * 8;
            src_y = mb_y * 16 + (motion_y >> 2) + (i >> 1) * 8;

            /* WARNING: do no forget half pels */
            src_x = av_clip(src_x, -16, s->width);
            if (src_x == s->width)
                dxy &= ~3;
            src_y = av_clip(src_y, -16, s->height);
            if (src_y == s->height)
                dxy &= ~12;

            ptr = ref_picture[0] + (src_y * s->linesize) + (src_x);
            if ((unsigned)src_x >= FFMAX(s->h_edge_pos - (motion_x & 3) - 7, 0) ||
                (unsigned)src_y >= FFMAX(s->v_edge_pos - (motion_y & 3) - 7, 0)) {
                s->vdsp.emulated_edge_mc(s->sc.edge_emu_buffer, ptr,
                                         s->linesize, s->linesize,
                                         9, 9,
                                         src_x, src_y,
                                         s->h_edge_pos,
                                         s->v_edge_pos);
                ptr = s->sc.edge_emu_buffer;
            }
            dest = dest_y + ((i & 1) * 8) + (i >> 1) * 8 * s->linesize;
            qpix_op[1][dxy](dest, ptr, s->linesize);

            mx += s->mv[dir][i][0] / 2;
            my += s->mv[dir][i][1] / 2;
        }
    } else {
        for (i = 0; i < 4; i++) {
            hpel_motion(s,
                        dest_y + ((i & 1) * 8) + (i >> 1) * 8 * s->linesize,
                        ref_picture[0],
                        mb_x * 16 + (i & 1) * 8,
                        mb_y * 16 + (i >> 1) * 8,
                        pix_op[1],
                        s->mv[dir][i][0],
                        s->mv[dir][i][1]);

            mx += s->mv[dir][i][0];
            my += s->mv[dir][i][1];
        }
    }

    if (!CONFIG_GRAY || !(s->avctx->flags & AV_CODEC_FLAG_GRAY))
        chroma_4mv_motion(s, dest_cb, dest_cr,
                          ref_picture, pix_op[1], mx, my);
}

/**
 * motion compensation of a single macroblock
 * @param s context
 * @param dest_y luma destination pointer
 * @param dest_cb chroma cb/u destination pointer
 * @param dest_cr chroma cr/v destination pointer
 * @param dir direction (0->forward, 1->backward)
 * @param ref_picture array[3] of pointers to the 3 planes of the reference picture
 * @param pix_op halfpel motion compensation function (average or put normally)
 * @param qpix_op qpel motion compensation function (average or put normally)
 * the motion vectors are taken from s->mv and the MV type from s->mv_type
 */
static av_always_inline void mpv_motion_internal(MpegEncContext *s,
                                                 uint8_t *dest_y,
                                                 uint8_t *dest_cb,
                                                 uint8_t *dest_cr,
                                                 int dir,
                                                 uint8_t **ref_picture,
                                                 op_pixels_func (*pix_op)[4],
                                                 qpel_mc_func (*qpix_op)[16],
                                                 int is_mpeg12)
{
    int i;
    int mb_y = s->mb_y;

    prefetch_motion(s, ref_picture, dir);

    if (!is_mpeg12 && s->obmc && s->pict_type != AV_PICTURE_TYPE_B) {
        apply_obmc(s, dest_y, dest_cb, dest_cr, ref_picture, pix_op);
        return;
    }

    switch (s->mv_type) {
    case MV_TYPE_16X16:
        if (s->mcsel) {
            if (s->real_sprite_warping_points == 1) {
                gmc1_motion(s, dest_y, dest_cb, dest_cr,
                            ref_picture);
            } else {
                gmc_motion(s, dest_y, dest_cb, dest_cr,
                           ref_picture);
            }
        } else if (!is_mpeg12 && s->quarter_sample) {
            qpel_motion(s, dest_y, dest_cb, dest_cr,
                        0, 0, 0,
                        ref_picture, pix_op, qpix_op,
                        s->mv[dir][0][0], s->mv[dir][0][1], 16);
        } else if (!is_mpeg12 && (CONFIG_WMV2_DECODER || CONFIG_WMV2_ENCODER) &&
                   s->mspel && s->codec_id == AV_CODEC_ID_WMV2) {
            ff_mspel_motion(s, dest_y, dest_cb, dest_cr,
                            ref_picture, pix_op,
                            s->mv[dir][0][0], s->mv[dir][0][1], 16);
        } else {
            mpeg_motion(s, dest_y, dest_cb, dest_cr, 0,
                        ref_picture, pix_op,
                        s->mv[dir][0][0], s->mv[dir][0][1], 16, mb_y);
        }
        break;
    case MV_TYPE_8X8:
        if (!is_mpeg12)
            apply_8x8(s, dest_y, dest_cb, dest_cr,
                      dir, ref_picture, qpix_op, pix_op);
        break;
    case MV_TYPE_FIELD:
        if (s->picture_structure == PICT_FRAME) {
            if (!is_mpeg12 && s->quarter_sample) {
                for (i = 0; i < 2; i++)
                    qpel_motion(s, dest_y, dest_cb, dest_cr,
                                1, i, s->field_select[dir][i],
                                ref_picture, pix_op, qpix_op,
                                s->mv[dir][i][0], s->mv[dir][i][1], 8);
            } else {
                /* top field */
                mpeg_motion_field(s, dest_y, dest_cb, dest_cr,
                                  0, s->field_select[dir][0],
                                  ref_picture, pix_op,
                                  s->mv[dir][0][0], s->mv[dir][0][1], 8, mb_y);
                /* bottom field */
                mpeg_motion_field(s, dest_y, dest_cb, dest_cr,
                                  1, s->field_select[dir][1],
                                  ref_picture, pix_op,
                                  s->mv[dir][1][0], s->mv[dir][1][1], 8, mb_y);
            }
        } else {
            if (   s->picture_structure != s->field_select[dir][0] + 1 && s->pict_type != AV_PICTURE_TYPE_B && !s->first_field
                || !ref_picture[0]) {
                ref_picture = s->current_picture_ptr->f->data;
            }

            mpeg_motion(s, dest_y, dest_cb, dest_cr,
                        s->field_select[dir][0],
                        ref_picture, pix_op,
                        s->mv[dir][0][0], s->mv[dir][0][1], 16, mb_y >> 1);
        }
        break;
    case MV_TYPE_16X8:
        for (i = 0; i < 2; i++) {
            uint8_t **ref2picture;

            if ((s->picture_structure == s->field_select[dir][i] + 1
                || s->pict_type == AV_PICTURE_TYPE_B || s->first_field) && ref_picture[0]) {
                ref2picture = ref_picture;
            } else {
                ref2picture = s->current_picture_ptr->f->data;
            }

            mpeg_motion(s, dest_y, dest_cb, dest_cr,
                        s->field_select[dir][i],
                        ref2picture, pix_op,
                        s->mv[dir][i][0], s->mv[dir][i][1] + 16 * i,
                        8, mb_y >> 1);

            dest_y  += 16 * s->linesize;
            dest_cb += (16 >> s->chroma_y_shift) * s->uvlinesize;
            dest_cr += (16 >> s->chroma_y_shift) * s->uvlinesize;
        }
        break;
    case MV_TYPE_DMV:
        if (s->picture_structure == PICT_FRAME) {
            for (i = 0; i < 2; i++) {
                int j;
                for (j = 0; j < 2; j++)
                    mpeg_motion_field(s, dest_y, dest_cb, dest_cr,
                                      j, j ^ i, ref_picture, pix_op,
                                      s->mv[dir][2 * i + j][0],
                                      s->mv[dir][2 * i + j][1], 8, mb_y);
                pix_op = s->hdsp.avg_pixels_tab;
            }
        } else {
            if (!ref_picture[0]) {
                ref_picture = s->current_picture_ptr->f->data;
            }
            for (i = 0; i < 2; i++) {
                mpeg_motion(s, dest_y, dest_cb, dest_cr,
                            s->picture_structure != i + 1,
                            ref_picture, pix_op,
                            s->mv[dir][2 * i][0], s->mv[dir][2 * i][1],
                            16, mb_y >> 1);

                // after put we make avg of the same block
                pix_op = s->hdsp.avg_pixels_tab;

                /* opposite parity is always in the same frame if this is
                 * second field */
                if (!s->first_field) {
                    ref_picture = s->current_picture_ptr->f->data;
                }
            }
        }
        break;
    default: av_assert2(0);
    }
}

void ff_mpv_motion(MpegEncContext *s,
                   uint8_t *dest_y, uint8_t *dest_cb,
                   uint8_t *dest_cr, int dir,
                   uint8_t **ref_picture,
                   op_pixels_func (*pix_op)[4],
                   qpel_mc_func (*qpix_op)[16])
{
#if !CONFIG_SMALL
    if (s->out_format == FMT_MPEG1)
        mpv_motion_internal(s, dest_y, dest_cb, dest_cr, dir,
                            ref_picture, pix_op, qpix_op, 1);
    else
#endif
        mpv_motion_internal(s, dest_y, dest_cb, dest_cr, dir,
                            ref_picture, pix_op, qpix_op, 0);
}