[cvs] / xvidcore / src / decoder.c Repository:
ViewVC logotype

View of /xvidcore/src/decoder.c

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1.87 - (download) (annotate)
Tue Dec 28 19:19:43 2010 UTC (13 years, 2 months ago) by Isibaar
Branch: MAIN
CVS Tags: HEAD
Changes since 1.86: +2 -2 lines
get rid off compiler warnings
/*****************************************************************************
 *
 *  XVID MPEG-4 VIDEO CODEC
 *  - Decoder Module -
 *
 *  Copyright(C) 2002      MinChen <chenm001@163.com>
 *               2002-2010 Peter Ross <pross@xvid.org>
 *
 *  This program is free software ; you can redistribute it and/or modify
 *  it under the terms of the GNU General Public License as published by
 *  the Free Software Foundation ; either version 2 of the License, or
 *  (at your option) any later version.
 *
 *  This program 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 General Public License for more details.
 *
 *  You should have received a copy of the GNU General Public License
 *  along with this program ; if not, write to the Free Software
 *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
 *
 * $Id: decoder.c,v 1.87 2010/12/28 19:19:43 Isibaar Exp $
 *
 ****************************************************************************/

#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#ifdef BFRAMES_DEC_DEBUG
#define BFRAMES_DEC
#endif

#include "xvid.h"
#include "portab.h"
#include "global.h"

#include "decoder.h"
#include "bitstream/bitstream.h"
#include "bitstream/mbcoding.h"

#include "quant/quant.h"
#include "quant/quant_matrix.h"
#include "dct/idct.h"
#include "dct/fdct.h"
#include "utils/mem_transfer.h"
#include "image/interpolate8x8.h"
#include "image/font.h"
#include "image/qpel.h"

#include "bitstream/mbcoding.h"
#include "prediction/mbprediction.h"
#include "utils/timer.h"
#include "utils/emms.h"
#include "motion/motion.h"
#include "motion/gmc.h"

#include "image/image.h"
#include "image/colorspace.h"
#include "image/postprocessing.h"
#include "utils/mem_align.h"

#define DIV2ROUND(n)  (((n)>>1)|((n)&1))
#define DIV2(n)       ((n)>>1)
#define DIVUVMOV(n) (((n) >> 1) + roundtab_79[(n) & 0x3]) //

static int
decoder_resize(DECODER * dec)
{
	/* free existing */
	image_destroy(&dec->cur, dec->edged_width, dec->edged_height);
	image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height);
	image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height);
	image_destroy(&dec->tmp, dec->edged_width, dec->edged_height);
	image_destroy(&dec->qtmp, dec->edged_width, dec->edged_height);

	image_destroy(&dec->gmc, dec->edged_width, dec->edged_height);

  image_null(&dec->cur);
  image_null(&dec->refn[0]);
  image_null(&dec->refn[1]);
  image_null(&dec->tmp);
  image_null(&dec->qtmp);
  image_null(&dec->gmc);


  xvid_free(dec->last_mbs);
  xvid_free(dec->mbs);
  xvid_free(dec->qscale);
  dec->last_mbs = NULL;
  dec->mbs = NULL;
  dec->qscale = NULL;

	/* realloc */
	dec->mb_width = (dec->width + 15) / 16;
	dec->mb_height = (dec->height + 15) / 16;

	dec->edged_width = 16 * dec->mb_width + 2 * EDGE_SIZE;
	dec->edged_height = 16 * dec->mb_height + 2 * EDGE_SIZE;

	if (   image_create(&dec->cur, dec->edged_width, dec->edged_height) 
	    || image_create(&dec->refn[0], dec->edged_width, dec->edged_height)
	    || image_create(&dec->refn[1], dec->edged_width, dec->edged_height) 	/* Support B-frame to reference last 2 frame */
	    || image_create(&dec->tmp, dec->edged_width, dec->edged_height)
	    || image_create(&dec->qtmp, dec->edged_width, dec->edged_height)
      || image_create(&dec->gmc, dec->edged_width, dec->edged_height) )
    goto memory_error;

	dec->mbs =
		xvid_malloc(sizeof(MACROBLOCK) * dec->mb_width * dec->mb_height,
					CACHE_LINE);
	if (dec->mbs == NULL)
	  goto memory_error;
	memset(dec->mbs, 0, sizeof(MACROBLOCK) * dec->mb_width * dec->mb_height);

	/* For skip MB flag */
	dec->last_mbs =
		xvid_malloc(sizeof(MACROBLOCK) * dec->mb_width * dec->mb_height,
					CACHE_LINE);
	if (dec->last_mbs == NULL)
	  goto memory_error;
	memset(dec->last_mbs, 0, sizeof(MACROBLOCK) * dec->mb_width * dec->mb_height);

	/* nothing happens if that fails */
	dec->qscale =
		xvid_malloc(sizeof(int) * dec->mb_width * dec->mb_height, CACHE_LINE);
	
	if (dec->qscale)
		memset(dec->qscale, 0, sizeof(int) * dec->mb_width * dec->mb_height);

	return 0;

memory_error:
        /* Most structures were deallocated / nullifieded, so it should be safe */
        /* decoder_destroy(dec) minus the write_timer */
  xvid_free(dec->mbs);
  image_destroy(&dec->cur, dec->edged_width, dec->edged_height);
  image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height);
  image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height);
  image_destroy(&dec->tmp, dec->edged_width, dec->edged_height);
  image_destroy(&dec->qtmp, dec->edged_width, dec->edged_height);

  xvid_free(dec);
  return XVID_ERR_MEMORY;
}


int
decoder_create(xvid_dec_create_t * create)
{
  DECODER *dec;

  if (XVID_VERSION_MAJOR(create->version) != 1) /* v1.x.x */
    return XVID_ERR_VERSION;

  dec = xvid_malloc(sizeof(DECODER), CACHE_LINE);
  if (dec == NULL) {
    return XVID_ERR_MEMORY;
  }

  memset(dec, 0, sizeof(DECODER));

  dec->mpeg_quant_matrices = xvid_malloc(sizeof(uint16_t) * 64 * 8, CACHE_LINE);
  if (dec->mpeg_quant_matrices == NULL) {
    xvid_free(dec);
    return XVID_ERR_MEMORY;
  }

  create->handle = dec;

  dec->width = create->width;
  dec->height = create->height;

  dec->num_threads = MAX(0, create->num_threads);

  image_null(&dec->cur);
  image_null(&dec->refn[0]);
  image_null(&dec->refn[1]);
  image_null(&dec->tmp);
  image_null(&dec->qtmp);

  /* image based GMC */
  image_null(&dec->gmc);

  dec->mbs = NULL;
  dec->last_mbs = NULL;
  dec->qscale = NULL;

  init_timer();
  init_postproc(&dec->postproc);
  init_mpeg_matrix(dec->mpeg_quant_matrices);

  /* For B-frame support (used to save reference frame's time */
  dec->frames = 0;
  dec->time = dec->time_base = dec->last_time_base = 0;
  dec->low_delay = 0;
  dec->packed_mode = 0;
  dec->time_inc_resolution = 1; /* until VOL header says otherwise */
  dec->ver_id = 1;

  if (create->fourcc == ((int)('X')|((int)('V')<<8)| 
                         ((int)('I')<<16)|((int)('D')<<24))) { /* XVID */
    dec->bs_version = 0; /* Initially assume oldest xvid version */ 
  }
  else {
	dec->bs_version = 0xffff; /* Initialize to very high value -> assume bugfree stream */
  }

  dec->fixed_dimensions = (dec->width > 0 && dec->height > 0);

  if (dec->fixed_dimensions) {
    int ret = decoder_resize(dec);
    if (ret == XVID_ERR_MEMORY) create->handle = NULL;
    return ret;
  }
  else
    return 0;
}


int
decoder_destroy(DECODER * dec)
{
  xvid_free(dec->last_mbs);
  xvid_free(dec->mbs);
  xvid_free(dec->qscale);

  /* image based GMC */
  image_destroy(&dec->gmc, dec->edged_width, dec->edged_height);

  image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height);
  image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height);
  image_destroy(&dec->tmp, dec->edged_width, dec->edged_height);
  image_destroy(&dec->qtmp, dec->edged_width, dec->edged_height);
  image_destroy(&dec->cur, dec->edged_width, dec->edged_height);
  xvid_free(dec->mpeg_quant_matrices);
  xvid_free(dec);

  write_timer();
  return 0;
}

static const int32_t dquant_table[4] = {
  -1, -2, 1, 2
};

/* decode an intra macroblock */
static void
decoder_mbintra(DECODER * dec,
        MACROBLOCK * pMB,
        const uint32_t x_pos,
        const uint32_t y_pos,
        const uint32_t acpred_flag,
        const uint32_t cbp,
        Bitstream * bs,
        const uint32_t quant,
        const uint32_t intra_dc_threshold,
        const unsigned int bound)
{

  DECLARE_ALIGNED_MATRIX(block, 6, 64, int16_t, CACHE_LINE);
  DECLARE_ALIGNED_MATRIX(data, 6, 64, int16_t, CACHE_LINE);

  uint32_t stride = dec->edged_width;
  uint32_t stride2 = stride / 2;
  uint32_t next_block = stride * 8;
  uint32_t i;
  uint32_t iQuant = pMB->quant;
  uint8_t *pY_Cur, *pU_Cur, *pV_Cur;

  pY_Cur = dec->cur.y + (y_pos << 4) * stride + (x_pos << 4);
  pU_Cur = dec->cur.u + (y_pos << 3) * stride2 + (x_pos << 3);
  pV_Cur = dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3);

  memset(block, 0, 6 * 64 * sizeof(int16_t)); /* clear */

  for (i = 0; i < 6; i++) {
    uint32_t iDcScaler = get_dc_scaler(iQuant, i < 4);
    int16_t predictors[8];
    int start_coeff;

    start_timer();
    predict_acdc(dec->mbs, x_pos, y_pos, dec->mb_width, i, &block[i * 64],
           iQuant, iDcScaler, predictors, bound);
    if (!acpred_flag) {
      pMB->acpred_directions[i] = 0;
    }
    stop_prediction_timer();

    if (quant < intra_dc_threshold) {
      int dc_size;
      int dc_dif;

      dc_size = i < 4 ? get_dc_size_lum(bs) : get_dc_size_chrom(bs);
      dc_dif = dc_size ? get_dc_dif(bs, dc_size) : 0;

      if (dc_size > 8) {
        BitstreamSkip(bs, 1); /* marker */
      }

      block[i * 64 + 0] = dc_dif;
      start_coeff = 1;

      DPRINTF(XVID_DEBUG_COEFF,"block[0] %i\n", dc_dif);
    } else {
      start_coeff = 0;
    }

    start_timer();
    if (cbp & (1 << (5 - i))) /* coded */
    {
      int direction = dec->alternate_vertical_scan ?
        2 : pMB->acpred_directions[i];

      get_intra_block(bs, &block[i * 64], direction, start_coeff);
    }
    stop_coding_timer();

    start_timer();
    add_acdc(pMB, i, &block[i * 64], iDcScaler, predictors, dec->bs_version);
    stop_prediction_timer();

    start_timer();
    if (dec->quant_type == 0) {
      dequant_h263_intra(&data[i * 64], &block[i * 64], iQuant, iDcScaler, dec->mpeg_quant_matrices);
    } else {
      dequant_mpeg_intra(&data[i * 64], &block[i * 64], iQuant, iDcScaler, dec->mpeg_quant_matrices);
    }
    stop_iquant_timer();

    start_timer();
    idct((short * const)&data[i * 64]);
    stop_idct_timer();

  }

  if (dec->interlacing && pMB->field_dct) {
    next_block = stride;
    stride *= 2;
  }

  start_timer();
  transfer_16to8copy(pY_Cur, &data[0 * 64], stride);
  transfer_16to8copy(pY_Cur + 8, &data[1 * 64], stride);
  transfer_16to8copy(pY_Cur + next_block, &data[2 * 64], stride);
  transfer_16to8copy(pY_Cur + 8 + next_block, &data[3 * 64], stride);
  transfer_16to8copy(pU_Cur, &data[4 * 64], stride2);
  transfer_16to8copy(pV_Cur, &data[5 * 64], stride2);
  stop_transfer_timer();
}

static void
decoder_mb_decode(DECODER * dec,
        const uint32_t cbp,
        Bitstream * bs,
        uint8_t * pY_Cur,
        uint8_t * pU_Cur,
        uint8_t * pV_Cur,
        const MACROBLOCK * pMB)
{
  DECLARE_ALIGNED_MATRIX(data, 1, 64, int16_t, CACHE_LINE);

  int stride = dec->edged_width;
  int i;
  const uint32_t iQuant = pMB->quant;
  const int direction = dec->alternate_vertical_scan ? 2 : 0;
  typedef void (*get_inter_block_function_t)(
      Bitstream * bs,
      int16_t * block,
      int direction,
      const int quant,
      const uint16_t *matrix);
  typedef void (*add_residual_function_t)(
      uint8_t *predicted_block,
      const int16_t *residual,
      int stride);

  const get_inter_block_function_t get_inter_block = (dec->quant_type == 0)
    ? (get_inter_block_function_t)get_inter_block_h263
    : (get_inter_block_function_t)get_inter_block_mpeg;

  uint8_t *dst[6];
  int strides[6];


  if (dec->interlacing && pMB->field_dct) {
    dst[0] = pY_Cur;
    dst[1] = pY_Cur + 8;
    dst[2] = pY_Cur + stride;
    dst[3] = dst[2] + 8;
    dst[4] = pU_Cur;
    dst[5] = pV_Cur;
    strides[0] = strides[1] = strides[2] = strides[3] = stride*2;
    strides[4] = stride/2;
    strides[5] = stride/2;
  } else {
    dst[0] = pY_Cur;
    dst[1] = pY_Cur + 8;
    dst[2] = pY_Cur + 8*stride;
    dst[3] = dst[2] + 8;
    dst[4] = pU_Cur;
    dst[5] = pV_Cur;
    strides[0] = strides[1] = strides[2] = strides[3] = stride;
    strides[4] = stride/2;
    strides[5] = stride/2;
  }

  for (i = 0; i < 6; i++) {
    /* Process only coded blocks */
    if (cbp & (1 << (5 - i))) {

      /* Clear the block */
      memset(&data[0], 0, 64*sizeof(int16_t));

      /* Decode coeffs and dequantize on the fly */
      start_timer();
      get_inter_block(bs, &data[0], direction, iQuant, get_inter_matrix(dec->mpeg_quant_matrices));
      stop_coding_timer();

      /* iDCT */
      start_timer();
      idct((short * const)&data[0]);
      stop_idct_timer();

      /* Add this residual to the predicted block */
      start_timer();
      transfer_16to8add(dst[i], &data[0], strides[i]);
      stop_transfer_timer();
    }
  }
}

static void __inline
validate_vector(VECTOR * mv, unsigned int x_pos, unsigned int y_pos, const DECODER * dec)
{
  /* clip a vector to valid range
     prevents crashes if bitstream is broken
  */
  int shift = 5 + dec->quarterpel;
  int xborder_high = (int)(dec->mb_width - x_pos) << shift;
  int xborder_low = (-(int)x_pos-1) << shift;
  int yborder_high = (int)(dec->mb_height - y_pos) << shift;
  int yborder_low = (-(int)y_pos-1) << shift;

#define CHECK_MV(mv) \
  do { \
  if ((mv).x > xborder_high) { \
    DPRINTF(XVID_DEBUG_MV, "mv.x > max -- %d > %d, MB %d, %d", (mv).x, xborder_high, x_pos, y_pos); \
    (mv).x = xborder_high; \
  } else if ((mv).x < xborder_low) { \
    DPRINTF(XVID_DEBUG_MV, "mv.x < min -- %d < %d, MB %d, %d", (mv).x, xborder_low, x_pos, y_pos); \
    (mv).x = xborder_low; \
  } \
  if ((mv).y > yborder_high) { \
    DPRINTF(XVID_DEBUG_MV, "mv.y > max -- %d > %d, MB %d, %d", (mv).y, yborder_high, x_pos, y_pos); \
    (mv).y = yborder_high; \
  } else if ((mv).y < yborder_low) { \
    DPRINTF(XVID_DEBUG_MV, "mv.y < min -- %d < %d, MB %d, %d", (mv).y, yborder_low, x_pos, y_pos); \
    (mv).y = yborder_low; \
  } \
  } while (0)

  CHECK_MV(mv[0]);
  CHECK_MV(mv[1]);
  CHECK_MV(mv[2]);
  CHECK_MV(mv[3]);
}

/* Up to this version, chroma rounding was wrong with qpel. 
 * So we try to be backward compatible to avoid artifacts */
#define BS_VERSION_BUGGY_CHROMA_ROUNDING 1

/* decode an inter macroblock */
static void
decoder_mbinter(DECODER * dec,
        const MACROBLOCK * pMB,
        const uint32_t x_pos,
        const uint32_t y_pos,
        const uint32_t cbp,
        Bitstream * bs,
        const uint32_t rounding,
        const int ref,
		const int bvop)
{
  uint32_t stride = dec->edged_width;
  uint32_t stride2 = stride / 2;
  uint32_t i;

  uint8_t *pY_Cur, *pU_Cur, *pV_Cur;

  int uv_dx, uv_dy;
  VECTOR mv[4]; /* local copy of mvs */

  pY_Cur = dec->cur.y + (y_pos << 4) * stride + (x_pos << 4);
  pU_Cur = dec->cur.u + (y_pos << 3) * stride2 + (x_pos << 3);
  pV_Cur = dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3);
  for (i = 0; i < 4; i++)
    mv[i] = pMB->mvs[i];

  validate_vector(mv, x_pos, y_pos, dec);

  start_timer();

  if ((pMB->mode != MODE_INTER4V) || (bvop)) { /* INTER, INTER_Q, NOT_CODED, FORWARD, BACKWARD */

    uv_dx = mv[0].x;
    uv_dy = mv[0].y;
    if (dec->quarterpel) {
			if (dec->bs_version <= BS_VERSION_BUGGY_CHROMA_ROUNDING) {
  				uv_dx = (uv_dx>>1) | (uv_dx&1);
				uv_dy = (uv_dy>>1) | (uv_dy&1);
			}
			else {
        uv_dx /= 2;
        uv_dy /= 2;
      }
    }
    uv_dx = (uv_dx >> 1) + roundtab_79[uv_dx & 0x3];
    uv_dy = (uv_dy >> 1) + roundtab_79[uv_dy & 0x3];

    if (dec->quarterpel)
      interpolate16x16_quarterpel(dec->cur.y, dec->refn[ref].y, dec->qtmp.y, dec->qtmp.y + 64,
                  dec->qtmp.y + 128, 16*x_pos, 16*y_pos,
                      mv[0].x, mv[0].y, stride, rounding);
    else
      interpolate16x16_switch(dec->cur.y, dec->refn[ref].y, 16*x_pos, 16*y_pos,
                  mv[0].x, mv[0].y, stride, rounding);

  } else {  /* MODE_INTER4V */

    if(dec->quarterpel) {
			if (dec->bs_version <= BS_VERSION_BUGGY_CHROMA_ROUNDING) {
				int z;
				uv_dx = 0; uv_dy = 0;
				for (z = 0; z < 4; z++) {
				  uv_dx += ((mv[z].x>>1) | (mv[z].x&1));
				  uv_dy += ((mv[z].y>>1) | (mv[z].y&1));
				}
			}
			else {
        uv_dx = (mv[0].x / 2) + (mv[1].x / 2) + (mv[2].x / 2) + (mv[3].x / 2);
        uv_dy = (mv[0].y / 2) + (mv[1].y / 2) + (mv[2].y / 2) + (mv[3].y / 2);
      }
    } else {
      uv_dx = mv[0].x + mv[1].x + mv[2].x + mv[3].x;
      uv_dy = mv[0].y + mv[1].y + mv[2].y + mv[3].y;
    }

    uv_dx = (uv_dx >> 3) + roundtab_76[uv_dx & 0xf];
    uv_dy = (uv_dy >> 3) + roundtab_76[uv_dy & 0xf];

    if (dec->quarterpel) {
      interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y , dec->qtmp.y, dec->qtmp.y + 64,
                  dec->qtmp.y + 128, 16*x_pos, 16*y_pos,
                  mv[0].x, mv[0].y, stride, rounding);
      interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y , dec->qtmp.y, dec->qtmp.y + 64,
                  dec->qtmp.y + 128, 16*x_pos + 8, 16*y_pos,
                  mv[1].x, mv[1].y, stride, rounding);
      interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y , dec->qtmp.y, dec->qtmp.y + 64,
                  dec->qtmp.y + 128, 16*x_pos, 16*y_pos + 8,
                  mv[2].x, mv[2].y, stride, rounding);
      interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y , dec->qtmp.y, dec->qtmp.y + 64,
                  dec->qtmp.y + 128, 16*x_pos + 8, 16*y_pos + 8,
                  mv[3].x, mv[3].y, stride, rounding);
    } else {
      interpolate8x8_switch(dec->cur.y, dec->refn[0].y , 16*x_pos, 16*y_pos,
                mv[0].x, mv[0].y, stride, rounding);
      interpolate8x8_switch(dec->cur.y, dec->refn[0].y , 16*x_pos + 8, 16*y_pos,
                mv[1].x, mv[1].y, stride, rounding);
      interpolate8x8_switch(dec->cur.y, dec->refn[0].y , 16*x_pos, 16*y_pos + 8,
                mv[2].x, mv[2].y, stride, rounding);
      interpolate8x8_switch(dec->cur.y, dec->refn[0].y , 16*x_pos + 8, 16*y_pos + 8,
                mv[3].x, mv[3].y, stride, rounding);
    }
  }

  /* chroma */
  interpolate8x8_switch(dec->cur.u, dec->refn[ref].u, 8 * x_pos, 8 * y_pos,
              uv_dx, uv_dy, stride2, rounding);
  interpolate8x8_switch(dec->cur.v, dec->refn[ref].v, 8 * x_pos, 8 * y_pos,
              uv_dx, uv_dy, stride2, rounding);

  stop_comp_timer();

  if (cbp)
    decoder_mb_decode(dec, cbp, bs, pY_Cur, pU_Cur, pV_Cur, pMB);
}

/* decode an inter macroblock in field mode */
static void
decoder_mbinter_field(DECODER * dec,
        const MACROBLOCK * pMB,
        const uint32_t x_pos,
        const uint32_t y_pos,
        const uint32_t cbp,
        Bitstream * bs,
        const uint32_t rounding,
        const int ref,
		const int bvop)
{
  uint32_t stride = dec->edged_width;
  uint32_t stride2 = stride / 2;

  uint8_t *pY_Cur, *pU_Cur, *pV_Cur;

  int uvtop_dx, uvtop_dy;
  int uvbot_dx, uvbot_dy;
  VECTOR mv[4]; /* local copy of mvs */

  /* Get pointer to memory areas */
  pY_Cur = dec->cur.y + (y_pos << 4) * stride + (x_pos << 4);
  pU_Cur = dec->cur.u + (y_pos << 3) * stride2 + (x_pos << 3);
  pV_Cur = dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3);
  
  mv[0] = pMB->mvs[0];
  mv[1] = pMB->mvs[1];
  memset(&mv[2],0,2*sizeof(VECTOR));

  validate_vector(mv, x_pos, y_pos, dec);

  start_timer();

  if((pMB->mode!=MODE_INTER4V) || (bvop))   /* INTER, INTER_Q, NOT_CODED, FORWARD, BACKWARD */
  { 
    /* Prepare top field vector */
    uvtop_dx = DIV2ROUND(mv[0].x);
    uvtop_dy = DIV2ROUND(mv[0].y);

    /* Prepare bottom field vector */
    uvbot_dx = DIV2ROUND(mv[1].x);
    uvbot_dy = DIV2ROUND(mv[1].y);

    if(dec->quarterpel)
    {
      /* NOT supported */
    }
    else
    {
      /* Interpolate top field left part(we use double stride for every 2nd line) */
      interpolate8x8_switch(dec->cur.y,dec->refn[ref].y+pMB->field_for_top*stride,
                            16*x_pos,8*y_pos,mv[0].x, mv[0].y>>1,2*stride, rounding);
      /* top field right part */
      interpolate8x8_switch(dec->cur.y,dec->refn[ref].y+pMB->field_for_top*stride,
                            16*x_pos+8,8*y_pos,mv[0].x, mv[0].y>>1,2*stride, rounding);

      /* Interpolate bottom field left part(we use double stride for every 2nd line) */
      interpolate8x8_switch(dec->cur.y+stride,dec->refn[ref].y+pMB->field_for_bot*stride,
                            16*x_pos,8*y_pos,mv[1].x, mv[1].y>>1,2*stride, rounding);
      /* Bottom field right part */
      interpolate8x8_switch(dec->cur.y+stride,dec->refn[ref].y+pMB->field_for_bot*stride,
                            16*x_pos+8,8*y_pos,mv[1].x, mv[1].y>>1,2*stride, rounding);

      /* Interpolate field1 U */
      interpolate8x4_switch(dec->cur.u,dec->refn[ref].u+pMB->field_for_top*stride2,
                            8*x_pos,4*y_pos,uvtop_dx,DIV2ROUND(uvtop_dy),stride,rounding);
      
      /* Interpolate field1 V */
      interpolate8x4_switch(dec->cur.v,dec->refn[ref].v+pMB->field_for_top*stride2,
                            8*x_pos,4*y_pos,uvtop_dx,DIV2ROUND(uvtop_dy),stride,rounding);
    
      /* Interpolate field2 U */
      interpolate8x4_switch(dec->cur.u+stride2,dec->refn[ref].u+pMB->field_for_bot*stride2,
                            8*x_pos,4*y_pos,uvbot_dx,DIV2ROUND(uvbot_dy),stride,rounding);
    
      /* Interpolate field2 V */
      interpolate8x4_switch(dec->cur.v+stride2,dec->refn[ref].v+pMB->field_for_bot*stride2,
                            8*x_pos,4*y_pos,uvbot_dx,DIV2ROUND(uvbot_dy),stride,rounding);
    }
  } 
  else 
  {
    /* We don't expect 4 motion vectors in interlaced mode */
  }

  stop_comp_timer();

  /* Must add error correction? */
  if(cbp)
   decoder_mb_decode(dec, cbp, bs, pY_Cur, pU_Cur, pV_Cur, pMB);
}

static void
decoder_mbgmc(DECODER * dec,
        MACROBLOCK * const pMB,
        const uint32_t x_pos,
        const uint32_t y_pos,
        const uint32_t fcode,
        const uint32_t cbp,
        Bitstream * bs,
        const uint32_t rounding)
{
  const uint32_t stride = dec->edged_width;
  const uint32_t stride2 = stride / 2;

  uint8_t *const pY_Cur=dec->cur.y + (y_pos << 4) * stride + (x_pos << 4);
  uint8_t *const pU_Cur=dec->cur.u + (y_pos << 3) * stride2 + (x_pos << 3);
  uint8_t *const pV_Cur=dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3);

  NEW_GMC_DATA * gmc_data = &dec->new_gmc_data;

  pMB->mvs[0] = pMB->mvs[1] = pMB->mvs[2] = pMB->mvs[3] = pMB->amv;

  start_timer();

/* this is where the calculations are done */

  gmc_data->predict_16x16(gmc_data,
      dec->cur.y + y_pos*16*stride + x_pos*16, dec->refn[0].y,
      stride, stride, x_pos, y_pos, rounding);

  gmc_data->predict_8x8(gmc_data,
      dec->cur.u + y_pos*8*stride2 + x_pos*8, dec->refn[0].u,
      dec->cur.v + y_pos*8*stride2 + x_pos*8, dec->refn[0].v,
      stride2, stride2, x_pos, y_pos, rounding);

  gmc_data->get_average_mv(gmc_data, &pMB->amv, x_pos, y_pos, dec->quarterpel);

  pMB->amv.x = gmc_sanitize(pMB->amv.x, dec->quarterpel, fcode);
  pMB->amv.y = gmc_sanitize(pMB->amv.y, dec->quarterpel, fcode);

  pMB->mvs[0] = pMB->mvs[1] = pMB->mvs[2] = pMB->mvs[3] = pMB->amv;

  stop_transfer_timer();

  if (cbp)
    decoder_mb_decode(dec, cbp, bs, pY_Cur, pU_Cur, pV_Cur, pMB);

}


static void
decoder_iframe(DECODER * dec,
        Bitstream * bs,
        int quant,
        int intra_dc_threshold)
{
  uint32_t bound;
  uint32_t x, y;
  const uint32_t mb_width = dec->mb_width;
  const uint32_t mb_height = dec->mb_height;

  bound = 0;

  for (y = 0; y < mb_height; y++) {
    for (x = 0; x < mb_width; x++) {
      MACROBLOCK *mb;
      uint32_t mcbpc;
      uint32_t cbpc;
      uint32_t acpred_flag;
      uint32_t cbpy;
      uint32_t cbp;

      while (BitstreamShowBits(bs, 9) == 1)
        BitstreamSkip(bs, 9);

      if (check_resync_marker(bs, 0))
      {
        bound = read_video_packet_header(bs, dec, 0,
              &quant, NULL, NULL, &intra_dc_threshold);
        x = bound % mb_width;
        y = MIN((bound / mb_width), (mb_height-1));
      }
      mb = &dec->mbs[y * dec->mb_width + x];

      DPRINTF(XVID_DEBUG_MB, "macroblock (%i,%i) %08x\n", x, y, BitstreamShowBits(bs, 32));

      mcbpc = get_mcbpc_intra(bs);
      mb->mode = mcbpc & 7;
      cbpc = (mcbpc >> 4);

      acpred_flag = BitstreamGetBit(bs);

      cbpy = get_cbpy(bs, 1);
      cbp = (cbpy << 2) | cbpc;

      if (mb->mode == MODE_INTRA_Q) {
        quant += dquant_table[BitstreamGetBits(bs, 2)];
        if (quant > 31) {
          quant = 31;
        } else if (quant < 1) {
          quant = 1;
        }
      }
      mb->quant = quant;
      mb->mvs[0].x = mb->mvs[0].y =
      mb->mvs[1].x = mb->mvs[1].y =
      mb->mvs[2].x = mb->mvs[2].y =
      mb->mvs[3].x = mb->mvs[3].y =0;

      if (dec->interlacing) {
        mb->field_dct = BitstreamGetBit(bs);
        DPRINTF(XVID_DEBUG_MB,"deci: field_dct: %i\n", mb->field_dct);
      }

      decoder_mbintra(dec, mb, x, y, acpred_flag, cbp, bs, quant,
              intra_dc_threshold, bound);

    }
    if(dec->out_frm)
      output_slice(&dec->cur, dec->edged_width,dec->width,dec->out_frm,0,y,mb_width);
  }

}


static void
get_motion_vector(DECODER * dec,
        Bitstream * bs,
        int x,
        int y,
        int k,
        VECTOR * ret_mv,
        int fcode,
        const int bound)
{

  const int scale_fac = 1 << (fcode - 1);
  const int high = (32 * scale_fac) - 1;
  const int low = ((-32) * scale_fac);
  const int range = (64 * scale_fac);

  const VECTOR pmv = get_pmv2(dec->mbs, dec->mb_width, bound, x, y, k);
  VECTOR mv;

  mv.x = get_mv(bs, fcode);
  mv.y = get_mv(bs, fcode);

  DPRINTF(XVID_DEBUG_MV,"mv_diff (%i,%i) pred (%i,%i) result (%i,%i)\n", mv.x, mv.y, pmv.x, pmv.y, mv.x+pmv.x, mv.y+pmv.y);

  mv.x += pmv.x;
  mv.y += pmv.y;

  if (mv.x < low) {
    mv.x += range;
  } else if (mv.x > high) {
    mv.x -= range;
  }

  if (mv.y < low) {
    mv.y += range;
  } else if (mv.y > high) {
    mv.y -= range;
  }

  ret_mv->x = mv.x;
  ret_mv->y = mv.y;
}

/* We use this when decoder runs interlaced -> different prediction */

static void get_motion_vector_interlaced(DECODER * dec,
        Bitstream * bs,
        int x,
        int y,
        int k,
        MACROBLOCK *pMB,
        int fcode,
        const int bound)
{
  const int scale_fac = 1 << (fcode - 1);
  const int high = (32 * scale_fac) - 1;
  const int low = ((-32) * scale_fac);
  const int range = (64 * scale_fac);
  
  /* Get interlaced prediction */
  const VECTOR pmv=get_pmv2_interlaced(dec->mbs,dec->mb_width,bound,x,y,k);
  VECTOR mv,mvf1,mvf2;

  if(!pMB->field_pred)
  {
    mv.x = get_mv(bs,fcode);
    mv.y = get_mv(bs,fcode);
    
    mv.x += pmv.x;
    mv.y += pmv.y;

    if(mv.x<low) {
      mv.x += range;
    } else if (mv.x>high) {
      mv.x-=range;
    }

    if (mv.y < low) {
      mv.y += range;
    } else if (mv.y > high) {
      mv.y -= range;
    }
    
    pMB->mvs[0]=pMB->mvs[1]=pMB->mvs[2]=pMB->mvs[3]=mv;
  }
  else
  {
    mvf1.x = get_mv(bs, fcode);
    mvf1.y = get_mv(bs, fcode);

    mvf1.x += pmv.x;
    mvf1.y = 2*(mvf1.y+pmv.y/2); /* It's multiple of 2 */

    if (mvf1.x < low) {
      mvf1.x += range;
    } else if (mvf1.x > high) {
      mvf1.x -= range;
    }

    if (mvf1.y < low) {
      mvf1.y += range;
    } else if (mvf1.y > high) {
      mvf1.y -= range;
    }

    mvf2.x = get_mv(bs, fcode);
    mvf2.y = get_mv(bs, fcode);

    mvf2.x += pmv.x;
    mvf2.y = 2*(mvf2.y+pmv.y/2); /* It's multiple of 2 */

    if (mvf2.x < low) {
      mvf2.x += range;
    } else if (mvf2.x > high) {
      mvf2.x -= range;
    }

    if (mvf2.y < low) {
      mvf2.y += range;
    } else if (mvf2.y > high) {
      mvf2.y -= range;
    }

    pMB->mvs[0]=mvf1;
    pMB->mvs[1]=mvf2;
    pMB->mvs[2].x=pMB->mvs[3].x=0;
    pMB->mvs[2].y=pMB->mvs[3].y=0;
  
    /* Calculate average for as it is field predicted */
    pMB->mvs_avg.x=DIV2ROUND(pMB->mvs[0].x+pMB->mvs[1].x);
    pMB->mvs_avg.y=DIV2ROUND(pMB->mvs[0].y+pMB->mvs[1].y);
  }
}

/* for P_VOP set gmc_warp to NULL */
static void
decoder_pframe(DECODER * dec,
        Bitstream * bs,
        int rounding,
        int quant,
        int fcode,
        int intra_dc_threshold,
        const WARPPOINTS *const gmc_warp)
{
  uint32_t x, y;
  uint32_t bound;
  int cp_mb, st_mb;
  const uint32_t mb_width = dec->mb_width;
  const uint32_t mb_height = dec->mb_height;

  if (!dec->is_edged[0]) {
    start_timer();
    image_setedges(&dec->refn[0], dec->edged_width, dec->edged_height,
            dec->width, dec->height, dec->bs_version);
    dec->is_edged[0] = 1;
    stop_edges_timer();
  }

  if (gmc_warp) {
    /* accuracy: 0==1/2, 1=1/4, 2=1/8, 3=1/16 */
    generate_GMCparameters( dec->sprite_warping_points,
        dec->sprite_warping_accuracy, gmc_warp,
        dec->width, dec->height, &dec->new_gmc_data);

    /* image warping is done block-based in decoder_mbgmc(), now */
  }

  bound = 0;

  for (y = 0; y < mb_height; y++) {
    cp_mb = st_mb = 0;
    for (x = 0; x < mb_width; x++) {
      MACROBLOCK *mb;

      /* skip stuffing */
      while (BitstreamShowBits(bs, 10) == 1)
        BitstreamSkip(bs, 10);

      if (check_resync_marker(bs, fcode - 1)) {
        bound = read_video_packet_header(bs, dec, fcode - 1,
          &quant, &fcode, NULL, &intra_dc_threshold);
        x = bound % mb_width;
        y = MIN((bound / mb_width), (mb_height-1));
      }
      mb = &dec->mbs[y * dec->mb_width + x];

      DPRINTF(XVID_DEBUG_MB, "macroblock (%i,%i) %08x\n", x, y, BitstreamShowBits(bs, 32));

      if (!(BitstreamGetBit(bs))) { /* block _is_ coded */
        uint32_t mcbpc, cbpc, cbpy, cbp;
        uint32_t intra, acpred_flag = 0;
        int mcsel = 0;    /* mcsel: '0'=local motion, '1'=GMC */

        cp_mb++;
        mcbpc = get_mcbpc_inter(bs);
        mb->mode = mcbpc & 7;
        cbpc = (mcbpc >> 4);

        DPRINTF(XVID_DEBUG_MB, "mode %i\n", mb->mode);
        DPRINTF(XVID_DEBUG_MB, "cbpc %i\n", cbpc);

        intra = (mb->mode == MODE_INTRA || mb->mode == MODE_INTRA_Q);

        if (gmc_warp && (mb->mode == MODE_INTER || mb->mode == MODE_INTER_Q))
          mcsel = BitstreamGetBit(bs);
        else if (intra)
          acpred_flag = BitstreamGetBit(bs);

        cbpy = get_cbpy(bs, intra);
        DPRINTF(XVID_DEBUG_MB, "cbpy %i mcsel %i \n", cbpy,mcsel);

        cbp = (cbpy << 2) | cbpc;

        if (mb->mode == MODE_INTER_Q || mb->mode == MODE_INTRA_Q) {
          int dquant = dquant_table[BitstreamGetBits(bs, 2)];
          DPRINTF(XVID_DEBUG_MB, "dquant %i\n", dquant);
          quant += dquant;
          if (quant > 31) {
            quant = 31;
          } else if (quant < 1) {
            quant = 1;
          }
          DPRINTF(XVID_DEBUG_MB, "quant %i\n", quant);
        }
        mb->quant = quant;

        mb->field_pred=0;
        if (dec->interlacing) {
          if (cbp || intra) {
            mb->field_dct = BitstreamGetBit(bs);
            DPRINTF(XVID_DEBUG_MB,"decp: field_dct: %i\n", mb->field_dct);
          }

          if ((mb->mode == MODE_INTER || mb->mode == MODE_INTER_Q) && !mcsel) {
            mb->field_pred = BitstreamGetBit(bs);
            DPRINTF(XVID_DEBUG_MB, "decp: field_pred: %i\n", mb->field_pred);

            if (mb->field_pred) {
              mb->field_for_top = BitstreamGetBit(bs);
              DPRINTF(XVID_DEBUG_MB,"decp: field_for_top: %i\n", mb->field_for_top);
              mb->field_for_bot = BitstreamGetBit(bs);
              DPRINTF(XVID_DEBUG_MB,"decp: field_for_bot: %i\n", mb->field_for_bot);
            }
          }
        }

        if (mcsel) {
          decoder_mbgmc(dec, mb, x, y, fcode, cbp, bs, rounding);
          continue;

        } else if (mb->mode == MODE_INTER || mb->mode == MODE_INTER_Q) {

          if(dec->interlacing) {
            /* Get motion vectors interlaced, field_pred is handled there */
            get_motion_vector_interlaced(dec, bs, x, y, 0, mb, fcode, bound);
          } else {
            get_motion_vector(dec, bs, x, y, 0, &mb->mvs[0], fcode, bound);
            mb->mvs[1] = mb->mvs[2] = mb->mvs[3] = mb->mvs[0];
          }
        } else if (mb->mode == MODE_INTER4V ) {
          /* interlaced missing here */
          get_motion_vector(dec, bs, x, y, 0, &mb->mvs[0], fcode, bound);
          get_motion_vector(dec, bs, x, y, 1, &mb->mvs[1], fcode, bound);
          get_motion_vector(dec, bs, x, y, 2, &mb->mvs[2], fcode, bound);
          get_motion_vector(dec, bs, x, y, 3, &mb->mvs[3], fcode, bound);
        } else { /* MODE_INTRA, MODE_INTRA_Q */
          mb->mvs[0].x = mb->mvs[1].x = mb->mvs[2].x = mb->mvs[3].x = 0;
          mb->mvs[0].y = mb->mvs[1].y = mb->mvs[2].y = mb->mvs[3].y = 0;
          decoder_mbintra(dec, mb, x, y, acpred_flag, cbp, bs, quant,
                  intra_dc_threshold, bound);
          continue;
        }

        /* See how to decode */
        if(!mb->field_pred)
         decoder_mbinter(dec, mb, x, y, cbp, bs, rounding, 0, 0);
        else 
         decoder_mbinter_field(dec, mb, x, y, cbp, bs, rounding, 0, 0);

      } else if (gmc_warp) {  /* a not coded S(GMC)-VOP macroblock */
        mb->mode = MODE_NOT_CODED_GMC;
        mb->quant = quant;
        decoder_mbgmc(dec, mb, x, y, fcode, 0x00, bs, rounding);

        if(dec->out_frm && cp_mb > 0) {
          output_slice(&dec->cur, dec->edged_width,dec->width,dec->out_frm,st_mb,y,cp_mb);
          cp_mb = 0;
        }
        st_mb = x+1;
      } else { /* not coded P_VOP macroblock */
        mb->mode = MODE_NOT_CODED;
        mb->quant = quant;

        mb->mvs[0].x = mb->mvs[1].x = mb->mvs[2].x = mb->mvs[3].x = 0;
        mb->mvs[0].y = mb->mvs[1].y = mb->mvs[2].y = mb->mvs[3].y = 0;
        mb->field_pred=0; /* (!) */

        decoder_mbinter(dec, mb, x, y, 0, bs, 
                                rounding, 0, 0);

        if(dec->out_frm && cp_mb > 0) {
          output_slice(&dec->cur, dec->edged_width,dec->width,dec->out_frm,st_mb,y,cp_mb);
          cp_mb = 0;
        }
        st_mb = x+1;
      }
    }

    if(dec->out_frm && cp_mb > 0)
      output_slice(&dec->cur, dec->edged_width,dec->width,dec->out_frm,st_mb,y,cp_mb);
  }
}


/* decode B-frame motion vector */
static void
get_b_motion_vector(Bitstream * bs,
          VECTOR * mv,
          int fcode,
          const VECTOR pmv,
          const DECODER * const dec,
          const int x, const int y)
{
  const int scale_fac = 1 << (fcode - 1);
  const int high = (32 * scale_fac) - 1;
  const int low = ((-32) * scale_fac);
  const int range = (64 * scale_fac);

  int mv_x = get_mv(bs, fcode);
  int mv_y = get_mv(bs, fcode);

  mv_x += pmv.x;
  mv_y += pmv.y;

  if (mv_x < low)
    mv_x += range;
  else if (mv_x > high)
    mv_x -= range;

  if (mv_y < low)
    mv_y += range;
  else if (mv_y > high)
    mv_y -= range;

  mv->x = mv_x;
  mv->y = mv_y;
}

/* decode an B-frame direct & interpolate macroblock */
static void
decoder_bf_interpolate_mbinter(DECODER * dec,
                IMAGE forward,
                IMAGE backward,
                MACROBLOCK * pMB,
                const uint32_t x_pos,
                const uint32_t y_pos,
                Bitstream * bs,
                const int direct)
{
  uint32_t stride = dec->edged_width;
  uint32_t stride2 = stride / 2;
  int uv_dx, uv_dy;
  int b_uv_dx, b_uv_dy;
  uint8_t *pY_Cur, *pU_Cur, *pV_Cur;
  const uint32_t cbp = pMB->cbp;

  pY_Cur = dec->cur.y + (y_pos << 4) * stride + (x_pos << 4);
  pU_Cur = dec->cur.u + (y_pos << 3) * stride2 + (x_pos << 3);
  pV_Cur = dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3);

  validate_vector(pMB->mvs, x_pos, y_pos, dec);
  validate_vector(pMB->b_mvs, x_pos, y_pos, dec);

  if (!direct) {
    uv_dx = pMB->mvs[0].x;
    uv_dy = pMB->mvs[0].y;
    b_uv_dx = pMB->b_mvs[0].x;
    b_uv_dy = pMB->b_mvs[0].y;

    if (dec->quarterpel) {
			if (dec->bs_version <= BS_VERSION_BUGGY_CHROMA_ROUNDING) {
				uv_dx = (uv_dx>>1) | (uv_dx&1);
				uv_dy = (uv_dy>>1) | (uv_dy&1);
				b_uv_dx = (b_uv_dx>>1) | (b_uv_dx&1);
				b_uv_dy = (b_uv_dy>>1) | (b_uv_dy&1);
			}
			else {
        uv_dx /= 2;
        uv_dy /= 2;
        b_uv_dx /= 2;
        b_uv_dy /= 2;
      }
    }

    uv_dx = (uv_dx >> 1) + roundtab_79[uv_dx & 0x3];
    uv_dy = (uv_dy >> 1) + roundtab_79[uv_dy & 0x3];
    b_uv_dx = (b_uv_dx >> 1) + roundtab_79[b_uv_dx & 0x3];
    b_uv_dy = (b_uv_dy >> 1) + roundtab_79[b_uv_dy & 0x3];

  } else {
	  if (dec->quarterpel) { /* for qpel the /2 shall be done before summation. We've done it right in the encoder in the past. */
							 /* TODO: figure out if we ever did it wrong on the encoder side. If yes, add some workaround */
		if (dec->bs_version <= BS_VERSION_BUGGY_CHROMA_ROUNDING) {
			int z;
			uv_dx = 0; uv_dy = 0;
			b_uv_dx = 0; b_uv_dy = 0;
			for (z = 0; z < 4; z++) {
			  uv_dx += ((pMB->mvs[z].x>>1) | (pMB->mvs[z].x&1));
			  uv_dy += ((pMB->mvs[z].y>>1) | (pMB->mvs[z].y&1));
			  b_uv_dx += ((pMB->b_mvs[z].x>>1) | (pMB->b_mvs[z].x&1));
			  b_uv_dy += ((pMB->b_mvs[z].y>>1) | (pMB->b_mvs[z].y&1));
			}
		}
		else {
			uv_dx = (pMB->mvs[0].x / 2) + (pMB->mvs[1].x / 2) + (pMB->mvs[2].x / 2) + (pMB->mvs[3].x / 2);
			uv_dy = (pMB->mvs[0].y / 2) + (pMB->mvs[1].y / 2) + (pMB->mvs[2].y / 2) + (pMB->mvs[3].y / 2);
			b_uv_dx = (pMB->b_mvs[0].x / 2) + (pMB->b_mvs[1].x / 2) + (pMB->b_mvs[2].x / 2) + (pMB->b_mvs[3].x / 2);
			b_uv_dy = (pMB->b_mvs[0].y / 2) + (pMB->b_mvs[1].y / 2) + (pMB->b_mvs[2].y / 2) + (pMB->b_mvs[3].y / 2);
		} 
	} else {
      uv_dx = pMB->mvs[0].x + pMB->mvs[1].x + pMB->mvs[2].x + pMB->mvs[3].x;
      uv_dy = pMB->mvs[0].y + pMB->mvs[1].y + pMB->mvs[2].y + pMB->mvs[3].y;
      b_uv_dx = pMB->b_mvs[0].x + pMB->b_mvs[1].x + pMB->b_mvs[2].x + pMB->b_mvs[3].x;
      b_uv_dy = pMB->b_mvs[0].y + pMB->b_mvs[1].y + pMB->b_mvs[2].y + pMB->b_mvs[3].y;
    }

    uv_dx = (uv_dx >> 3) + roundtab_76[uv_dx & 0xf];
    uv_dy = (uv_dy >> 3) + roundtab_76[uv_dy & 0xf];
    b_uv_dx = (b_uv_dx >> 3) + roundtab_76[b_uv_dx & 0xf];
    b_uv_dy = (b_uv_dy >> 3) + roundtab_76[b_uv_dy & 0xf];
  }

  start_timer();
  if(dec->quarterpel) {
    if(!direct) {
      interpolate16x16_quarterpel(dec->cur.y, forward.y, dec->qtmp.y, dec->qtmp.y + 64,
                    dec->qtmp.y + 128, 16*x_pos, 16*y_pos,
                    pMB->mvs[0].x, pMB->mvs[0].y, stride, 0);
    } else {
      interpolate8x8_quarterpel(dec->cur.y, forward.y, dec->qtmp.y, dec->qtmp.y + 64,
                    dec->qtmp.y + 128, 16*x_pos, 16*y_pos,
                    pMB->mvs[0].x, pMB->mvs[0].y, stride, 0);
      interpolate8x8_quarterpel(dec->cur.y, forward.y, dec->qtmp.y, dec->qtmp.y + 64,
                    dec->qtmp.y + 128, 16*x_pos + 8, 16*y_pos,
                    pMB->mvs[1].x, pMB->mvs[1].y, stride, 0);
      interpolate8x8_quarterpel(dec->cur.y, forward.y, dec->qtmp.y, dec->qtmp.y + 64,
                    dec->qtmp.y + 128, 16*x_pos, 16*y_pos + 8,
                    pMB->mvs[2].x, pMB->mvs[2].y, stride, 0);
      interpolate8x8_quarterpel(dec->cur.y, forward.y, dec->qtmp.y, dec->qtmp.y + 64,
                    dec->qtmp.y + 128, 16*x_pos + 8, 16*y_pos + 8,
                    pMB->mvs[3].x, pMB->mvs[3].y, stride, 0);
    }
  } else {
    interpolate8x8_switch(dec->cur.y, forward.y, 16 * x_pos, 16 * y_pos,
              pMB->mvs[0].x, pMB->mvs[0].y, stride, 0);
    interpolate8x8_switch(dec->cur.y, forward.y, 16 * x_pos + 8, 16 * y_pos,
              pMB->mvs[1].x, pMB->mvs[1].y, stride, 0);
    interpolate8x8_switch(dec->cur.y, forward.y, 16 * x_pos, 16 * y_pos + 8,
              pMB->mvs[2].x, pMB->mvs[2].y, stride, 0);
    interpolate8x8_switch(dec->cur.y, forward.y, 16 * x_pos + 8, 16 * y_pos + 8,
              pMB->mvs[3].x, pMB->mvs[3].y, stride, 0);
  }

  interpolate8x8_switch(dec->cur.u, forward.u, 8 * x_pos, 8 * y_pos, uv_dx,
            uv_dy, stride2, 0);
  interpolate8x8_switch(dec->cur.v, forward.v, 8 * x_pos, 8 * y_pos, uv_dx,
            uv_dy, stride2, 0);


  if(dec->quarterpel) {
    if(!direct) {
      interpolate16x16_add_quarterpel(dec->cur.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64,
          dec->qtmp.y + 128, 16*x_pos, 16*y_pos,
          pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0);
    } else {
      interpolate8x8_add_quarterpel(dec->cur.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64,
          dec->qtmp.y + 128, 16*x_pos, 16*y_pos,
          pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0);
      interpolate8x8_add_quarterpel(dec->cur.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64,
          dec->qtmp.y + 128, 16*x_pos + 8, 16*y_pos,
          pMB->b_mvs[1].x, pMB->b_mvs[1].y, stride, 0);
      interpolate8x8_add_quarterpel(dec->cur.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64,
          dec->qtmp.y + 128, 16*x_pos, 16*y_pos + 8,
          pMB->b_mvs[2].x, pMB->b_mvs[2].y, stride, 0);
      interpolate8x8_add_quarterpel(dec->cur.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64,
          dec->qtmp.y + 128, 16*x_pos + 8, 16*y_pos + 8,
          pMB->b_mvs[3].x, pMB->b_mvs[3].y, stride, 0);
    }
  } else {
    interpolate8x8_add_switch(dec->cur.y, backward.y, 16 * x_pos, 16 * y_pos,
        pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0);
    interpolate8x8_add_switch(dec->cur.y, backward.y, 16 * x_pos + 8,
        16 * y_pos, pMB->b_mvs[1].x, pMB->b_mvs[1].y, stride, 0);
    interpolate8x8_add_switch(dec->cur.y, backward.y, 16 * x_pos,
        16 * y_pos + 8, pMB->b_mvs[2].x, pMB->b_mvs[2].y, stride, 0);
    interpolate8x8_add_switch(dec->cur.y, backward.y, 16 * x_pos + 8,
        16 * y_pos + 8, pMB->b_mvs[3].x, pMB->b_mvs[3].y, stride, 0);
  }

  interpolate8x8_add_switch(dec->cur.u, backward.u, 8 * x_pos, 8 * y_pos,
      b_uv_dx, b_uv_dy, stride2, 0);
  interpolate8x8_add_switch(dec->cur.v, backward.v, 8 * x_pos, 8 * y_pos,
      b_uv_dx, b_uv_dy, stride2, 0);

  stop_comp_timer();

  if (cbp)
    decoder_mb_decode(dec, cbp, bs, pY_Cur, pU_Cur, pV_Cur, pMB);
}

/* for decode B-frame dbquant */
static __inline int32_t
get_dbquant(Bitstream * bs)
{
  if (!BitstreamGetBit(bs))   /*  '0' */
    return (0);
  else if (!BitstreamGetBit(bs))  /* '10' */
    return (-2);
  else              /* '11' */
    return (2);
}

/*
 * decode B-frame mb_type
 * bit    ret_value
 * 1    0
 * 01   1
 * 001    2
 * 0001   3
 */
static int32_t __inline
get_mbtype(Bitstream * bs)
{
  int32_t mb_type;

  for (mb_type = 0; mb_type <= 3; mb_type++)
    if (BitstreamGetBit(bs))
      return (mb_type);

  return -1;
}

static int __inline get_resync_len_b(const int fcode_backward,
                                     const int fcode_forward) {
  int resync_len = ((fcode_forward>fcode_backward) ? fcode_forward : fcode_backward) - 1;
  if (resync_len < 1) resync_len = 1;
  return resync_len;
}

static void
decoder_bframe(DECODER * dec,
        Bitstream * bs,
        int quant,
        int fcode_forward,
        int fcode_backward)
{
  uint32_t x, y;
  VECTOR mv;
  const VECTOR zeromv = {0,0};
  int i;
  int resync_len;

  if (!dec->is_edged[0]) {
    start_timer();
    image_setedges(&dec->refn[0], dec->edged_width, dec->edged_height,
            dec->width, dec->height, dec->bs_version);
    dec->is_edged[0] = 1;
    stop_edges_timer();
  }

  if (!dec->is_edged[1]) {
    start_timer();
    image_setedges(&dec->refn[1], dec->edged_width, dec->edged_height,
            dec->width, dec->height, dec->bs_version);
    dec->is_edged[1] = 1;
    stop_edges_timer();
  }

  resync_len = get_resync_len_b(fcode_backward, fcode_forward);
  for (y = 0; y < dec->mb_height; y++) {
    /* Initialize Pred Motion Vector */
    dec->p_fmv = dec->p_bmv = zeromv;
    for (x = 0; x < dec->mb_width; x++) {
      MACROBLOCK *mb = &dec->mbs[y * dec->mb_width + x];
      MACROBLOCK *last_mb = &dec->last_mbs[y * dec->mb_width + x];
      int intra_dc_threshold; /* fake variable */

      mv =
      mb->b_mvs[0] = mb->b_mvs[1] = mb->b_mvs[2] = mb->b_mvs[3] =
      mb->mvs[0] = mb->mvs[1] = mb->mvs[2] = mb->mvs[3] = zeromv;
      mb->quant = quant;

      /*
       * skip if the co-located P_VOP macroblock is not coded
       * if not codec in co-located S_VOP macroblock is _not_
       * automatically skipped
       */

      if (last_mb->mode == MODE_NOT_CODED) {
        mb->cbp = 0;
        mb->mode = MODE_FORWARD;
        decoder_mbinter(dec, mb, x, y, mb->cbp, bs, 0, 1, 1);
        continue;
      }

      if (check_resync_marker(bs, resync_len)) {
        int bound = read_video_packet_header(bs, dec, resync_len, &quant,
                           &fcode_forward, &fcode_backward, &intra_dc_threshold);

		bound = MAX(0, bound-1); /* valid bound must always be >0 */
        x = bound % dec->mb_width;
        y = MIN((bound / dec->mb_width), (dec->mb_height-1));
        /* reset predicted macroblocks */
        dec->p_fmv = dec->p_bmv = zeromv;
        /* update resync len with new fcodes */
        resync_len = get_resync_len_b(fcode_backward, fcode_forward);
		continue; /* re-init loop */
	  }

      if (!BitstreamGetBit(bs)) { /* modb=='0' */
        const uint8_t modb2 = BitstreamGetBit(bs);

        mb->mode = get_mbtype(bs);

        if (!modb2)   /* modb=='00' */
          mb->cbp = BitstreamGetBits(bs, 6);
        else
          mb->cbp = 0;

        if (mb->mode && mb->cbp) {
          quant += get_dbquant(bs);
          if (quant > 31)
            quant = 31;
          else if (quant < 1)
            quant = 1;
        }
        mb->quant = quant;

        if (dec->interlacing) {
          if (mb->cbp) {
            mb->field_dct = BitstreamGetBit(bs);
            DPRINTF(XVID_DEBUG_MB,"decp: field_dct: %i\n", mb->field_dct);
          }

          if (mb->mode) {
            mb->field_pred = BitstreamGetBit(bs);
            DPRINTF(XVID_DEBUG_MB, "decp: field_pred: %i\n", mb->field_pred);

            if (mb->field_pred) {
              mb->field_for_top = BitstreamGetBit(bs);
              DPRINTF(XVID_DEBUG_MB,"decp: field_for_top: %i\n", mb->field_for_top);
              mb->field_for_bot = BitstreamGetBit(bs);
              DPRINTF(XVID_DEBUG_MB,"decp: field_for_bot: %i\n", mb->field_for_bot);
            }
          }
        }

      } else {
        mb->mode = MODE_DIRECT_NONE_MV;
        mb->cbp = 0;
      }

      switch (mb->mode) {
      case MODE_DIRECT:
        get_b_motion_vector(bs, &mv, 1, zeromv, dec, x, y);

      case MODE_DIRECT_NONE_MV:
        for (i = 0; i < 4; i++) {
          mb->mvs[i].x = last_mb->mvs[i].x*dec->time_bp/dec->time_pp + mv.x;
          mb->mvs[i].y = last_mb->mvs[i].y*dec->time_bp/dec->time_pp + mv.y;

          mb->b_mvs[i].x = (mv.x)
            ?  mb->mvs[i].x - last_mb->mvs[i].x
            : last_mb->mvs[i].x*(dec->time_bp - dec->time_pp)/dec->time_pp;
          mb->b_mvs[i].y = (mv.y)
            ? mb->mvs[i].y - last_mb->mvs[i].y
            : last_mb->mvs[i].y*(dec->time_bp - dec->time_pp)/dec->time_pp;
        }

        decoder_bf_interpolate_mbinter(dec, dec->refn[1], dec->refn[0],
                        mb, x, y, bs, 1);
        break;

      case MODE_INTERPOLATE:
        get_b_motion_vector(bs, &mb->mvs[0], fcode_forward, dec->p_fmv, dec, x, y);
        dec->p_fmv = mb->mvs[1] = mb->mvs[2] = mb->mvs[3] = mb->mvs[0];

        get_b_motion_vector(bs, &mb->b_mvs[0], fcode_backward, dec->p_bmv, dec, x, y);
        dec->p_bmv = mb->b_mvs[1] = mb->b_mvs[2] = mb->b_mvs[3] = mb->b_mvs[0];

        decoder_bf_interpolate_mbinter(dec, dec->refn[1], dec->refn[0],
                      mb, x, y, bs, 0);
        break;

      case MODE_BACKWARD:
        get_b_motion_vector(bs, &mb->mvs[0], fcode_backward, dec->p_bmv, dec, x, y);
        dec->p_bmv = mb->mvs[1] = mb->mvs[2] = mb->mvs[3] = mb->mvs[0];

        decoder_mbinter(dec, mb, x, y, mb->cbp, bs, 0, 0, 1);
        break;

      case MODE_FORWARD:
        get_b_motion_vector(bs, &mb->mvs[0], fcode_forward, dec->p_fmv, dec, x, y);
        dec->p_fmv = mb->mvs[1] = mb->mvs[2] = mb->mvs[3] = mb->mvs[0];

        decoder_mbinter(dec, mb, x, y, mb->cbp, bs, 0, 1, 1);
        break;

      default:
        DPRINTF(XVID_DEBUG_ERROR,"Not supported B-frame mb_type = %i\n", mb->mode);
      }
    } /* End of for */
  }
}

/* perform post processing if necessary, and output the image */
static void decoder_output(DECODER * dec, IMAGE * img, MACROBLOCK * mbs,
          xvid_dec_frame_t * frame, xvid_dec_stats_t * stats,
          int coding_type, int quant)
{
  const int brightness = XVID_VERSION_MINOR(frame->version) >= 1 ? frame->brightness : 0;

  if (dec->cartoon_mode)
    frame->general &= ~XVID_FILMEFFECT;

  if ((frame->general & (XVID_DEBLOCKY|XVID_DEBLOCKUV|XVID_FILMEFFECT) || brightness!=0)
    && mbs != NULL) /* post process */
  {
    /* note: image is stored to tmp */
    image_copy(&dec->tmp, img, dec->edged_width, dec->height);
    image_postproc(&dec->postproc, &dec->tmp, dec->edged_width,
             mbs, dec->mb_width, dec->mb_height, dec->mb_width,
             frame->general, brightness, dec->frames, (coding_type == B_VOP), dec->num_threads);
    img = &dec->tmp;
  }

  image_output(img, dec->width, dec->height,
         dec->edged_width, (uint8_t**)frame->output.plane, frame->output.stride,
         frame->output.csp, dec->interlacing);

  if (stats) {
    stats->type = coding2type(coding_type);
    stats->data.vop.time_base = (int)dec->time_base;
    stats->data.vop.time_increment = 0; /* XXX: todo */
    stats->data.vop.qscale_stride = dec->mb_width;
    stats->data.vop.qscale = dec->qscale;
    if (stats->data.vop.qscale != NULL && mbs != NULL) {
      unsigned int i;
      for (i = 0; i < dec->mb_width*dec->mb_height; i++)
        stats->data.vop.qscale[i] = mbs[i].quant;
    } else
      stats->data.vop.qscale = NULL;
  }
}

int
decoder_decode(DECODER * dec,
        xvid_dec_frame_t * frame, xvid_dec_stats_t * stats)
{

  Bitstream bs;
  uint32_t rounding;
  uint32_t quant = 2;
  uint32_t fcode_forward;
  uint32_t fcode_backward;
  uint32_t intra_dc_threshold;
  WARPPOINTS gmc_warp;
  int coding_type;
  int success, output, seen_something;

  if (XVID_VERSION_MAJOR(frame->version) != 1 || (stats && XVID_VERSION_MAJOR(stats->version) != 1))  /* v1.x.x */
    return XVID_ERR_VERSION;

  start_global_timer();

  dec->low_delay_default = (frame->general & XVID_LOWDELAY);
  if ((frame->general & XVID_DISCONTINUITY))
    dec->frames = 0;
  dec->out_frm = (frame->output.csp == XVID_CSP_SLICE) ? &frame->output : NULL;

  if(frame->length<0) {  /* decoder flush */
    int ret;
    /* if not decoding "low_delay/packed", and this isn't low_delay and
      we have a reference frame, then outout the reference frame */
    if (!(dec->low_delay_default && dec->packed_mode) && !dec->low_delay && dec->frames>0) {
      decoder_output(dec, &dec->refn[0], dec->last_mbs, frame, stats, dec->last_coding_type, quant);
      dec->frames = 0;
      ret = 0;
    } else {
      if (stats) stats->type = XVID_TYPE_NOTHING;
      ret = XVID_ERR_END;
    }

    emms();
    stop_global_timer();
    return ret;
  }

  BitstreamInit(&bs, frame->bitstream, frame->length);

  /* XXX: 0x7f is only valid whilst decoding vfw xvid/divx5 avi's */
  if(dec->low_delay_default && frame->length == 1 && BitstreamShowBits(&bs, 8) == 0x7f)
  {
    image_output(&dec->refn[0], dec->width, dec->height, dec->edged_width,
           (uint8_t**)frame->output.plane, frame->output.stride, frame->output.csp, dec->interlacing);
    if (stats) stats->type = XVID_TYPE_NOTHING;
    emms();
    return 1; /* one byte consumed */
  }

  success = 0;
  output = 0;
  seen_something = 0;

repeat:

  coding_type = BitstreamReadHeaders(&bs, dec, &rounding,
      &quant, &fcode_forward, &fcode_backward, &intra_dc_threshold, &gmc_warp);

  DPRINTF(XVID_DEBUG_HEADER, "coding_type=%i,  packed=%i,  time=%"
#if defined(_MSC_VER)
    "I64"
#else
    "ll"
#endif
    "i,  time_pp=%i,  time_bp=%i\n",
              coding_type,  dec->packed_mode, dec->time, dec->time_pp, dec->time_bp);

  if (coding_type == -1) { /* nothing */
    if (success) goto done;
    if (stats) stats->type = XVID_TYPE_NOTHING;
    emms();
    return BitstreamPos(&bs)/8;
  }

  if (coding_type == -2 || coding_type == -3) { /* vol and/or resize */

    if (coding_type == -3)
      if (decoder_resize(dec)) return XVID_ERR_MEMORY;

    if(stats) {
      stats->type = XVID_TYPE_VOL;
      stats->data.vol.general = 0;
      /*XXX: if (dec->interlacing)
        stats->data.vol.general |= ++INTERLACING; */
      stats->data.vol.width = dec->width;
      stats->data.vol.height = dec->height;
      stats->data.vol.par = dec->aspect_ratio;
      stats->data.vol.par_width = dec->par_width;
      stats->data.vol.par_height = dec->par_height;
      emms();
      return BitstreamPos(&bs)/8; /* number of bytes consumed */
    }
    goto repeat;
  }

  if(dec->frames == 0 && coding_type != I_VOP) {
    /* 1st frame is not an i-vop */
    goto repeat;
  }

  dec->p_bmv.x = dec->p_bmv.y = dec->p_fmv.x = dec->p_fmv.y = 0;  /* init pred vector to 0 */

  /* packed_mode: special-N_VOP treament */
  if (dec->packed_mode && coding_type == N_VOP) {
    if (dec->low_delay_default && dec->frames > 0) {
      decoder_output(dec, &dec->refn[0], dec->last_mbs, frame, stats, dec->last_coding_type, quant);
      output = 1;
    }
    /* ignore otherwise */
  } else if (coding_type != B_VOP) {
    switch(coding_type) {
    case I_VOP :
      decoder_iframe(dec, &bs, quant, intra_dc_threshold);
      break;
    case P_VOP :
      decoder_pframe(dec, &bs, rounding, quant,
                        fcode_forward, intra_dc_threshold, NULL);
      break;
    case S_VOP :
      decoder_pframe(dec, &bs, rounding, quant,
                        fcode_forward, intra_dc_threshold, &gmc_warp);
      break;
    case N_VOP :
      /* XXX: not_coded vops are not used for forward prediction */
      /* we should not swap(last_mbs,mbs) */
      image_copy(&dec->cur, &dec->refn[0], dec->edged_width, dec->height);
      SWAP(MACROBLOCK *, dec->mbs, dec->last_mbs); /* it will be swapped back */
      break;
    }

    /* note: for packed_mode, output is performed when the special-N_VOP is decoded */
    if (!(dec->low_delay_default && dec->packed_mode)) {
      if(dec->low_delay) {
        decoder_output(dec, &dec->cur, dec->mbs, frame, stats, coding_type, quant);
        output = 1;
      } else if (dec->frames > 0) { /* is the reference frame valid? */
        /* output the reference frame */
        decoder_output(dec, &dec->refn[0], dec->last_mbs, frame, stats, dec->last_coding_type, quant);
        output = 1;
      }
    }
    
    image_swap(&dec->refn[0], &dec->refn[1]);
    dec->is_edged[1] = dec->is_edged[0];
    image_swap(&dec->cur, &dec->refn[0]);
    dec->is_edged[0] = 0;
    SWAP(MACROBLOCK *, dec->mbs, dec->last_mbs);
    dec->last_coding_type = coding_type;

    dec->frames++;
    seen_something = 1;

  } else {  /* B_VOP */

    if (dec->low_delay) {
      DPRINTF(XVID_DEBUG_ERROR, "warning: bvop found in low_delay==1 stream\n");
      dec->low_delay = 0;
    }

    if (dec->frames < 2) {
      /* attemping to decode a bvop without atleast 2 reference frames */
      image_printf(&dec->cur, dec->edged_width, dec->height, 16, 16,
            "broken b-frame, mising ref frames");
      if (stats) stats->type = XVID_TYPE_NOTHING;
    } else if (dec->time_pp <= dec->time_bp) {
      /* this occurs when dx50_bvop_compatibility==0 sequences are
      decoded in vfw. */
      image_printf(&dec->cur, dec->edged_width, dec->height, 16, 16,
            "broken b-frame, tpp=%i tbp=%i", dec->time_pp, dec->time_bp);
      if (stats) stats->type = XVID_TYPE_NOTHING;
    } else {
      decoder_bframe(dec, &bs, quant, fcode_forward, fcode_backward);
      decoder_output(dec, &dec->cur, dec->mbs, frame, stats, coding_type, quant);
    }

    output = 1;
    dec->frames++;
  }

#if 0 /* Avoids to read to much data because of 32bit reads in our BS functions */
   BitstreamByteAlign(&bs);
#endif

  /* low_delay_default mode: repeat in packed_mode */
  if (dec->low_delay_default && dec->packed_mode && output == 0 && success == 0) {
    success = 1;
    goto repeat;
  }

done :

  /* if we reach here without outputing anything _and_
     the calling application has specified low_delay_default,
     we *must* output something.
     this always occurs on the first call to decode() call
     when bframes are present in the bitstream. it may also
     occur if no vops  were seen in the bitstream

     if packed_mode is enabled, then we output the recently
     decoded frame (the very first ivop). otherwise we have
     nothing to display, and therefore output a black screen.
  */
  if (dec->low_delay_default && output == 0) {
    if (dec->packed_mode && seen_something) {
      decoder_output(dec, &dec->refn[0], dec->last_mbs, frame, stats, dec->last_coding_type, quant);
    } else {
      image_clear(&dec->cur, dec->width, dec->height, dec->edged_width, 0, 128, 128);
      decoder_output(dec, &dec->cur, NULL, frame, stats, P_VOP, quant);
      if (stats) stats->type = XVID_TYPE_NOTHING;
    }
  }

  emms();
  stop_global_timer();

  return (BitstreamPos(&bs)+7)/8; /* number of bytes consumed */
}

No admin address has been configured
ViewVC Help
Powered by ViewVC 1.0.4