annotate ulti.c @ 4980:800a543a2513 libavcodec

tweak mmx2 sad. 40% faster on core2, 18% faster on k8, 5% faster on p4.

/*
 * IBM Ultimotion Video Decoder
 * Copyright (C) 2004 Konstantin Shishkov
 *
 * 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
 *
 */

/**
 * @file ulti.c
 * IBM Ultimotion Video Decoder.
 */

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

#include "avcodec.h"

#include "ulti_cb.h"

typedef struct UltimotionDecodeContext {
    AVCodecContext *avctx;
    int width, height, blocks;
    AVFrame frame;
    const uint8_t *ulti_codebook;
} UltimotionDecodeContext;

static int ulti_decode_init(AVCodecContext *avctx)
{
    UltimotionDecodeContext *s = avctx->priv_data;

    s->avctx = avctx;
    s->width = avctx->width;
    s->height = avctx->height;
    s->blocks = (s->width / 8) * (s->height / 8);
    avctx->pix_fmt = PIX_FMT_YUV410P;
    avctx->coded_frame = (AVFrame*) &s->frame;
    s->ulti_codebook = ulti_codebook;

    return 0;
}

static int block_coords[8] = // 4x4 block coords in 8x8 superblock
    { 0, 0, 0, 4, 4, 4, 4, 0};

static int angle_by_index[4] = { 0, 2, 6, 12};

/* Lookup tables for luma and chroma - used by ulti_convert_yuv() */
static uint8_t ulti_lumas[64] =
    { 0x10, 0x13, 0x17, 0x1A, 0x1E, 0x21, 0x25, 0x28,
      0x2C, 0x2F, 0x33, 0x36, 0x3A, 0x3D, 0x41, 0x44,
      0x48, 0x4B, 0x4F, 0x52, 0x56, 0x59, 0x5C, 0x60,
      0x63, 0x67, 0x6A, 0x6E, 0x71, 0x75, 0x78, 0x7C,
      0x7F, 0x83, 0x86, 0x8A, 0x8D, 0x91, 0x94, 0x98,
      0x9B, 0x9F, 0xA2, 0xA5, 0xA9, 0xAC, 0xB0, 0xB3,
      0xB7, 0xBA, 0xBE, 0xC1, 0xC5, 0xC8, 0xCC, 0xCF,
      0xD3, 0xD6, 0xDA, 0xDD, 0xE1, 0xE4, 0xE8, 0xEB};

static uint8_t ulti_chromas[16] =
    { 0x60, 0x67, 0x6D, 0x73, 0x7A, 0x80, 0x86, 0x8D,
      0x93, 0x99, 0xA0, 0xA6, 0xAC, 0xB3, 0xB9, 0xC0};

/* convert Ultimotion YUV block (sixteen 6-bit Y samples and
 two 4-bit chroma samples) into standard YUV and put it into frame */
static void ulti_convert_yuv(AVFrame *frame, int x, int y,
                             uint8_t *luma,int chroma)
{
    uint8_t *y_plane, *cr_plane, *cb_plane;
    int i;

    y_plane = frame->data[0] + x + y * frame->linesize[0];
    cr_plane = frame->data[1] + (x / 4) + (y / 4) * frame->linesize[1];
    cb_plane = frame->data[2] + (x / 4) + (y / 4) * frame->linesize[2];

    cr_plane[0] = ulti_chromas[chroma >> 4];

    cb_plane[0] = ulti_chromas[chroma & 0xF];


    for(i = 0; i < 16; i++){
        y_plane[i & 3] = ulti_lumas[luma[i]];
        if((i & 3) == 3) { //next row
            y_plane += frame->linesize[0];
        }
    }
}

/* generate block like in MS Video1 */
static void ulti_pattern(AVFrame *frame, int x, int y,
                         int f0, int f1, int Y0, int Y1, int chroma)
{
    uint8_t Luma[16];
    int mask, i;
    for(mask = 0x80, i = 0; mask; mask >>= 1, i++) {
        if(f0 & mask)
            Luma[i] = Y1;
        else
            Luma[i] = Y0;
    }

    for(mask = 0x80, i = 8; mask; mask >>= 1, i++) {
        if(f1 & mask)
            Luma[i] = Y1;
        else
            Luma[i] = Y0;
    }

    ulti_convert_yuv(frame, x, y, Luma, chroma);
}

/* fill block with some gradient */
static void ulti_grad(AVFrame *frame, int x, int y, uint8_t *Y, int chroma, int angle)
{
    uint8_t Luma[16];
    if(angle & 8) { //reverse order
        int t;
        angle &= 0x7;
        t = Y[0];
        Y[0] = Y[3];
        Y[3] = t;
        t = Y[1];
        Y[1] = Y[2];
        Y[2] = t;
    }
    switch(angle){
    case 0:
        Luma[0]  = Y[0]; Luma[1]  = Y[1]; Luma[2]  = Y[2]; Luma[3]  = Y[3];
        Luma[4]  = Y[0]; Luma[5]  = Y[1]; Luma[6]  = Y[2]; Luma[7]  = Y[3];
        Luma[8]  = Y[0]; Luma[9]  = Y[1]; Luma[10] = Y[2]; Luma[11] = Y[3];
        Luma[12] = Y[0]; Luma[13] = Y[1]; Luma[14] = Y[2]; Luma[15] = Y[3];
        break;
    case 1:
        Luma[0]  = Y[1]; Luma[1]  = Y[2]; Luma[2]  = Y[3]; Luma[3]  = Y[3];
        Luma[4]  = Y[0]; Luma[5]  = Y[1]; Luma[6]  = Y[2]; Luma[7]  = Y[3];
        Luma[8]  = Y[0]; Luma[9]  = Y[1]; Luma[10] = Y[2]; Luma[11] = Y[3];
        Luma[12] = Y[0]; Luma[13] = Y[0]; Luma[14] = Y[1]; Luma[15] = Y[2];
        break;
    case 2:
        Luma[0]  = Y[1]; Luma[1]  = Y[2]; Luma[2]  = Y[3]; Luma[3]  = Y[3];
        Luma[4]  = Y[1]; Luma[5]  = Y[2]; Luma[6]  = Y[2]; Luma[7]  = Y[3];
        Luma[8]  = Y[0]; Luma[9]  = Y[1]; Luma[10] = Y[1]; Luma[11] = Y[2];
        Luma[12] = Y[0]; Luma[13] = Y[0]; Luma[14] = Y[1]; Luma[15] = Y[2];
        break;
    case 3:
        Luma[0]  = Y[2]; Luma[1]  = Y[3]; Luma[2]  = Y[3]; Luma[3]  = Y[3];
        Luma[4]  = Y[1]; Luma[5]  = Y[2]; Luma[6]  = Y[2]; Luma[7]  = Y[3];
        Luma[8]  = Y[0]; Luma[9]  = Y[1]; Luma[10] = Y[1]; Luma[11] = Y[2];
        Luma[12] = Y[0]; Luma[13] = Y[0]; Luma[14] = Y[0]; Luma[15] = Y[1];
        break;
    case 4:
        Luma[0]  = Y[3]; Luma[1]  = Y[3]; Luma[2]  = Y[3]; Luma[3]  = Y[3];
        Luma[4]  = Y[2]; Luma[5]  = Y[2]; Luma[6]  = Y[2]; Luma[7]  = Y[2];
        Luma[8]  = Y[1]; Luma[9]  = Y[1]; Luma[10] = Y[1]; Luma[11] = Y[1];
        Luma[12] = Y[0]; Luma[13] = Y[0]; Luma[14] = Y[0]; Luma[15] = Y[0];
        break;
    case 5:
        Luma[0]  = Y[3]; Luma[1]  = Y[3]; Luma[2]  = Y[3]; Luma[3]  = Y[2];
        Luma[4]  = Y[3]; Luma[5]  = Y[2]; Luma[6]  = Y[2]; Luma[7]  = Y[1];
        Luma[8]  = Y[2]; Luma[9]  = Y[1]; Luma[10] = Y[1]; Luma[11] = Y[0];
        Luma[12] = Y[1]; Luma[13] = Y[0]; Luma[14] = Y[0]; Luma[15] = Y[0];
        break;
    case 6:
        Luma[0]  = Y[3]; Luma[1]  = Y[3]; Luma[2]  = Y[2]; Luma[3]  = Y[2];
        Luma[4]  = Y[3]; Luma[5]  = Y[2]; Luma[6]  = Y[1]; Luma[7]  = Y[1];
        Luma[8]  = Y[2]; Luma[9]  = Y[2]; Luma[10] = Y[1]; Luma[11] = Y[0];
        Luma[12] = Y[1]; Luma[13] = Y[1]; Luma[14] = Y[0]; Luma[15] = Y[0];
        break;
    case 7:
        Luma[0]  = Y[3]; Luma[1]  = Y[3]; Luma[2]  = Y[2]; Luma[3]  = Y[1];
        Luma[4]  = Y[3]; Luma[5]  = Y[2]; Luma[6]  = Y[1]; Luma[7]  = Y[0];
        Luma[8]  = Y[3]; Luma[9]  = Y[2]; Luma[10] = Y[1]; Luma[11] = Y[0];
        Luma[12] = Y[2]; Luma[13] = Y[1]; Luma[14] = Y[0]; Luma[15] = Y[0];
        break;
    default:
        Luma[0]  = Y[0]; Luma[1]  = Y[0]; Luma[2]  = Y[1]; Luma[3]  = Y[1];
        Luma[4]  = Y[0]; Luma[5]  = Y[0]; Luma[6]  = Y[1]; Luma[7]  = Y[1];
        Luma[8]  = Y[2]; Luma[9]  = Y[2]; Luma[10] = Y[3]; Luma[11] = Y[3];
        Luma[12] = Y[2]; Luma[13] = Y[2]; Luma[14] = Y[3]; Luma[15] = Y[3];
        break;
    }

    ulti_convert_yuv(frame, x, y, Luma, chroma);
}

static int ulti_decode_frame(AVCodecContext *avctx,
                             void *data, int *data_size,
                             uint8_t *buf, int buf_size)
{
    UltimotionDecodeContext *s=avctx->priv_data;
    int modifier = 0;
    int uniq = 0;
    int mode = 0;
    int blocks = 0;
    int done = 0;
    int x = 0, y = 0;
    int i;
    int skip;
    int tmp;

    if(s->frame.data[0])
        avctx->release_buffer(avctx, &s->frame);

    s->frame.reference = 1;
    s->frame.buffer_hints = FF_BUFFER_HINTS_VALID | FF_BUFFER_HINTS_PRESERVE | FF_BUFFER_HINTS_REUSABLE;
    if(avctx->get_buffer(avctx, &s->frame) < 0) {
        av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
        return -1;
    }

    while(!done) {
        int idx;
        if(blocks >= s->blocks || y >= s->height)
            break;//all blocks decoded

        idx = *buf++;
        if((idx & 0xF8) == 0x70) {
            switch(idx) {
            case 0x70: //change modifier
                modifier = *buf++;
                if(modifier>1)
                    av_log(avctx, AV_LOG_INFO, "warning: modifier must be 0 or 1, got %i\n", modifier);
                break;
            case 0x71: // set uniq flag
                uniq = 1;
                break;
            case 0x72: //toggle mode
                mode = !mode;
                break;
            case 0x73: //end-of-frame
                done = 1;
                break;
            case 0x74: //skip some blocks
                skip = *buf++;
                if ((blocks + skip) >= s->blocks)
                    break;
                blocks += skip;
                x += skip * 8;
                while(x >= s->width) {
                    x -= s->width;
                    y += 8;
                }
                break;
            default:
                av_log(avctx, AV_LOG_INFO, "warning: unknown escape 0x%02X\n", idx);
            }
        } else { //handle one block
            int code;
            int cf;
            int angle = 0;
            uint8_t Y[4]; // luma samples of block
            int tx = 0, ty = 0; //coords of subblock
            int chroma = 0;
            if (mode || uniq) {
                uniq = 0;
                cf = 1;
                chroma = 0;
            } else {
                cf = 0;
                if (idx)
                    chroma = *buf++;
            }
            for (i = 0; i < 4; i++) { // for every subblock
                code = (idx >> (6 - i*2)) & 3; //extract 2 bits
                if(!code) //skip subblock
                    continue;
                if(cf)
                    chroma = *buf++;
                tx = x + block_coords[i * 2];
                ty = y + block_coords[(i * 2) + 1];
                switch(code) {
                case 1:
                    tmp = *buf++;

                    angle = angle_by_index[(tmp >> 6) & 0x3];

                    Y[0] = tmp & 0x3F;
                    Y[1] = Y[0];

                    if (angle) {
                        Y[2] = Y[0]+1;
                        if (Y[2] > 0x3F)
                            Y[2] = 0x3F;
                        Y[3] = Y[2];
                    } else {
                        Y[2] = Y[0];
                        Y[3] = Y[0];
                    }
                    break;

                case 2:
                    if (modifier) { // unpack four luma samples
                        tmp = (*buf++) << 16;
                        tmp += (*buf++) << 8;
                        tmp += *buf++;

                        Y[0] = (tmp >> 18) & 0x3F;
                        Y[1] = (tmp >> 12) & 0x3F;
                        Y[2] = (tmp >> 6) & 0x3F;
                        Y[3] = tmp & 0x3F;
                        angle = 16;
                    } else { // retrieve luma samples from codebook
                        tmp = (*buf++) << 8;
                        tmp += (*buf++);

                        angle = (tmp >> 12) & 0xF;
                        tmp &= 0xFFF;
                        tmp <<= 2;
                        Y[0] = s->ulti_codebook[tmp];
                        Y[1] = s->ulti_codebook[tmp + 1];
                        Y[2] = s->ulti_codebook[tmp + 2];
                        Y[3] = s->ulti_codebook[tmp + 3];
                    }
                    break;

                case 3:
                    if (modifier) { // all 16 luma samples
                        uint8_t Luma[16];

                        tmp = (*buf++) << 16;
                        tmp += (*buf++) << 8;
                        tmp += *buf++;
                        Luma[0] = (tmp >> 18) & 0x3F;
                        Luma[1] = (tmp >> 12) & 0x3F;
                        Luma[2] = (tmp >> 6) & 0x3F;
                        Luma[3] = tmp & 0x3F;

                        tmp = (*buf++) << 16;
                        tmp += (*buf++) << 8;
                        tmp += *buf++;
                        Luma[4] = (tmp >> 18) & 0x3F;
                        Luma[5] = (tmp >> 12) & 0x3F;
                        Luma[6] = (tmp >> 6) & 0x3F;
                        Luma[7] = tmp & 0x3F;

                        tmp = (*buf++) << 16;
                        tmp += (*buf++) << 8;
                        tmp += *buf++;
                        Luma[8] = (tmp >> 18) & 0x3F;
                        Luma[9] = (tmp >> 12) & 0x3F;
                        Luma[10] = (tmp >> 6) & 0x3F;
                        Luma[11] = tmp & 0x3F;

                        tmp = (*buf++) << 16;
                        tmp += (*buf++) << 8;
                        tmp += *buf++;
                        Luma[12] = (tmp >> 18) & 0x3F;
                        Luma[13] = (tmp >> 12) & 0x3F;
                        Luma[14] = (tmp >> 6) & 0x3F;
                        Luma[15] = tmp & 0x3F;

                        ulti_convert_yuv(&s->frame, tx, ty, Luma, chroma);
                    } else {
                        tmp = *buf++;
                        if(tmp & 0x80) {
                            angle = (tmp >> 4) & 0x7;
                            tmp = (tmp << 8) + *buf++;
                            Y[0] = (tmp >> 6) & 0x3F;
                            Y[1] = tmp & 0x3F;
                            Y[2] = (*buf++) & 0x3F;
                            Y[3] = (*buf++) & 0x3F;
                            ulti_grad(&s->frame, tx, ty, Y, chroma, angle); //draw block
                        } else { // some patterns
                            int f0, f1;
                            f0 = *buf++;
                            f1 = tmp;
                            Y[0] = (*buf++) & 0x3F;
                            Y[1] = (*buf++) & 0x3F;
                            ulti_pattern(&s->frame, tx, ty, f1, f0, Y[0], Y[1], chroma);
                        }
                    }
                    break;
                }
                if(code != 3)
                    ulti_grad(&s->frame, tx, ty, Y, chroma, angle); // draw block
            }
            blocks++;
                x += 8;
            if(x >= s->width) {
                x = 0;
                y += 8;
            }
        }
    }

    *data_size=sizeof(AVFrame);
    *(AVFrame*)data= s->frame;

    return buf_size;
}

static int ulti_decode_end(AVCodecContext *avctx)
{
/*    UltimotionDecodeContext *s = avctx->priv_data;*/

    return 0;
}

AVCodec ulti_decoder = {
    "ultimotion",
    CODEC_TYPE_VIDEO,
    CODEC_ID_ULTI,
    sizeof(UltimotionDecodeContext),
    ulti_decode_init,
    NULL,
    ulti_decode_end,
    ulti_decode_frame,
    CODEC_CAP_DR1,
    NULL
};