view dxa.c @ 11032:01bd040f8607 libavcodec

Unroll main loop so the edge==0 case is seperate. This allows many things to be simplified away. h264 decoder is overall 1% faster with a mbaff sample and 0.1% slower with the cathedral sample, probably because the slow loop filter code must be loaded into the code cache for each first MB of each row but isnt used for the following MBs.

/*
 * Feeble Files/ScummVM DXA decoder
 * Copyright (c) 2007 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 libavcodec/dxa.c
 * DXA Video decoder
 */

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

#include "libavutil/intreadwrite.h"
#include "avcodec.h"

#include <zlib.h>

/*
 * Decoder context
 */
typedef struct DxaDecContext {
    AVCodecContext *avctx;
    AVFrame pic, prev;

    int dsize;
    uint8_t *decomp_buf;
    uint32_t pal[256];
} DxaDecContext;

static const int shift1[6] = { 0, 8, 8, 8, 4, 4 };
static const int shift2[6] = { 0, 0, 8, 4, 0, 4 };

static int decode_13(AVCodecContext *avctx, DxaDecContext *c, uint8_t* dst, uint8_t *src, uint8_t *ref)
{
    uint8_t *code, *data, *mv, *msk, *tmp, *tmp2;
    int i, j, k;
    int type, x, y, d, d2;
    int stride = c->pic.linesize[0];
    uint32_t mask;

    code = src  + 12;
    data = code + ((avctx->width * avctx->height) >> 4);
    mv   = data + AV_RB32(src + 0);
    msk  = mv   + AV_RB32(src + 4);

    for(j = 0; j < avctx->height; j += 4){
        for(i = 0; i < avctx->width; i += 4){
            tmp  = dst + i;
            tmp2 = ref + i;
            type = *code++;
            switch(type){
            case 4: // motion compensation
                x = (*mv) >> 4;    if(x & 8) x = 8 - x;
                y = (*mv++) & 0xF; if(y & 8) y = 8 - y;
                tmp2 += x + y*stride;
            case 0: // skip
            case 5: // skip in method 12
                for(y = 0; y < 4; y++){
                    memcpy(tmp, tmp2, 4);
                    tmp  += stride;
                    tmp2 += stride;
                }
                break;
            case 1:  // masked change
            case 10: // masked change with only half of pixels changed
            case 11: // cases 10-15 are for method 12 only
            case 12:
            case 13:
            case 14:
            case 15:
                if(type == 1){
                    mask = AV_RB16(msk);
                    msk += 2;
                }else{
                    type -= 10;
                    mask = ((msk[0] & 0xF0) << shift1[type]) | ((msk[0] & 0xF) << shift2[type]);
                    msk++;
                }
                for(y = 0; y < 4; y++){
                    for(x = 0; x < 4; x++){
                        tmp[x] = (mask & 0x8000) ? *data++ : tmp2[x];
                        mask <<= 1;
                    }
                    tmp  += stride;
                    tmp2 += stride;
                }
                break;
            case 2: // fill block
                for(y = 0; y < 4; y++){
                    memset(tmp, data[0], 4);
                    tmp += stride;
                }
                data++;
                break;
            case 3: // raw block
                for(y = 0; y < 4; y++){
                    memcpy(tmp, data, 4);
                    data += 4;
                    tmp  += stride;
                }
                break;
            case 8: // subblocks - method 13 only
                mask = *msk++;
                for(k = 0; k < 4; k++){
                    d  = ((k & 1) << 1) + ((k & 2) * stride);
                    d2 = ((k & 1) << 1) + ((k & 2) * stride);
                    tmp2 = ref + i + d2;
                    switch(mask & 0xC0){
                    case 0x80: // motion compensation
                        x = (*mv) >> 4;    if(x & 8) x = 8 - x;
                        y = (*mv++) & 0xF; if(y & 8) y = 8 - y;
                        tmp2 += x + y*stride;
                    case 0x00: // skip
                        tmp[d + 0         ] = tmp2[0];
                        tmp[d + 1         ] = tmp2[1];
                        tmp[d + 0 + stride] = tmp2[0 + stride];
                        tmp[d + 1 + stride] = tmp2[1 + stride];
                        break;
                    case 0x40: // fill
                        tmp[d + 0         ] = data[0];
                        tmp[d + 1         ] = data[0];
                        tmp[d + 0 + stride] = data[0];
                        tmp[d + 1 + stride] = data[0];
                        data++;
                        break;
                    case 0xC0: // raw
                        tmp[d + 0         ] = *data++;
                        tmp[d + 1         ] = *data++;
                        tmp[d + 0 + stride] = *data++;
                        tmp[d + 1 + stride] = *data++;
                        break;
                    }
                    mask <<= 2;
                }
                break;
            case 32: // vector quantization - 2 colors
                mask = AV_RB16(msk);
                msk += 2;
                for(y = 0; y < 4; y++){
                    for(x = 0; x < 4; x++){
                        tmp[x] = data[mask & 1];
                        mask >>= 1;
                    }
                    tmp  += stride;
                    tmp2 += stride;
                }
                data += 2;
                break;
            case 33: // vector quantization - 3 or 4 colors
            case 34:
                mask = AV_RB32(msk);
                msk += 4;
                for(y = 0; y < 4; y++){
                    for(x = 0; x < 4; x++){
                        tmp[x] = data[mask & 3];
                        mask >>= 2;
                    }
                    tmp  += stride;
                    tmp2 += stride;
                }
                data += type - 30;
                break;
            default:
                av_log(avctx, AV_LOG_ERROR, "Unknown opcode %d\n", type);
                return -1;
            }
        }
        dst += stride * 4;
        ref += stride * 4;
    }
    return 0;
}

static int decode_frame(AVCodecContext *avctx, void *data, int *data_size, AVPacket *avpkt)
{
    const uint8_t *buf = avpkt->data;
    int buf_size = avpkt->size;
    DxaDecContext * const c = avctx->priv_data;
    uint8_t *outptr, *srcptr, *tmpptr;
    unsigned long dsize;
    int i, j, compr;
    int stride;
    int orig_buf_size = buf_size;
    int pc = 0;

    /* make the palette available on the way out */
    if(buf[0]=='C' && buf[1]=='M' && buf[2]=='A' && buf[3]=='P'){
        int r, g, b;

        buf += 4;
        for(i = 0; i < 256; i++){
            r = *buf++;
            g = *buf++;
            b = *buf++;
            c->pal[i] = (r << 16) | (g << 8) | b;
        }
        pc = 1;
        buf_size -= 768+4;
    }

    if(avctx->get_buffer(avctx, &c->pic) < 0){
        av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
        return -1;
    }
    memcpy(c->pic.data[1], c->pal, AVPALETTE_SIZE);
    c->pic.palette_has_changed = pc;

    outptr = c->pic.data[0];
    srcptr = c->decomp_buf;
    tmpptr = c->prev.data[0];
    stride = c->pic.linesize[0];

    if(buf[0]=='N' && buf[1]=='U' && buf[2]=='L' && buf[3]=='L')
        compr = -1;
    else
        compr = buf[4];

    dsize = c->dsize;
    if((compr != 4 && compr != -1) && uncompress(c->decomp_buf, &dsize, buf + 9, buf_size - 9) != Z_OK){
        av_log(avctx, AV_LOG_ERROR, "Uncompress failed!\n");
        return -1;
    }
    switch(compr){
    case -1:
        c->pic.key_frame = 0;
        c->pic.pict_type = FF_P_TYPE;
        if(c->prev.data[0])
            memcpy(c->pic.data[0], c->prev.data[0], c->pic.linesize[0] * avctx->height);
        else{ // Should happen only when first frame is 'NULL'
            memset(c->pic.data[0], 0, c->pic.linesize[0] * avctx->height);
            c->pic.key_frame = 1;
            c->pic.pict_type = FF_I_TYPE;
        }
        break;
    case 2:
    case 3:
    case 4:
    case 5:
        c->pic.key_frame = !(compr & 1);
        c->pic.pict_type = (compr & 1) ? FF_P_TYPE : FF_I_TYPE;
        for(j = 0; j < avctx->height; j++){
            if(compr & 1){
                for(i = 0; i < avctx->width; i++)
                    outptr[i] = srcptr[i] ^ tmpptr[i];
                tmpptr += stride;
            }else
                memcpy(outptr, srcptr, avctx->width);
            outptr += stride;
            srcptr += avctx->width;
        }
        break;
    case 12: // ScummVM coding
    case 13:
        c->pic.key_frame = 0;
        c->pic.pict_type = FF_P_TYPE;
        decode_13(avctx, c, c->pic.data[0], srcptr, c->prev.data[0]);
        break;
    default:
        av_log(avctx, AV_LOG_ERROR, "Unknown/unsupported compression type %d\n", buf[4]);
        return -1;
    }

    FFSWAP(AVFrame, c->pic, c->prev);
    if(c->pic.data[0])
        avctx->release_buffer(avctx, &c->pic);

    *data_size = sizeof(AVFrame);
    *(AVFrame*)data = c->prev;

    /* always report that the buffer was completely consumed */
    return orig_buf_size;
}

static av_cold int decode_init(AVCodecContext *avctx)
{
    DxaDecContext * const c = avctx->priv_data;

    c->avctx = avctx;
    avctx->pix_fmt = PIX_FMT_PAL8;

    c->dsize = avctx->width * avctx->height * 2;
    if((c->decomp_buf = av_malloc(c->dsize)) == NULL) {
        av_log(avctx, AV_LOG_ERROR, "Can't allocate decompression buffer.\n");
        return -1;
    }

    return 0;
}

static av_cold int decode_end(AVCodecContext *avctx)
{
    DxaDecContext * const c = avctx->priv_data;

    av_freep(&c->decomp_buf);
    if(c->prev.data[0])
        avctx->release_buffer(avctx, &c->prev);
    if(c->pic.data[0])
        avctx->release_buffer(avctx, &c->pic);

    return 0;
}

AVCodec dxa_decoder = {
    "dxa",
    CODEC_TYPE_VIDEO,
    CODEC_ID_DXA,
    sizeof(DxaDecContext),
    decode_init,
    NULL,
    decode_end,
    decode_frame,
    CODEC_CAP_DR1,
    .long_name = NULL_IF_CONFIG_SMALL("Feeble Files/ScummVM DXA"),
};