view ljpegenc.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.

/*
 * lossless JPEG encoder
 * Copyright (c) 2000, 2001 Fabrice Bellard
 * Copyright (c) 2003 Alex Beregszaszi
 * Copyright (c) 2003-2004 Michael Niedermayer
 *
 * Support for external huffman table, various fixes (AVID workaround),
 * aspecting, new decode_frame mechanism and apple mjpeg-b support
 *                                  by Alex Beregszaszi
 *
 * 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/ljpegenc.c
 * lossless JPEG encoder.
 */

#include "avcodec.h"
#include "dsputil.h"
#include "mpegvideo.h"
#include "mjpeg.h"
#include "mjpegenc.h"


static int encode_picture_lossless(AVCodecContext *avctx, unsigned char *buf, int buf_size, void *data){
    MpegEncContext * const s = avctx->priv_data;
    MJpegContext * const m = s->mjpeg_ctx;
    AVFrame *pict = data;
    const int width= s->width;
    const int height= s->height;
    AVFrame * const p= (AVFrame*)&s->current_picture;
    const int predictor= avctx->prediction_method+1;

    init_put_bits(&s->pb, buf, buf_size);

    *p = *pict;
    p->pict_type= FF_I_TYPE;
    p->key_frame= 1;

    ff_mjpeg_encode_picture_header(s);

    s->header_bits= put_bits_count(&s->pb);

    if(avctx->pix_fmt == PIX_FMT_BGRA){
        int x, y, i;
        const int linesize= p->linesize[0];
        uint16_t (*buffer)[4]= (void *) s->rd_scratchpad;
        int left[3], top[3], topleft[3];

        for(i=0; i<3; i++){
            buffer[0][i]= 1 << (9 - 1);
        }

        for(y = 0; y < height; y++) {
            const int modified_predictor= y ? predictor : 1;
            uint8_t *ptr = p->data[0] + (linesize * y);

            if(s->pb.buf_end - s->pb.buf - (put_bits_count(&s->pb)>>3) < width*3*4){
                av_log(s->avctx, AV_LOG_ERROR, "encoded frame too large\n");
                return -1;
            }

            for(i=0; i<3; i++){
                top[i]= left[i]= topleft[i]= buffer[0][i];
            }
            for(x = 0; x < width; x++) {
                buffer[x][1] = ptr[4*x+0] - ptr[4*x+1] + 0x100;
                buffer[x][2] = ptr[4*x+2] - ptr[4*x+1] + 0x100;
                buffer[x][0] = (ptr[4*x+0] + 2*ptr[4*x+1] + ptr[4*x+2])>>2;

                for(i=0;i<3;i++) {
                    int pred, diff;

                    PREDICT(pred, topleft[i], top[i], left[i], modified_predictor);

                    topleft[i]= top[i];
                    top[i]= buffer[x+1][i];

                    left[i]= buffer[x][i];

                    diff= ((left[i] - pred + 0x100)&0x1FF) - 0x100;

                    if(i==0)
                        ff_mjpeg_encode_dc(s, diff, m->huff_size_dc_luminance, m->huff_code_dc_luminance); //FIXME ugly
                    else
                        ff_mjpeg_encode_dc(s, diff, m->huff_size_dc_chrominance, m->huff_code_dc_chrominance);
                }
            }
        }
    }else{
        int mb_x, mb_y, i;
        const int mb_width  = (width  + s->mjpeg_hsample[0] - 1) / s->mjpeg_hsample[0];
        const int mb_height = (height + s->mjpeg_vsample[0] - 1) / s->mjpeg_vsample[0];

        for(mb_y = 0; mb_y < mb_height; mb_y++) {
            if(s->pb.buf_end - s->pb.buf - (put_bits_count(&s->pb)>>3) < mb_width * 4 * 3 * s->mjpeg_hsample[0] * s->mjpeg_vsample[0]){
                av_log(s->avctx, AV_LOG_ERROR, "encoded frame too large\n");
                return -1;
            }
            for(mb_x = 0; mb_x < mb_width; mb_x++) {
                if(mb_x==0 || mb_y==0){
                    for(i=0;i<3;i++) {
                        uint8_t *ptr;
                        int x, y, h, v, linesize;
                        h = s->mjpeg_hsample[i];
                        v = s->mjpeg_vsample[i];
                        linesize= p->linesize[i];

                        for(y=0; y<v; y++){
                            for(x=0; x<h; x++){
                                int pred;

                                ptr = p->data[i] + (linesize * (v * mb_y + y)) + (h * mb_x + x); //FIXME optimize this crap
                                if(y==0 && mb_y==0){
                                    if(x==0 && mb_x==0){
                                        pred= 128;
                                    }else{
                                        pred= ptr[-1];
                                    }
                                }else{
                                    if(x==0 && mb_x==0){
                                        pred= ptr[-linesize];
                                    }else{
                                        PREDICT(pred, ptr[-linesize-1], ptr[-linesize], ptr[-1], predictor);
                                    }
                                }

                                if(i==0)
                                    ff_mjpeg_encode_dc(s, *ptr - pred, m->huff_size_dc_luminance, m->huff_code_dc_luminance); //FIXME ugly
                                else
                                    ff_mjpeg_encode_dc(s, *ptr - pred, m->huff_size_dc_chrominance, m->huff_code_dc_chrominance);
                            }
                        }
                    }
                }else{
                    for(i=0;i<3;i++) {
                        uint8_t *ptr;
                        int x, y, h, v, linesize;
                        h = s->mjpeg_hsample[i];
                        v = s->mjpeg_vsample[i];
                        linesize= p->linesize[i];

                        for(y=0; y<v; y++){
                            for(x=0; x<h; x++){
                                int pred;

                                ptr = p->data[i] + (linesize * (v * mb_y + y)) + (h * mb_x + x); //FIXME optimize this crap
//printf("%d %d %d %d %8X\n", mb_x, mb_y, x, y, ptr);
                                PREDICT(pred, ptr[-linesize-1], ptr[-linesize], ptr[-1], predictor);

                                if(i==0)
                                    ff_mjpeg_encode_dc(s, *ptr - pred, m->huff_size_dc_luminance, m->huff_code_dc_luminance); //FIXME ugly
                                else
                                    ff_mjpeg_encode_dc(s, *ptr - pred, m->huff_size_dc_chrominance, m->huff_code_dc_chrominance);
                            }
                        }
                    }
                }
            }
        }
    }

    emms_c();

    ff_mjpeg_encode_picture_trailer(s);
    s->picture_number++;

    flush_put_bits(&s->pb);
    return put_bits_ptr(&s->pb) - s->pb.buf;
//    return (put_bits_count(&f->pb)+7)/8;
}


AVCodec ljpeg_encoder = { //FIXME avoid MPV_* lossless JPEG should not need them
    "ljpeg",
    CODEC_TYPE_VIDEO,
    CODEC_ID_LJPEG,
    sizeof(MpegEncContext),
    MPV_encode_init,
    encode_picture_lossless,
    MPV_encode_end,
    .long_name = NULL_IF_CONFIG_SMALL("Lossless JPEG"),
};