libavcodec/pngenc.c

   1 /*
   2  * PNG image format
   3  * Copyright (c) 2003 Fabrice Bellard
   4  *
   5  * This file is part of Libav.
   6  *
   7  * Libav is free software; you can redistribute it and/or
   8  * modify it under the terms of the GNU Lesser General Public
   9  * License as published by the Free Software Foundation; either
  10  * version 2.1 of the License, or (at your option) any later version.
  11  *
  12  * Libav is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15  * Lesser General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Lesser General Public
  18  * License along with Libav; if not, write to the Free Software
  19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  20  */
  21 #include "avcodec.h"
  22 #include "bytestream.h"
  23 #include "dsputil.h"
  24 #include "png.h"
  25
  26 /* TODO:
  27  * - add 2, 4 and 16 bit depth support
  28  */
  29
  30 #include <zlib.h>
  31
  32 //#define DEBUG
  33
  34 #define IOBUF_SIZE 4096
  35
  36 typedef struct PNGEncContext {
  37     DSPContext dsp;
  38
  39     uint8_t *bytestream;
  40     uint8_t *bytestream_start;
  41     uint8_t *bytestream_end;
  42     AVFrame picture;
  43
  44     int filter_type;
  45
  46     z_stream zstream;
  47     uint8_t buf[IOBUF_SIZE];
  48 } PNGEncContext;
  49
  50 static void png_get_interlaced_row(uint8_t *dst, int row_size,
  51                                    int bits_per_pixel, int pass,
  52                                    const uint8_t *src, int width)
  53 {
  54     int x, mask, dst_x, j, b, bpp;
  55     uint8_t *d;
  56     const uint8_t *s;
  57
  58     mask = ff_png_pass_mask[pass];
  59     switch(bits_per_pixel) {
  60     case 1:
  61         memset(dst, 0, row_size);
  62         dst_x = 0;
  63         for(x = 0; x < width; x++) {
  64             j = (x & 7);
  65             if ((mask << j) & 0x80) {
  66                 b = (src[x >> 3] >> (7 - j)) & 1;
  67                 dst[dst_x >> 3] |= b << (7 - (dst_x & 7));
  68                 dst_x++;
  69             }
  70         }
  71         break;
  72     default:
  73         bpp = bits_per_pixel >> 3;
  74         d = dst;
  75         s = src;
  76         for(x = 0; x < width; x++) {
  77             j = x & 7;
  78             if ((mask << j) & 0x80) {
  79                 memcpy(d, s, bpp);
  80                 d += bpp;
  81             }
  82             s += bpp;
  83         }
  84         break;
  85     }
  86 }
  87
  88 static void sub_png_paeth_prediction(uint8_t *dst, uint8_t *src, uint8_t *top, int w, int bpp)
  89 {
  90     int i;
  91     for(i = 0; i < w; i++) {
  92         int a, b, c, p, pa, pb, pc;
  93
  94         a = src[i - bpp];
  95         b = top[i];
  96         c = top[i - bpp];
  97
  98         p = b - c;
  99         pc = a - c;
 100
 101         pa = abs(p);
 102         pb = abs(pc);
 103         pc = abs(p + pc);
 104
 105         if (pa <= pb && pa <= pc)
 106             p = a;
 107         else if (pb <= pc)
 108             p = b;
 109         else
 110             p = c;
 111         dst[i] = src[i] - p;
 112     }
 113 }
 114
 115 static void png_filter_row(DSPContext *dsp, uint8_t *dst, int filter_type,
 116                            uint8_t *src, uint8_t *top, int size, int bpp)
 117 {
 118     int i;
 119
 120     switch(filter_type) {
 121     case PNG_FILTER_VALUE_NONE:
 122         memcpy(dst, src, size);
 123         break;
 124     case PNG_FILTER_VALUE_SUB:
 125         dsp->diff_bytes(dst, src, src-bpp, size);
 126         memcpy(dst, src, bpp);
 127         break;
 128     case PNG_FILTER_VALUE_UP:
 129         dsp->diff_bytes(dst, src, top, size);
 130         break;
 131     case PNG_FILTER_VALUE_AVG:
 132         for(i = 0; i < bpp; i++)
 133             dst[i] = src[i] - (top[i] >> 1);
 134         for(; i < size; i++)
 135             dst[i] = src[i] - ((src[i-bpp] + top[i]) >> 1);
 136         break;
 137     case PNG_FILTER_VALUE_PAETH:
 138         for(i = 0; i < bpp; i++)
 139             dst[i] = src[i] - top[i];
 140         sub_png_paeth_prediction(dst+i, src+i, top+i, size-i, bpp);
 141         break;
 142     }
 143 }
 144
 145 static uint8_t *png_choose_filter(PNGEncContext *s, uint8_t *dst,
 146                                   uint8_t *src, uint8_t *top, int size, int bpp)
 147 {
 148     int pred = s->filter_type;
 149     assert(bpp || !pred);
 150     if(!top && pred)
 151         pred = PNG_FILTER_VALUE_SUB;
 152     if(pred == PNG_FILTER_VALUE_MIXED) {
 153         int i;
 154         int cost, bcost = INT_MAX;
 155         uint8_t *buf1 = dst, *buf2 = dst + size + 16;
 156         for(pred=0; pred<5; pred++) {
 157             png_filter_row(&s->dsp, buf1+1, pred, src, top, size, bpp);
 158             buf1[0] = pred;
 159             cost = 0;
 160             for(i=0; i<=size; i++)
 161                 cost += abs((int8_t)buf1[i]);
 162             if(cost < bcost) {
 163                 bcost = cost;
 164                 FFSWAP(uint8_t*, buf1, buf2);
 165             }
 166         }
 167         return buf2;
 168     } else {
 169         png_filter_row(&s->dsp, dst+1, pred, src, top, size, bpp);
 170         dst[0] = pred;
 171         return dst;
 172     }
 173 }
 174
 175 static void convert_from_rgb32(uint8_t *dst, const uint8_t *src, int width)
 176 {
 177     uint8_t *d;
 178     int j;
 179     unsigned int v;
 180
 181     d = dst;
 182     for(j = 0; j < width; j++) {
 183         v = ((const uint32_t *)src)[j];
 184         d[0] = v >> 16;
 185         d[1] = v >> 8;
 186         d[2] = v;
 187         d[3] = v >> 24;
 188         d += 4;
 189     }
 190 }
 191
 192 static void png_write_chunk(uint8_t **f, uint32_t tag,
 193                             const uint8_t *buf, int length)
 194 {
 195     uint32_t crc;
 196     uint8_t tagbuf[4];
 197
 198     bytestream_put_be32(f, length);
 199     crc = crc32(0, Z_NULL, 0);
 200     AV_WL32(tagbuf, tag);
 201     crc = crc32(crc, tagbuf, 4);
 202     bytestream_put_be32(f, av_bswap32(tag));
 203     if (length > 0) {
 204         crc = crc32(crc, buf, length);
 205         memcpy(*f, buf, length);
 206         *f += length;
 207     }
 208     bytestream_put_be32(f, crc);
 209 }
 210
 211 /* XXX: do filtering */
 212 static int png_write_row(PNGEncContext *s, const uint8_t *data, int size)
 213 {
 214     int ret;
 215
 216     s->zstream.avail_in = size;
 217     s->zstream.next_in = (uint8_t *)data;
 218     while (s->zstream.avail_in > 0) {
 219         ret = deflate(&s->zstream, Z_NO_FLUSH);
 220         if (ret != Z_OK)
 221             return -1;
 222         if (s->zstream.avail_out == 0) {
 223             if(s->bytestream_end - s->bytestream > IOBUF_SIZE + 100)
 224                 png_write_chunk(&s->bytestream, MKTAG('I', 'D', 'A', 'T'), s->buf, IOBUF_SIZE);
 225             s->zstream.avail_out = IOBUF_SIZE;
 226             s->zstream.next_out = s->buf;
 227         }
 228     }
 229     return 0;
 230 }
 231
 232 static int encode_frame(AVCodecContext *avctx, AVPacket *pkt,
 233                         const AVFrame *pict, int *got_packet)
 234 {
 235     PNGEncContext *s = avctx->priv_data;
 236     AVFrame * const p= &s->picture;
 237     int bit_depth, color_type, y, len, row_size, ret, is_progressive;
 238     int bits_per_pixel, pass_row_size, enc_row_size, max_packet_size;
 239     int compression_level;
 240     uint8_t *ptr, *top;
 241     uint8_t *crow_base = NULL, *crow_buf, *crow;
 242     uint8_t *progressive_buf = NULL;
 243     uint8_t *rgba_buf = NULL;
 244     uint8_t *top_buf = NULL;
 245
 246     *p = *pict;
 247     p->pict_type= AV_PICTURE_TYPE_I;
 248     p->key_frame= 1;
 249
 250     is_progressive = !!(avctx->flags & CODEC_FLAG_INTERLACED_DCT);
 251     switch(avctx->pix_fmt) {
 252     case AV_PIX_FMT_RGB32:
 253         bit_depth = 8;
 254         color_type = PNG_COLOR_TYPE_RGB_ALPHA;
 255         break;
 256     case AV_PIX_FMT_RGB24:
 257         bit_depth = 8;
 258         color_type = PNG_COLOR_TYPE_RGB;
 259         break;
 260     case AV_PIX_FMT_GRAY8:
 261         bit_depth = 8;
 262         color_type = PNG_COLOR_TYPE_GRAY;
 263         break;
 264     case AV_PIX_FMT_MONOBLACK:
 265         bit_depth = 1;
 266         color_type = PNG_COLOR_TYPE_GRAY;
 267         break;
 268     case AV_PIX_FMT_PAL8:
 269         bit_depth = 8;
 270         color_type = PNG_COLOR_TYPE_PALETTE;
 271         break;
 272     default:
 273         return -1;
 274     }
 275     bits_per_pixel = ff_png_get_nb_channels(color_type) * bit_depth;
 276     row_size = (avctx->width * bits_per_pixel + 7) >> 3;
 277
 278     s->zstream.zalloc = ff_png_zalloc;
 279     s->zstream.zfree = ff_png_zfree;
 280     s->zstream.opaque = NULL;
 281     compression_level = avctx->compression_level == FF_COMPRESSION_DEFAULT ?
 282                             Z_DEFAULT_COMPRESSION :
 283                             av_clip(avctx->compression_level, 0, 9);
 284     ret = deflateInit2(&s->zstream, compression_level,
 285                        Z_DEFLATED, 15, 8, Z_DEFAULT_STRATEGY);
 286     if (ret != Z_OK)
 287         return -1;
 288
 289     enc_row_size    = deflateBound(&s->zstream, row_size);
 290     max_packet_size = avctx->height * (enc_row_size +
 291                                        ((enc_row_size + IOBUF_SIZE - 1) / IOBUF_SIZE) * 12)
 292                       + FF_MIN_BUFFER_SIZE;
 293     if (!pkt->data &&
 294         (ret = av_new_packet(pkt, max_packet_size)) < 0) {
 295         av_log(avctx, AV_LOG_ERROR, "Could not allocate output packet of size %d.\n",
 296                max_packet_size);
 297         return ret;
 298     }
 299
 300     s->bytestream_start =
 301     s->bytestream       = pkt->data;
 302     s->bytestream_end   = pkt->data + pkt->size;
 303
 304     crow_base = av_malloc((row_size + 32) << (s->filter_type == PNG_FILTER_VALUE_MIXED));
 305     if (!crow_base)
 306         goto fail;
 307     crow_buf = crow_base + 15; // pixel data should be aligned, but there's a control byte before it
 308     if (is_progressive) {
 309         progressive_buf = av_malloc(row_size + 1);
 310         if (!progressive_buf)
 311             goto fail;
 312     }
 313     if (color_type == PNG_COLOR_TYPE_RGB_ALPHA) {
 314         rgba_buf = av_malloc(row_size + 1);
 315         if (!rgba_buf)
 316             goto fail;
 317     }
 318     if (is_progressive || color_type == PNG_COLOR_TYPE_RGB_ALPHA) {
 319         top_buf = av_malloc(row_size + 1);
 320         if (!top_buf)
 321             goto fail;
 322     }
 323
 324     /* write png header */
 325     memcpy(s->bytestream, ff_pngsig, 8);
 326     s->bytestream += 8;
 327
 328     AV_WB32(s->buf, avctx->width);
 329     AV_WB32(s->buf + 4, avctx->height);
 330     s->buf[8] = bit_depth;
 331     s->buf[9] = color_type;
 332     s->buf[10] = 0; /* compression type */
 333     s->buf[11] = 0; /* filter type */
 334     s->buf[12] = is_progressive; /* interlace type */
 335
 336     png_write_chunk(&s->bytestream, MKTAG('I', 'H', 'D', 'R'), s->buf, 13);
 337
 338     /* put the palette if needed */
 339     if (color_type == PNG_COLOR_TYPE_PALETTE) {
 340         int has_alpha, alpha, i;
 341         unsigned int v;
 342         uint32_t *palette;
 343         uint8_t *alpha_ptr;
 344
 345         palette = (uint32_t *)p->data[1];
 346         ptr = s->buf;
 347         alpha_ptr = s->buf + 256 * 3;
 348         has_alpha = 0;
 349         for(i = 0; i < 256; i++) {
 350             v = palette[i];
 351             alpha = v >> 24;
 352             if (alpha && alpha != 0xff)
 353                 has_alpha = 1;
 354             *alpha_ptr++ = alpha;
 355             bytestream_put_be24(&ptr, v);
 356         }
 357         png_write_chunk(&s->bytestream, MKTAG('P', 'L', 'T', 'E'), s->buf, 256 * 3);
 358         if (has_alpha) {
 359             png_write_chunk(&s->bytestream, MKTAG('t', 'R', 'N', 'S'), s->buf + 256 * 3, 256);
 360         }
 361     }
 362
 363     /* now put each row */
 364     s->zstream.avail_out = IOBUF_SIZE;
 365     s->zstream.next_out = s->buf;
 366     if (is_progressive) {
 367         int pass;
 368
 369         for(pass = 0; pass < NB_PASSES; pass++) {
 370             /* NOTE: a pass is completely omitted if no pixels would be
 371                output */
 372             pass_row_size = ff_png_pass_row_size(pass, bits_per_pixel, avctx->width);
 373             if (pass_row_size > 0) {
 374                 top = NULL;
 375                 for(y = 0; y < avctx->height; y++) {
 376                     if ((ff_png_pass_ymask[pass] << (y & 7)) & 0x80) {
 377                         ptr = p->data[0] + y * p->linesize[0];
 378                         FFSWAP(uint8_t*, progressive_buf, top_buf);
 379                         if (color_type == PNG_COLOR_TYPE_RGB_ALPHA) {
 380                             convert_from_rgb32(rgba_buf, ptr, avctx->width);
 381                             ptr = rgba_buf;
 382                         }
 383                         png_get_interlaced_row(progressive_buf, pass_row_size,
 384                                                bits_per_pixel, pass,
 385                                                ptr, avctx->width);
 386                         crow = png_choose_filter(s, crow_buf, progressive_buf, top, pass_row_size, bits_per_pixel>>3);
 387                         png_write_row(s, crow, pass_row_size + 1);
 388                         top = progressive_buf;
 389                     }
 390                 }
 391             }
 392         }
 393     } else {
 394         top = NULL;
 395         for(y = 0; y < avctx->height; y++) {
 396             ptr = p->data[0] + y * p->linesize[0];
 397             if (color_type == PNG_COLOR_TYPE_RGB_ALPHA) {
 398                 FFSWAP(uint8_t*, rgba_buf, top_buf);
 399                 convert_from_rgb32(rgba_buf, ptr, avctx->width);
 400                 ptr = rgba_buf;
 401             }
 402             crow = png_choose_filter(s, crow_buf, ptr, top, row_size, bits_per_pixel>>3);
 403             png_write_row(s, crow, row_size + 1);
 404             top = ptr;
 405         }
 406     }
 407     /* compress last bytes */
 408     for(;;) {
 409         ret = deflate(&s->zstream, Z_FINISH);
 410         if (ret == Z_OK || ret == Z_STREAM_END) {
 411             len = IOBUF_SIZE - s->zstream.avail_out;
 412             if (len > 0 && s->bytestream_end - s->bytestream > len + 100) {
 413                 png_write_chunk(&s->bytestream, MKTAG('I', 'D', 'A', 'T'), s->buf, len);
 414             }
 415             s->zstream.avail_out = IOBUF_SIZE;
 416             s->zstream.next_out = s->buf;
 417             if (ret == Z_STREAM_END)
 418                 break;
 419         } else {
 420             goto fail;
 421         }
 422     }
 423     png_write_chunk(&s->bytestream, MKTAG('I', 'E', 'N', 'D'), NULL, 0);
 424
 425     pkt->size   = s->bytestream - s->bytestream_start;
 426     pkt->flags |= AV_PKT_FLAG_KEY;
 427     *got_packet = 1;
 428     ret         = 0;
 429
 430  the_end:
 431     av_free(crow_base);
 432     av_free(progressive_buf);
 433     av_free(rgba_buf);
 434     av_free(top_buf);
 435     deflateEnd(&s->zstream);
 436     return ret;
 437  fail:
 438     ret = -1;
 439     goto the_end;
 440 }
 441
 442 static av_cold int png_enc_init(AVCodecContext *avctx){
 443     PNGEncContext *s = avctx->priv_data;
 444
 445     avcodec_get_frame_defaults(&s->picture);
 446     avctx->coded_frame= &s->picture;
 447     ff_dsputil_init(&s->dsp, avctx);
 448
 449     s->filter_type = av_clip(avctx->prediction_method, PNG_FILTER_VALUE_NONE, PNG_FILTER_VALUE_MIXED);
 450     if(avctx->pix_fmt == AV_PIX_FMT_MONOBLACK)
 451         s->filter_type = PNG_FILTER_VALUE_NONE;
 452
 453     return 0;
 454 }
 455
 456 AVCodec ff_png_encoder = {
 457     .name           = "png",
 458     .type           = AVMEDIA_TYPE_VIDEO,
 459     .id             = AV_CODEC_ID_PNG,
 460     .priv_data_size = sizeof(PNGEncContext),
 461     .init           = png_enc_init,
 462     .encode2        = encode_frame,
 463     .pix_fmts       = (const enum AVPixelFormat[]){
 464         AV_PIX_FMT_RGB24, AV_PIX_FMT_RGB32, AV_PIX_FMT_PAL8, AV_PIX_FMT_GRAY8,
 465         AV_PIX_FMT_MONOBLACK, AV_PIX_FMT_NONE
 466     },
 467     .long_name      = NULL_IF_CONFIG_SMALL("PNG (Portable Network Graphics) image"),
 468 };