-
Notifications
You must be signed in to change notification settings - Fork 0
/
lodepng.h
1452 lines (1186 loc) · 67.6 KB
/
lodepng.h
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/*
LodePNG version 20200306
Copyright (c) 2005-2020 Lode Vandevenne
This software is provided 'as-is', without any express or implied
warranty. In no event will the authors be held liable for any damages
arising from the use of this software.
Permission is granted to anyone to use this software for any purpose,
including commercial applications, and to alter it and redistribute it
freely, subject to the following restrictions:
1. The origin of this software must not be misrepresented; you must not
claim that you wrote the original software. If you use this software
in a product, an acknowledgment in the product documentation would be
appreciated but is not required.
2. Altered source versions must be plainly marked as such, and must not be
misrepresented as being the original software.
3. This notice may not be removed or altered from any source
distribution.
*/
#ifndef LODEPNG_H
#define LODEPNG_H
#include <string.h> /*for size_t*/
extern const char* LODEPNG_VERSION_STRING;
extern unsigned IDAT_buffer_size;
/*
The following #defines are used to create code sections. They can be disabled
to disable code sections, which can give faster compile time and smaller binary.
The "NO_COMPILE" defines are designed to be used to pass as defines to the
compiler command to disable them without modifying this header, e.g.
-DLODEPNG_NO_COMPILE_ZLIB for gcc.
In addition to those below, you can also define LODEPNG_NO_COMPILE_CRC to
allow implementing a custom lodepng_crc32.
*/
/*deflate & zlib. If disabled, you must specify alternative zlib functions in
the custom_zlib field of the compress and decompress settings*/
#ifndef LODEPNG_NO_COMPILE_ZLIB
#define LODEPNG_COMPILE_ZLIB
#endif
/*png encoder and png decoder*/
#ifndef LODEPNG_NO_COMPILE_PNG
#define LODEPNG_COMPILE_PNG
#endif
/*deflate&zlib encoder and png encoder*/
#ifndef LODEPNG_NO_COMPILE_ENCODER
#define LODEPNG_COMPILE_ENCODER
#endif
/*the optional built in harddisk file loading and saving functions*/
#ifndef LODEPNG_NO_COMPILE_DISK
#define LODEPNG_COMPILE_DISK
#endif
/*ability to convert error numerical codes to English text string*/
#ifndef LODEPNG_NO_COMPILE_ERROR_TEXT
#define LODEPNG_COMPILE_ERROR_TEXT
#endif
/*Compile the default allocators (C's free, malloc and realloc). If you disable this,
you can define the functions lodepng_free, lodepng_malloc and lodepng_realloc in your
source files with custom allocators.*/
#ifndef LODEPNG_NO_COMPILE_ALLOCATORS
#define LODEPNG_COMPILE_ALLOCATORS
#endif
#ifdef LODEPNG_COMPILE_PNG
/*The PNG color types (also used for raw image).*/
typedef enum LodePNGColorType {
LCT_GREY = 0, /*grayscale: 1,2,4,8,16 bit*/
LCT_RGB = 2, /*RGB: 8,16 bit*/
LCT_PALETTE = 3, /*palette: 1,2,4,8 bit*/
LCT_GREY_ALPHA = 4, /*grayscale with alpha: 8,16 bit*/
LCT_RGBA = 6, /*RGB with alpha: 8,16 bit*/
/*LCT_MAX_OCTET_VALUE lets the compiler allow this enum to represent any invalid
byte value from 0 to 255 that could be present in an invalid PNG file header. Do
not use, compare with or set the name LCT_MAX_OCTET_VALUE, instead either use
the valid color type names above, or numeric values like 1 or 7 when checking for
particular disallowed color type byte values, or cast to integer to print it.*/
LCT_MAX_OCTET_VALUE = 255
} LodePNGColorType;
#ifdef LODEPNG_COMPILE_ENCODER
/*
Converts raw pixel data into a PNG image in memory. The colortype and bitdepth
of the output PNG image cannot be chosen, they are automatically determined
by the colortype, bitdepth and content of the input pixel data.
Note: for 16-bit per channel colors, needs big endian format like PNG does.
out: Output parameter. Pointer to buffer that will contain the PNG image data.
Must be freed after usage with free(*out).
outsize: Output parameter. Pointer to the size in bytes of the out buffer.
image: The raw pixel data to encode. The size of this buffer should be
w * h * (bytes per pixel), bytes per pixel depends on colortype and bitdepth.
w: width of the raw pixel data in pixels.
h: height of the raw pixel data in pixels.
colortype: the color type of the raw input image. See explanation on PNG color types.
bitdepth: the bit depth of the raw input image. See explanation on PNG color types.
Return value: LodePNG error code (0 means no error).
*/
unsigned lodepng_encode_memory(unsigned char** out, size_t* outsize,
const unsigned char* image, unsigned w, unsigned h,
LodePNGColorType colortype, unsigned bitdepth, unsigned windowsize);
/*Same as lodepng_encode_memory, but always encodes from 32-bit RGBA raw image.*/
unsigned lodepng_encode32(unsigned char** out, size_t* outsize,
const unsigned char* image, unsigned w, unsigned h);
/*Same as lodepng_encode_memory, but always encodes from 24-bit RGB raw image.*/
unsigned lodepng_encode24(unsigned char** out, size_t* outsize,
const unsigned char* image, unsigned w, unsigned h);
#ifdef LODEPNG_COMPILE_DISK
/*
Converts raw pixel data into a PNG file on disk.
Same as the other encode functions, but instead takes a filename as output.
NOTE: This overwrites existing files without warning!
*/
unsigned lodepng_encode_file(const char* filename,
const unsigned char* image, unsigned w, unsigned h,
LodePNGColorType colortype, unsigned bitdepth, unsigned windowsize);
/*Same as lodepng_encode_file, but always encodes from 32-bit RGBA raw image.*/
unsigned lodepng_encode32_file(const char* filename,
const unsigned char* image, unsigned w, unsigned h);
/*Same as lodepng_encode_file, but always encodes from 24-bit RGB raw image.*/
unsigned lodepng_encode24_file(const char* filename,
const unsigned char* image, unsigned w, unsigned h);
#endif /*LODEPNG_COMPILE_DISK*/
#endif /*LODEPNG_COMPILE_ENCODER*/
#endif /*LODEPNG_COMPILE_PNG*/
#ifdef LODEPNG_COMPILE_ERROR_TEXT
/*Returns an English description of the numerical error code.*/
const char* lodepng_error_text(unsigned code);
#endif /*LODEPNG_COMPILE_ERROR_TEXT*/
#ifdef LODEPNG_COMPILE_ENCODER
/*
Settings for zlib compression. Tweaking these settings tweaks the balance
between speed and compression ratio.
*/
typedef struct LodePNGCompressSettings LodePNGCompressSettings;
struct LodePNGCompressSettings /*deflate = compress*/ {
/*LZ77 related settings*/
unsigned btype; /*the block type for LZ (0, 1, 2 or 3, see zlib standard). Should be 2 for proper compression.*/
unsigned use_lz77; /*whether or not to use LZ77. Should be 1 for proper compression.*/
unsigned windowsize; /*must be a power of two <= 32768. higher compresses more but is slower. Default value: 2048.*/
unsigned minmatch; /*minimum lz77 length. 3 is normally best, 6 can be better for some PNGs. Default: 0*/
unsigned nicematch; /*stop searching if >= this length found. Set to 258 for best compression. Default: 128*/
unsigned lazymatching; /*use lazy matching: better compression but a bit slower. Default: true*/
/*use custom zlib encoder instead of built in one (default: null)*/
unsigned (*custom_zlib)(unsigned char**, size_t*,
const unsigned char*, size_t,
const LodePNGCompressSettings*);
/*use custom deflate encoder instead of built in one (default: null)
if custom_zlib is used, custom_deflate is ignored since only the built in
zlib function will call custom_deflate*/
unsigned (*custom_deflate)(unsigned char**, size_t*,
const unsigned char*, size_t,
const LodePNGCompressSettings*);
const void* custom_context; /*optional custom settings for custom functions*/
};
extern const LodePNGCompressSettings lodepng_default_compress_settings;
void lodepng_compress_settings_init(LodePNGCompressSettings* settings);
#endif /*LODEPNG_COMPILE_ENCODER*/
#ifdef LODEPNG_COMPILE_PNG
/*
Color mode of an image. Contains all information required to decode the pixel
bits to RGBA colors. This information is the same as used in the PNG file
format, and is used both for PNG and raw image data in LodePNG.
*/
typedef struct LodePNGColorMode {
/*header (IHDR)*/
LodePNGColorType colortype; /*color type, see PNG standard or documentation further in this header file*/
unsigned bitdepth; /*bits per sample, see PNG standard or documentation further in this header file*/
/*
palette (PLTE and tRNS)
Dynamically allocated with the colors of the palette, including alpha.
This field may not be allocated directly, use lodepng_color_mode_init first,
then lodepng_palette_add per color to correctly initialize it (to ensure size
of exactly 1024 bytes).
The alpha channels must be set as well, set them to 255 for opaque images.
When decoding, by default you can ignore this palette, since LodePNG already
fills the palette colors in the pixels of the raw RGBA output.
The palette is only supported for color type 3.
*/
unsigned char* palette; /*palette in RGBARGBA... order. Must be either 0, or when allocated must have 1024 bytes*/
size_t palettesize; /*palette size in number of colors (amount of used bytes is 4 * palettesize)*/
/*
transparent color key (tRNS)
This color uses the same bit depth as the bitdepth value in this struct, which can be 1-bit to 16-bit.
For grayscale PNGs, r, g and b will all 3 be set to the same.
When decoding, by default you can ignore this information, since LodePNG sets
pixels with this key to transparent already in the raw RGBA output.
The color key is only supported for color types 0 and 2.
*/
unsigned key_defined; /*is a transparent color key given? 0 = false, 1 = true*/
unsigned key_r; /*red/grayscale component of color key*/
unsigned key_g; /*green component of color key*/
unsigned key_b; /*blue component of color key*/
} LodePNGColorMode;
/*init, cleanup and copy functions to use with this struct*/
void lodepng_color_mode_init(LodePNGColorMode* info);
void lodepng_color_mode_cleanup(LodePNGColorMode* info);
/*return value is error code (0 means no error)*/
unsigned lodepng_color_mode_copy(LodePNGColorMode* dest, const LodePNGColorMode* source);
/* Makes a temporary LodePNGColorMode that does not need cleanup (no palette) */
LodePNGColorMode lodepng_color_mode_make(LodePNGColorType colortype, unsigned bitdepth);
void lodepng_palette_clear(LodePNGColorMode* info);
/*add 1 color to the palette*/
unsigned lodepng_palette_add(LodePNGColorMode* info,
unsigned char r, unsigned char g, unsigned char b, unsigned char a);
/*get the total amount of bits per pixel, based on colortype and bitdepth in the struct*/
unsigned lodepng_get_bpp(const LodePNGColorMode* info);
/*get the amount of color channels used, based on colortype in the struct.
If a palette is used, it counts as 1 channel.*/
unsigned lodepng_get_channels(const LodePNGColorMode* info);
/*is it a grayscale type? (only colortype 0 or 4)*/
unsigned lodepng_is_greyscale_type(const LodePNGColorMode* info);
/*has it got an alpha channel? (only colortype 2 or 6)*/
unsigned lodepng_is_alpha_type(const LodePNGColorMode* info);
/*has it got a palette? (only colortype 3)*/
unsigned lodepng_is_palette_type(const LodePNGColorMode* info);
/*only returns true if there is a palette and there is a value in the palette with alpha < 255.
Loops through the palette to check this.*/
unsigned lodepng_has_palette_alpha(const LodePNGColorMode* info);
/*
Check if the given color info indicates the possibility of having non-opaque pixels in the PNG image.
Returns true if the image can have translucent or invisible pixels (it still be opaque if it doesn't use such pixels).
Returns false if the image can only have opaque pixels.
In detail, it returns true only if it's a color type with alpha, or has a palette with non-opaque values,
or if "key_defined" is true.
*/
unsigned lodepng_can_have_alpha(const LodePNGColorMode* info);
/*Returns the byte size of a raw image buffer with given width, height and color mode*/
size_t lodepng_get_raw_size(unsigned w, unsigned h, const LodePNGColorMode* color);
/*Information about the PNG image, except pixels, width and height.*/
typedef struct LodePNGInfo {
/*header (IHDR), palette (PLTE) and transparency (tRNS) chunks*/
unsigned compression_method;/*compression method of the original file. Always 0.*/
unsigned filter_method; /*filter method of the original file*/
unsigned interlace_method; /*interlace method of the original file: 0=none, 1=Adam7*/
LodePNGColorMode color; /*color type and bits, palette and transparency of the PNG file*/
} LodePNGInfo;
/*init, cleanup and copy functions to use with this struct*/
void lodepng_info_init(LodePNGInfo* info);
void lodepng_info_cleanup(LodePNGInfo* info);
/*return value is error code (0 means no error)*/
unsigned lodepng_info_copy(LodePNGInfo* dest, const LodePNGInfo* source);
/*
Converts raw buffer from one color type to another color type, based on
LodePNGColorMode structs to describe the input and output color type.
See the reference manual at the end of this header file to see which color conversions are supported.
return value = LodePNG error code (0 if all went ok, an error if the conversion isn't supported)
The out buffer must have size (w * h * bpp + 7) / 8, where bpp is the bits per pixel
of the output color type (lodepng_get_bpp).
For < 8 bpp images, there should not be padding bits at the end of scanlines.
For 16-bit per channel colors, uses big endian format like PNG does.
Return value is LodePNG error code
*/
unsigned lodepng_convert(unsigned char* out, const unsigned char* in,
const LodePNGColorMode* mode_out, const LodePNGColorMode* mode_in,
unsigned w, unsigned h);
#ifdef LODEPNG_COMPILE_ENCODER
/*automatically use color type with less bits per pixel if losslessly possible. Default: AUTO*/
typedef enum LodePNGFilterStrategy {
/*every filter at zero*/
LFS_ZERO = 0,
/*every filter at 1, 2, 3 or 4 (paeth), unlike LFS_ZERO not a good choice, but for testing*/
LFS_ONE = 1,
LFS_TWO = 2,
LFS_THREE = 3,
LFS_FOUR = 4,
/*Use filter that gives minimum sum, as described in the official PNG filter heuristic.*/
LFS_MINSUM,
/*Use the filter type that gives smallest Shannon entropy for this scanline. Depending
on the image, this is better or worse than minsum.*/
LFS_ENTROPY,
/*
Brute-force-search PNG filters by compressing each filter for each scanline.
Experimental, very slow, and only rarely gives better compression than MINSUM.
*/
LFS_BRUTE_FORCE,
/*use predefined_filters buffer: you specify the filter type for each scanline*/
LFS_PREDEFINED
} LodePNGFilterStrategy;
/*Gives characteristics about the integer RGBA colors of the image (count, alpha channel usage, bit depth, ...),
which helps decide which color model to use for encoding.
Used internally by default if "auto_convert" is enabled. Public because it's useful for custom algorithms.*/
typedef struct LodePNGColorStats {
unsigned colored; /*not grayscale*/
unsigned key; /*image is not opaque and color key is possible instead of full alpha*/
unsigned short key_r; /*key values, always as 16-bit, in 8-bit case the byte is duplicated, e.g. 65535 means 255*/
unsigned short key_g;
unsigned short key_b;
unsigned alpha; /*image is not opaque and alpha channel or alpha palette required*/
unsigned numcolors; /*amount of colors, up to 257. Not valid if bits == 16 or allow_palette is disabled.*/
unsigned char palette[1024]; /*Remembers up to the first 256 RGBA colors, in no particular order, only valid when numcolors is valid*/
unsigned bits; /*bits per channel (not for palette). 1,2 or 4 for grayscale only. 16 if 16-bit per channel required.*/
size_t numpixels;
/*user settings for computing/using the stats*/
unsigned allow_palette; /*default 1. if 0, disallow choosing palette colortype in auto_choose_color, and don't count numcolors*/
unsigned allow_greyscale; /*default 1. if 0, choose RGB or RGBA even if the image only has gray colors*/
} LodePNGColorStats;
void lodepng_color_stats_init(LodePNGColorStats* stats);
/*Get a LodePNGColorStats of the image. The stats must already have been inited.
Returns error code (e.g. alloc fail) or 0 if ok.*/
unsigned lodepng_compute_color_stats(LodePNGColorStats* stats,
const unsigned char* image, unsigned w, unsigned h,
const LodePNGColorMode* mode_in);
/*Settings for the encoder.*/
typedef struct LodePNGEncoderSettings {
LodePNGCompressSettings zlibsettings; /*settings for the zlib encoder, such as window size, ...*/
unsigned auto_convert; /*automatically choose output PNG color type. Default: true*/
/*If true, follows the official PNG heuristic: if the PNG uses a palette or lower than
8 bit depth, set all filters to zero. Otherwise use the filter_strategy. Note that to
completely follow the official PNG heuristic, filter_palette_zero must be true and
filter_strategy must be LFS_MINSUM*/
unsigned filter_palette_zero;
/*Which filter strategy to use when not using zeroes due to filter_palette_zero.
Set filter_palette_zero to 0 to ensure always using your chosen strategy. Default: LFS_MINSUM*/
LodePNGFilterStrategy filter_strategy;
/*used if filter_strategy is LFS_PREDEFINED. In that case, this must point to a buffer with
the same length as the amount of scanlines in the image, and each value must <= 5. You
have to cleanup this buffer, LodePNG will never free it. Don't forget that filter_palette_zero
must be set to 0 to ensure this is also used on palette or low bitdepth images.*/
const unsigned char* predefined_filters;
/*force creating a PLTE chunk if colortype is 2 or 6 (= a suggested palette).
If colortype is 3, PLTE is _always_ created.*/
unsigned force_palette;
} LodePNGEncoderSettings;
void lodepng_encoder_settings_init(LodePNGEncoderSettings* settings);
#endif /*LODEPNG_COMPILE_ENCODER*/
#ifdef LODEPNG_COMPILE_ENCODER
/*The settings, state and information for extended encoding and decoding.*/
typedef struct LodePNGState {
#ifdef LODEPNG_COMPILE_ENCODER
LodePNGEncoderSettings encoder; /*the encoding settings*/
#endif /*LODEPNG_COMPILE_ENCODER*/
LodePNGColorMode info_raw; /*specifies the format in which you would like to get the raw pixel buffer*/
LodePNGInfo info_png; /*info of the PNG image obtained after decoding*/
unsigned error;
} LodePNGState;
/*init, cleanup and copy functions to use with this struct*/
void lodepng_state_init(LodePNGState* state);
void lodepng_state_cleanup(LodePNGState* state);
void lodepng_state_copy(LodePNGState* dest, const LodePNGState* source);
#endif /*LODEPNG_COMPILE_ENCODER*/
/*
Reads one metadata chunk (other than IHDR) of the PNG file and outputs what it
read in the state. Returns error code on failure.
Use lodepng_inspect first with a new state, then e.g. lodepng_chunk_find_const
to find the desired chunk type, and if non null use lodepng_inspect_chunk (with
chunk_pointer - start_of_file as pos).
Supports most metadata chunks from the PNG standard (gAMA, bKGD, tEXt, ...).
Ignores unsupported, unknown, non-metadata or IHDR chunks (without error).
Requirements: &in[pos] must point to start of a chunk, must use regular
lodepng_inspect first since format of most other chunks depends on IHDR, and if
there is a PLTE chunk, that one must be inspected before tRNS or bKGD.
*/
unsigned lodepng_inspect_chunk(LodePNGState* state, size_t pos,
const unsigned char* in, size_t insize);
#ifdef LODEPNG_COMPILE_ENCODER
/*This function allocates the out buffer with standard malloc and stores the size in *outsize.*/
unsigned lodepng_encode(unsigned char** out, size_t* outsize,
const unsigned char* image, unsigned w, unsigned h,
LodePNGState* state);
#endif /*LODEPNG_COMPILE_ENCODER*/
/*
The lodepng_chunk functions are normally not needed, except to traverse the
unknown chunks stored in the LodePNGInfo struct, or add new ones to it.
It also allows traversing the chunks of an encoded PNG file yourself.
The chunk pointer always points to the beginning of the chunk itself, that is
the first byte of the 4 length bytes.
In the PNG file format, chunks have the following format:
-4 bytes length: length of the data of the chunk in bytes (chunk itself is 12 bytes longer)
-4 bytes chunk type (ASCII a-z,A-Z only, see below)
-length bytes of data (may be 0 bytes if length was 0)
-4 bytes of CRC, computed on chunk name + data
The first chunk starts at the 8th byte of the PNG file, the entire rest of the file
exists out of concatenated chunks with the above format.
PNG standard chunk ASCII naming conventions:
-First byte: uppercase = critical, lowercase = ancillary
-Second byte: uppercase = public, lowercase = private
-Third byte: must be uppercase
-Fourth byte: uppercase = unsafe to copy, lowercase = safe to copy
*/
/*
Gets the length of the data of the chunk. Total chunk length has 12 bytes more.
There must be at least 4 bytes to read from. If the result value is too large,
it may be corrupt data.
*/
unsigned lodepng_chunk_length(const unsigned char* chunk);
/*puts the 4-byte type in null terminated string*/
void lodepng_chunk_type(char type[5], const unsigned char* chunk);
/*check if the type is the given type*/
unsigned char lodepng_chunk_type_equals(const unsigned char* chunk, const char* type);
/*0: it's one of the critical chunk types, 1: it's an ancillary chunk (see PNG standard)*/
unsigned char lodepng_chunk_ancillary(const unsigned char* chunk);
/*0: public, 1: private (see PNG standard)*/
unsigned char lodepng_chunk_private(const unsigned char* chunk);
/*0: the chunk is unsafe to copy, 1: the chunk is safe to copy (see PNG standard)*/
unsigned char lodepng_chunk_safetocopy(const unsigned char* chunk);
/*get pointer to the data of the chunk, where the input points to the header of the chunk*/
unsigned char* lodepng_chunk_data(unsigned char* chunk);
const unsigned char* lodepng_chunk_data_const(const unsigned char* chunk);
/*returns 0 if the crc is correct, 1 if it's incorrect (0 for OK as usual!)*/
unsigned lodepng_chunk_check_crc(const unsigned char* chunk);
/*generates the correct CRC from the data and puts it in the last 4 bytes of the chunk*/
void lodepng_chunk_generate_crc(unsigned char* chunk);
/*
Iterate to next chunks, allows iterating through all chunks of the PNG file.
Input must be at the beginning of a chunk (result of a previous lodepng_chunk_next call,
or the 8th byte of a PNG file which always has the first chunk), or alternatively may
point to the first byte of the PNG file (which is not a chunk but the magic header, the
function will then skip over it and return the first real chunk).
Will output pointer to the start of the next chunk, or at or beyond end of the file if there
is no more chunk after this or possibly if the chunk is corrupt.
Start this process at the 8th byte of the PNG file.
In a non-corrupt PNG file, the last chunk should have name "IEND".
*/
unsigned char* lodepng_chunk_next(unsigned char* chunk, unsigned char* end);
const unsigned char* lodepng_chunk_next_const(const unsigned char* chunk, const unsigned char* end);
/*Finds the first chunk with the given type in the range [chunk, end), or returns NULL if not found.*/
unsigned char* lodepng_chunk_find(unsigned char* chunk, unsigned char* end, const char type[5]);
const unsigned char* lodepng_chunk_find_const(const unsigned char* chunk, const unsigned char* end, const char type[5]);
/*
Appends chunk to the data in out. The given chunk should already have its chunk header.
The out variable and outsize are updated to reflect the new reallocated buffer.
Returns error code (0 if it went ok)
*/
unsigned lodepng_chunk_append(unsigned char** out, size_t* outsize, const unsigned char* chunk);
/*
Appends new chunk to out. The chunk to append is given by giving its length, type
and data separately. The type is a 4-letter string.
The out variable and outsize are updated to reflect the new reallocated buffer.
Returne error code (0 if it went ok)
*/
unsigned lodepng_chunk_create(unsigned char** out, size_t* outsize, unsigned length,
const char* type, const unsigned char* data);
/*Calculate CRC32 of buffer*/
unsigned lodepng_crc32(const unsigned char* buf, size_t len);
#endif /*LODEPNG_COMPILE_PNG*/
#ifdef LODEPNG_COMPILE_ZLIB
/*
This zlib part can be used independently to zlib compress and decompress a
buffer. It cannot be used to create gzip files however, and it only supports the
part of zlib that is required for PNG, it does not support dictionaries.
*/
#ifdef LODEPNG_COMPILE_ENCODER
/*
Compresses data with Zlib. Reallocates the out buffer and appends the data.
Zlib adds a small header and trailer around the deflate data.
The data is output in the format of the zlib specification.
Either, *out must be NULL and *outsize must be 0, or, *out must be a valid
buffer and *outsize its size in bytes. out must be freed by user after usage.
*/
unsigned lodepng_zlib_compress(unsigned char** out, size_t* outsize,
const unsigned char* in, size_t insize,
const LodePNGCompressSettings* settings);
/*
Find length-limited Huffman code for given frequencies. This function is in the
public interface only for tests, it's used internally by lodepng_deflate.
*/
unsigned lodepng_huffman_code_lengths(unsigned* lengths, const unsigned* frequencies,
size_t numcodes, unsigned maxbitlen);
/*Compress a buffer with deflate. See RFC 1951. Out buffer must be freed after use.*/
unsigned lodepng_deflate(unsigned char** out, size_t* outsize,
const unsigned char* in, size_t insize,
const LodePNGCompressSettings* settings);
#endif /*LODEPNG_COMPILE_ENCODER*/
#endif /*LODEPNG_COMPILE_ZLIB*/
#ifdef LODEPNG_COMPILE_DISK
/*
Load a file from disk into buffer. The function allocates the out buffer, and
after usage you should free it.
out: output parameter, contains pointer to loaded buffer.
outsize: output parameter, size of the allocated out buffer
filename: the path to the file to load
return value: error code (0 means ok)
*/
unsigned lodepng_load_file(unsigned char** out, size_t* outsize, const char* filename);
/*
Save a file from buffer to disk. Warning, if it exists, this function overwrites
the file without warning!
buffer: the buffer to write
buffersize: size of the buffer to write
filename: the path to the file to save to
return value: error code (0 means ok)
*/
unsigned lodepng_save_file(const unsigned char* buffer, size_t buffersize, const char* filename);
#endif /*LODEPNG_COMPILE_DISK*/
#endif /*LODEPNG_H inclusion guard*/
/*
LodePNG Documentation
---------------------
0. table of contents
--------------------
1. about
1.1. supported features
1.2. features not supported
2. C and C++ version
3. security
4. decoding
5. encoding
6. color conversions
6.1. PNG color types
6.2. color conversions
6.3. padding bits
6.4. A note about 16-bits per channel and endianness
7. error values
8. chunks and PNG editing
9. compiler support
10. examples
10.1. decoder C++ example
10.2. decoder C example
11. state settings reference
12. changes
13. contact information
1. about
--------
PNG is a file format to store raster images losslessly with good compression,
supporting different color types and alpha channel.
LodePNG is a PNG codec according to the Portable Network Graphics (PNG)
Specification (Second Edition) - W3C Recommendation 10 November 2003.
The specifications used are:
*) Portable Network Graphics (PNG) Specification (Second Edition):
http://www.w3.org/TR/2003/REC-PNG-20031110
*) RFC 1950 ZLIB Compressed Data Format version 3.3:
http://www.gzip.org/zlib/rfc-zlib.html
*) RFC 1951 DEFLATE Compressed Data Format Specification ver 1.3:
http://www.gzip.org/zlib/rfc-deflate.html
The most recent version of LodePNG can currently be found at
http://lodev.org/lodepng/
LodePNG works both in C (ISO C90) and C++, with a C++ wrapper that adds
extra functionality.
LodePNG exists out of two files:
-lodepng.h: the header file for both C and C++
-lodepng.c(pp): give it the name lodepng.c or lodepng.cpp (or .cc) depending on your usage
If you want to start using LodePNG right away without reading this doc, get the
examples from the LodePNG website to see how to use it in code, or check the
smaller examples in chapter 13 here.
LodePNG is simple but only supports the basic requirements. To achieve
simplicity, the following design choices were made: There are no dependencies
on any external library. There are functions to decode and encode a PNG with
a single function call, and extended versions of these functions taking a
LodePNGState struct allowing to specify or get more information. By default
the colors of the raw image are always RGB or RGBA, no matter what color type
the PNG file uses. To read and write files, there are simple functions to
convert the files to/from buffers in memory.
This all makes LodePNG suitable for loading textures in games, demos and small
programs, ... It's less suitable for full fledged image editors, loading PNGs
over network (it requires all the image data to be available before decoding can
begin), life-critical systems, ...
1.1. supported features
-----------------------
The following features are supported by the decoder:
*) decoding of PNGs with any color type, bit depth and interlace mode, to a 24- or 32-bit color raw image,
or the same color type as the PNG
*) encoding of PNGs, from any raw image to 24- or 32-bit color, or the same color type as the raw image
*) Adam7 interlace and deinterlace for any color type
*) loading the image from harddisk or decoding it from a buffer from other sources than harddisk
*) support for alpha channels, including RGBA color model, translucent palettes and color keying
*) zlib decompression (inflate)
*) zlib compression (deflate)
*) CRC32 and ADLER32 checksums
*) colorimetric color profile conversions: currently experimentally available in lodepng_util.cpp only,
plus alternatively ability to pass on chroma/gamma/ICC profile information to other color management system.
*) handling of unknown chunks, allowing making a PNG editor that stores custom and unknown chunks.
*) the following chunks are supported by both encoder and decoder:
IHDR: header information
PLTE: color palette
IDAT: pixel data
IEND: the final chunk
tRNS: transparency for palettized images
tEXt: textual information
zTXt: compressed textual information
iTXt: international textual information
bKGD: suggested background color
pHYs: physical dimensions
tIME: modification time
cHRM: RGB chromaticities
gAMA: RGB gamma correction
iCCP: ICC color profile
sRGB: rendering intent
1.2. features not supported
---------------------------
The following features are _not_ supported:
*) some features needed to make a conformant PNG-Editor might be still missing.
*) partial loading/stream processing. All data must be available and is processed in one call.
*) The following public chunks are not (yet) supported but treated as unknown chunks by LodePNG:
sBIT
hIST
sPLT
2. C and C++ version
--------------------
The C version uses buffers allocated with alloc that you need to free()
yourself. You need to use init and cleanup functions for each struct whenever
using a struct from the C version to avoid exploits and memory leaks.
The C++ version has extra functions with std::vectors in the interface and the
lodepng::State class which is a LodePNGState with constructor and destructor.
These files work without modification for both C and C++ compilers because all
the additional C++ code is in "#ifdef __cplusplus" blocks that make C-compilers
ignore it, and the C code is made to compile both with strict ISO C90 and C++.
To use the C++ version, you need to rename the source file to lodepng.cpp
(instead of lodepng.c), and compile it with a C++ compiler.
To use the C version, you need to rename the source file to lodepng.c (instead
of lodepng.cpp), and compile it with a C compiler.
3. Security
-----------
Even if carefully designed, it's always possible that LodePNG contains possible
exploits. If you discover one, please let me know, and it will be fixed.
When using LodePNG, care has to be taken with the C version of LodePNG, as well
as the C-style structs when working with C++. The following conventions are used
for all C-style structs:
-if a struct has a corresponding init function, always call the init function when making a new one
-if a struct has a corresponding cleanup function, call it before the struct disappears to avoid memory leaks
-if a struct has a corresponding copy function, use the copy function instead of "=".
The destination must also be inited already.
4. Decoding
-----------
Decoding converts a PNG compressed image to a raw pixel buffer.
Most documentation on using the decoder is at its declarations in the header
above. For C, simple decoding can be done with functions such as
lodepng_decode32, and more advanced decoding can be done with the struct
LodePNGState and lodepng_decode. For C++, all decoding can be done with the
various lodepng::decode functions, and lodepng::State can be used for advanced
features.
When using the LodePNGState, it uses the following fields for decoding:
*) LodePNGInfo info_png: it stores extra information about the PNG (the input) in here
*) LodePNGColorMode info_raw: here you can say what color mode of the raw image (the output) you want to get
*) LodePNGDecoderSettings decoder: you can specify a few extra settings for the decoder to use
LodePNGInfo info_png
--------------------
After decoding, this contains extra information of the PNG image, except the actual
pixels, width and height because these are already gotten directly from the decoder
functions.
It contains for example the original color type of the PNG image, text comments,
suggested background color, etc... More details about the LodePNGInfo struct are
at its declaration documentation.
LodePNGColorMode info_raw
-------------------------
When decoding, here you can specify which color type you want
the resulting raw image to be. If this is different from the colortype of the
PNG, then the decoder will automatically convert the result. This conversion
always works, except if you want it to convert a color PNG to grayscale or to
a palette with missing colors.
By default, 32-bit color is used for the result.
LodePNGDecoderSettings decoder
------------------------------
The settings can be used to ignore the errors created by invalid CRC and Adler32
chunks, and to disable the decoding of tEXt chunks.
There's also a setting color_convert, true by default. If false, no conversion
is done, the resulting data will be as it was in the PNG (after decompression)
and you'll have to puzzle the colors of the pixels together yourself using the
color type information in the LodePNGInfo.
5. Encoding
-----------
Encoding converts a raw pixel buffer to a PNG compressed image.
Most documentation on using the encoder is at its declarations in the header
above. For C, simple encoding can be done with functions such as
lodepng_encode32, and more advanced decoding can be done with the struct
LodePNGState and lodepng_encode. For C++, all encoding can be done with the
various lodepng::encode functions, and lodepng::State can be used for advanced
features.
Like the decoder, the encoder can also give errors. However it gives less errors
since the encoder input is trusted, the decoder input (a PNG image that could
be forged by anyone) is not trusted.
When using the LodePNGState, it uses the following fields for encoding:
*) LodePNGInfo info_png: here you specify how you want the PNG (the output) to be.
*) LodePNGColorMode info_raw: here you say what color type of the raw image (the input) has
*) LodePNGEncoderSettings encoder: you can specify a few settings for the encoder to use
LodePNGInfo info_png
--------------------
When encoding, you use this the opposite way as when decoding: for encoding,
you fill in the values you want the PNG to have before encoding. By default it's
not needed to specify a color type for the PNG since it's automatically chosen,
but it's possible to choose it yourself given the right settings.
The encoder will not always exactly match the LodePNGInfo struct you give,
it tries as close as possible. Some things are ignored by the encoder. The
encoder uses, for example, the following settings from it when applicable:
colortype and bitdepth, text chunks, time chunk, the color key, the palette, the
background color, the interlace method, unknown chunks, ...
When encoding to a PNG with colortype 3, the encoder will generate a PLTE chunk.
If the palette contains any colors for which the alpha channel is not 255 (so
there are translucent colors in the palette), it'll add a tRNS chunk.
LodePNGColorMode info_raw
-------------------------
You specify the color type of the raw image that you give to the input here,
including a possible transparent color key and palette you happen to be using in
your raw image data.
By default, 32-bit color is assumed, meaning your input has to be in RGBA
format with 4 bytes (unsigned chars) per pixel.
LodePNGEncoderSettings encoder
------------------------------
The following settings are supported (some are in sub-structs):
*) auto_convert: when this option is enabled, the encoder will
automatically choose the smallest possible color mode (including color key) that
can encode the colors of all pixels without information loss.
*) btype: the block type for LZ77. 0 = uncompressed, 1 = fixed huffman tree,
2 = dynamic huffman tree (best compression). Should be 2 for proper
compression.
*) use_lz77: whether or not to use LZ77 for compressed block types. Should be
true for proper compression.
*) windowsize: the window size used by the LZ77 encoder (1 - 32768). Has value
2048 by default, but can be set to 32768 for better, but slow, compression.
*) force_palette: if colortype is 2 or 6, you can make the encoder write a PLTE
chunk if force_palette is true. This can used as suggested palette to convert
to by viewers that don't support more than 256 colors (if those still exist)
*) add_id: add text chunk "Encoder: LodePNG <version>" to the image.
*) text_compression: default 1. If 1, it'll store texts as zTXt instead of tEXt chunks.
zTXt chunks use zlib compression on the text. This gives a smaller result on
large texts but a larger result on small texts (such as a single program name).
It's all tEXt or all zTXt though, there's no separate setting per text yet.
6. color conversions
--------------------
An important thing to note about LodePNG, is that the color type of the PNG, and
the color type of the raw image, are completely independent. By default, when
you decode a PNG, you get the result as a raw image in the color type you want,
no matter whether the PNG was encoded with a palette, grayscale or RGBA color.
And if you encode an image, by default LodePNG will automatically choose the PNG
color type that gives good compression based on the values of colors and amount
of colors in the image. It can be configured to let you control it instead as
well, though.
To be able to do this, LodePNG does conversions from one color mode to another.
It can convert from almost any color type to any other color type, except the
following conversions: RGB to grayscale is not supported, and converting to a
palette when the palette doesn't have a required color is not supported. This is
not supported on purpose: this is information loss which requires a color
reduction algorithm that is beyond the scope of a PNG encoder (yes, RGB to gray
is easy, but there are multiple ways if you want to give some channels more
weight).
By default, when decoding, you get the raw image in 32-bit RGBA or 24-bit RGB
color, no matter what color type the PNG has. And by default when encoding,
LodePNG automatically picks the best color model for the output PNG, and expects
the input image to be 32-bit RGBA or 24-bit RGB. So, unless you want to control
the color format of the images yourself, you can skip this chapter.
6.1. PNG color types
--------------------
A PNG image can have many color types, ranging from 1-bit color to 64-bit color,
as well as palettized color modes. After the zlib decompression and unfiltering
in the PNG image is done, the raw pixel data will have that color type and thus
a certain amount of bits per pixel. If you want the output raw image after
decoding to have another color type, a conversion is done by LodePNG.
The PNG specification gives the following color types:
0: grayscale, bit depths 1, 2, 4, 8, 16
2: RGB, bit depths 8 and 16
3: palette, bit depths 1, 2, 4 and 8
4: grayscale with alpha, bit depths 8 and 16
6: RGBA, bit depths 8 and 16
Bit depth is the amount of bits per pixel per color channel. So the total amount
of bits per pixel is: amount of channels * bitdepth.
6.2. color conversions
----------------------
As explained in the sections about the encoder and decoder, you can specify
color types and bit depths in info_png and info_raw to change the default
behaviour.
If, when decoding, you want the raw image to be something else than the default,
you need to set the color type and bit depth you want in the LodePNGColorMode,
or the parameters colortype and bitdepth of the simple decoding function.
If, when encoding, you use another color type than the default in the raw input
image, you need to specify its color type and bit depth in the LodePNGColorMode
of the raw image, or use the parameters colortype and bitdepth of the simple
encoding function.
If, when encoding, you don't want LodePNG to choose the output PNG color type
but control it yourself, you need to set auto_convert in the encoder settings
to false, and specify the color type you want in the LodePNGInfo of the
encoder (including palette: it can generate a palette if auto_convert is true,
otherwise not).
If the input and output color type differ (whether user chosen or auto chosen),
LodePNG will do a color conversion, which follows the rules below, and may
sometimes result in an error.
To avoid some confusion:
-the decoder converts from PNG to raw image
-the encoder converts from raw image to PNG
-the colortype and bitdepth in LodePNGColorMode info_raw, are those of the raw image
-the colortype and bitdepth in the color field of LodePNGInfo info_png, are those of the PNG
-when encoding, the color type in LodePNGInfo is ignored if auto_convert
is enabled, it is automatically generated instead
-when decoding, the color type in LodePNGInfo is set by the decoder to that of the original
PNG image, but it can be ignored since the raw image has the color type you requested instead
-if the color type of the LodePNGColorMode and PNG image aren't the same, a conversion
between the color types is done if the color types are supported. If it is not
supported, an error is returned. If the types are the same, no conversion is done.
-even though some conversions aren't supported, LodePNG supports loading PNGs from any
colortype and saving PNGs to any colortype, sometimes it just requires preparing
the raw image correctly before encoding.
-both encoder and decoder use the same color converter.
The function lodepng_convert does the color conversion. It is available in the
interface but normally isn't needed since the encoder and decoder already call
it.
Non supported color conversions:
-color to grayscale when non-gray pixels are present: no error is thrown, but
the result will look ugly because only the red channel is taken (it assumes all
three channels are the same in this case so ignores green and blue). The reason
no error is given is to allow converting from three-channel grayscale images to
one-channel even if there are numerical imprecisions.
-anything to palette when the palette does not have an exact match for a from-color
in it: in this case an error is thrown
Supported color conversions:
-anything to 8-bit RGB, 8-bit RGBA, 16-bit RGB, 16-bit RGBA
-any gray or gray+alpha, to gray or gray+alpha
-anything to a palette, as long as the palette has the requested colors in it
-removing alpha channel
-higher to smaller bitdepth, and vice versa
If you want no color conversion to be done (e.g. for speed or control):
-In the encoder, you can make it save a PNG with any color type by giving the
raw color mode and LodePNGInfo the same color mode, and setting auto_convert to
false.
-In the decoder, you can make it store the pixel data in the same color type
as the PNG has, by setting the color_convert setting to false. Settings in
info_raw are then ignored.
6.3. padding bits
-----------------
In the PNG file format, if a less than 8-bit per pixel color type is used and the scanlines
have a bit amount that isn't a multiple of 8, then padding bits are used so that each
scanline starts at a fresh byte. But that is NOT true for the LodePNG raw input and output.
The raw input image you give to the encoder, and the raw output image you get from the decoder
will NOT have these padding bits, e.g. in the case of a 1-bit image with a width
of 7 pixels, the first pixel of the second scanline will the 8th bit of the first byte,
not the first bit of a new byte.
6.4. A note about 16-bits per channel and endianness
----------------------------------------------------
LodePNG uses unsigned char arrays for 16-bit per channel colors too, just like
for any other color format. The 16-bit values are stored in big endian (most
significant byte first) in these arrays. This is the opposite order of the
little endian used by x86 CPU's.
LodePNG always uses big endian because the PNG file format does so internally.
Conversions to other formats than PNG uses internally are not supported by
LodePNG on purpose, there are myriads of formats, including endianness of 16-bit
colors, the order in which you store R, G, B and A, and so on. Supporting and
converting to/from all that is outside the scope of LodePNG.
This may mean that, depending on your use case, you may want to convert the big
endian output of LodePNG to little endian with a for loop. This is certainly not
always needed, many applications and libraries support big endian 16-bit colors
anyway, but it means you cannot simply cast the unsigned char* buffer to an
unsigned short* buffer on x86 CPUs.