1 libpng-manual.txt - A description on how to use and modify libpng
3 libpng version 1.6.32 - August 24, 2017
4 Updated and distributed by Glenn Randers-Pehrson
5 <glennrp at users.sourceforge.net>
6 Copyright (c) 1998-2017 Glenn Randers-Pehrson
8 This document is released under the libpng license.
9 For conditions of distribution and use, see the disclaimer
14 libpng versions 0.97, January 1998, through 1.6.32 - August 24, 2017
15 Updated and distributed by Glenn Randers-Pehrson
16 Copyright (c) 1998-2017 Glenn Randers-Pehrson
18 libpng 1.0 beta 6 - version 0.96 - May 28, 1997
19 Updated and distributed by Andreas Dilger
20 Copyright (c) 1996, 1997 Andreas Dilger
22 libpng 1.0 beta 2 - version 0.88 - January 26, 1996
23 For conditions of distribution and use, see copyright
24 notice in png.h. Copyright (c) 1995, 1996 Guy Eric
25 Schalnat, Group 42, Inc.
27 Updated/rewritten per request in the libpng FAQ
28 Copyright (c) 1995, 1996 Frank J. T. Wojcik
29 December 18, 1995 & January 20, 1996
38 VI. Modifying/Customizing libpng
40 VIII. Changes to Libpng from version 0.88
41 IX. Changes to Libpng from version 1.0.x to 1.2.x
42 X. Changes to Libpng from version 1.0.x/1.2.x to 1.4.x
43 XI. Changes to Libpng from version 1.4.x to 1.5.x
44 XII. Changes to Libpng from version 1.5.x to 1.6.x
45 XIII. Detecting libpng
46 XIV. Source code repository
48 XVI. Y2K Compliance in libpng
52 This file describes how to use and modify the PNG reference library
53 (known as libpng) for your own use. In addition to this
54 file, example.c is a good starting point for using the library, as
55 it is heavily commented and should include everything most people
56 will need. We assume that libpng is already installed; see the
57 INSTALL file for instructions on how to configure and install libpng.
59 For examples of libpng usage, see the files "example.c", "pngtest.c",
60 and the files in the "contrib" directory, all of which are included in
61 the libpng distribution.
63 Libpng was written as a companion to the PNG specification, as a way
64 of reducing the amount of time and effort it takes to support the PNG
65 file format in application programs.
67 The PNG specification (second edition), November 2003, is available as
68 a W3C Recommendation and as an ISO Standard (ISO/IEC 15948:2004 (E)) at
69 <https://www.w3.org/TR/2003/REC-PNG-20031110/
70 The W3C and ISO documents have identical technical content.
72 The PNG-1.2 specification is available at
73 <https://png-mng.sourceforge.io/pub/png/spec/1.2/>.
74 It is technically equivalent
75 to the PNG specification (second edition) but has some additional material.
77 The PNG-1.0 specification is available as RFC 2083
78 <https://png-mng.sourceforge.io/pub/png/spec/1.0/> and as a
79 W3C Recommendation <https://www.w3.org/TR/REC-png-961001>.
81 Some additional chunks are described in the special-purpose public chunks
82 documents at <http://www.libpng.org/pub/png/spec/register/>
85 about PNG, and the latest version of libpng, can be found at the PNG home
86 page, <http://www.libpng.org/pub/png/>.
88 Most users will not have to modify the library significantly; advanced
89 users may want to modify it more. All attempts were made to make it as
90 complete as possible, while keeping the code easy to understand.
91 Currently, this library only supports C. Support for other languages
94 Libpng has been designed to handle multiple sessions at one time,
95 to be easily modifiable, to be portable to the vast majority of
96 machines (ANSI, K&R, 16-, 32-, and 64-bit) available, and to be easy
97 to use. The ultimate goal of libpng is to promote the acceptance of
98 the PNG file format in whatever way possible. While there is still
99 work to be done (see the TODO file), libpng should cover the
100 majority of the needs of its users.
102 Libpng uses zlib for its compression and decompression of PNG files.
103 Further information about zlib, and the latest version of zlib, can
104 be found at the zlib home page, <https://zlib.net/>.
105 The zlib compression utility is a general purpose utility that is
106 useful for more than PNG files, and can be used without libpng.
107 See the documentation delivered with zlib for more details.
108 You can usually find the source files for the zlib utility wherever you
109 find the libpng source files.
111 Libpng is thread safe, provided the threads are using different
112 instances of the structures. Each thread should have its own
113 png_struct and png_info instances, and thus its own image.
114 Libpng does not protect itself against two threads using the
115 same instance of a structure.
119 There are two main structures that are important to libpng, png_struct
120 and png_info. Both are internal structures that are no longer exposed
121 in the libpng interface (as of libpng 1.5.0).
123 The png_info structure is designed to provide information about the
124 PNG file. At one time, the fields of png_info were intended to be
125 directly accessible to the user. However, this tended to cause problems
126 with applications using dynamically loaded libraries, and as a result
127 a set of interface functions for png_info (the png_get_*() and png_set_*()
128 functions) was developed, and direct access to the png_info fields was
131 The png_struct structure is the object used by the library to decode a
132 single image. As of 1.5.0 this structure is also not exposed.
134 Almost all libpng APIs require a pointer to a png_struct as the first argument.
135 Many (in particular the png_set and png_get APIs) also require a pointer
136 to png_info as the second argument. Some application visible macros
137 defined in png.h designed for basic data access (reading and writing
138 integers in the PNG format) don't take a png_info pointer, but it's almost
139 always safe to assume that a (png_struct*) has to be passed to call an API
142 You can have more than one png_info structure associated with an image,
143 as illustrated in pngtest.c, one for information valid prior to the
144 IDAT chunks and another (called "end_info" below) for things after them.
146 The png.h header file is an invaluable reference for programming with libpng.
147 And while I'm on the topic, make sure you include the libpng header file:
151 and also (as of libpng-1.5.0) the zlib header file, if you need it:
157 The png.h header file defines a number of integral types used by the
158 APIs. Most of these are fairly obvious; for example types corresponding
159 to integers of particular sizes and types for passing color values.
161 One exception is how non-integral numbers are handled. For application
162 convenience most APIs that take such numbers have C (double) arguments;
163 however, internally PNG, and libpng, use 32 bit signed integers and encode
164 the value by multiplying by 100,000. As of libpng 1.5.0 a convenience
165 macro PNG_FP_1 is defined in png.h along with a type (png_fixed_point)
166 which is simply (png_int_32).
168 All APIs that take (double) arguments also have a matching API that
169 takes the corresponding fixed point integer arguments. The fixed point
170 API has the same name as the floating point one with "_fixed" appended.
171 The actual range of values permitted in the APIs is frequently less than
172 the full range of (png_fixed_point) (-21474 to +21474). When APIs require
173 a non-negative argument the type is recorded as png_uint_32 above. Consult
174 the header file and the text below for more information.
176 Special care must be take with sCAL chunk handling because the chunk itself
177 uses non-integral values encoded as strings containing decimal floating point
178 numbers. See the comments in the header file.
182 The main header file function declarations are frequently protected by C
183 preprocessing directives of the form:
185 #ifdef PNG_feature_SUPPORTED
189 #ifdef PNG_feature_SUPPORTED
193 The library can be built without support for these APIs, although a
194 standard build will have all implemented APIs. Application programs
195 should check the feature macros before using an API for maximum
196 portability. From libpng 1.5.0 the feature macros set during the build
197 of libpng are recorded in the header file "pnglibconf.h" and this file
198 is always included by png.h.
200 If you don't need to change the library configuration from the default, skip to
201 the next section ("Reading").
203 Notice that some of the makefiles in the 'scripts' directory and (in 1.5.0) all
204 of the build project files in the 'projects' directory simply copy
205 scripts/pnglibconf.h.prebuilt to pnglibconf.h. This means that these build
206 systems do not permit easy auto-configuration of the library - they only
207 support the default configuration.
209 The easiest way to make minor changes to the libpng configuration when
210 auto-configuration is supported is to add definitions to the command line
211 using (typically) CPPFLAGS. For example:
213 CPPFLAGS=-DPNG_NO_FLOATING_ARITHMETIC
215 will change the internal libpng math implementation for gamma correction and
216 other arithmetic calculations to fixed point, avoiding the need for fast
217 floating point support. The result can be seen in the generated pnglibconf.h -
218 make sure it contains the changed feature macro setting.
220 If you need to make more extensive configuration changes - more than one or two
221 feature macro settings - you can either add -DPNG_USER_CONFIG to the build
222 command line and put a list of feature macro settings in pngusr.h or you can set
223 DFA_XTRA (a makefile variable) to a file containing the same information in the
224 form of 'option' settings.
226 A. Changing pnglibconf.h
228 A variety of methods exist to build libpng. Not all of these support
229 reconfiguration of pnglibconf.h. To reconfigure pnglibconf.h it must either be
230 rebuilt from scripts/pnglibconf.dfa using awk or it must be edited by hand.
232 Hand editing is achieved by copying scripts/pnglibconf.h.prebuilt to
233 pnglibconf.h and changing the lines defining the supported features, paying
234 very close attention to the 'option' information in scripts/pnglibconf.dfa
235 that describes those features and their requirements. This is easy to get
238 B. Configuration using DFA_XTRA
240 Rebuilding from pnglibconf.dfa is easy if a functioning 'awk', or a later
241 variant such as 'nawk' or 'gawk', is available. The configure build will
242 automatically find an appropriate awk and build pnglibconf.h.
243 The scripts/pnglibconf.mak file contains a set of make rules for doing the
244 same thing if configure is not used, and many of the makefiles in the scripts
245 directory use this approach.
247 When rebuilding simply write a new file containing changed options and set
248 DFA_XTRA to the name of this file. This causes the build to append the new file
249 to the end of scripts/pnglibconf.dfa. The pngusr.dfa file should contain lines
250 of the following forms:
254 This turns all optional features off. Include it at the start of pngusr.dfa to
255 make it easier to build a minimal configuration. You will need to turn at least
256 some features on afterward to enable either reading or writing code, or both.
261 Enable or disable a single feature. This will automatically enable other
262 features required by a feature that is turned on or disable other features that
263 require a feature which is turned off. Conflicting settings will cause an error
264 message to be emitted by awk.
266 setting feature default value
268 Changes the default value of setting 'feature' to 'value'. There are a small
269 number of settings listed at the top of pnglibconf.h, they are documented in the
270 source code. Most of these values have performance implications for the library
271 but most of them have no visible effect on the API. Some can also be overridden
274 This method of building a customized pnglibconf.h is illustrated in
275 contrib/pngminim/*. See the "$(PNGCONF):" target in the makefile and
276 pngusr.dfa in these directories.
278 C. Configuration using PNG_USER_CONFIG
280 If -DPNG_USER_CONFIG is added to the CPPFLAGS when pnglibconf.h is built,
281 the file pngusr.h will automatically be included before the options in
282 scripts/pnglibconf.dfa are processed. Your pngusr.h file should contain only
283 macro definitions turning features on or off or setting settings.
285 Apart from the global setting "everything = off" all the options listed above
286 can be set using macros in pngusr.h:
288 #define PNG_feature_SUPPORTED
294 #define PNG_NO_feature
300 #define PNG_feature value
304 setting feature default value
306 Notice that in both cases, pngusr.dfa and pngusr.h, the contents of the
307 pngusr file you supply override the contents of scripts/pnglibconf.dfa
309 If confusing or incomprehensible behavior results it is possible to
310 examine the intermediate file pnglibconf.dfn to find the full set of
311 dependency information for each setting and option. Simply locate the
312 feature in the file and read the C comments that precede it.
314 This method is also illustrated in the contrib/pngminim/* makefiles and
319 We'll now walk you through the possible functions to call when reading
320 in a PNG file sequentially, briefly explaining the syntax and purpose
321 of each one. See example.c and png.h for more detail. While
322 progressive reading is covered in the next section, you will still
323 need some of the functions discussed in this section to read a PNG
328 You will want to do the I/O initialization(*) before you get into libpng,
329 so if it doesn't work, you don't have much to undo. Of course, you
330 will also want to insure that you are, in fact, dealing with a PNG
331 file. Libpng provides a simple check to see if a file is a PNG file.
332 To use it, pass in the first 1 to 8 bytes of the file to the function
333 png_sig_cmp(), and it will return 0 (false) if the bytes match the
334 corresponding bytes of the PNG signature, or nonzero (true) otherwise.
335 Of course, the more bytes you pass in, the greater the accuracy of the
338 If you are intending to keep the file pointer open for use in libpng,
339 you must ensure you don't read more than 8 bytes from the beginning
340 of the file, and you also have to make a call to png_set_sig_bytes()
341 with the number of bytes you read from the beginning. Libpng will
342 then only check the bytes (if any) that your program didn't read.
344 (*): If you are not using the standard I/O functions, you will need
345 to replace them with custom functions. See the discussion under
348 FILE *fp = fopen(file_name, "rb");
354 if (fread(header, 1, number, fp) != number)
359 is_png = !png_sig_cmp(header, 0, number);
365 Next, png_struct and png_info need to be allocated and initialized. In
366 order to ensure that the size of these structures is correct even with a
367 dynamically linked libpng, there are functions to initialize and
368 allocate the structures. We also pass the library version, optional
369 pointers to error handling functions, and a pointer to a data struct for
370 use by the error functions, if necessary (the pointer and functions can
371 be NULL if the default error handlers are to be used). See the section
372 on Changes to Libpng below regarding the old initialization functions.
373 The structure allocation functions quietly return NULL if they fail to
374 create the structure, so your application should check for that.
376 png_structp png_ptr = png_create_read_struct
377 (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
378 user_error_fn, user_warning_fn);
383 png_infop info_ptr = png_create_info_struct(png_ptr);
387 png_destroy_read_struct(&png_ptr,
388 (png_infopp)NULL, (png_infopp)NULL);
392 If you want to use your own memory allocation routines,
393 use a libpng that was built with PNG_USER_MEM_SUPPORTED defined, and use
394 png_create_read_struct_2() instead of png_create_read_struct():
396 png_structp png_ptr = png_create_read_struct_2
397 (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
398 user_error_fn, user_warning_fn, (png_voidp)
399 user_mem_ptr, user_malloc_fn, user_free_fn);
401 The error handling routines passed to png_create_read_struct()
402 and the memory alloc/free routines passed to png_create_struct_2()
403 are only necessary if you are not using the libpng supplied error
404 handling and memory alloc/free functions.
406 When libpng encounters an error, it expects to longjmp back
407 to your routine. Therefore, you will need to call setjmp and pass
408 your png_jmpbuf(png_ptr). If you read the file from different
409 routines, you will need to update the longjmp buffer every time you enter
410 a new routine that will call a png_*() function.
412 See your documentation of setjmp/longjmp for your compiler for more
413 information on setjmp/longjmp. See the discussion on libpng error
414 handling in the Customizing Libpng section below for more information
415 on the libpng error handling. If an error occurs, and libpng longjmp's
416 back to your setjmp, you will want to call png_destroy_read_struct() to
419 if (setjmp(png_jmpbuf(png_ptr)))
421 png_destroy_read_struct(&png_ptr, &info_ptr,
427 Pass (png_infopp)NULL instead of &end_info if you didn't create
428 an end_info structure.
430 If you would rather avoid the complexity of setjmp/longjmp issues,
431 you can compile libpng with PNG_NO_SETJMP, in which case
432 errors will result in a call to PNG_ABORT() which defaults to abort().
434 You can #define PNG_ABORT() to a function that does something
435 more useful than abort(), as long as your function does not
438 Now you need to set up the input code. The default for libpng is to
439 use the C function fread(). If you use this, you will need to pass a
440 valid FILE * in the function png_init_io(). Be sure that the file is
441 opened in binary mode. If you wish to handle reading data in another
442 way, you need not call the png_init_io() function, but you must then
443 implement the libpng I/O methods discussed in the Customizing Libpng
446 png_init_io(png_ptr, fp);
448 If you had previously opened the file and read any of the signature from
449 the beginning in order to see if this was a PNG file, you need to let
450 libpng know that there are some bytes missing from the start of the file.
452 png_set_sig_bytes(png_ptr, number);
454 You can change the zlib compression buffer size to be used while
455 reading compressed data with
457 png_set_compression_buffer_size(png_ptr, buffer_size);
459 where the default size is 8192 bytes. Note that the buffer size
460 is changed immediately and the buffer is reallocated immediately,
461 instead of setting a flag to be acted upon later.
463 If you want CRC errors to be handled in a different manner than
466 png_set_crc_action(png_ptr, crit_action, ancil_action);
468 The values for png_set_crc_action() say how libpng is to handle CRC errors in
469 ancillary and critical chunks, and whether to use the data contained
470 therein. Starting with libpng-1.6.26, this also governs how an ADLER32 error
471 is handled while reading the IDAT chunk. Note that it is impossible to
472 "discard" data in a critical chunk.
474 Choices for (int) crit_action are
475 PNG_CRC_DEFAULT 0 error/quit
476 PNG_CRC_ERROR_QUIT 1 error/quit
477 PNG_CRC_WARN_USE 3 warn/use data
478 PNG_CRC_QUIET_USE 4 quiet/use data
479 PNG_CRC_NO_CHANGE 5 use the current value
481 Choices for (int) ancil_action are
482 PNG_CRC_DEFAULT 0 error/quit
483 PNG_CRC_ERROR_QUIT 1 error/quit
484 PNG_CRC_WARN_DISCARD 2 warn/discard data
485 PNG_CRC_WARN_USE 3 warn/use data
486 PNG_CRC_QUIET_USE 4 quiet/use data
487 PNG_CRC_NO_CHANGE 5 use the current value
489 When the setting for crit_action is PNG_CRC_QUIET_USE, the CRC and ADLER32
490 checksums are not only ignored, but they are not evaluated.
492 Setting up callback code
494 You can set up a callback function to handle any unknown chunks in the
495 input stream. You must supply the function
497 read_chunk_callback(png_structp png_ptr,
498 png_unknown_chunkp chunk);
500 /* The unknown chunk structure contains your
501 chunk data, along with similar data for any other
508 /* Note that libpng has already taken care of
511 /* put your code here. Search for your chunk in the
512 unknown chunk structure, process it, and return one
515 return (-n); /* chunk had an error */
516 return (0); /* did not recognize */
517 return (n); /* success */
520 (You can give your function another name that you like instead of
521 "read_chunk_callback")
523 To inform libpng about your function, use
525 png_set_read_user_chunk_fn(png_ptr, user_chunk_ptr,
526 read_chunk_callback);
528 This names not only the callback function, but also a user pointer that
529 you can retrieve with
531 png_get_user_chunk_ptr(png_ptr);
533 If you call the png_set_read_user_chunk_fn() function, then all unknown
534 chunks which the callback does not handle will be saved when read. You can
535 cause them to be discarded by returning '1' ("handled") instead of '0'. This
536 behavior will change in libpng 1.7 and the default handling set by the
537 png_set_keep_unknown_chunks() function, described below, will be used when the
538 callback returns 0. If you want the existing behavior you should set the global
539 default to PNG_HANDLE_CHUNK_IF_SAFE now; this is compatible with all current
540 versions of libpng and with 1.7. Libpng 1.6 issues a warning if you keep the
541 default, or PNG_HANDLE_CHUNK_NEVER, and the callback returns 0.
543 At this point, you can set up a callback function that will be
544 called after each row has been read, which you can use to control
545 a progress meter or the like. It's demonstrated in pngtest.c.
546 You must supply a function
548 void read_row_callback(png_structp png_ptr,
549 png_uint_32 row, int pass);
551 /* put your code here */
554 (You can give it another name that you like instead of "read_row_callback")
556 To inform libpng about your function, use
558 png_set_read_status_fn(png_ptr, read_row_callback);
560 When this function is called the row has already been completely processed and
561 the 'row' and 'pass' refer to the next row to be handled. For the
562 non-interlaced case the row that was just handled is simply one less than the
563 passed in row number, and pass will always be 0. For the interlaced case the
564 same applies unless the row value is 0, in which case the row just handled was
565 the last one from one of the preceding passes. Because interlacing may skip a
566 pass you cannot be sure that the preceding pass is just 'pass-1'; if you really
567 need to know what the last pass is record (row,pass) from the callback and use
568 the last recorded value each time.
570 As with the user transform you can find the output row using the
571 PNG_ROW_FROM_PASS_ROW macro.
573 Unknown-chunk handling
575 Now you get to set the way the library processes unknown chunks in the
576 input PNG stream. Both known and unknown chunks will be read. Normal
577 behavior is that known chunks will be parsed into information in
578 various info_ptr members while unknown chunks will be discarded. This
579 behavior can be wasteful if your application will never use some known
580 chunk types. To change this, you can call:
582 png_set_keep_unknown_chunks(png_ptr, keep,
583 chunk_list, num_chunks);
585 keep - 0: default unknown chunk handling
586 1: ignore; do not keep
587 2: keep only if safe-to-copy
588 3: keep even if unsafe-to-copy
590 You can use these definitions:
591 PNG_HANDLE_CHUNK_AS_DEFAULT 0
592 PNG_HANDLE_CHUNK_NEVER 1
593 PNG_HANDLE_CHUNK_IF_SAFE 2
594 PNG_HANDLE_CHUNK_ALWAYS 3
596 chunk_list - list of chunks affected (a byte string,
597 five bytes per chunk, NULL or '\0' if
598 num_chunks is positive; ignored if
601 num_chunks - number of chunks affected; if 0, all
602 unknown chunks are affected. If positive,
603 only the chunks in the list are affected,
604 and if negative all unknown chunks and
605 all known chunks except for the IHDR,
606 PLTE, tRNS, IDAT, and IEND chunks are
609 Unknown chunks declared in this way will be saved as raw data onto a
610 list of png_unknown_chunk structures. If a chunk that is normally
611 known to libpng is named in the list, it will be handled as unknown,
612 according to the "keep" directive. If a chunk is named in successive
613 instances of png_set_keep_unknown_chunks(), the final instance will
614 take precedence. The IHDR and IEND chunks should not be named in
615 chunk_list; if they are, libpng will process them normally anyway.
616 If you know that your application will never make use of some particular
617 chunks, use PNG_HANDLE_CHUNK_NEVER (or 1) as demonstrated below.
619 Here is an example of the usage of png_set_keep_unknown_chunks(),
620 where the private "vpAg" chunk will later be processed by a user chunk
623 png_byte vpAg[5]={118, 112, 65, 103, (png_byte) '\0'};
625 #if defined(PNG_UNKNOWN_CHUNKS_SUPPORTED)
626 png_byte unused_chunks[]=
628 104, 73, 83, 84, (png_byte) '\0', /* hIST */
629 105, 84, 88, 116, (png_byte) '\0', /* iTXt */
630 112, 67, 65, 76, (png_byte) '\0', /* pCAL */
631 115, 67, 65, 76, (png_byte) '\0', /* sCAL */
632 115, 80, 76, 84, (png_byte) '\0', /* sPLT */
633 116, 73, 77, 69, (png_byte) '\0', /* tIME */
639 #if defined(PNG_UNKNOWN_CHUNKS_SUPPORTED)
640 /* ignore all unknown chunks
641 * (use global setting "2" for libpng16 and earlier):
643 png_set_keep_unknown_chunks(read_ptr, 2, NULL, 0);
645 /* except for vpAg: */
646 png_set_keep_unknown_chunks(read_ptr, 2, vpAg, 1);
648 /* also ignore unused known chunks: */
649 png_set_keep_unknown_chunks(read_ptr, 1, unused_chunks,
650 (int)(sizeof unused_chunks)/5);
655 The PNG specification allows the width and height of an image to be as
656 large as 2^31-1 (0x7fffffff), or about 2.147 billion rows and columns.
657 For safety, libpng imposes a default limit of 1 million rows and columns.
658 Larger images will be rejected immediately with a png_error() call. If
659 you wish to change these limits, you can use
661 png_set_user_limits(png_ptr, width_max, height_max);
663 to set your own limits (libpng may reject some very wide images
664 anyway because of potential buffer overflow conditions).
666 You should put this statement after you create the PNG structure and
667 before calling png_read_info(), png_read_png(), or png_process_data().
669 When writing a PNG datastream, put this statement before calling
670 png_write_info() or png_write_png().
672 If you need to retrieve the limits that are being applied, use
674 width_max = png_get_user_width_max(png_ptr);
675 height_max = png_get_user_height_max(png_ptr);
677 The PNG specification sets no limit on the number of ancillary chunks
678 allowed in a PNG datastream. By default, libpng imposes a limit of
679 a total of 1000 sPLT, tEXt, iTXt, zTXt, and unknown chunks to be stored.
680 If you have set up both info_ptr and end_info_ptr, the limit applies
681 separately to each. You can change the limit on the total number of such
682 chunks that will be stored, with
684 png_set_chunk_cache_max(png_ptr, user_chunk_cache_max);
686 where 0x7fffffffL means unlimited. You can retrieve this limit with
688 chunk_cache_max = png_get_chunk_cache_max(png_ptr);
690 Libpng imposes a limit of 8 Megabytes (8,000,000 bytes) on the amount of
691 memory that any chunk other than IDAT can occupy, originally or when
692 decompressed (prior to libpng-1.6.32 the limit was only applied to compressed
693 chunks after decompression). You can change this limit with
695 png_set_chunk_malloc_max(png_ptr, user_chunk_malloc_max);
697 and you can retrieve the limit with
699 chunk_malloc_max = png_get_chunk_malloc_max(png_ptr);
701 Any chunks that would cause either of these limits to be exceeded will
704 Information about your system
706 If you intend to display the PNG or to incorporate it in other image data you
707 need to tell libpng information about your display or drawing surface so that
708 libpng can convert the values in the image to match the display.
710 From libpng-1.5.4 this information can be set before reading the PNG file
711 header. In earlier versions png_set_gamma() existed but behaved incorrectly if
712 called before the PNG file header had been read and png_set_alpha_mode() did not
715 If you need to support versions prior to libpng-1.5.4 test the version number
716 as illustrated below using "PNG_LIBPNG_VER >= 10504" and follow the procedures
717 described in the appropriate manual page.
719 You give libpng the encoding expected by your system expressed as a 'gamma'
720 value. You can also specify a default encoding for the PNG file in
721 case the required information is missing from the file. By default libpng
722 assumes that the PNG data matches your system, to keep this default call:
724 png_set_gamma(png_ptr, screen_gamma, output_gamma);
726 or you can use the fixed point equivalent:
728 png_set_gamma_fixed(png_ptr, PNG_FP_1*screen_gamma,
729 PNG_FP_1*output_gamma);
731 If you don't know the gamma for your system it is probably 2.2 - a good
732 approximation to the IEC standard for display systems (sRGB). If images are
733 too contrasty or washed out you got the value wrong - check your system
736 Many systems permit the system gamma to be changed via a lookup table in the
737 display driver, a few systems, including older Macs, change the response by
738 default. As of 1.5.4 three special values are available to handle common
741 PNG_DEFAULT_sRGB: Indicates that the system conforms to the
742 IEC 61966-2-1 standard. This matches almost
744 PNG_GAMMA_MAC_18: Indicates that the system is an older
745 (pre Mac OS 10.6) Apple Macintosh system with
746 the default settings.
747 PNG_GAMMA_LINEAR: Just the fixed point value for 1.0 - indicates
748 that the system expects data with no gamma
751 You would use the linear (unencoded) value if you need to process the pixel
752 values further because this avoids the need to decode and re-encode each
753 component value whenever arithmetic is performed. A lot of graphics software
754 uses linear values for this reason, often with higher precision component values
755 to preserve overall accuracy.
758 The output_gamma value expresses how to decode the output values, not how
759 they are encoded. The values used correspond to the normal numbers used to
760 describe the overall gamma of a computer display system; for example 2.2 for
761 an sRGB conformant system. The values are scaled by 100000 in the _fixed
762 version of the API (so 220000 for sRGB.)
764 The inverse of the value is always used to provide a default for the PNG file
765 encoding if it has no gAMA chunk and if png_set_gamma() has not been called
766 to override the PNG gamma information.
768 When the ALPHA_OPTIMIZED mode is selected the output gamma is used to encode
769 opaque pixels however pixels with lower alpha values are not encoded,
770 regardless of the output gamma setting.
772 When the standard Porter Duff handling is requested with mode 1 the output
773 encoding is set to be linear and the output_gamma value is only relevant
774 as a default for input data that has no gamma information. The linear output
775 encoding will be overridden if png_set_gamma() is called - the results may be
778 The following numbers are derived from the sRGB standard and the research
779 behind it. sRGB is defined to be approximated by a PNG gAMA chunk value of
780 0.45455 (1/2.2) for PNG. The value implicitly includes any viewing
781 correction required to take account of any differences in the color
782 environment of the original scene and the intended display environment; the
783 value expresses how to *decode* the image for display, not how the original
786 sRGB provides a peg for the PNG standard by defining a viewing environment.
787 sRGB itself, and earlier TV standards, actually use a more complex transform
788 (a linear portion then a gamma 2.4 power law) than PNG can express. (PNG is
789 limited to simple power laws.) By saying that an image for direct display on
790 an sRGB conformant system should be stored with a gAMA chunk value of 45455
791 (11.3.3.2 and 11.3.3.5 of the ISO PNG specification) the PNG specification
792 makes it possible to derive values for other display systems and
795 The Mac value is deduced from the sRGB based on an assumption that the actual
796 extra viewing correction used in early Mac display systems was implemented as
797 a power 1.45 lookup table.
799 Any system where a programmable lookup table is used or where the behavior of
800 the final display device characteristics can be changed requires system
801 specific code to obtain the current characteristic. However this can be
802 difficult and most PNG gamma correction only requires an approximate value.
804 By default, if png_set_alpha_mode() is not called, libpng assumes that all
805 values are unencoded, linear, values and that the output device also has a
806 linear characteristic. This is only very rarely correct - it is invariably
807 better to call png_set_alpha_mode() with PNG_DEFAULT_sRGB than rely on the
808 default if you don't know what the right answer is!
810 The special value PNG_GAMMA_MAC_18 indicates an older Mac system (pre Mac OS
811 10.6) which used a correction table to implement a somewhat lower gamma on an
812 otherwise sRGB system.
814 Both these values are reserved (not simple gamma values) in order to allow
815 more precise correction internally in the future.
817 NOTE: the values can be passed to either the fixed or floating
818 point APIs, but the floating point API will also accept floating point
821 The second thing you may need to tell libpng about is how your system handles
822 alpha channel information. Some, but not all, PNG files contain an alpha
823 channel. To display these files correctly you need to compose the data onto a
824 suitable background, as described in the PNG specification.
826 Libpng only supports composing onto a single color (using png_set_background;
827 see below). Otherwise you must do the composition yourself and, in this case,
828 you may need to call png_set_alpha_mode:
830 #if PNG_LIBPNG_VER >= 10504
831 png_set_alpha_mode(png_ptr, mode, screen_gamma);
833 png_set_gamma(png_ptr, screen_gamma, 1.0/screen_gamma);
836 The screen_gamma value is the same as the argument to png_set_gamma; however,
837 how it affects the output depends on the mode. png_set_alpha_mode() sets the
838 file gamma default to 1/screen_gamma, so normally you don't need to call
839 png_set_gamma. If you need different defaults call png_set_gamma() before
840 png_set_alpha_mode() - if you call it after it will override the settings made
841 by png_set_alpha_mode().
843 The mode is as follows:
845 PNG_ALPHA_PNG: The data is encoded according to the PNG
846 specification. Red, green and blue, or gray, components are
847 gamma encoded color values and are not premultiplied by the
848 alpha value. The alpha value is a linear measure of the
849 contribution of the pixel to the corresponding final output pixel.
851 You should normally use this format if you intend to perform
852 color correction on the color values; most, maybe all, color
853 correction software has no handling for the alpha channel and,
854 anyway, the math to handle pre-multiplied component values is
855 unnecessarily complex.
857 Before you do any arithmetic on the component values you need
858 to remove the gamma encoding and multiply out the alpha
859 channel. See the PNG specification for more detail. It is
860 important to note that when an image with an alpha channel is
861 scaled, linear encoded, pre-multiplied component values must
864 The remaining modes assume you don't need to do any further color correction or
865 that if you do, your color correction software knows all about alpha (it
866 probably doesn't!). They 'associate' the alpha with the color information by
867 storing color channel values that have been scaled by the alpha. The
868 advantage is that the color channels can be resampled (the image can be
869 scaled) in this form. The disadvantage is that normal practice is to store
870 linear, not (gamma) encoded, values and this requires 16-bit channels for
871 still images rather than the 8-bit channels that are just about sufficient if
872 gamma encoding is used. In addition all non-transparent pixel values,
873 including completely opaque ones, must be gamma encoded to produce the final
874 image. These are the 'STANDARD', 'ASSOCIATED' or 'PREMULTIPLIED' modes
875 described below (the latter being the two common names for associated alpha
876 color channels). Note that PNG files always contain non-associated color
877 channels; png_set_alpha_mode() with one of the modes causes the decoder to
878 convert the pixels to an associated form before returning them to your
881 Since it is not necessary to perform arithmetic on opaque color values so
882 long as they are not to be resampled and are in the final color space it is
883 possible to optimize the handling of alpha by storing the opaque pixels in
884 the PNG format (adjusted for the output color space) while storing partially
885 opaque pixels in the standard, linear, format. The accuracy required for
886 standard alpha composition is relatively low, because the pixels are
887 isolated, therefore typically the accuracy loss in storing 8-bit linear
888 values is acceptable. (This is not true if the alpha channel is used to
889 simulate transparency over large areas - use 16 bits or the PNG mode in
890 this case!) This is the 'OPTIMIZED' mode. For this mode a pixel is
891 treated as opaque only if the alpha value is equal to the maximum value.
893 PNG_ALPHA_STANDARD: The data libpng produces is encoded in the
894 standard way assumed by most correctly written graphics software.
895 The gamma encoding will be removed by libpng and the
896 linear component values will be pre-multiplied by the
899 With this format the final image must be re-encoded to
900 match the display gamma before the image is displayed.
901 If your system doesn't do that, yet still seems to
902 perform arithmetic on the pixels without decoding them,
903 it is broken - check out the modes below.
905 With PNG_ALPHA_STANDARD libpng always produces linear
906 component values, whatever screen_gamma you supply. The
907 screen_gamma value is, however, used as a default for
908 the file gamma if the PNG file has no gamma information.
910 If you call png_set_gamma() after png_set_alpha_mode() you
911 will override the linear encoding. Instead the
912 pre-multiplied pixel values will be gamma encoded but
913 the alpha channel will still be linear. This may
914 actually match the requirements of some broken software,
917 While linear 8-bit data is often used it has
918 insufficient precision for any image with a reasonable
919 dynamic range. To avoid problems, and if your software
920 supports it, use png_set_expand_16() to force all
921 components to 16 bits.
923 PNG_ALPHA_OPTIMIZED: This mode is the same as PNG_ALPHA_STANDARD
924 except that completely opaque pixels are gamma encoded according to
925 the screen_gamma value. Pixels with alpha less than 1.0
926 will still have linear components.
928 Use this format if you have control over your
929 compositing software and so don't do other arithmetic
930 (such as scaling) on the data you get from libpng. Your
931 compositing software can simply copy opaque pixels to
932 the output but still has linear values for the
935 In normal compositing, where the alpha channel encodes
936 partial pixel coverage (as opposed to broad area
937 translucency), the inaccuracies of the 8-bit
938 representation of non-opaque pixels are irrelevant.
940 You can also try this format if your software is broken;
941 it might look better.
943 PNG_ALPHA_BROKEN: This is PNG_ALPHA_STANDARD; however, all component
944 values, including the alpha channel are gamma encoded. This is
945 broken because, in practice, no implementation that uses this choice
946 correctly undoes the encoding before handling alpha composition. Use this
947 choice only if other serious errors in the software or hardware you use
948 mandate it. In most cases of broken software or hardware the bug in the
949 final display manifests as a subtle halo around composited parts of the
950 image. You may not even perceive this as a halo; the composited part of
951 the image may simply appear separate from the background, as though it had
952 been cut out of paper and pasted on afterward.
954 If you don't have to deal with bugs in software or hardware, or if you can fix
955 them, there are three recommended ways of using png_set_alpha_mode():
957 png_set_alpha_mode(png_ptr, PNG_ALPHA_PNG,
960 You can do color correction on the result (libpng does not currently
961 support color correction internally). When you handle the alpha channel
962 you need to undo the gamma encoding and multiply out the alpha.
964 png_set_alpha_mode(png_ptr, PNG_ALPHA_STANDARD,
966 png_set_expand_16(png_ptr);
968 If you are using the high level interface, don't call png_set_expand_16();
969 instead pass PNG_TRANSFORM_EXPAND_16 to the interface.
971 With this mode you can't do color correction, but you can do arithmetic,
972 including composition and scaling, on the data without further processing.
974 png_set_alpha_mode(png_ptr, PNG_ALPHA_OPTIMIZED,
977 You can avoid the expansion to 16-bit components with this mode, but you
978 lose the ability to scale the image or perform other linear arithmetic.
979 All you can do is compose the result onto a matching output. Since this
980 mode is libpng-specific you also need to write your own composition
983 The following are examples of calls to png_set_alpha_mode to achieve the
984 required overall gamma correction and, where necessary, alpha
987 png_set_alpha_mode(pp, PNG_ALPHA_PNG, PNG_DEFAULT_sRGB);
989 This is the default libpng handling of the alpha channel - it is not
990 pre-multiplied into the color components. In addition the call states
991 that the output is for a sRGB system and causes all PNG files without gAMA
992 chunks to be assumed to be encoded using sRGB.
994 png_set_alpha_mode(pp, PNG_ALPHA_PNG, PNG_GAMMA_MAC);
996 In this case the output is assumed to be something like an sRGB conformant
997 display preceeded by a power-law lookup table of power 1.45. This is how
998 early Mac systems behaved.
1000 png_set_alpha_mode(pp, PNG_ALPHA_STANDARD, PNG_GAMMA_LINEAR);
1002 This is the classic Jim Blinn approach and will work in academic
1003 environments where everything is done by the book. It has the shortcoming
1004 of assuming that input PNG data with no gamma information is linear - this
1005 is unlikely to be correct unless the PNG files where generated locally.
1006 Most of the time the output precision will be so low as to show
1007 significant banding in dark areas of the image.
1009 png_set_expand_16(pp);
1010 png_set_alpha_mode(pp, PNG_ALPHA_STANDARD, PNG_DEFAULT_sRGB);
1012 This is a somewhat more realistic Jim Blinn inspired approach. PNG files
1013 are assumed to have the sRGB encoding if not marked with a gamma value and
1014 the output is always 16 bits per component. This permits accurate scaling
1015 and processing of the data. If you know that your input PNG files were
1016 generated locally you might need to replace PNG_DEFAULT_sRGB with the
1017 correct value for your system.
1019 png_set_alpha_mode(pp, PNG_ALPHA_OPTIMIZED, PNG_DEFAULT_sRGB);
1021 If you just need to composite the PNG image onto an existing background
1022 and if you control the code that does this you can use the optimization
1023 setting. In this case you just copy completely opaque pixels to the
1024 output. For pixels that are not completely transparent (you just skip
1025 those) you do the composition math using png_composite or png_composite_16
1026 below then encode the resultant 8-bit or 16-bit values to match the output
1031 If neither the PNG nor the standard linear encoding work for you because
1032 of the software or hardware you use then you have a big problem. The PNG
1033 case will probably result in halos around the image. The linear encoding
1034 will probably result in a washed out, too bright, image (it's actually too
1035 contrasty.) Try the ALPHA_OPTIMIZED mode above - this will probably
1036 substantially reduce the halos. Alternatively try:
1038 png_set_alpha_mode(pp, PNG_ALPHA_BROKEN, PNG_DEFAULT_sRGB);
1040 This option will also reduce the halos, but there will be slight dark
1041 halos round the opaque parts of the image where the background is light.
1042 In the OPTIMIZED mode the halos will be light halos where the background
1043 is dark. Take your pick - the halos are unavoidable unless you can get
1044 your hardware/software fixed! (The OPTIMIZED approach is slightly
1047 When the default gamma of PNG files doesn't match the output gamma.
1048 If you have PNG files with no gamma information png_set_alpha_mode allows
1049 you to provide a default gamma, but it also sets the ouput gamma to the
1050 matching value. If you know your PNG files have a gamma that doesn't
1051 match the output you can take advantage of the fact that
1052 png_set_alpha_mode always sets the output gamma but only sets the PNG
1053 default if it is not already set:
1055 png_set_alpha_mode(pp, PNG_ALPHA_PNG, PNG_DEFAULT_sRGB);
1056 png_set_alpha_mode(pp, PNG_ALPHA_PNG, PNG_GAMMA_MAC);
1058 The first call sets both the default and the output gamma values, the
1059 second call overrides the output gamma without changing the default. This
1060 is easier than achieving the same effect with png_set_gamma. You must use
1061 PNG_ALPHA_PNG for the first call - internal checking in png_set_alpha will
1062 fire if more than one call to png_set_alpha_mode and png_set_background is
1063 made in the same read operation, however multiple calls with PNG_ALPHA_PNG
1066 If you don't need, or can't handle, the alpha channel you can call
1067 png_set_background() to remove it by compositing against a fixed color. Don't
1068 call png_set_strip_alpha() to do this - it will leave spurious pixel values in
1069 transparent parts of this image.
1071 png_set_background(png_ptr, &background_color,
1072 PNG_BACKGROUND_GAMMA_SCREEN, 0, 1);
1074 The background_color is an RGB or grayscale value according to the data format
1075 libpng will produce for you. Because you don't yet know the format of the PNG
1076 file, if you call png_set_background at this point you must arrange for the
1077 format produced by libpng to always have 8-bit or 16-bit components and then
1078 store the color as an 8-bit or 16-bit color as appropriate. The color contains
1079 separate gray and RGB component values, so you can let libpng produce gray or
1080 RGB output according to the input format, but low bit depth grayscale images
1081 must always be converted to at least 8-bit format. (Even though low bit depth
1082 grayscale images can't have an alpha channel they can have a transparent
1085 You set the transforms you need later, either as flags to the high level
1086 interface or libpng API calls for the low level interface. For reference the
1087 settings and API calls required are:
1090 PNG_TRANSFORM_SCALE_16 | PNG_EXPAND
1091 png_set_expand(png_ptr); png_set_scale_16(png_ptr);
1093 If you must get exactly the same inaccurate results
1094 produced by default in versions prior to libpng-1.5.4,
1095 use PNG_TRANSFORM_STRIP_16 and png_set_strip_16(png_ptr)
1099 PNG_TRANSFORM_EXPAND_16
1100 png_set_expand_16(png_ptr);
1102 In either case palette image data will be expanded to RGB. If you just want
1103 color data you can add PNG_TRANSFORM_GRAY_TO_RGB or png_set_gray_to_rgb(png_ptr)
1106 Calling png_set_background before the PNG file header is read will not work
1107 prior to libpng-1.5.4. Because the failure may result in unexpected warnings or
1108 errors it is therefore much safer to call png_set_background after the head has
1109 been read. Unfortunately this means that prior to libpng-1.5.4 it cannot be
1110 used with the high level interface.
1112 The high-level read interface
1114 At this point there are two ways to proceed; through the high-level
1115 read interface, or through a sequence of low-level read operations.
1116 You can use the high-level interface if (a) you are willing to read
1117 the entire image into memory, and (b) the input transformations
1118 you want to do are limited to the following set:
1120 PNG_TRANSFORM_IDENTITY No transformation
1121 PNG_TRANSFORM_SCALE_16 Strip 16-bit samples to
1123 PNG_TRANSFORM_STRIP_16 Chop 16-bit samples to
1124 8-bit less accurately
1125 PNG_TRANSFORM_STRIP_ALPHA Discard the alpha channel
1126 PNG_TRANSFORM_PACKING Expand 1, 2 and 4-bit
1128 PNG_TRANSFORM_PACKSWAP Change order of packed
1130 PNG_TRANSFORM_EXPAND Perform set_expand()
1131 PNG_TRANSFORM_INVERT_MONO Invert monochrome images
1132 PNG_TRANSFORM_SHIFT Normalize pixels to the
1134 PNG_TRANSFORM_BGR Flip RGB to BGR, RGBA
1136 PNG_TRANSFORM_SWAP_ALPHA Flip RGBA to ARGB or GA
1138 PNG_TRANSFORM_INVERT_ALPHA Change alpha from opacity
1140 PNG_TRANSFORM_SWAP_ENDIAN Byte-swap 16-bit samples
1141 PNG_TRANSFORM_GRAY_TO_RGB Expand grayscale samples
1142 to RGB (or GA to RGBA)
1143 PNG_TRANSFORM_EXPAND_16 Expand samples to 16 bits
1145 (This excludes setting a background color, doing gamma transformation,
1146 quantizing, and setting filler.) If this is the case, simply do this:
1148 png_read_png(png_ptr, info_ptr, png_transforms, NULL)
1150 where png_transforms is an integer containing the bitwise OR of some
1151 set of transformation flags. This call is equivalent to png_read_info(),
1152 followed the set of transformations indicated by the transform mask,
1153 then png_read_image(), and finally png_read_end().
1155 (The final parameter of this call is not yet used. Someday it might point
1156 to transformation parameters required by some future input transform.)
1158 You must use png_transforms and not call any png_set_transform() functions
1159 when you use png_read_png().
1161 After you have called png_read_png(), you can retrieve the image data
1164 row_pointers = png_get_rows(png_ptr, info_ptr);
1166 where row_pointers is an array of pointers to the pixel data for each row:
1168 png_bytep row_pointers[height];
1170 If you know your image size and pixel size ahead of time, you can allocate
1171 row_pointers prior to calling png_read_png() with
1173 if (height > PNG_UINT_32_MAX/(sizeof (png_byte)))
1175 "Image is too tall to process in memory");
1177 if (width > PNG_UINT_32_MAX/pixel_size)
1179 "Image is too wide to process in memory");
1181 row_pointers = png_malloc(png_ptr,
1182 height*(sizeof (png_bytep)));
1184 for (int i=0; i<height, i++)
1185 row_pointers[i]=NULL; /* security precaution */
1187 for (int i=0; i<height, i++)
1188 row_pointers[i]=png_malloc(png_ptr,
1191 png_set_rows(png_ptr, info_ptr, &row_pointers);
1193 Alternatively you could allocate your image in one big block and define
1194 row_pointers[i] to point into the proper places in your block, but first
1195 be sure that your platform is able to allocate such a large buffer:
1197 /* Guard against integer overflow */
1198 if (height > PNG_SIZE_MAX/(width*pixel_size)) {
1199 png_error(png_ptr,"image_data buffer would be too large");
1202 png_bytep buffer=png_malloc(png_ptr,height*width*pixel_size);
1204 for (int i=0; i<height, i++)
1205 row_pointers[i]=buffer+i*width*pixel_size;
1207 png_set_rows(png_ptr, info_ptr, &row_pointers);
1209 If you use png_set_rows(), the application is responsible for freeing
1210 row_pointers (and row_pointers[i], if they were separately allocated).
1212 If you don't allocate row_pointers ahead of time, png_read_png() will
1213 do it, and it'll be free'ed by libpng when you call png_destroy_*().
1215 The low-level read interface
1217 If you are going the low-level route, you are now ready to read all
1218 the file information up to the actual image data. You do this with a
1219 call to png_read_info().
1221 png_read_info(png_ptr, info_ptr);
1223 This will process all chunks up to but not including the image data.
1225 This also copies some of the data from the PNG file into the decode structure
1226 for use in later transformations. Important information copied in is:
1228 1) The PNG file gamma from the gAMA chunk. This overwrites the default value
1229 provided by an earlier call to png_set_gamma or png_set_alpha_mode.
1231 2) Prior to libpng-1.5.4 the background color from a bKGd chunk. This
1232 damages the information provided by an earlier call to png_set_background
1233 resulting in unexpected behavior. Libpng-1.5.4 no longer does this.
1235 3) The number of significant bits in each component value. Libpng uses this to
1236 optimize gamma handling by reducing the internal lookup table sizes.
1238 4) The transparent color information from a tRNS chunk. This can be modified by
1239 a later call to png_set_tRNS.
1241 Querying the info structure
1243 Functions are used to get the information from the info_ptr once it
1244 has been read. Note that these fields may not be completely filled
1245 in until png_read_end() has read the chunk data following the image.
1247 png_get_IHDR(png_ptr, info_ptr, &width, &height,
1248 &bit_depth, &color_type, &interlace_type,
1249 &compression_type, &filter_method);
1251 width - holds the width of the image
1252 in pixels (up to 2^31).
1254 height - holds the height of the image
1255 in pixels (up to 2^31).
1257 bit_depth - holds the bit depth of one of the
1258 image channels. (valid values are
1259 1, 2, 4, 8, 16 and depend also on
1260 the color_type. See also
1261 significant bits (sBIT) below).
1263 color_type - describes which color/alpha channels
1266 (bit depths 1, 2, 4, 8, 16)
1267 PNG_COLOR_TYPE_GRAY_ALPHA
1269 PNG_COLOR_TYPE_PALETTE
1270 (bit depths 1, 2, 4, 8)
1273 PNG_COLOR_TYPE_RGB_ALPHA
1276 PNG_COLOR_MASK_PALETTE
1277 PNG_COLOR_MASK_COLOR
1278 PNG_COLOR_MASK_ALPHA
1280 interlace_type - (PNG_INTERLACE_NONE or
1281 PNG_INTERLACE_ADAM7)
1283 compression_type - (must be PNG_COMPRESSION_TYPE_BASE
1286 filter_method - (must be PNG_FILTER_TYPE_BASE
1287 for PNG 1.0, and can also be
1288 PNG_INTRAPIXEL_DIFFERENCING if
1289 the PNG datastream is embedded in
1290 a MNG-1.0 datastream)
1292 Any of width, height, color_type, bit_depth,
1293 interlace_type, compression_type, or filter_method can
1294 be NULL if you are not interested in their values.
1296 Note that png_get_IHDR() returns 32-bit data into
1297 the application's width and height variables.
1298 This is an unsafe situation if these are not png_uint_32
1299 variables. In such situations, the
1300 png_get_image_width() and png_get_image_height()
1301 functions described below are safer.
1303 width = png_get_image_width(png_ptr,
1306 height = png_get_image_height(png_ptr,
1309 bit_depth = png_get_bit_depth(png_ptr,
1312 color_type = png_get_color_type(png_ptr,
1315 interlace_type = png_get_interlace_type(png_ptr,
1318 compression_type = png_get_compression_type(png_ptr,
1321 filter_method = png_get_filter_type(png_ptr,
1324 channels = png_get_channels(png_ptr, info_ptr);
1326 channels - number of channels of info for the
1327 color type (valid values are 1 (GRAY,
1328 PALETTE), 2 (GRAY_ALPHA), 3 (RGB),
1329 4 (RGB_ALPHA or RGB + filler byte))
1331 rowbytes = png_get_rowbytes(png_ptr, info_ptr);
1333 rowbytes - number of bytes needed to hold a row
1334 This value, the bit_depth, color_type,
1335 and the number of channels can change
1336 if you use transforms such as
1337 png_set_expand(). See
1338 png_read_update_info(), below.
1340 signature = png_get_signature(png_ptr, info_ptr);
1342 signature - holds the signature read from the
1343 file (if any). The data is kept in
1344 the same offset it would be if the
1345 whole signature were read (i.e. if an
1346 application had already read in 4
1347 bytes of signature before starting
1348 libpng, the remaining 4 bytes would
1349 be in signature[4] through signature[7]
1350 (see png_set_sig_bytes())).
1352 These are also important, but their validity depends on whether the chunk
1353 has been read. The png_get_valid(png_ptr, info_ptr, PNG_INFO_<chunk>) and
1354 png_get_<chunk>(png_ptr, info_ptr, ...) functions return non-zero if the
1355 data has been read, or zero if it is missing. The parameters to the
1356 png_get_<chunk> are set directly if they are simple data types, or a
1357 pointer into the info_ptr is returned for any complex types.
1359 The colorspace data from gAMA, cHRM, sRGB, iCCP, and sBIT chunks
1360 is simply returned to give the application information about how the
1361 image was encoded. Libpng itself only does transformations using the file
1362 gamma when combining semitransparent pixels with the background color, and,
1363 since libpng-1.6.0, when converting between 8-bit sRGB and 16-bit linear pixels
1364 within the simplified API. Libpng also uses the file gamma when converting
1365 RGB to gray, beginning with libpng-1.0.5, if the application calls
1366 png_set_rgb_to_gray()).
1368 png_get_PLTE(png_ptr, info_ptr, &palette,
1371 palette - the palette for the file
1372 (array of png_color)
1374 num_palette - number of entries in the palette
1376 png_get_gAMA(png_ptr, info_ptr, &file_gamma);
1377 png_get_gAMA_fixed(png_ptr, info_ptr, &int_file_gamma);
1379 file_gamma - the gamma at which the file is
1380 written (PNG_INFO_gAMA)
1382 int_file_gamma - 100,000 times the gamma at which the
1385 png_get_cHRM(png_ptr, info_ptr, &white_x, &white_y, &red_x,
1386 &red_y, &green_x, &green_y, &blue_x, &blue_y)
1387 png_get_cHRM_XYZ(png_ptr, info_ptr, &red_X, &red_Y, &red_Z,
1388 &green_X, &green_Y, &green_Z, &blue_X, &blue_Y,
1390 png_get_cHRM_fixed(png_ptr, info_ptr, &int_white_x,
1391 &int_white_y, &int_red_x, &int_red_y,
1392 &int_green_x, &int_green_y, &int_blue_x,
1394 png_get_cHRM_XYZ_fixed(png_ptr, info_ptr, &int_red_X, &int_red_Y,
1395 &int_red_Z, &int_green_X, &int_green_Y,
1396 &int_green_Z, &int_blue_X, &int_blue_Y,
1399 {white,red,green,blue}_{x,y}
1400 A color space encoding specified using the
1401 chromaticities of the end points and the
1402 white point. (PNG_INFO_cHRM)
1404 {red,green,blue}_{X,Y,Z}
1405 A color space encoding specified using the
1406 encoding end points - the CIE tristimulus
1407 specification of the intended color of the red,
1408 green and blue channels in the PNG RGB data.
1409 The white point is simply the sum of the three
1410 end points. (PNG_INFO_cHRM)
1412 png_get_sRGB(png_ptr, info_ptr, &srgb_intent);
1414 srgb_intent - the rendering intent (PNG_INFO_sRGB)
1415 The presence of the sRGB chunk
1416 means that the pixel data is in the
1417 sRGB color space. This chunk also
1418 implies specific values of gAMA and
1421 png_get_iCCP(png_ptr, info_ptr, &name,
1422 &compression_type, &profile, &proflen);
1424 name - The profile name.
1426 compression_type - The compression type; always
1427 PNG_COMPRESSION_TYPE_BASE for PNG 1.0.
1428 You may give NULL to this argument to
1431 profile - International Color Consortium color
1432 profile data. May contain NULs.
1434 proflen - length of profile data in bytes.
1436 png_get_sBIT(png_ptr, info_ptr, &sig_bit);
1438 sig_bit - the number of significant bits for
1439 (PNG_INFO_sBIT) each of the gray,
1440 red, green, and blue channels,
1441 whichever are appropriate for the
1442 given color type (png_color_16)
1444 png_get_tRNS(png_ptr, info_ptr, &trans_alpha,
1445 &num_trans, &trans_color);
1447 trans_alpha - array of alpha (transparency)
1448 entries for palette (PNG_INFO_tRNS)
1450 num_trans - number of transparent entries
1453 trans_color - graylevel or color sample values of
1454 the single transparent color for
1455 non-paletted images (PNG_INFO_tRNS)
1457 png_get_eXIf_1(png_ptr, info_ptr, &num_exif, &exif);
1460 exif - Exif profile (array of png_byte)
1462 png_get_hIST(png_ptr, info_ptr, &hist);
1465 hist - histogram of palette (array of
1468 png_get_tIME(png_ptr, info_ptr, &mod_time);
1470 mod_time - time image was last modified
1473 png_get_bKGD(png_ptr, info_ptr, &background);
1475 background - background color (of type
1476 png_color_16p) (PNG_VALID_bKGD)
1477 valid 16-bit red, green and blue
1478 values, regardless of color_type
1480 num_comments = png_get_text(png_ptr, info_ptr,
1481 &text_ptr, &num_text);
1483 num_comments - number of comments
1485 text_ptr - array of png_text holding image
1488 text_ptr[i].compression - type of compression used
1489 on "text" PNG_TEXT_COMPRESSION_NONE
1490 PNG_TEXT_COMPRESSION_zTXt
1491 PNG_ITXT_COMPRESSION_NONE
1492 PNG_ITXT_COMPRESSION_zTXt
1494 text_ptr[i].key - keyword for comment. Must contain
1497 text_ptr[i].text - text comments for current
1498 keyword. Can be empty.
1500 text_ptr[i].text_length - length of text string,
1501 after decompression, 0 for iTXt
1503 text_ptr[i].itxt_length - length of itxt string,
1504 after decompression, 0 for tEXt/zTXt
1506 text_ptr[i].lang - language of comment (empty
1507 string for unknown).
1509 text_ptr[i].lang_key - keyword in UTF-8
1510 (empty string for unknown).
1512 Note that the itxt_length, lang, and lang_key
1513 members of the text_ptr structure only exist when the
1514 library is built with iTXt chunk support. Prior to
1515 libpng-1.4.0 the library was built by default without
1516 iTXt support. Also note that when iTXt is supported,
1517 they contain NULL pointers when the "compression"
1518 field contains PNG_TEXT_COMPRESSION_NONE or
1519 PNG_TEXT_COMPRESSION_zTXt.
1521 num_text - number of comments (same as
1522 num_comments; you can put NULL here
1523 to avoid the duplication)
1525 Note while png_set_text() will accept text, language,
1526 and translated keywords that can be NULL pointers, the
1527 structure returned by png_get_text will always contain
1528 regular zero-terminated C strings. They might be
1529 empty strings but they will never be NULL pointers.
1531 num_spalettes = png_get_sPLT(png_ptr, info_ptr,
1534 num_spalettes - number of sPLT chunks read.
1536 palette_ptr - array of palette structures holding
1537 contents of one or more sPLT chunks
1540 png_get_oFFs(png_ptr, info_ptr, &offset_x, &offset_y,
1543 offset_x - positive offset from the left edge
1544 of the screen (can be negative)
1546 offset_y - positive offset from the top edge
1547 of the screen (can be negative)
1549 unit_type - PNG_OFFSET_PIXEL, PNG_OFFSET_MICROMETER
1551 png_get_pHYs(png_ptr, info_ptr, &res_x, &res_y,
1554 res_x - pixels/unit physical resolution in
1557 res_y - pixels/unit physical resolution in
1560 unit_type - PNG_RESOLUTION_UNKNOWN,
1561 PNG_RESOLUTION_METER
1563 png_get_sCAL(png_ptr, info_ptr, &unit, &width,
1566 unit - physical scale units (an integer)
1568 width - width of a pixel in physical scale units
1570 height - height of a pixel in physical scale units
1571 (width and height are doubles)
1573 png_get_sCAL_s(png_ptr, info_ptr, &unit, &width,
1576 unit - physical scale units (an integer)
1578 width - width of a pixel in physical scale units
1579 (expressed as a string)
1581 height - height of a pixel in physical scale units
1582 (width and height are strings like "2.54")
1584 num_unknown_chunks = png_get_unknown_chunks(png_ptr,
1585 info_ptr, &unknowns)
1587 unknowns - array of png_unknown_chunk
1588 structures holding unknown chunks
1590 unknowns[i].name - name of unknown chunk
1592 unknowns[i].data - data of unknown chunk
1594 unknowns[i].size - size of unknown chunk's data
1596 unknowns[i].location - position of chunk in file
1598 The value of "i" corresponds to the order in which the
1599 chunks were read from the PNG file or inserted with the
1600 png_set_unknown_chunks() function.
1602 The value of "location" is a bitwise "or" of
1604 PNG_HAVE_IHDR (0x01)
1605 PNG_HAVE_PLTE (0x02)
1606 PNG_AFTER_IDAT (0x08)
1608 The data from the pHYs chunk can be retrieved in several convenient
1611 res_x = png_get_x_pixels_per_meter(png_ptr,
1614 res_y = png_get_y_pixels_per_meter(png_ptr,
1617 res_x_and_y = png_get_pixels_per_meter(png_ptr,
1620 res_x = png_get_x_pixels_per_inch(png_ptr,
1623 res_y = png_get_y_pixels_per_inch(png_ptr,
1626 res_x_and_y = png_get_pixels_per_inch(png_ptr,
1629 aspect_ratio = png_get_pixel_aspect_ratio(png_ptr,
1632 Each of these returns 0 [signifying "unknown"] if
1633 the data is not present or if res_x is 0;
1634 res_x_and_y is 0 if res_x != res_y
1636 Note that because of the way the resolutions are
1637 stored internally, the inch conversions won't
1638 come out to exactly even number. For example,
1639 72 dpi is stored as 0.28346 pixels/meter, and
1640 when this is retrieved it is 71.9988 dpi, so
1641 be sure to round the returned value appropriately
1642 if you want to display a reasonable-looking result.
1644 The data from the oFFs chunk can be retrieved in several convenient
1647 x_offset = png_get_x_offset_microns(png_ptr, info_ptr);
1649 y_offset = png_get_y_offset_microns(png_ptr, info_ptr);
1651 x_offset = png_get_x_offset_inches(png_ptr, info_ptr);
1653 y_offset = png_get_y_offset_inches(png_ptr, info_ptr);
1655 Each of these returns 0 [signifying "unknown" if both
1656 x and y are 0] if the data is not present or if the
1657 chunk is present but the unit is the pixel. The
1658 remark about inexact inch conversions applies here
1659 as well, because a value in inches can't always be
1660 converted to microns and back without some loss
1663 For more information, see the
1664 PNG specification for chunk contents. Be careful with trusting
1665 rowbytes, as some of the transformations could increase the space
1666 needed to hold a row (expand, filler, gray_to_rgb, etc.).
1667 See png_read_update_info(), below.
1669 A quick word about text_ptr and num_text. PNG stores comments in
1670 keyword/text pairs, one pair per chunk, with no limit on the number
1671 of text chunks, and a 2^31 byte limit on their size. While there are
1672 suggested keywords, there is no requirement to restrict the use to these
1673 strings. It is strongly suggested that keywords and text be sensible
1674 to humans (that's the point), so don't use abbreviations. Non-printing
1675 symbols are not allowed. See the PNG specification for more details.
1676 There is also no requirement to have text after the keyword.
1678 Keywords should be limited to 79 Latin-1 characters without leading or
1679 trailing spaces, but non-consecutive spaces are allowed within the
1680 keyword. It is possible to have the same keyword any number of times.
1681 The text_ptr is an array of png_text structures, each holding a
1682 pointer to a language string, a pointer to a keyword and a pointer to
1683 a text string. The text string, language code, and translated
1684 keyword may be empty or NULL pointers. The keyword/text
1685 pairs are put into the array in the order that they are received.
1686 However, some or all of the text chunks may be after the image, so, to
1687 make sure you have read all the text chunks, don't mess with these
1688 until after you read the stuff after the image. This will be
1689 mentioned again below in the discussion that goes with png_read_end().
1691 Input transformations
1693 After you've read the header information, you can set up the library
1694 to handle any special transformations of the image data. The various
1695 ways to transform the data will be described in the order that they
1696 should occur. This is important, as some of these change the color
1697 type and/or bit depth of the data, and some others only work on
1698 certain color types and bit depths.
1700 Transformations you request are ignored if they don't have any meaning for a
1701 particular input data format. However some transformations can have an effect
1702 as a result of a previous transformation. If you specify a contradictory set of
1703 transformations, for example both adding and removing the alpha channel, you
1704 cannot predict the final result.
1706 The color used for the transparency values should be supplied in the same
1707 format/depth as the current image data. It is stored in the same format/depth
1708 as the image data in a tRNS chunk, so this is what libpng expects for this data.
1710 The color used for the background value depends on the need_expand argument as
1713 Data will be decoded into the supplied row buffers packed into bytes
1714 unless the library has been told to transform it into another format.
1715 For example, 4 bit/pixel paletted or grayscale data will be returned
1716 2 pixels/byte with the leftmost pixel in the high-order bits of the byte,
1717 unless png_set_packing() is called. 8-bit RGB data will be stored
1718 in RGB RGB RGB format unless png_set_filler() or png_set_add_alpha()
1719 is called to insert filler bytes, either before or after each RGB triplet.
1721 16-bit RGB data will be returned RRGGBB RRGGBB, with the most significant
1722 byte of the color value first, unless png_set_scale_16() is called to
1723 transform it to regular RGB RGB triplets, or png_set_filler() or
1724 png_set_add alpha() is called to insert two filler bytes, either before
1725 or after each RRGGBB triplet. Similarly, 8-bit or 16-bit grayscale data can
1726 be modified with png_set_filler(), png_set_add_alpha(), png_set_strip_16(),
1727 or png_set_scale_16().
1729 The following code transforms grayscale images of less than 8 to 8 bits,
1730 changes paletted images to RGB, and adds a full alpha channel if there is
1731 transparency information in a tRNS chunk. This is most useful on
1732 grayscale images with bit depths of 2 or 4 or if there is a multiple-image
1733 viewing application that wishes to treat all images in the same way.
1735 if (color_type == PNG_COLOR_TYPE_PALETTE)
1736 png_set_palette_to_rgb(png_ptr);
1738 if (png_get_valid(png_ptr, info_ptr,
1739 PNG_INFO_tRNS)) png_set_tRNS_to_alpha(png_ptr);
1741 if (color_type == PNG_COLOR_TYPE_GRAY &&
1742 bit_depth < 8) png_set_expand_gray_1_2_4_to_8(png_ptr);
1744 The first two functions are actually aliases for png_set_expand(), added
1745 in libpng version 1.0.4, with the function names expanded to improve code
1746 readability. In some future version they may actually do different
1749 As of libpng version 1.2.9, png_set_expand_gray_1_2_4_to_8() was
1750 added. It expands the sample depth without changing tRNS to alpha.
1752 As of libpng version 1.5.2, png_set_expand_16() was added. It behaves as
1753 png_set_expand(); however, the resultant channels have 16 bits rather than 8.
1754 Use this when the output color or gray channels are made linear to avoid fairly
1755 severe accuracy loss.
1758 png_set_expand_16(png_ptr);
1760 PNG can have files with 16 bits per channel. If you only can handle
1761 8 bits per channel, this will strip the pixels down to 8-bit.
1763 if (bit_depth == 16)
1764 #if PNG_LIBPNG_VER >= 10504
1765 png_set_scale_16(png_ptr);
1767 png_set_strip_16(png_ptr);
1770 (The more accurate "png_set_scale_16()" API became available in libpng version
1773 If you need to process the alpha channel on the image separately from the image
1774 data (for example if you convert it to a bitmap mask) it is possible to have
1775 libpng strip the channel leaving just RGB or gray data:
1777 if (color_type & PNG_COLOR_MASK_ALPHA)
1778 png_set_strip_alpha(png_ptr);
1780 If you strip the alpha channel you need to find some other way of dealing with
1781 the information. If, instead, you want to convert the image to an opaque
1782 version with no alpha channel use png_set_background; see below.
1784 As of libpng version 1.5.2, almost all useful expansions are supported, the
1785 major ommissions are conversion of grayscale to indexed images (which can be
1786 done trivially in the application) and conversion of indexed to grayscale (which
1787 can be done by a trivial manipulation of the palette.)
1789 In the following table, the 01 means grayscale with depth<8, 31 means
1790 indexed with depth<8, other numerals represent the color type, "T" means
1791 the tRNS chunk is present, A means an alpha channel is present, and O
1792 means tRNS or alpha is present but all pixels in the image are opaque.
1794 FROM 01 31 0 0T 0O 2 2T 2O 3 3T 3O 4A 4O 6A 6O
1796 01 - [G] - - - - - - - - - - - - -
1797 31 [Q] Q [Q] [Q] [Q] Q Q Q Q Q Q [Q] [Q] Q Q
1798 0 1 G + . . G G G G G G B B GB GB
1799 0T lt Gt t + . Gt G G Gt G G Bt Bt GBt GBt
1800 0O lt Gt t . + Gt Gt G Gt Gt G Bt Bt GBt GBt
1801 2 C P C C C + . . C - - CB CB B B
1802 2T Ct - Ct C C t + t - - - CBt CBt Bt Bt
1803 2O Ct - Ct C C t t + - - - CBt CBt Bt Bt
1804 3 [Q] p [Q] [Q] [Q] Q Q Q + . . [Q] [Q] Q Q
1805 3T [Qt] p [Qt][Q] [Q] Qt Qt Qt t + t [Qt][Qt] Qt Qt
1806 3O [Qt] p [Qt][Q] [Q] Qt Qt Qt t t + [Qt][Qt] Qt Qt
1807 4A lA G A T T GA GT GT GA GT GT + BA G GBA
1808 4O lA GBA A T T GA GT GT GA GT GT BA + GBA G
1809 6A CA PA CA C C A T tT PA P P C CBA + BA
1810 6O CA PBA CA C C A tT T PA P P CBA C BA +
1813 "+" identifies entries where 'from' and 'to' are the same.
1814 "-" means the transformation is not supported.
1815 "." means nothing is necessary (a tRNS chunk can just be ignored).
1816 "t" means the transformation is obtained by png_set_tRNS.
1817 "A" means the transformation is obtained by png_set_add_alpha().
1818 "X" means the transformation is obtained by png_set_expand().
1819 "1" means the transformation is obtained by
1820 png_set_expand_gray_1_2_4_to_8() (and by png_set_expand()
1821 if there is no transparency in the original or the final
1823 "C" means the transformation is obtained by png_set_gray_to_rgb().
1824 "G" means the transformation is obtained by png_set_rgb_to_gray().
1825 "P" means the transformation is obtained by
1826 png_set_expand_palette_to_rgb().
1827 "p" means the transformation is obtained by png_set_packing().
1828 "Q" means the transformation is obtained by png_set_quantize().
1829 "T" means the transformation is obtained by
1830 png_set_tRNS_to_alpha().
1831 "B" means the transformation is obtained by
1832 png_set_background(), or png_strip_alpha().
1834 When an entry has multiple transforms listed all are required to cause the
1835 right overall transformation. When two transforms are separated by a comma
1836 either will do the job. When transforms are enclosed in [] the transform should
1837 do the job but this is currently unimplemented - a different format will result
1838 if the suggested transformations are used.
1840 In PNG files, the alpha channel in an image
1841 is the level of opacity. If you need the alpha channel in an image to
1842 be the level of transparency instead of opacity, you can invert the
1843 alpha channel (or the tRNS chunk data) after it's read, so that 0 is
1844 fully opaque and 255 (in 8-bit or paletted images) or 65535 (in 16-bit
1845 images) is fully transparent, with
1847 png_set_invert_alpha(png_ptr);
1849 PNG files pack pixels of bit depths 1, 2, and 4 into bytes as small as
1850 they can, resulting in, for example, 8 pixels per byte for 1 bit
1851 files. This code expands to 1 pixel per byte without changing the
1852 values of the pixels:
1855 png_set_packing(png_ptr);
1857 PNG files have possible bit depths of 1, 2, 4, 8, and 16. All pixels
1858 stored in a PNG image have been "scaled" or "shifted" up to the next
1859 higher possible bit depth (e.g. from 5 bits/sample in the range [0,31]
1860 to 8 bits/sample in the range [0, 255]). However, it is also possible
1861 to convert the PNG pixel data back to the original bit depth of the
1862 image. This call reduces the pixels back down to the original bit depth:
1864 png_color_8p sig_bit;
1866 if (png_get_sBIT(png_ptr, info_ptr, &sig_bit))
1867 png_set_shift(png_ptr, sig_bit);
1869 PNG files store 3-color pixels in red, green, blue order. This code
1870 changes the storage of the pixels to blue, green, red:
1872 if (color_type == PNG_COLOR_TYPE_RGB ||
1873 color_type == PNG_COLOR_TYPE_RGB_ALPHA)
1874 png_set_bgr(png_ptr);
1876 PNG files store RGB pixels packed into 3 or 6 bytes. This code expands them
1877 into 4 or 8 bytes for windowing systems that need them in this format:
1879 if (color_type == PNG_COLOR_TYPE_RGB)
1880 png_set_filler(png_ptr, filler, PNG_FILLER_BEFORE);
1882 where "filler" is the 8-bit or 16-bit number to fill with, and the location
1883 is either PNG_FILLER_BEFORE or PNG_FILLER_AFTER, depending upon whether
1884 you want the filler before the RGB or after. When filling an 8-bit pixel,
1885 the least significant 8 bits of the number are used, if a 16-bit number is
1886 supplied. This transformation does not affect images that already have full
1887 alpha channels. To add an opaque alpha channel, use filler=0xffff and
1888 PNG_FILLER_AFTER which will generate RGBA pixels.
1890 Note that png_set_filler() does not change the color type. If you want
1891 to do that, you can add a true alpha channel with
1893 if (color_type == PNG_COLOR_TYPE_RGB ||
1894 color_type == PNG_COLOR_TYPE_GRAY)
1895 png_set_add_alpha(png_ptr, filler, PNG_FILLER_AFTER);
1897 where "filler" contains the alpha value to assign to each pixel.
1898 The png_set_add_alpha() function was added in libpng-1.2.7.
1900 If you are reading an image with an alpha channel, and you need the
1901 data as ARGB instead of the normal PNG format RGBA:
1903 if (color_type == PNG_COLOR_TYPE_RGB_ALPHA)
1904 png_set_swap_alpha(png_ptr);
1906 For some uses, you may want a grayscale image to be represented as
1907 RGB. This code will do that conversion:
1909 if (color_type == PNG_COLOR_TYPE_GRAY ||
1910 color_type == PNG_COLOR_TYPE_GRAY_ALPHA)
1911 png_set_gray_to_rgb(png_ptr);
1913 Conversely, you can convert an RGB or RGBA image to grayscale or grayscale
1916 if (color_type == PNG_COLOR_TYPE_RGB ||
1917 color_type == PNG_COLOR_TYPE_RGB_ALPHA)
1918 png_set_rgb_to_gray(png_ptr, error_action,
1919 double red_weight, double green_weight);
1921 error_action = 1: silently do the conversion
1923 error_action = 2: issue a warning if the original
1924 image has any pixel where
1925 red != green or red != blue
1927 error_action = 3: issue an error and abort the
1928 conversion if the original
1929 image has any pixel where
1930 red != green or red != blue
1932 red_weight: weight of red component
1934 green_weight: weight of green component
1935 If either weight is negative, default
1938 In the corresponding fixed point API the red_weight and green_weight values are
1939 simply scaled by 100,000:
1941 png_set_rgb_to_gray(png_ptr, error_action,
1942 png_fixed_point red_weight,
1943 png_fixed_point green_weight);
1945 If you have set error_action = 1 or 2, you can
1946 later check whether the image really was gray, after processing
1947 the image rows, with the png_get_rgb_to_gray_status(png_ptr) function.
1948 It will return a png_byte that is zero if the image was gray or
1949 1 if there were any non-gray pixels. Background and sBIT data
1950 will be silently converted to grayscale, using the green channel
1951 data for sBIT, regardless of the error_action setting.
1953 The default values come from the PNG file cHRM chunk if present; otherwise, the
1954 defaults correspond to the ITU-R recommendation 709, and also the sRGB color
1955 space, as recommended in the Charles Poynton's Colour FAQ,
1956 Copyright (c) 2006-11-28 Charles Poynton, in section 9:
1958 <http://www.poynton.com/notes/colour_and_gamma/ColorFAQ.html#RTFToC9>
1960 Y = 0.2126 * R + 0.7152 * G + 0.0722 * B
1962 Previous versions of this document, 1998 through 2002, recommended a slightly
1965 Y = 0.212671 * R + 0.715160 * G + 0.072169 * B
1967 Libpng uses an integer approximation:
1969 Y = (6968 * R + 23434 * G + 2366 * B)/32768
1971 The calculation is done in a linear colorspace, if the image gamma
1974 The png_set_background() function has been described already; it tells libpng to
1975 composite images with alpha or simple transparency against the supplied
1976 background color. For compatibility with versions of libpng earlier than
1977 libpng-1.5.4 it is recommended that you call the function after reading the file
1978 header, even if you don't want to use the color in a bKGD chunk, if one exists.
1980 If the PNG file contains a bKGD chunk (PNG_INFO_bKGD valid),
1981 you may use this color, or supply another color more suitable for
1982 the current display (e.g., the background color from a web page). You
1983 need to tell libpng how the color is represented, both the format of the
1984 component values in the color (the number of bits) and the gamma encoding of the
1985 color. The function takes two arguments, background_gamma_mode and need_expand
1986 to convey this information; however, only two combinations are likely to be
1989 png_color_16 my_background;
1990 png_color_16p image_background;
1992 if (png_get_bKGD(png_ptr, info_ptr, &image_background))
1993 png_set_background(png_ptr, image_background,
1994 PNG_BACKGROUND_GAMMA_FILE, 1/*needs to be expanded*/, 1);
1996 png_set_background(png_ptr, &my_background,
1997 PNG_BACKGROUND_GAMMA_SCREEN, 0/*do not expand*/, 1);
1999 The second call was described above - my_background is in the format of the
2000 final, display, output produced by libpng. Because you now know the format of
2001 the PNG it is possible to avoid the need to choose either 8-bit or 16-bit
2002 output and to retain palette images (the palette colors will be modified
2003 appropriately and the tRNS chunk removed.) However, if you are doing this,
2004 take great care not to ask for transformations without checking first that
2007 In the first call the background color has the original bit depth and color type
2008 of the PNG file. So, for palette images the color is supplied as a palette
2009 index and for low bit greyscale images the color is a reduced bit value in
2010 image_background->gray.
2012 If you didn't call png_set_gamma() before reading the file header, for example
2013 if you need your code to remain compatible with older versions of libpng prior
2014 to libpng-1.5.4, this is the place to call it.
2016 Do not call it if you called png_set_alpha_mode(); doing so will damage the
2017 settings put in place by png_set_alpha_mode(). (If png_set_alpha_mode() is
2018 supported then you can certainly do png_set_gamma() before reading the PNG
2021 This API unconditionally sets the screen and file gamma values, so it will
2022 override the value in the PNG file unless it is called before the PNG file
2023 reading starts. For this reason you must always call it with the PNG file
2024 value when you call it in this position:
2026 if (png_get_gAMA(png_ptr, info_ptr, &file_gamma))
2027 png_set_gamma(png_ptr, screen_gamma, file_gamma);
2030 png_set_gamma(png_ptr, screen_gamma, 0.45455);
2032 If you need to reduce an RGB file to a paletted file, or if a paletted
2033 file has more entries than will fit on your screen, png_set_quantize()
2034 will do that. Note that this is a simple match quantization that merely
2035 finds the closest color available. This should work fairly well with
2036 optimized palettes, but fairly badly with linear color cubes. If you
2037 pass a palette that is larger than maximum_colors, the file will
2038 reduce the number of colors in the palette so it will fit into
2039 maximum_colors. If there is a histogram, libpng will use it to make
2040 more intelligent choices when reducing the palette. If there is no
2041 histogram, it may not do as good a job.
2043 if (color_type & PNG_COLOR_MASK_COLOR)
2045 if (png_get_valid(png_ptr, info_ptr,
2048 png_uint_16p histogram = NULL;
2050 png_get_hIST(png_ptr, info_ptr,
2052 png_set_quantize(png_ptr, palette, num_palette,
2053 max_screen_colors, histogram, 1);
2058 png_color std_color_cube[MAX_SCREEN_COLORS] =
2061 png_set_quantize(png_ptr, std_color_cube,
2062 MAX_SCREEN_COLORS, MAX_SCREEN_COLORS,
2067 PNG files describe monochrome as black being zero and white being one.
2068 The following code will reverse this (make black be one and white be
2071 if (bit_depth == 1 && color_type == PNG_COLOR_TYPE_GRAY)
2072 png_set_invert_mono(png_ptr);
2074 This function can also be used to invert grayscale and gray-alpha images:
2076 if (color_type == PNG_COLOR_TYPE_GRAY ||
2077 color_type == PNG_COLOR_TYPE_GRAY_ALPHA)
2078 png_set_invert_mono(png_ptr);
2080 PNG files store 16-bit pixels in network byte order (big-endian,
2081 ie. most significant bits first). This code changes the storage to the
2082 other way (little-endian, i.e. least significant bits first, the
2083 way PCs store them):
2085 if (bit_depth == 16)
2086 png_set_swap(png_ptr);
2088 If you are using packed-pixel images (1, 2, or 4 bits/pixel), and you
2089 need to change the order the pixels are packed into bytes, you can use:
2092 png_set_packswap(png_ptr);
2094 Finally, you can write your own transformation function if none of
2095 the existing ones meets your needs. This is done by setting a callback
2098 png_set_read_user_transform_fn(png_ptr,
2101 You must supply the function
2103 void read_transform_fn(png_structp png_ptr, png_row_infop
2104 row_info, png_bytep data)
2106 See pngtest.c for a working example. Your function will be called
2107 after all of the other transformations have been processed. Take care with
2108 interlaced images if you do the interlace yourself - the width of the row is the
2109 width in 'row_info', not the overall image width.
2111 If supported, libpng provides two information routines that you can use to find
2112 where you are in processing the image:
2114 png_get_current_pass_number(png_structp png_ptr);
2115 png_get_current_row_number(png_structp png_ptr);
2117 Don't try using these outside a transform callback - firstly they are only
2118 supported if user transforms are supported, secondly they may well return
2119 unexpected results unless the row is actually being processed at the moment they
2123 images the value returned is the row in the input sub-image image. Use
2124 PNG_ROW_FROM_PASS_ROW(row, pass) and PNG_COL_FROM_PASS_COL(col, pass) to
2125 find the output pixel (x,y) given an interlaced sub-image pixel (row,col,pass).
2127 The discussion of interlace handling above contains more information on how to
2130 You can also set up a pointer to a user structure for use by your
2131 callback function, and you can inform libpng that your transform
2132 function will change the number of channels or bit depth with the
2135 png_set_user_transform_info(png_ptr, user_ptr,
2136 user_depth, user_channels);
2138 The user's application, not libpng, is responsible for allocating and
2139 freeing any memory required for the user structure.
2141 You can retrieve the pointer via the function
2142 png_get_user_transform_ptr(). For example:
2144 voidp read_user_transform_ptr =
2145 png_get_user_transform_ptr(png_ptr);
2147 The last thing to handle is interlacing; this is covered in detail below,
2148 but you must call the function here if you want libpng to handle expansion
2149 of the interlaced image.
2151 number_of_passes = png_set_interlace_handling(png_ptr);
2153 After setting the transformations, libpng can update your png_info
2154 structure to reflect any transformations you've requested with this
2157 png_read_update_info(png_ptr, info_ptr);
2159 This is most useful to update the info structure's rowbytes
2160 field so you can use it to allocate your image memory. This function
2161 will also update your palette with the correct screen_gamma and
2162 background if these have been given with the calls above. You may
2163 only call png_read_update_info() once with a particular info_ptr.
2165 After you call png_read_update_info(), you can allocate any
2166 memory you need to hold the image. The row data is simply
2167 raw byte data for all forms of images. As the actual allocation
2168 varies among applications, no example will be given. If you
2169 are allocating one large chunk, you will need to build an
2170 array of pointers to each row, as it will be needed for some
2171 of the functions below.
2173 Be sure that your platform can allocate the buffer that you'll need.
2174 libpng internally checks for oversize width, but you'll need to
2175 do your own check for number_of_rows*width*pixel_size if you are using
2176 a multiple-row buffer:
2178 /* Guard against integer overflow */
2179 if (number_of_rows > PNG_SIZE_MAX/(width*pixel_size)) {
2180 png_error(png_ptr,"image_data buffer would be too large");
2183 Remember: Before you call png_read_update_info(), the png_get_*()
2184 functions return the values corresponding to the original PNG image.
2185 After you call png_read_update_info the values refer to the image
2186 that libpng will output. Consequently you must call all the png_set_
2187 functions before you call png_read_update_info(). This is particularly
2188 important for png_set_interlace_handling() - if you are going to call
2189 png_read_update_info() you must call png_set_interlace_handling() before
2190 it unless you want to receive interlaced output.
2194 After you've allocated memory, you can read the image data.
2195 The simplest way to do this is in one function call. If you are
2196 allocating enough memory to hold the whole image, you can just
2197 call png_read_image() and libpng will read in all the image data
2198 and put it in the memory area supplied. You will need to pass in
2199 an array of pointers to each row.
2201 This function automatically handles interlacing, so you don't
2202 need to call png_set_interlace_handling() (unless you call
2203 png_read_update_info()) or call this function multiple times, or any
2204 of that other stuff necessary with png_read_rows().
2206 png_read_image(png_ptr, row_pointers);
2208 where row_pointers is:
2210 png_bytep row_pointers[height];
2212 You can point to void or char or whatever you use for pixels.
2214 If you don't want to read in the whole image at once, you can
2215 use png_read_rows() instead. If there is no interlacing (check
2216 interlace_type == PNG_INTERLACE_NONE), this is simple:
2218 png_read_rows(png_ptr, row_pointers, NULL,
2221 where row_pointers is the same as in the png_read_image() call.
2223 If you are doing this just one row at a time, you can do this with
2224 a single row_pointer instead of an array of row_pointers:
2226 png_bytep row_pointer = row;
2227 png_read_row(png_ptr, row_pointer, NULL);
2229 If the file is interlaced (interlace_type != 0 in the IHDR chunk), things
2230 get somewhat harder. The only current (PNG Specification version 1.2)
2231 interlacing type for PNG is (interlace_type == PNG_INTERLACE_ADAM7);
2232 a somewhat complicated 2D interlace scheme, known as Adam7, that
2233 breaks down an image into seven smaller images of varying size, based
2234 on an 8x8 grid. This number is defined (from libpng 1.5) as
2235 PNG_INTERLACE_ADAM7_PASSES in png.h
2237 libpng can fill out those images or it can give them to you "as is".
2238 It is almost always better to have libpng handle the interlacing for you.
2239 If you want the images filled out, there are two ways to do that. The one
2240 mentioned in the PNG specification is to expand each pixel to cover
2241 those pixels that have not been read yet (the "rectangle" method).
2242 This results in a blocky image for the first pass, which gradually
2243 smooths out as more pixels are read. The other method is the "sparkle"
2244 method, where pixels are drawn only in their final locations, with the
2245 rest of the image remaining whatever colors they were initialized to
2246 before the start of the read. The first method usually looks better,
2247 but tends to be slower, as there are more pixels to put in the rows.
2249 If, as is likely, you want libpng to expand the images, call this before
2250 calling png_start_read_image() or png_read_update_info():
2252 if (interlace_type == PNG_INTERLACE_ADAM7)
2254 = png_set_interlace_handling(png_ptr);
2256 This will return the number of passes needed. Currently, this is seven,
2257 but may change if another interlace type is added. This function can be
2258 called even if the file is not interlaced, where it will return one pass.
2259 You then need to read the whole image 'number_of_passes' times. Each time
2260 will distribute the pixels from the current pass to the correct place in
2261 the output image, so you need to supply the same rows to png_read_rows in
2264 If you are not going to display the image after each pass, but are
2265 going to wait until the entire image is read in, use the sparkle
2266 effect. This effect is faster and the end result of either method
2267 is exactly the same. If you are planning on displaying the image
2268 after each pass, the "rectangle" effect is generally considered the
2271 If you only want the "sparkle" effect, just call png_read_row() or
2273 normal, with the third parameter NULL. Make sure you make pass over
2274 the image number_of_passes times, and you don't change the data in the
2275 rows between calls. You can change the locations of the data, just
2276 not the data. Each pass only writes the pixels appropriate for that
2277 pass, and assumes the data from previous passes is still valid.
2279 png_read_rows(png_ptr, row_pointers, NULL,
2282 png_read_row(png_ptr, row_pointers, NULL);
2284 If you only want the first effect (the rectangles), do the same as
2285 before except pass the row buffer in the third parameter, and leave
2286 the second parameter NULL.
2288 png_read_rows(png_ptr, NULL, row_pointers,
2291 png_read_row(png_ptr, NULL, row_pointers);
2293 If you don't want libpng to handle the interlacing details, just call
2294 png_read_rows() PNG_INTERLACE_ADAM7_PASSES times to read in all the images.
2295 Each of the images is a valid image by itself; however, you will almost
2296 certainly need to distribute the pixels from each sub-image to the
2297 correct place. This is where everything gets very tricky.
2299 If you want to retrieve the separate images you must pass the correct
2300 number of rows to each successive call of png_read_rows(). The calculation
2301 gets pretty complicated for small images, where some sub-images may
2302 not even exist because either their width or height ends up zero.
2303 libpng provides two macros to help you in 1.5 and later versions:
2305 png_uint_32 width = PNG_PASS_COLS(image_width, pass_number);
2306 png_uint_32 height = PNG_PASS_ROWS(image_height, pass_number);
2308 Respectively these tell you the width and height of the sub-image
2309 corresponding to the numbered pass. 'pass' is in in the range 0 to 6 -
2310 this can be confusing because the specification refers to the same passes
2311 as 1 to 7! Be careful, you must check both the width and height before
2312 calling png_read_rows() and not call it for that pass if either is zero.
2314 You can, of course, read each sub-image row by row. If you want to
2315 produce optimal code to make a pixel-by-pixel transformation of an
2316 interlaced image this is the best approach; read each row of each pass,
2317 transform it, and write it out to a new interlaced image.
2319 If you want to de-interlace the image yourself libpng provides further
2320 macros to help that tell you where to place the pixels in the output image.
2321 Because the interlacing scheme is rectangular - sub-image pixels are always
2322 arranged on a rectangular grid - all you need to know for each pass is the
2323 starting column and row in the output image of the first pixel plus the
2324 spacing between each pixel. As of libpng 1.5 there are four macros to
2325 retrieve this information:
2327 png_uint_32 x = PNG_PASS_START_COL(pass);
2328 png_uint_32 y = PNG_PASS_START_ROW(pass);
2329 png_uint_32 xStep = 1U << PNG_PASS_COL_SHIFT(pass);
2330 png_uint_32 yStep = 1U << PNG_PASS_ROW_SHIFT(pass);
2332 These allow you to write the obvious loop:
2334 png_uint_32 input_y = 0;
2335 png_uint_32 output_y = PNG_PASS_START_ROW(pass);
2337 while (output_y < output_image_height)
2339 png_uint_32 input_x = 0;
2340 png_uint_32 output_x = PNG_PASS_START_COL(pass);
2342 while (output_x < output_image_width)
2344 image[output_y][output_x] =
2345 subimage[pass][input_y][input_x++];
2354 Notice that the steps between successive output rows and columns are
2355 returned as shifts. This is possible because the pixels in the subimages
2356 are always a power of 2 apart - 1, 2, 4 or 8 pixels - in the original
2357 image. In practice you may need to directly calculate the output coordinate
2358 given an input coordinate. libpng provides two further macros for this
2361 png_uint_32 output_x = PNG_COL_FROM_PASS_COL(input_x, pass);
2362 png_uint_32 output_y = PNG_ROW_FROM_PASS_ROW(input_y, pass);
2364 Finally a pair of macros are provided to tell you if a particular image
2365 row or column appears in a given pass:
2367 int col_in_pass = PNG_COL_IN_INTERLACE_PASS(output_x, pass);
2368 int row_in_pass = PNG_ROW_IN_INTERLACE_PASS(output_y, pass);
2370 Bear in mind that you will probably also need to check the width and height
2371 of the pass in addition to the above to be sure the pass even exists!
2373 With any luck you are convinced by now that you don't want to do your own
2374 interlace handling. In reality normally the only good reason for doing this
2375 is if you are processing PNG files on a pixel-by-pixel basis and don't want
2376 to load the whole file into memory when it is interlaced.
2378 libpng includes a test program, pngvalid, that illustrates reading and
2379 writing of interlaced images. If you can't get interlacing to work in your
2380 code and don't want to leave it to libpng (the recommended approach), see
2381 how pngvalid.c does it.
2383 Finishing a sequential read
2385 After you are finished reading the image through the
2386 low-level interface, you can finish reading the file.
2388 If you want to use a different crc action for handling CRC errors in
2389 chunks after the image data, you can call png_set_crc_action()
2390 again at this point.
2392 If you are interested in comments or time, which may be stored either
2393 before or after the image data, you should pass the separate png_info
2394 struct if you want to keep the comments from before and after the image
2397 png_infop end_info = png_create_info_struct(png_ptr);
2401 png_destroy_read_struct(&png_ptr, &info_ptr,
2406 png_read_end(png_ptr, end_info);
2408 If you are not interested, you should still call png_read_end()
2409 but you can pass NULL, avoiding the need to create an end_info structure.
2410 If you do this, libpng will not process any chunks after IDAT other than
2411 skipping over them and perhaps (depending on whether you have called
2412 png_set_crc_action) checking their CRCs while looking for the IEND chunk.
2414 png_read_end(png_ptr, (png_infop)NULL);
2416 If you don't call png_read_end(), then your file pointer will be
2417 left pointing to the first chunk after the last IDAT, which is probably
2418 not what you want if you expect to read something beyond the end of
2421 When you are done, you can free all memory allocated by libpng like this:
2423 png_destroy_read_struct(&png_ptr, &info_ptr,
2426 or, if you didn't create an end_info structure,
2428 png_destroy_read_struct(&png_ptr, &info_ptr,
2431 It is also possible to individually free the info_ptr members that
2432 point to libpng-allocated storage with the following function:
2434 png_free_data(png_ptr, info_ptr, mask, seq)
2436 mask - identifies data to be freed, a mask
2437 containing the bitwise OR of one or
2439 PNG_FREE_PLTE, PNG_FREE_TRNS,
2440 PNG_FREE_HIST, PNG_FREE_ICCP,
2441 PNG_FREE_PCAL, PNG_FREE_ROWS,
2442 PNG_FREE_SCAL, PNG_FREE_SPLT,
2443 PNG_FREE_TEXT, PNG_FREE_UNKN,
2444 or simply PNG_FREE_ALL
2446 seq - sequence number of item to be freed
2449 This function may be safely called when the relevant storage has
2450 already been freed, or has not yet been allocated, or was allocated
2451 by the user and not by libpng, and will in those cases do nothing.
2452 The "seq" parameter is ignored if only one item of the selected data
2453 type, such as PLTE, is allowed. If "seq" is not -1, and multiple items
2454 are allowed for the data type identified in the mask, such as text or
2455 sPLT, only the n'th item in the structure is freed, where n is "seq".
2457 The default behavior is only to free data that was allocated internally
2458 by libpng. This can be changed, so that libpng will not free the data,
2459 or so that it will free data that was allocated by the user with png_malloc()
2460 or png_calloc() and passed in via a png_set_*() function, with
2462 png_data_freer(png_ptr, info_ptr, freer, mask)
2465 PNG_DESTROY_WILL_FREE_DATA
2466 PNG_SET_WILL_FREE_DATA
2467 PNG_USER_WILL_FREE_DATA
2469 mask - which data elements are affected
2470 same choices as in png_free_data()
2472 This function only affects data that has already been allocated.
2473 You can call this function after reading the PNG data but before calling
2474 any png_set_*() functions, to control whether the user or the png_set_*()
2475 function is responsible for freeing any existing data that might be present,
2476 and again after the png_set_*() functions to control whether the user
2477 or png_destroy_*() is supposed to free the data. When the user assumes
2478 responsibility for libpng-allocated data, the application must use
2479 png_free() to free it, and when the user transfers responsibility to libpng
2480 for data that the user has allocated, the user must have used png_malloc()
2481 or png_calloc() to allocate it.
2483 If you allocated your row_pointers in a single block, as suggested above in
2484 the description of the high level read interface, you must not transfer
2485 responsibility for freeing it to the png_set_rows or png_read_destroy function,
2486 because they would also try to free the individual row_pointers[i].
2488 If you allocated text_ptr.text, text_ptr.lang, and text_ptr.translated_keyword
2489 separately, do not transfer responsibility for freeing text_ptr to libpng,
2490 because when libpng fills a png_text structure it combines these members with
2491 the key member, and png_free_data() will free only text_ptr.key. Similarly,
2492 if you transfer responsibility for free'ing text_ptr from libpng to your
2493 application, your application must not separately free those members.
2495 The png_free_data() function will turn off the "valid" flag for anything
2496 it frees. If you need to turn the flag off for a chunk that was freed by
2497 your application instead of by libpng, you can use
2499 png_set_invalid(png_ptr, info_ptr, mask);
2501 mask - identifies the chunks to be made invalid,
2502 containing the bitwise OR of one or
2504 PNG_INFO_gAMA, PNG_INFO_sBIT,
2505 PNG_INFO_cHRM, PNG_INFO_PLTE,
2506 PNG_INFO_tRNS, PNG_INFO_bKGD,
2508 PNG_INFO_hIST, PNG_INFO_pHYs,
2509 PNG_INFO_oFFs, PNG_INFO_tIME,
2510 PNG_INFO_pCAL, PNG_INFO_sRGB,
2511 PNG_INFO_iCCP, PNG_INFO_sPLT,
2512 PNG_INFO_sCAL, PNG_INFO_IDAT
2514 For a more compact example of reading a PNG image, see the file example.c.
2516 Reading PNG files progressively
2518 The progressive reader is slightly different from the non-progressive
2519 reader. Instead of calling png_read_info(), png_read_rows(), and
2520 png_read_end(), you make one call to png_process_data(), which calls
2521 callbacks when it has the info, a row, or the end of the image. You
2522 set up these callbacks with png_set_progressive_read_fn(). You don't
2523 have to worry about the input/output functions of libpng, as you are
2524 giving the library the data directly in png_process_data(). I will
2525 assume that you have read the section on reading PNG files above,
2526 so I will only highlight the differences (although I will show
2529 png_structp png_ptr;
2532 /* An example code fragment of how you would
2533 initialize the progressive reader in your
2536 initialize_png_reader()
2538 png_ptr = png_create_read_struct
2539 (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
2540 user_error_fn, user_warning_fn);
2545 info_ptr = png_create_info_struct(png_ptr);
2549 png_destroy_read_struct(&png_ptr,
2550 (png_infopp)NULL, (png_infopp)NULL);
2554 if (setjmp(png_jmpbuf(png_ptr)))
2556 png_destroy_read_struct(&png_ptr, &info_ptr,
2561 /* This one's new. You can provide functions
2562 to be called when the header info is valid,
2563 when each row is completed, and when the image
2564 is finished. If you aren't using all functions,
2565 you can specify NULL parameters. Even when all
2566 three functions are NULL, you need to call
2567 png_set_progressive_read_fn(). You can use
2568 any struct as the user_ptr (cast to a void pointer
2569 for the function call), and retrieve the pointer
2570 from inside the callbacks using the function
2572 png_get_progressive_ptr(png_ptr);
2574 which will return a void pointer, which you have
2575 to cast appropriately.
2577 png_set_progressive_read_fn(png_ptr, (void *)user_ptr,
2578 info_callback, row_callback, end_callback);
2583 /* A code fragment that you call as you receive blocks
2586 process_data(png_bytep buffer, png_uint_32 length)
2588 if (setjmp(png_jmpbuf(png_ptr)))
2590 png_destroy_read_struct(&png_ptr, &info_ptr,
2595 /* This one's new also. Simply give it a chunk
2596 of data from the file stream (in order, of
2597 course). On machines with segmented memory
2598 models machines, don't give it any more than
2599 64K. The library seems to run fine with sizes
2600 of 4K. Although you can give it much less if
2601 necessary (I assume you can give it chunks of
2602 1 byte, I haven't tried less than 256 bytes
2603 yet). When this function returns, you may
2604 want to display any rows that were generated
2605 in the row callback if you don't already do
2608 png_process_data(png_ptr, info_ptr, buffer, length);
2610 /* At this point you can call png_process_data_skip if
2611 you want to handle data the library will skip yourself;
2612 it simply returns the number of bytes to skip (and stops
2613 libpng skipping that number of bytes on the next
2614 png_process_data call).
2618 /* This function is called (as set by
2619 png_set_progressive_read_fn() above) when enough data
2620 has been supplied so all of the header has been
2624 info_callback(png_structp png_ptr, png_infop info)
2626 /* Do any setup here, including setting any of
2627 the transformations mentioned in the Reading
2628 PNG files section. For now, you _must_ call
2629 either png_start_read_image() or
2630 png_read_update_info() after all the
2631 transformations are set (even if you don't set
2632 any). You may start getting rows before
2633 png_process_data() returns, so this is your
2634 last chance to prepare for that.
2636 This is where you turn on interlace handling,
2637 assuming you don't want to do it yourself.
2639 If you need to you can stop the processing of
2640 your original input data at this point by calling
2641 png_process_data_pause. This returns the number
2642 of unprocessed bytes from the last png_process_data
2643 call - it is up to you to ensure that the next call
2644 sees these bytes again. If you don't want to bother
2645 with this you can get libpng to cache the unread
2646 bytes by setting the 'save' parameter (see png.h) but
2647 then libpng will have to copy the data internally.
2651 /* This function is called when each row of image
2654 row_callback(png_structp png_ptr, png_bytep new_row,
2655 png_uint_32 row_num, int pass)
2657 /* If the image is interlaced, and you turned
2658 on the interlace handler, this function will
2659 be called for every row in every pass. Some
2660 of these rows will not be changed from the
2661 previous pass. When the row is not changed,
2662 the new_row variable will be NULL. The rows
2663 and passes are called in order, so you don't
2664 really need the row_num and pass, but I'm
2665 supplying them because it may make your life
2668 If you did not turn on interlace handling then
2669 the callback is called for each row of each
2670 sub-image when the image is interlaced. In this
2671 case 'row_num' is the row in the sub-image, not
2672 the row in the output image as it is in all other
2675 For the non-NULL rows of interlaced images when
2676 you have switched on libpng interlace handling,
2677 you must call png_progressive_combine_row()
2678 passing in the row and the old row. You can
2679 call this function for NULL rows (it will just
2680 return) and for non-interlaced images (it just
2681 does the memcpy for you) if it will make the
2682 code easier. Thus, you can just do this for
2683 all cases if you switch on interlace handling;
2686 png_progressive_combine_row(png_ptr, old_row,
2689 /* where old_row is what was displayed
2690 previously for the row. Note that the first
2691 pass (pass == 0, really) will completely cover
2692 the old row, so the rows do not have to be
2693 initialized. After the first pass (and only
2694 for interlaced images), you will have to pass
2695 the current row, and the function will combine
2696 the old row and the new row.
2698 You can also call png_process_data_pause in this
2699 callback - see above.
2704 end_callback(png_structp png_ptr, png_infop info)
2706 /* This function is called after the whole image
2707 has been read, including any chunks after the
2708 image (up to and including the IEND). You
2709 will usually have the same info chunk as you
2710 had in the header, although some data may have
2711 been added to the comments and time fields.
2713 Most people won't do much here, perhaps setting
2714 a flag that marks the image as finished.
2722 Much of this is very similar to reading. However, everything of
2723 importance is repeated here, so you won't have to constantly look
2724 back up in the reading section to understand writing.
2728 You will want to do the I/O initialization before you get into libpng,
2729 so if it doesn't work, you don't have anything to undo. If you are not
2730 using the standard I/O functions, you will need to replace them with
2731 custom writing functions. See the discussion under Customizing libpng.
2733 FILE *fp = fopen(file_name, "wb");
2738 Next, png_struct and png_info need to be allocated and initialized.
2739 As these can be both relatively large, you may not want to store these
2740 on the stack, unless you have stack space to spare. Of course, you
2741 will want to check if they return NULL. If you are also reading,
2742 you won't want to name your read structure and your write structure
2743 both "png_ptr"; you can call them anything you like, such as
2744 "read_ptr" and "write_ptr". Look at pngtest.c, for example.
2746 png_structp png_ptr = png_create_write_struct
2747 (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
2748 user_error_fn, user_warning_fn);
2753 png_infop info_ptr = png_create_info_struct(png_ptr);
2756 png_destroy_write_struct(&png_ptr,
2761 If you want to use your own memory allocation routines,
2762 define PNG_USER_MEM_SUPPORTED and use
2763 png_create_write_struct_2() instead of png_create_write_struct():
2765 png_structp png_ptr = png_create_write_struct_2
2766 (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
2767 user_error_fn, user_warning_fn, (png_voidp)
2768 user_mem_ptr, user_malloc_fn, user_free_fn);
2770 After you have these structures, you will need to set up the
2771 error handling. When libpng encounters an error, it expects to
2772 longjmp() back to your routine. Therefore, you will need to call
2773 setjmp() and pass the png_jmpbuf(png_ptr). If you
2774 write the file from different routines, you will need to update
2775 the png_jmpbuf(png_ptr) every time you enter a new routine that will
2776 call a png_*() function. See your documentation of setjmp/longjmp
2777 for your compiler for more information on setjmp/longjmp. See
2778 the discussion on libpng error handling in the Customizing Libpng
2779 section below for more information on the libpng error handling.
2781 if (setjmp(png_jmpbuf(png_ptr)))
2783 png_destroy_write_struct(&png_ptr, &info_ptr);
2790 If you would rather avoid the complexity of setjmp/longjmp issues,
2791 you can compile libpng with PNG_NO_SETJMP, in which case
2792 errors will result in a call to PNG_ABORT() which defaults to abort().
2794 You can #define PNG_ABORT() to a function that does something
2795 more useful than abort(), as long as your function does not
2798 Checking for invalid palette index on write was added at libpng
2799 1.5.10. If a pixel contains an invalid (out-of-range) index libpng issues
2800 a benign error. This is enabled by default because this condition is an
2801 error according to the PNG specification, Clause 11.3.2, but the error can
2802 be ignored in each png_ptr with
2804 png_set_check_for_invalid_index(png_ptr, 0);
2806 If the error is ignored, or if png_benign_error() treats it as a warning,
2807 any invalid pixels are written as-is by the encoder, resulting in an
2808 invalid PNG datastream as output. In this case the application is
2809 responsible for ensuring that the pixel indexes are in range when it writes
2810 a PLTE chunk with fewer entries than the bit depth would allow.
2812 Now you need to set up the output code. The default for libpng is to
2813 use the C function fwrite(). If you use this, you will need to pass a
2814 valid FILE * in the function png_init_io(). Be sure that the file is
2815 opened in binary mode. Again, if you wish to handle writing data in
2816 another way, see the discussion on libpng I/O handling in the Customizing
2817 Libpng section below.
2819 png_init_io(png_ptr, fp);
2821 If you are embedding your PNG into a datastream such as MNG, and don't
2822 want libpng to write the 8-byte signature, or if you have already
2823 written the signature in your application, use
2825 png_set_sig_bytes(png_ptr, 8);
2827 to inform libpng that it should not write a signature.
2831 At this point, you can set up a callback function that will be
2832 called after each row has been written, which you can use to control
2833 a progress meter or the like. It's demonstrated in pngtest.c.
2834 You must supply a function
2836 void write_row_callback(png_structp png_ptr, png_uint_32 row,
2839 /* put your code here */
2842 (You can give it another name that you like instead of "write_row_callback")
2844 To inform libpng about your function, use
2846 png_set_write_status_fn(png_ptr, write_row_callback);
2848 When this function is called the row has already been completely processed and
2849 it has also been written out. The 'row' and 'pass' refer to the next row to be
2851 non-interlaced case the row that was just handled is simply one less than the
2852 passed in row number, and pass will always be 0. For the interlaced case the
2853 same applies unless the row value is 0, in which case the row just handled was
2854 the last one from one of the preceding passes. Because interlacing may skip a
2855 pass you cannot be sure that the preceding pass is just 'pass-1', if you really
2856 need to know what the last pass is record (row,pass) from the callback and use
2857 the last recorded value each time.
2859 As with the user transform you can find the output row using the
2860 PNG_ROW_FROM_PASS_ROW macro.
2862 You now have the option of modifying how the compression library will
2863 run. The following functions are mainly for testing, but may be useful
2864 in some cases, like if you need to write PNG files extremely fast and
2865 are willing to give up some compression, or if you want to get the
2866 maximum possible compression at the expense of slower writing. If you
2867 have no special needs in this area, let the library do what it wants by
2868 not calling this function at all, as it has been tuned to deliver a good
2869 speed/compression ratio. The second parameter to png_set_filter() is
2870 the filter method, for which the only valid values are 0 (as of the
2871 July 1999 PNG specification, version 1.2) or 64 (if you are writing
2872 a PNG datastream that is to be embedded in a MNG datastream). The third
2873 parameter is a flag that indicates which filter type(s) are to be tested
2874 for each scanline. See the PNG specification for details on the specific
2878 /* turn on or off filtering, and/or choose
2879 specific filters. You can use either a single
2880 PNG_FILTER_VALUE_NAME or the bitwise OR of one
2881 or more PNG_FILTER_NAME masks.
2883 png_set_filter(png_ptr, 0,
2884 PNG_FILTER_NONE | PNG_FILTER_VALUE_NONE |
2885 PNG_FILTER_SUB | PNG_FILTER_VALUE_SUB |
2886 PNG_FILTER_UP | PNG_FILTER_VALUE_UP |
2887 PNG_FILTER_AVG | PNG_FILTER_VALUE_AVG |
2888 PNG_FILTER_PAETH | PNG_FILTER_VALUE_PAETH|
2889 PNG_ALL_FILTERS | PNG_FAST_FILTERS);
2891 If an application wants to start and stop using particular filters during
2892 compression, it should start out with all of the filters (to ensure that
2893 the previous row of pixels will be stored in case it's needed later),
2894 and then add and remove them after the start of compression.
2896 If you are writing a PNG datastream that is to be embedded in a MNG
2897 datastream, the second parameter can be either 0 or 64.
2899 The png_set_compression_*() functions interface to the zlib compression
2900 library, and should mostly be ignored unless you really know what you are
2901 doing. The only generally useful call is png_set_compression_level()
2902 which changes how much time zlib spends on trying to compress the image
2903 data. See the Compression Library (zlib.h and algorithm.txt, distributed
2904 with zlib) for details on the compression levels.
2908 /* Set the zlib compression level */
2909 png_set_compression_level(png_ptr,
2910 Z_BEST_COMPRESSION);
2912 /* Set other zlib parameters for compressing IDAT */
2913 png_set_compression_mem_level(png_ptr, 8);
2914 png_set_compression_strategy(png_ptr,
2915 Z_DEFAULT_STRATEGY);
2916 png_set_compression_window_bits(png_ptr, 15);
2917 png_set_compression_method(png_ptr, 8);
2918 png_set_compression_buffer_size(png_ptr, 8192)
2920 /* Set zlib parameters for text compression
2921 * If you don't call these, the parameters
2922 * fall back on those defined for IDAT chunks
2924 png_set_text_compression_mem_level(png_ptr, 8);
2925 png_set_text_compression_strategy(png_ptr,
2926 Z_DEFAULT_STRATEGY);
2927 png_set_text_compression_window_bits(png_ptr, 15);
2928 png_set_text_compression_method(png_ptr, 8);
2930 Setting the contents of info for output
2932 You now need to fill in the png_info structure with all the data you
2933 wish to write before the actual image. Note that the only thing you
2934 are allowed to write after the image is the text chunks and the time
2935 chunk (as of PNG Specification 1.2, anyway). See png_write_end() and
2936 the latest PNG specification for more information on that. If you
2937 wish to write them before the image, fill them in now, and flag that
2938 data as being valid. If you want to wait until after the data, don't
2939 fill them until png_write_end(). For all the fields in png_info and
2940 their data types, see png.h. For explanations of what the fields
2941 contain, see the PNG specification.
2943 Some of the more important parts of the png_info are:
2945 png_set_IHDR(png_ptr, info_ptr, width, height,
2946 bit_depth, color_type, interlace_type,
2947 compression_type, filter_method)
2949 width - holds the width of the image
2950 in pixels (up to 2^31).
2952 height - holds the height of the image
2953 in pixels (up to 2^31).
2955 bit_depth - holds the bit depth of one of the
2957 (valid values are 1, 2, 4, 8, 16
2958 and depend also on the
2959 color_type. See also significant
2962 color_type - describes which color/alpha
2963 channels are present.
2965 (bit depths 1, 2, 4, 8, 16)
2966 PNG_COLOR_TYPE_GRAY_ALPHA
2968 PNG_COLOR_TYPE_PALETTE
2969 (bit depths 1, 2, 4, 8)
2972 PNG_COLOR_TYPE_RGB_ALPHA
2975 PNG_COLOR_MASK_PALETTE
2976 PNG_COLOR_MASK_COLOR
2977 PNG_COLOR_MASK_ALPHA
2979 interlace_type - PNG_INTERLACE_NONE or
2982 compression_type - (must be
2983 PNG_COMPRESSION_TYPE_DEFAULT)
2985 filter_method - (must be PNG_FILTER_TYPE_DEFAULT
2986 or, if you are writing a PNG to
2987 be embedded in a MNG datastream,
2989 PNG_INTRAPIXEL_DIFFERENCING)
2991 If you call png_set_IHDR(), the call must appear before any of the
2992 other png_set_*() functions, because they might require access to some of
2993 the IHDR settings. The remaining png_set_*() functions can be called
2996 If you wish, you can reset the compression_type, interlace_type, or
2997 filter_method later by calling png_set_IHDR() again; if you do this, the
2998 width, height, bit_depth, and color_type must be the same in each call.
3000 png_set_PLTE(png_ptr, info_ptr, palette,
3003 palette - the palette for the file
3004 (array of png_color)
3005 num_palette - number of entries in the palette
3008 png_set_gAMA(png_ptr, info_ptr, file_gamma);
3009 png_set_gAMA_fixed(png_ptr, info_ptr, int_file_gamma);
3011 file_gamma - the gamma at which the image was
3012 created (PNG_INFO_gAMA)
3014 int_file_gamma - 100,000 times the gamma at which
3015 the image was created
3017 png_set_cHRM(png_ptr, info_ptr, white_x, white_y, red_x, red_y,
3018 green_x, green_y, blue_x, blue_y)
3019 png_set_cHRM_XYZ(png_ptr, info_ptr, red_X, red_Y, red_Z, green_X,
3020 green_Y, green_Z, blue_X, blue_Y, blue_Z)
3021 png_set_cHRM_fixed(png_ptr, info_ptr, int_white_x, int_white_y,
3022 int_red_x, int_red_y, int_green_x, int_green_y,
3023 int_blue_x, int_blue_y)
3024 png_set_cHRM_XYZ_fixed(png_ptr, info_ptr, int_red_X, int_red_Y,
3025 int_red_Z, int_green_X, int_green_Y, int_green_Z,
3026 int_blue_X, int_blue_Y, int_blue_Z)
3028 {white,red,green,blue}_{x,y}
3029 A color space encoding specified using the chromaticities
3030 of the end points and the white point.
3032 {red,green,blue}_{X,Y,Z}
3033 A color space encoding specified using the encoding end
3034 points - the CIE tristimulus specification of the intended
3035 color of the red, green and blue channels in the PNG RGB
3036 data. The white point is simply the sum of the three end
3039 png_set_sRGB(png_ptr, info_ptr, srgb_intent);
3041 srgb_intent - the rendering intent
3042 (PNG_INFO_sRGB) The presence of
3043 the sRGB chunk means that the pixel
3044 data is in the sRGB color space.
3045 This chunk also implies specific
3046 values of gAMA and cHRM. Rendering
3047 intent is the CSS-1 property that
3048 has been defined by the International
3050 (http://www.color.org).
3052 PNG_sRGB_INTENT_SATURATION,
3053 PNG_sRGB_INTENT_PERCEPTUAL,
3054 PNG_sRGB_INTENT_ABSOLUTE, or
3055 PNG_sRGB_INTENT_RELATIVE.
3058 png_set_sRGB_gAMA_and_cHRM(png_ptr, info_ptr,
3061 srgb_intent - the rendering intent
3062 (PNG_INFO_sRGB) The presence of the
3063 sRGB chunk means that the pixel
3064 data is in the sRGB color space.
3065 This function also causes gAMA and
3066 cHRM chunks with the specific values
3067 that are consistent with sRGB to be
3070 png_set_iCCP(png_ptr, info_ptr, name, compression_type,
3073 name - The profile name.
3075 compression_type - The compression type; always
3076 PNG_COMPRESSION_TYPE_BASE for PNG 1.0.
3077 You may give NULL to this argument to
3080 profile - International Color Consortium color
3081 profile data. May contain NULs.
3083 proflen - length of profile data in bytes.
3085 png_set_sBIT(png_ptr, info_ptr, sig_bit);
3087 sig_bit - the number of significant bits for
3088 (PNG_INFO_sBIT) each of the gray, red,
3089 green, and blue channels, whichever are
3090 appropriate for the given color type
3093 png_set_tRNS(png_ptr, info_ptr, trans_alpha,
3094 num_trans, trans_color);
3096 trans_alpha - array of alpha (transparency)
3097 entries for palette (PNG_INFO_tRNS)
3099 num_trans - number of transparent entries
3102 trans_color - graylevel or color sample values
3103 (in order red, green, blue) of the
3104 single transparent color for
3105 non-paletted images (PNG_INFO_tRNS)
3107 png_set_eXIf_1(png_ptr, info_ptr, num_exif, exif);
3109 exif - Exif profile (array of
3110 png_byte) (PNG_INFO_eXIf)
3112 png_set_hIST(png_ptr, info_ptr, hist);
3114 hist - histogram of palette (array of
3115 png_uint_16) (PNG_INFO_hIST)
3117 png_set_tIME(png_ptr, info_ptr, mod_time);
3119 mod_time - time image was last modified
3122 png_set_bKGD(png_ptr, info_ptr, background);
3124 background - background color (of type
3125 png_color_16p) (PNG_VALID_bKGD)
3127 png_set_text(png_ptr, info_ptr, text_ptr, num_text);
3129 text_ptr - array of png_text holding image
3132 text_ptr[i].compression - type of compression used
3133 on "text" PNG_TEXT_COMPRESSION_NONE
3134 PNG_TEXT_COMPRESSION_zTXt
3135 PNG_ITXT_COMPRESSION_NONE
3136 PNG_ITXT_COMPRESSION_zTXt
3137 text_ptr[i].key - keyword for comment. Must contain
3139 text_ptr[i].text - text comments for current
3140 keyword. Can be NULL or empty.
3141 text_ptr[i].text_length - length of text string,
3142 after decompression, 0 for iTXt
3143 text_ptr[i].itxt_length - length of itxt string,
3144 after decompression, 0 for tEXt/zTXt
3145 text_ptr[i].lang - language of comment (NULL or
3147 text_ptr[i].translated_keyword - keyword in UTF-8 (NULL
3148 or empty for unknown).
3150 Note that the itxt_length, lang, and lang_key
3151 members of the text_ptr structure only exist when the
3152 library is built with iTXt chunk support. Prior to
3153 libpng-1.4.0 the library was built by default without
3154 iTXt support. Also note that when iTXt is supported,
3155 they contain NULL pointers when the "compression"
3156 field contains PNG_TEXT_COMPRESSION_NONE or
3157 PNG_TEXT_COMPRESSION_zTXt.
3159 num_text - number of comments
3161 png_set_sPLT(png_ptr, info_ptr, &palette_ptr,
3164 palette_ptr - array of png_sPLT_struct structures
3165 to be added to the list of palettes
3166 in the info structure.
3167 num_spalettes - number of palette structures to be
3170 png_set_oFFs(png_ptr, info_ptr, offset_x, offset_y,
3173 offset_x - positive offset from the left
3176 offset_y - positive offset from the top
3179 unit_type - PNG_OFFSET_PIXEL, PNG_OFFSET_MICROMETER
3181 png_set_pHYs(png_ptr, info_ptr, res_x, res_y,
3184 res_x - pixels/unit physical resolution
3187 res_y - pixels/unit physical resolution
3190 unit_type - PNG_RESOLUTION_UNKNOWN,
3191 PNG_RESOLUTION_METER
3193 png_set_sCAL(png_ptr, info_ptr, unit, width, height)
3195 unit - physical scale units (an integer)
3197 width - width of a pixel in physical scale units
3199 height - height of a pixel in physical scale units
3200 (width and height are doubles)
3202 png_set_sCAL_s(png_ptr, info_ptr, unit, width, height)
3204 unit - physical scale units (an integer)
3206 width - width of a pixel in physical scale units
3207 expressed as a string
3209 height - height of a pixel in physical scale units
3210 (width and height are strings like "2.54")
3212 png_set_unknown_chunks(png_ptr, info_ptr, &unknowns,
3215 unknowns - array of png_unknown_chunk
3216 structures holding unknown chunks
3217 unknowns[i].name - name of unknown chunk
3218 unknowns[i].data - data of unknown chunk
3219 unknowns[i].size - size of unknown chunk's data
3220 unknowns[i].location - position to write chunk in file
3221 0: do not write chunk
3222 PNG_HAVE_IHDR: before PLTE
3223 PNG_HAVE_PLTE: before IDAT
3224 PNG_AFTER_IDAT: after IDAT
3226 The "location" member is set automatically according to
3227 what part of the output file has already been written.
3228 You can change its value after calling png_set_unknown_chunks()
3229 as demonstrated in pngtest.c. Within each of the "locations",
3230 the chunks are sequenced according to their position in the
3231 structure (that is, the value of "i", which is the order in which
3232 the chunk was either read from the input file or defined with
3233 png_set_unknown_chunks).
3235 A quick word about text and num_text. text is an array of png_text
3236 structures. num_text is the number of valid structures in the array.
3237 Each png_text structure holds a language code, a keyword, a text value,
3238 and a compression type.
3240 The compression types have the same valid numbers as the compression
3241 types of the image data. Currently, the only valid number is zero.
3242 However, you can store text either compressed or uncompressed, unlike
3243 images, which always have to be compressed. So if you don't want the
3244 text compressed, set the compression type to PNG_TEXT_COMPRESSION_NONE.
3245 Because tEXt and zTXt chunks don't have a language field, if you
3246 specify PNG_TEXT_COMPRESSION_NONE or PNG_TEXT_COMPRESSION_zTXt
3247 any language code or translated keyword will not be written out.
3249 Until text gets around a few hundred bytes, it is not worth compressing it.
3250 After the text has been written out to the file, the compression type
3251 is set to PNG_TEXT_COMPRESSION_NONE_WR or PNG_TEXT_COMPRESSION_zTXt_WR,
3252 so that it isn't written out again at the end (in case you are calling
3253 png_write_end() with the same struct).
3255 The keywords that are given in the PNG Specification are:
3257 Title Short (one line) title or
3260 Author Name of image's creator
3262 Description Description of image (possibly long)
3264 Copyright Copyright notice
3266 Creation Time Time of original image creation
3267 (usually RFC 1123 format, see below)
3269 Software Software used to create the image
3271 Disclaimer Legal disclaimer
3273 Warning Warning of nature of content
3275 Source Device used to create the image
3277 Comment Miscellaneous comment; conversion
3278 from other image format
3280 The keyword-text pairs work like this. Keywords should be short
3281 simple descriptions of what the comment is about. Some typical
3282 keywords are found in the PNG specification, as is some recommendations
3283 on keywords. You can repeat keywords in a file. You can even write
3284 some text before the image and some after. For example, you may want
3285 to put a description of the image before the image, but leave the
3286 disclaimer until after, so viewers working over modem connections
3287 don't have to wait for the disclaimer to go over the modem before
3288 they start seeing the image. Finally, keywords should be full
3289 words, not abbreviations. Keywords and text are in the ISO 8859-1
3290 (Latin-1) character set (a superset of regular ASCII) and can not
3291 contain NUL characters, and should not contain control or other
3292 unprintable characters. To make the comments widely readable, stick
3293 with basic ASCII, and avoid machine specific character set extensions
3294 like the IBM-PC character set. The keyword must be present, but
3295 you can leave off the text string on non-compressed pairs.
3296 Compressed pairs must have a text string, as only the text string
3297 is compressed anyway, so the compression would be meaningless.
3299 PNG supports modification time via the png_time structure. Two
3300 conversion routines are provided, png_convert_from_time_t() for
3301 time_t and png_convert_from_struct_tm() for struct tm. The
3302 time_t routine uses gmtime(). You don't have to use either of
3303 these, but if you wish to fill in the png_time structure directly,
3304 you should provide the time in universal time (GMT) if possible
3305 instead of your local time. Note that the year number is the full
3306 year (e.g. 1998, rather than 98 - PNG is year 2000 compliant!), and
3307 that months start with 1.
3309 If you want to store the time of the original image creation, you should
3310 use a plain tEXt chunk with the "Creation Time" keyword. This is
3311 necessary because the "creation time" of a PNG image is somewhat vague,
3312 depending on whether you mean the PNG file, the time the image was
3313 created in a non-PNG format, a still photo from which the image was
3314 scanned, or possibly the subject matter itself. In order to facilitate
3315 machine-readable dates, it is recommended that the "Creation Time"
3316 tEXt chunk use RFC 1123 format dates (e.g. "22 May 1997 18:07:10 GMT"),
3317 although this isn't a requirement. Unlike the tIME chunk, the
3318 "Creation Time" tEXt chunk is not expected to be automatically changed
3319 by the software. To facilitate the use of RFC 1123 dates, a function
3320 png_convert_to_rfc1123_buffer(buffer, png_timep) is provided to
3321 convert from PNG time to an RFC 1123 format string. The caller must provide
3322 a writeable buffer of at least 29 bytes.
3324 Writing unknown chunks
3326 You can use the png_set_unknown_chunks function to queue up private chunks
3327 for writing. You give it a chunk name, location, raw data, and a size. You
3328 also must use png_set_keep_unknown_chunks() to ensure that libpng will
3329 handle them. That's all there is to it. The chunks will be written by the
3330 next following png_write_info_before_PLTE, png_write_info, or png_write_end
3331 function, depending upon the specified location. Any chunks previously
3332 read into the info structure's unknown-chunk list will also be written out
3333 in a sequence that satisfies the PNG specification's ordering rules.
3335 Here is an example of writing two private chunks, prVt and miNE:
3337 #ifdef PNG_WRITE_UNKNOWN_CHUNKS_SUPPORTED
3338 /* Set unknown chunk data */
3339 png_unknown_chunk unk_chunk[2];
3340 strcpy((char *) unk_chunk[0].name, "prVt";
3341 unk_chunk[0].data = (unsigned char *) "PRIVATE DATA";
3342 unk_chunk[0].size = strlen(unk_chunk[0].data)+1;
3343 unk_chunk[0].location = PNG_HAVE_IHDR;
3344 strcpy((char *) unk_chunk[1].name, "miNE";
3345 unk_chunk[1].data = (unsigned char *) "MY CHUNK DATA";
3346 unk_chunk[1].size = strlen(unk_chunk[0].data)+1;
3347 unk_chunk[1].location = PNG_AFTER_IDAT;
3348 png_set_unknown_chunks(write_ptr, write_info_ptr,
3350 /* Needed because miNE is not safe-to-copy */
3351 png_set_keep_unknown_chunks(png, PNG_HANDLE_CHUNK_ALWAYS,
3352 (png_bytep) "miNE", 1);
3353 # if PNG_LIBPNG_VER < 10600
3354 /* Deal with unknown chunk location bug in 1.5.x and earlier */
3355 png_set_unknown_chunk_location(png, info, 0, PNG_HAVE_IHDR);
3356 png_set_unknown_chunk_location(png, info, 1, PNG_AFTER_IDAT);
3358 # if PNG_LIBPNG_VER < 10500
3359 /* PNG_AFTER_IDAT writes two copies of the chunk prior to libpng-1.5.0,
3360 * one before IDAT and another after IDAT, so don't use it; only use
3361 * PNG_HAVE_IHDR location. This call resets the location previously
3362 * set by assignment and png_set_unknown_chunk_location() for chunk 1.
3364 png_set_unknown_chunk_location(png, info, 1, PNG_HAVE_IHDR);
3368 The high-level write interface
3370 At this point there are two ways to proceed; through the high-level
3371 write interface, or through a sequence of low-level write operations.
3372 You can use the high-level interface if your image data is present
3373 in the info structure. All defined output
3374 transformations are permitted, enabled by the following masks.
3376 PNG_TRANSFORM_IDENTITY No transformation
3377 PNG_TRANSFORM_PACKING Pack 1, 2 and 4-bit samples
3378 PNG_TRANSFORM_PACKSWAP Change order of packed
3380 PNG_TRANSFORM_INVERT_MONO Invert monochrome images
3381 PNG_TRANSFORM_SHIFT Normalize pixels to the
3383 PNG_TRANSFORM_BGR Flip RGB to BGR, RGBA
3385 PNG_TRANSFORM_SWAP_ALPHA Flip RGBA to ARGB or GA
3387 PNG_TRANSFORM_INVERT_ALPHA Change alpha from opacity
3389 PNG_TRANSFORM_SWAP_ENDIAN Byte-swap 16-bit samples
3390 PNG_TRANSFORM_STRIP_FILLER Strip out filler
3392 PNG_TRANSFORM_STRIP_FILLER_BEFORE Strip out leading
3394 PNG_TRANSFORM_STRIP_FILLER_AFTER Strip out trailing
3397 If you have valid image data in the info structure (you can use
3398 png_set_rows() to put image data in the info structure), simply do this:
3400 png_write_png(png_ptr, info_ptr, png_transforms, NULL)
3402 where png_transforms is an integer containing the bitwise OR of some set of
3403 transformation flags. This call is equivalent to png_write_info(),
3404 followed the set of transformations indicated by the transform mask,
3405 then png_write_image(), and finally png_write_end().
3407 (The final parameter of this call is not yet used. Someday it might point
3408 to transformation parameters required by some future output transform.)
3410 You must use png_transforms and not call any png_set_transform() functions
3411 when you use png_write_png().
3413 The low-level write interface
3415 If you are going the low-level route instead, you are now ready to
3416 write all the file information up to the actual image data. You do
3417 this with a call to png_write_info().
3419 png_write_info(png_ptr, info_ptr);
3421 Note that there is one transformation you may need to do before
3422 png_write_info(). In PNG files, the alpha channel in an image is the
3423 level of opacity. If your data is supplied as a level of transparency,
3424 you can invert the alpha channel before you write it, so that 0 is
3425 fully transparent and 255 (in 8-bit or paletted images) or 65535
3426 (in 16-bit images) is fully opaque, with
3428 png_set_invert_alpha(png_ptr);
3430 This must appear before png_write_info() instead of later with the
3431 other transformations because in the case of paletted images the tRNS
3432 chunk data has to be inverted before the tRNS chunk is written. If
3433 your image is not a paletted image, the tRNS data (which in such cases
3434 represents a single color to be rendered as transparent) won't need to
3435 be changed, and you can safely do this transformation after your
3436 png_write_info() call.
3438 If you need to write a private chunk that you want to appear before
3439 the PLTE chunk when PLTE is present, you can write the PNG info in
3440 two steps, and insert code to write your own chunk between them:
3442 png_write_info_before_PLTE(png_ptr, info_ptr);
3443 png_set_unknown_chunks(png_ptr, info_ptr, ...);
3444 png_write_info(png_ptr, info_ptr);
3446 After you've written the file information, you can set up the library
3447 to handle any special transformations of the image data. The various
3448 ways to transform the data will be described in the order that they
3449 should occur. This is important, as some of these change the color
3450 type and/or bit depth of the data, and some others only work on
3451 certain color types and bit depths. Even though each transformation
3452 checks to see if it has data that it can do something with, you should
3453 make sure to only enable a transformation if it will be valid for the
3454 data. For example, don't swap red and blue on grayscale data.
3456 PNG files store RGB pixels packed into 3 or 6 bytes. This code tells
3457 the library to strip input data that has 4 or 8 bytes per pixel down
3458 to 3 or 6 bytes (or strip 2 or 4-byte grayscale+filler data to 1 or 2
3461 png_set_filler(png_ptr, 0, PNG_FILLER_BEFORE);
3463 where the 0 is unused, and the location is either PNG_FILLER_BEFORE or
3464 PNG_FILLER_AFTER, depending upon whether the filler byte in the pixel
3465 is stored XRGB or RGBX.
3467 PNG files pack pixels of bit depths 1, 2, and 4 into bytes as small as
3468 they can, resulting in, for example, 8 pixels per byte for 1 bit files.
3469 If the data is supplied at 1 pixel per byte, use this code, which will
3470 correctly pack the pixels into a single byte:
3472 png_set_packing(png_ptr);
3474 PNG files reduce possible bit depths to 1, 2, 4, 8, and 16. If your
3475 data is of another bit depth, you can write an sBIT chunk into the
3476 file so that decoders can recover the original data if desired.
3478 /* Set the true bit depth of the image data */
3479 if (color_type & PNG_COLOR_MASK_COLOR)
3481 sig_bit.red = true_bit_depth;
3482 sig_bit.green = true_bit_depth;
3483 sig_bit.blue = true_bit_depth;
3488 sig_bit.gray = true_bit_depth;
3491 if (color_type & PNG_COLOR_MASK_ALPHA)
3493 sig_bit.alpha = true_bit_depth;
3496 png_set_sBIT(png_ptr, info_ptr, &sig_bit);
3498 If the data is stored in the row buffer in a bit depth other than
3499 one supported by PNG (e.g. 3 bit data in the range 0-7 for a 4-bit PNG),
3500 this will scale the values to appear to be the correct bit depth as
3503 png_set_shift(png_ptr, &sig_bit);
3505 PNG files store 16-bit pixels in network byte order (big-endian,
3506 ie. most significant bits first). This code would be used if they are
3507 supplied the other way (little-endian, i.e. least significant bits
3508 first, the way PCs store them):
3511 png_set_swap(png_ptr);
3513 If you are using packed-pixel images (1, 2, or 4 bits/pixel), and you
3514 need to change the order the pixels are packed into bytes, you can use:
3517 png_set_packswap(png_ptr);
3519 PNG files store 3 color pixels in red, green, blue order. This code
3520 would be used if they are supplied as blue, green, red:
3522 png_set_bgr(png_ptr);
3524 PNG files describe monochrome as black being zero and white being
3525 one. This code would be used if the pixels are supplied with this reversed
3526 (black being one and white being zero):
3528 png_set_invert_mono(png_ptr);
3530 Finally, you can write your own transformation function if none of
3531 the existing ones meets your needs. This is done by setting a callback
3534 png_set_write_user_transform_fn(png_ptr,
3535 write_transform_fn);
3537 You must supply the function
3539 void write_transform_fn(png_structp png_ptr, png_row_infop
3540 row_info, png_bytep data)
3542 See pngtest.c for a working example. Your function will be called
3543 before any of the other transformations are processed. If supported
3544 libpng also supplies an information routine that may be called from
3547 png_get_current_row_number(png_ptr);
3548 png_get_current_pass_number(png_ptr);
3550 This returns the current row passed to the transform. With interlaced
3551 images the value returned is the row in the input sub-image image. Use
3552 PNG_ROW_FROM_PASS_ROW(row, pass) and PNG_COL_FROM_PASS_COL(col, pass) to
3553 find the output pixel (x,y) given an interlaced sub-image pixel (row,col,pass).
3555 The discussion of interlace handling above contains more information on how to
3558 You can also set up a pointer to a user structure for use by your
3561 png_set_user_transform_info(png_ptr, user_ptr, 0, 0);
3563 The user_channels and user_depth parameters of this function are ignored
3564 when writing; you can set them to zero as shown.
3566 You can retrieve the pointer via the function png_get_user_transform_ptr().
3569 voidp write_user_transform_ptr =
3570 png_get_user_transform_ptr(png_ptr);
3572 It is possible to have libpng flush any pending output, either manually,
3573 or automatically after a certain number of lines have been written. To
3574 flush the output stream a single time call:
3576 png_write_flush(png_ptr);
3578 and to have libpng flush the output stream periodically after a certain
3579 number of scanlines have been written, call:
3581 png_set_flush(png_ptr, nrows);
3583 Note that the distance between rows is from the last time png_write_flush()
3584 was called, or the first row of the image if it has never been called.
3585 So if you write 50 lines, and then png_set_flush 25, it will flush the
3586 output on the next scanline, and every 25 lines thereafter, unless
3587 png_write_flush() is called before 25 more lines have been written.
3588 If nrows is too small (less than about 10 lines for a 640 pixel wide
3589 RGB image) the image compression may decrease noticeably (although this
3590 may be acceptable for real-time applications). Infrequent flushing will
3591 only degrade the compression performance by a few percent over images
3592 that do not use flushing.
3594 Writing the image data
3596 That's it for the transformations. Now you can write the image data.
3597 The simplest way to do this is in one function call. If you have the
3598 whole image in memory, you can just call png_write_image() and libpng
3599 will write the image. You will need to pass in an array of pointers to
3600 each row. This function automatically handles interlacing, so you don't
3601 need to call png_set_interlace_handling() or call this function multiple
3602 times, or any of that other stuff necessary with png_write_rows().
3604 png_write_image(png_ptr, row_pointers);
3606 where row_pointers is:
3608 png_byte *row_pointers[height];
3610 You can point to void or char or whatever you use for pixels.
3612 If you don't want to write the whole image at once, you can
3613 use png_write_rows() instead. If the file is not interlaced,
3616 png_write_rows(png_ptr, row_pointers,
3619 row_pointers is the same as in the png_write_image() call.
3621 If you are just writing one row at a time, you can do this with
3622 a single row_pointer instead of an array of row_pointers:
3624 png_bytep row_pointer = row;
3626 png_write_row(png_ptr, row_pointer);
3628 When the file is interlaced, things can get a good deal more complicated.
3629 The only currently (as of the PNG Specification version 1.2, dated July
3630 1999) defined interlacing scheme for PNG files is the "Adam7" interlace
3631 scheme, that breaks down an image into seven smaller images of varying
3632 size. libpng will build these images for you, or you can do them
3633 yourself. If you want to build them yourself, see the PNG specification
3634 for details of which pixels to write when.
3636 If you don't want libpng to handle the interlacing details, just
3637 use png_set_interlace_handling() and call png_write_rows() the
3638 correct number of times to write all the sub-images
3639 (png_set_interlace_handling() returns the number of sub-images.)
3641 If you want libpng to build the sub-images, call this before you start
3644 number_of_passes = png_set_interlace_handling(png_ptr);
3646 This will return the number of passes needed. Currently, this is seven,
3647 but may change if another interlace type is added.
3649 Then write the complete image number_of_passes times.
3651 png_write_rows(png_ptr, row_pointers, number_of_rows);
3653 Think carefully before you write an interlaced image. Typically code that
3654 reads such images reads all the image data into memory, uncompressed, before
3655 doing any processing. Only code that can display an image on the fly can
3656 take advantage of the interlacing and even then the image has to be exactly
3657 the correct size for the output device, because scaling an image requires
3658 adjacent pixels and these are not available until all the passes have been
3661 If you do write an interlaced image you will hardly ever need to handle
3662 the interlacing yourself. Call png_set_interlace_handling() and use the
3663 approach described above.
3665 The only time it is conceivable that you will really need to write an
3666 interlaced image pass-by-pass is when you have read one pass by pass and
3667 made some pixel-by-pixel transformation to it, as described in the read
3668 code above. In this case use the PNG_PASS_ROWS and PNG_PASS_COLS macros
3669 to determine the size of each sub-image in turn and simply write the rows
3670 you obtained from the read code.
3672 Finishing a sequential write
3674 After you are finished writing the image, you should finish writing
3675 the file. If you are interested in writing comments or time, you should
3676 pass an appropriately filled png_info pointer. If you are not interested,
3679 png_write_end(png_ptr, info_ptr);
3681 When you are done, you can free all memory used by libpng like this:
3683 png_destroy_write_struct(&png_ptr, &info_ptr);
3685 It is also possible to individually free the info_ptr members that
3686 point to libpng-allocated storage with the following function:
3688 png_free_data(png_ptr, info_ptr, mask, seq)
3690 mask - identifies data to be freed, a mask
3691 containing the bitwise OR of one or
3693 PNG_FREE_PLTE, PNG_FREE_TRNS,
3694 PNG_FREE_HIST, PNG_FREE_ICCP,
3695 PNG_FREE_PCAL, PNG_FREE_ROWS,
3696 PNG_FREE_SCAL, PNG_FREE_SPLT,
3697 PNG_FREE_TEXT, PNG_FREE_UNKN,
3698 or simply PNG_FREE_ALL
3700 seq - sequence number of item to be freed
3703 This function may be safely called when the relevant storage has
3704 already been freed, or has not yet been allocated, or was allocated
3705 by the user and not by libpng, and will in those cases do nothing.
3706 The "seq" parameter is ignored if only one item of the selected data
3707 type, such as PLTE, is allowed. If "seq" is not -1, and multiple items
3708 are allowed for the data type identified in the mask, such as text or
3709 sPLT, only the n'th item in the structure is freed, where n is "seq".
3711 If you allocated data such as a palette that you passed in to libpng
3712 with png_set_*, you must not free it until just before the call to
3713 png_destroy_write_struct().
3715 The default behavior is only to free data that was allocated internally
3716 by libpng. This can be changed, so that libpng will not free the data,
3717 or so that it will free data that was allocated by the user with png_malloc()
3718 or png_calloc() and passed in via a png_set_*() function, with
3720 png_data_freer(png_ptr, info_ptr, freer, mask)
3723 PNG_DESTROY_WILL_FREE_DATA
3724 PNG_SET_WILL_FREE_DATA
3725 PNG_USER_WILL_FREE_DATA
3727 mask - which data elements are affected
3728 same choices as in png_free_data()
3730 For example, to transfer responsibility for some data from a read structure
3731 to a write structure, you could use
3733 png_data_freer(read_ptr, read_info_ptr,
3734 PNG_USER_WILL_FREE_DATA,
3735 PNG_FREE_PLTE|PNG_FREE_tRNS|PNG_FREE_hIST)
3737 png_data_freer(write_ptr, write_info_ptr,
3738 PNG_DESTROY_WILL_FREE_DATA,
3739 PNG_FREE_PLTE|PNG_FREE_tRNS|PNG_FREE_hIST)
3741 thereby briefly reassigning responsibility for freeing to the user but
3742 immediately afterwards reassigning it once more to the write_destroy
3743 function. Having done this, it would then be safe to destroy the read
3744 structure and continue to use the PLTE, tRNS, and hIST data in the write
3747 This function only affects data that has already been allocated.
3748 You can call this function before calling after the png_set_*() functions
3749 to control whether the user or png_destroy_*() is supposed to free the data.
3750 When the user assumes responsibility for libpng-allocated data, the
3751 application must use
3752 png_free() to free it, and when the user transfers responsibility to libpng
3753 for data that the user has allocated, the user must have used png_malloc()
3754 or png_calloc() to allocate it.
3756 If you allocated text_ptr.text, text_ptr.lang, and text_ptr.translated_keyword
3757 separately, do not transfer responsibility for freeing text_ptr to libpng,
3758 because when libpng fills a png_text structure it combines these members with
3759 the key member, and png_free_data() will free only text_ptr.key. Similarly,
3760 if you transfer responsibility for free'ing text_ptr from libpng to your
3761 application, your application must not separately free those members.
3762 For a more compact example of writing a PNG image, see the file example.c.
3766 The simplified API, which became available in libpng-1.6.0, hides the details
3767 of both libpng and the PNG file format itself.
3768 It allows PNG files to be read into a very limited number of
3769 in-memory bitmap formats or to be written from the same formats. If these
3770 formats do not accommodate your needs then you can, and should, use the more
3771 sophisticated APIs above - these support a wide variety of in-memory formats
3772 and a wide variety of sophisticated transformations to those formats as well
3773 as a wide variety of APIs to manipulate ancilliary information.
3775 To read a PNG file using the simplified API:
3777 1) Declare a 'png_image' structure (see below) on the stack, set the
3778 version field to PNG_IMAGE_VERSION and the 'opaque' pointer to NULL
3779 (this is REQUIRED, your program may crash if you don't do it.)
3781 2) Call the appropriate png_image_begin_read... function.
3783 3) Set the png_image 'format' member to the required sample format.
3785 4) Allocate a buffer for the image and, if required, the color-map.
3787 5) Call png_image_finish_read to read the image and, if required, the
3788 color-map into your buffers.
3790 There are no restrictions on the format of the PNG input itself; all valid
3791 color types, bit depths, and interlace methods are acceptable, and the
3792 input image is transformed as necessary to the requested in-memory format
3793 during the png_image_finish_read() step. The only caveat is that if you
3794 request a color-mapped image from a PNG that is full-color or makes
3795 complex use of an alpha channel the transformation is extremely lossy and the
3796 result may look terrible.
3798 To write a PNG file using the simplified API:
3800 1) Declare a 'png_image' structure on the stack and memset()
3803 2) Initialize the members of the structure that describe the
3804 image, setting the 'format' member to the format of the
3807 3) Call the appropriate png_image_write... function with a
3808 pointer to the image and, if necessary, the color-map to write
3811 png_image is a structure that describes the in-memory format of an image
3812 when it is being read or defines the in-memory format of an image that you
3813 need to write. The "png_image" structure contains the following members:
3815 png_controlp opaque Initialize to NULL, free with png_image_free
3816 png_uint_32 version Set to PNG_IMAGE_VERSION
3817 png_uint_32 width Image width in pixels (columns)
3818 png_uint_32 height Image height in pixels (rows)
3819 png_uint_32 format Image format as defined below
3820 png_uint_32 flags A bit mask containing informational flags
3821 png_uint_32 colormap_entries; Number of entries in the color-map
3822 png_uint_32 warning_or_error;
3825 In the event of an error or warning the "warning_or_error"
3826 field will be set to a non-zero value and the 'message' field will contain
3827 a '\0' terminated string with the libpng error or warning message. If both
3828 warnings and an error were encountered, only the error is recorded. If there
3829 are multiple warnings, only the first one is recorded.
3831 The upper 30 bits of the "warning_or_error" value are reserved; the low two
3832 bits contain a two bit code such that a value more than 1 indicates a failure
3833 in the API just called:
3835 0 - no warning or error
3838 3 - error preceded by warning
3840 The pixels (samples) of the image have one to four channels whose components
3841 have original values in the range 0 to 1.0:
3843 1: A single gray or luminance channel (G).
3844 2: A gray/luminance channel and an alpha channel (GA).
3845 3: Three red, green, blue color channels (RGB).
3846 4: Three color channels and an alpha channel (RGBA).
3848 The channels are encoded in one of two ways:
3850 a) As a small integer, value 0..255, contained in a single byte. For the
3851 alpha channel the original value is simply value/255. For the color or
3852 luminance channels the value is encoded according to the sRGB specification
3853 and matches the 8-bit format expected by typical display devices.
3855 The color/gray channels are not scaled (pre-multiplied) by the alpha
3856 channel and are suitable for passing to color management software.
3858 b) As a value in the range 0..65535, contained in a 2-byte integer, in
3859 the native byte order of the platform on which the application is running.
3860 All channels can be converted to the original value by dividing by 65535; all
3861 channels are linear. Color channels use the RGB encoding (RGB end-points) of
3862 the sRGB specification. This encoding is identified by the
3863 PNG_FORMAT_FLAG_LINEAR flag below.
3865 When the simplified API needs to convert between sRGB and linear colorspaces,
3866 the actual sRGB transfer curve defined in the sRGB specification (see the
3867 article at https://en.wikipedia.org/wiki/SRGB) is used, not the gamma=1/2.2
3868 approximation used elsewhere in libpng.
3870 When an alpha channel is present it is expected to denote pixel coverage
3871 of the color or luminance channels and is returned as an associated alpha
3872 channel: the color/gray channels are scaled (pre-multiplied) by the alpha
3875 The samples are either contained directly in the image data, between 1 and 8
3876 bytes per pixel according to the encoding, or are held in a color-map indexed
3877 by bytes in the image data. In the case of a color-map the color-map entries
3878 are individual samples, encoded as above, and the image data has one byte per
3879 pixel to select the relevant sample from the color-map.
3883 The #defines to be used in png_image::format. Each #define identifies a
3884 particular layout of channel data and, if present, alpha values. There are
3885 separate defines for each of the two component encodings.
3887 A format is built up using single bit flag values. All combinations are
3888 valid. Formats can be built up from the flag values or you can use one of
3889 the predefined values below. When testing formats always use the FORMAT_FLAG
3890 macros to test for individual features - future versions of the library may
3893 When reading or writing color-mapped images the format should be set to the
3894 format of the entries in the color-map then png_image_{read,write}_colormap
3895 called to read or write the color-map and set the format correctly for the
3896 image data. Do not set the PNG_FORMAT_FLAG_COLORMAP bit directly!
3898 NOTE: libpng can be built with particular features disabled. If you see
3899 compiler errors because the definition of one of the following flags has been
3900 compiled out it is because libpng does not have the required support. It is
3901 possible, however, for the libpng configuration to enable the format on just
3902 read or just write; in that case you may see an error at run time.
3903 You can guard against this by checking for the definition of the
3904 appropriate "_SUPPORTED" macro, one of:
3906 PNG_SIMPLIFIED_{READ,WRITE}_{BGR,AFIRST}_SUPPORTED
3908 PNG_FORMAT_FLAG_ALPHA format with an alpha channel
3909 PNG_FORMAT_FLAG_COLOR color format: otherwise grayscale
3910 PNG_FORMAT_FLAG_LINEAR 2-byte channels else 1-byte
3911 PNG_FORMAT_FLAG_COLORMAP image data is color-mapped
3912 PNG_FORMAT_FLAG_BGR BGR colors, else order is RGB
3913 PNG_FORMAT_FLAG_AFIRST alpha channel comes first
3915 Supported formats are as follows. Future versions of libpng may support more
3916 formats; for compatibility with older versions simply check if the format
3917 macro is defined using #ifdef. These defines describe the in-memory layout
3918 of the components of the pixels of the image.
3920 First the single byte (sRGB) formats:
3932 Then the linear 2-byte formats. When naming these "Y" is used to
3933 indicate a luminance (gray) channel. The component order within the pixel
3934 is always the same - there is no provision for swapping the order of the
3935 components in the linear format. The components are 16-bit integers in
3936 the native byte order for your platform, and there is no provision for
3937 swapping the bytes to a different endian condition.
3940 PNG_FORMAT_LINEAR_Y_ALPHA
3941 PNG_FORMAT_LINEAR_RGB
3942 PNG_FORMAT_LINEAR_RGB_ALPHA
3944 With color-mapped formats the image data is one byte for each pixel. The byte
3945 is an index into the color-map which is formatted as above. To obtain a
3946 color-mapped format it is sufficient just to add the PNG_FOMAT_FLAG_COLORMAP
3947 to one of the above definitions, or you can use one of the definitions below.
3949 PNG_FORMAT_RGB_COLORMAP
3950 PNG_FORMAT_BGR_COLORMAP
3951 PNG_FORMAT_RGBA_COLORMAP
3952 PNG_FORMAT_ARGB_COLORMAP
3953 PNG_FORMAT_BGRA_COLORMAP
3954 PNG_FORMAT_ABGR_COLORMAP
3958 These are convenience macros to derive information from a png_image
3959 structure. The PNG_IMAGE_SAMPLE_ macros return values appropriate to the
3960 actual image sample values - either the entries in the color-map or the
3961 pixels in the image. The PNG_IMAGE_PIXEL_ macros return corresponding values
3962 for the pixels and will always return 1 for color-mapped formats. The
3963 remaining macros return information about the rows in the image and the
3966 NOTE: All the macros that take a png_image::format parameter are compile time
3967 constants if the format parameter is, itself, a constant. Therefore these
3968 macros can be used in array declarations and case labels where required.
3969 Similarly the macros are also pre-processor constants (sizeof is not used) so
3970 they can be used in #if tests.
3972 PNG_IMAGE_SAMPLE_CHANNELS(fmt)
3973 Returns the total number of channels in a given format: 1..4
3975 PNG_IMAGE_SAMPLE_COMPONENT_SIZE(fmt)
3976 Returns the size in bytes of a single component of a pixel or color-map
3977 entry (as appropriate) in the image: 1 or 2.
3979 PNG_IMAGE_SAMPLE_SIZE(fmt)
3980 This is the size of the sample data for one sample. If the image is
3981 color-mapped it is the size of one color-map entry (and image pixels are
3982 one byte in size), otherwise it is the size of one image pixel.
3984 PNG_IMAGE_MAXIMUM_COLORMAP_COMPONENTS(fmt)
3985 The maximum size of the color-map required by the format expressed in a
3986 count of components. This can be used to compile-time allocate a
3989 png_uint_16 colormap[PNG_IMAGE_MAXIMUM_COLORMAP_COMPONENTS(linear_fmt)];
3991 png_byte colormap[PNG_IMAGE_MAXIMUM_COLORMAP_COMPONENTS(sRGB_fmt)];
3993 Alternatively use the PNG_IMAGE_COLORMAP_SIZE macro below to use the
3994 information from one of the png_image_begin_read_ APIs and dynamically
3995 allocate the required memory.
3997 PNG_IMAGE_COLORMAP_SIZE(fmt)
3998 The size of the color-map required by the format; this is the size of the
3999 color-map buffer passed to the png_image_{read,write}_colormap APIs. It is
4000 a fixed number determined by the format so can easily be allocated on the
4003 Corresponding information about the pixels
4005 PNG_IMAGE_PIXEL_CHANNELS(fmt)
4006 The number of separate channels (components) in a pixel; 1 for a
4009 PNG_IMAGE_PIXEL_COMPONENT_SIZE(fmt)\
4010 The size, in bytes, of each component in a pixel; 1 for a color-mapped
4013 PNG_IMAGE_PIXEL_SIZE(fmt)
4014 The size, in bytes, of a complete pixel; 1 for a color-mapped image.
4016 Information about the whole row, or whole image
4018 PNG_IMAGE_ROW_STRIDE(image)
4019 Returns the total number of components in a single row of the image; this
4020 is the minimum 'row stride', the minimum count of components between each
4021 row. For a color-mapped image this is the minimum number of bytes in a
4024 If you need the stride measured in bytes, row_stride_bytes is
4025 PNG_IMAGE_ROW_STRIDE(image) * PNG_IMAGE_PIXEL_COMPONENT_SIZE(fmt)
4026 plus any padding bytes that your application might need, for example
4027 to start the next row on a 4-byte boundary.
4029 PNG_IMAGE_BUFFER_SIZE(image, row_stride)
4030 Return the size, in bytes, of an image buffer given a png_image and a row
4031 stride - the number of components to leave space for in each row.
4033 PNG_IMAGE_SIZE(image)
4034 Return the size, in bytes, of the image in memory given just a png_image;
4035 the row stride is the minimum stride required for the image.
4037 PNG_IMAGE_COLORMAP_SIZE(image)
4038 Return the size, in bytes, of the color-map of this image. If the image
4039 format is not a color-map format this will return a size sufficient for
4040 256 entries in the given format; check PNG_FORMAT_FLAG_COLORMAP if
4041 you don't want to allocate a color-map in this case.
4045 Flags containing additional information about the image are held in
4046 the 'flags' field of png_image.
4048 PNG_IMAGE_FLAG_COLORSPACE_NOT_sRGB == 0x01
4049 This indicates the the RGB values of the in-memory bitmap do not
4050 correspond to the red, green and blue end-points defined by sRGB.
4052 PNG_IMAGE_FLAG_FAST == 0x02
4053 On write emphasise speed over compression; the resultant PNG file will be
4054 larger but will be produced significantly faster, particular for large
4055 images. Do not use this option for images which will be distributed, only
4056 used it when producing intermediate files that will be read back in
4057 repeatedly. For a typical 24-bit image the option will double the read
4058 speed at the cost of increasing the image size by 25%, however for many
4059 more compressible images the PNG file can be 10 times larger with only a
4062 PNG_IMAGE_FLAG_16BIT_sRGB == 0x04
4063 On read if the image is a 16-bit per component image and there is no gAMA
4064 or sRGB chunk assume that the components are sRGB encoded. Notice that
4065 images output by the simplified API always have gamma information; setting
4066 this flag only affects the interpretation of 16-bit images from an
4067 external source. It is recommended that the application expose this flag
4068 to the user; the user can normally easily recognize the difference between
4069 linear and sRGB encoding. This flag has no effect on write - the data
4070 passed to the write APIs must have the correct encoding (as defined
4073 If the flag is not set (the default) input 16-bit per component data is
4074 assumed to be linear.
4076 NOTE: the flag can only be set after the png_image_begin_read_ call,
4077 because that call initializes the 'flags' field.
4081 The png_image passed to the read APIs must have been initialized by setting
4082 the png_controlp field 'opaque' to NULL (or, better, memset the whole thing.)
4084 int png_image_begin_read_from_file( png_imagep image,
4085 const char *file_name)
4087 The named file is opened for read and the image header
4088 is filled in from the PNG header in the file.
4090 int png_image_begin_read_from_stdio (png_imagep image,
4093 The PNG header is read from the stdio FILE object.
4095 int png_image_begin_read_from_memory(png_imagep image,
4096 png_const_voidp memory, png_size_t size)
4098 The PNG header is read from the given memory buffer.
4100 int png_image_finish_read(png_imagep image,
4101 png_colorp background, void *buffer,
4102 png_int_32 row_stride, void *colormap));
4104 Finish reading the image into the supplied buffer and
4105 clean up the png_image structure.
4107 row_stride is the step, in png_byte or png_uint_16 units
4108 as appropriate, between adjacent rows. A positive stride
4109 indicates that the top-most row is first in the buffer -
4110 the normal top-down arrangement. A negative stride
4111 indicates that the bottom-most row is first in the buffer.
4113 background need only be supplied if an alpha channel must
4114 be removed from a png_byte format and the removal is to be
4115 done by compositing on a solid color; otherwise it may be
4116 NULL and any composition will be done directly onto the
4117 buffer. The value is an sRGB color to use for the
4118 background, for grayscale output the green channel is used.
4120 For linear output removing the alpha channel is always done
4121 by compositing on black.
4123 void png_image_free(png_imagep image)
4125 Free any data allocated by libpng in image->opaque,
4126 setting the pointer to NULL. May be called at any time
4127 after the structure is initialized.
4129 When the simplified API needs to convert between sRGB and linear colorspaces,
4130 the actual sRGB transfer curve defined in the sRGB specification (see the
4131 article at https://en.wikipedia.org/wiki/SRGB) is used, not the gamma=1/2.2
4132 approximation used elsewhere in libpng.
4136 For write you must initialize a png_image structure to describe the image to
4139 version: must be set to PNG_IMAGE_VERSION
4140 opaque: must be initialized to NULL
4141 width: image width in pixels
4142 height: image height in rows
4143 format: the format of the data you wish to write
4144 flags: set to 0 unless one of the defined flags applies; set
4145 PNG_IMAGE_FLAG_COLORSPACE_NOT_sRGB for color format images
4146 where the RGB values do not correspond to the colors in sRGB.
4147 colormap_entries: set to the number of entries in the color-map (0 to 256)
4149 int png_image_write_to_file, (png_imagep image,
4150 const char *file, int convert_to_8bit, const void *buffer,
4151 png_int_32 row_stride, const void *colormap));
4153 Write the image to the named file.
4155 int png_image_write_to_memory (png_imagep image, void *memory,
4156 png_alloc_size_t * PNG_RESTRICT memory_bytes,
4157 int convert_to_8_bit, const void *buffer, ptrdiff_t row_stride,
4158 const void *colormap));
4160 Write the image to memory.
4162 int png_image_write_to_stdio(png_imagep image, FILE *file,
4163 int convert_to_8_bit, const void *buffer,
4164 png_int_32 row_stride, const void *colormap)
4166 Write the image to the given (FILE*).
4168 With all write APIs if image is in one of the linear formats with
4169 (png_uint_16) data then setting convert_to_8_bit will cause the output to be
4170 a (png_byte) PNG gamma encoded according to the sRGB specification, otherwise
4171 a 16-bit linear encoded PNG file is written.
4173 With all APIs row_stride is handled as in the read APIs - it is the spacing
4174 from one row to the next in component sized units (float) and if negative
4175 indicates a bottom-up row layout in the buffer. If you pass zero, libpng will
4176 calculate the row_stride for you from the width and number of channels.
4178 Note that the write API does not support interlacing, sub-8-bit pixels,
4179 indexed (paletted) images, or most ancillary chunks.
4181 VI. Modifying/Customizing libpng
4183 There are two issues here. The first is changing how libpng does
4184 standard things like memory allocation, input/output, and error handling.
4185 The second deals with more complicated things like adding new chunks,
4186 adding new transformations, and generally changing how libpng works.
4187 Both of those are compile-time issues; that is, they are generally
4188 determined at the time the code is written, and there is rarely a need
4189 to provide the user with a means of changing them.
4191 Memory allocation, input/output, and error handling
4193 All of the memory allocation, input/output, and error handling in libpng
4194 goes through callbacks that are user-settable. The default routines are
4195 in pngmem.c, pngrio.c, pngwio.c, and pngerror.c, respectively. To change
4196 these functions, call the appropriate png_set_*_fn() function.
4198 Memory allocation is done through the functions png_malloc(), png_calloc(),
4199 and png_free(). The png_malloc() and png_free() functions currently just
4200 call the standard C functions and png_calloc() calls png_malloc() and then
4201 clears the newly allocated memory to zero; note that png_calloc(png_ptr, size)
4202 is not the same as the calloc(number, size) function provided by stdlib.h.
4203 There is limited support for certain systems with segmented memory
4204 architectures and the types of pointers declared by png.h match this; you
4205 will have to use appropriate pointers in your application. If you prefer
4206 to use a different method of allocating and freeing data, you can use
4207 png_create_read_struct_2() or png_create_write_struct_2() to register your
4208 own functions as described above. These functions also provide a void
4209 pointer that can be retrieved via
4211 mem_ptr=png_get_mem_ptr(png_ptr);
4213 Your replacement memory functions must have prototypes as follows:
4215 png_voidp malloc_fn(png_structp png_ptr,
4216 png_alloc_size_t size);
4218 void free_fn(png_structp png_ptr, png_voidp ptr);
4220 Your malloc_fn() must return NULL in case of failure. The png_malloc()
4221 function will normally call png_error() if it receives a NULL from the
4222 system memory allocator or from your replacement malloc_fn().
4224 Your free_fn() will never be called with a NULL ptr, since libpng's
4225 png_free() checks for NULL before calling free_fn().
4227 Input/Output in libpng is done through png_read() and png_write(),
4228 which currently just call fread() and fwrite(). The FILE * is stored in
4229 png_struct and is initialized via png_init_io(). If you wish to change
4230 the method of I/O, the library supplies callbacks that you can set
4231 through the function png_set_read_fn() and png_set_write_fn() at run
4232 time, instead of calling the png_init_io() function. These functions
4233 also provide a void pointer that can be retrieved via the function
4234 png_get_io_ptr(). For example:
4236 png_set_read_fn(png_structp read_ptr,
4237 voidp read_io_ptr, png_rw_ptr read_data_fn)
4239 png_set_write_fn(png_structp write_ptr,
4240 voidp write_io_ptr, png_rw_ptr write_data_fn,
4241 png_flush_ptr output_flush_fn);
4243 voidp read_io_ptr = png_get_io_ptr(read_ptr);
4244 voidp write_io_ptr = png_get_io_ptr(write_ptr);
4246 The replacement I/O functions must have prototypes as follows:
4248 void user_read_data(png_structp png_ptr,
4249 png_bytep data, png_size_t length);
4251 void user_write_data(png_structp png_ptr,
4252 png_bytep data, png_size_t length);
4254 void user_flush_data(png_structp png_ptr);
4256 The user_read_data() function is responsible for detecting and
4257 handling end-of-data errors.
4259 Supplying NULL for the read, write, or flush functions sets them back
4260 to using the default C stream functions, which expect the io_ptr to
4261 point to a standard *FILE structure. It is probably a mistake
4262 to use NULL for one of write_data_fn and output_flush_fn but not both
4263 of them, unless you have built libpng with PNG_NO_WRITE_FLUSH defined.
4264 It is an error to read from a write stream, and vice versa.
4266 Error handling in libpng is done through png_error() and png_warning().
4267 Errors handled through png_error() are fatal, meaning that png_error()
4268 should never return to its caller. Currently, this is handled via
4269 setjmp() and longjmp() (unless you have compiled libpng with
4270 PNG_NO_SETJMP, in which case it is handled via PNG_ABORT()),
4271 but you could change this to do things like exit() if you should wish,
4272 as long as your function does not return.
4274 On non-fatal errors, png_warning() is called
4275 to print a warning message, and then control returns to the calling code.
4276 By default png_error() and png_warning() print a message on stderr via
4277 fprintf() unless the library is compiled with PNG_NO_CONSOLE_IO defined
4278 (because you don't want the messages) or PNG_NO_STDIO defined (because
4279 fprintf() isn't available). If you wish to change the behavior of the error
4280 functions, you will need to set up your own message callbacks. These
4281 functions are normally supplied at the time that the png_struct is created.
4282 It is also possible to redirect errors and warnings to your own replacement
4283 functions after png_create_*_struct() has been called by calling:
4285 png_set_error_fn(png_structp png_ptr,
4286 png_voidp error_ptr, png_error_ptr error_fn,
4287 png_error_ptr warning_fn);
4289 If NULL is supplied for either error_fn or warning_fn, then the libpng
4290 default function will be used, calling fprintf() and/or longjmp() if a
4291 problem is encountered. The replacement error functions should have
4292 parameters as follows:
4294 void user_error_fn(png_structp png_ptr,
4295 png_const_charp error_msg);
4297 void user_warning_fn(png_structp png_ptr,
4298 png_const_charp warning_msg);
4300 Then, within your user_error_fn or user_warning_fn, you can retrieve
4301 the error_ptr if you need it, by calling
4303 png_voidp error_ptr = png_get_error_ptr(png_ptr);
4305 The motivation behind using setjmp() and longjmp() is the C++ throw and
4306 catch exception handling methods. This makes the code much easier to write,
4307 as there is no need to check every return code of every function call.
4308 However, there are some uncertainties about the status of local variables
4309 after a longjmp, so the user may want to be careful about doing anything
4310 after setjmp returns non-zero besides returning itself. Consult your
4311 compiler documentation for more details. For an alternative approach, you
4312 may wish to use the "cexcept" facility (see https://cexcept.sourceforge.io/),
4313 which is illustrated in pngvalid.c and in contrib/visupng.
4315 Beginning in libpng-1.4.0, the png_set_benign_errors() API became available.
4316 You can use this to handle certain errors (normally handled as errors)
4319 png_set_benign_errors (png_ptr, int allowed);
4321 allowed: 0: treat png_benign_error() as an error.
4322 1: treat png_benign_error() as a warning.
4324 As of libpng-1.6.0, the default condition is to treat benign errors as
4325 warnings while reading and as errors while writing.
4329 If you need to read or write custom chunks, you may need to get deeper
4330 into the libpng code. The library now has mechanisms for storing
4331 and writing chunks of unknown type; you can even declare callbacks
4332 for custom chunks. However, this may not be good enough if the
4333 library code itself needs to know about interactions between your
4334 chunk and existing `intrinsic' chunks.
4336 If you need to write a new intrinsic chunk, first read the PNG
4337 specification. Acquire a first level of understanding of how it works.
4338 Pay particular attention to the sections that describe chunk names,
4339 and look at how other chunks were designed, so you can do things
4340 similarly. Second, check out the sections of libpng that read and
4341 write chunks. Try to find a chunk that is similar to yours and use
4342 it as a template. More details can be found in the comments inside
4343 the code. It is best to handle private or unknown chunks in a generic method,
4344 via callback functions, instead of by modifying libpng functions. This
4345 is illustrated in pngtest.c, which uses a callback function to handle a
4346 private "vpAg" chunk and the new "sTER" chunk, which are both unknown to
4349 If you wish to write your own transformation for the data, look through
4350 the part of the code that does the transformations, and check out some of
4351 the simpler ones to get an idea of how they work. Try to find a similar
4352 transformation to the one you want to add and copy off of it. More details
4353 can be found in the comments inside the code itself.
4355 Configuring for gui/windowing platforms:
4357 You will need to write new error and warning functions that use the GUI
4358 interface, as described previously, and set them to be the error and
4359 warning functions at the time that png_create_*_struct() is called,
4360 in order to have them available during the structure initialization.
4361 They can be changed later via png_set_error_fn(). On some compilers,
4362 you may also have to change the memory allocators (png_malloc, etc.).
4366 There are special functions to configure the compression. Perhaps the
4367 most useful one changes the compression level, which currently uses
4368 input compression values in the range 0 - 9. The library normally
4369 uses the default compression level (Z_DEFAULT_COMPRESSION = 6). Tests
4370 have shown that for a large majority of images, compression values in
4371 the range 3-6 compress nearly as well as higher levels, and do so much
4372 faster. For online applications it may be desirable to have maximum speed
4373 (Z_BEST_SPEED = 1). With versions of zlib after v0.99, you can also
4374 specify no compression (Z_NO_COMPRESSION = 0), but this would create
4375 files larger than just storing the raw bitmap. You can specify the
4376 compression level by calling:
4379 png_set_compression_level(png_ptr, level);
4381 Another useful one is to reduce the memory level used by the library.
4382 The memory level defaults to 8, but it can be lowered if you are
4383 short on memory (running DOS, for example, where you only have 640K).
4384 Note that the memory level does have an effect on compression; among
4385 other things, lower levels will result in sections of incompressible
4386 data being emitted in smaller stored blocks, with a correspondingly
4387 larger relative overhead of up to 15% in the worst case.
4390 png_set_compression_mem_level(png_ptr, level);
4392 The other functions are for configuring zlib. They are not recommended
4393 for normal use and may result in writing an invalid PNG file. See
4394 zlib.h for more information on what these mean.
4397 png_set_compression_strategy(png_ptr,
4400 png_set_compression_window_bits(png_ptr,
4403 png_set_compression_method(png_ptr, method);
4405 This controls the size of the IDAT chunks (default 8192):
4407 png_set_compression_buffer_size(png_ptr, size);
4409 As of libpng version 1.5.4, additional APIs became
4410 available to set these separately for non-IDAT
4411 compressed chunks such as zTXt, iTXt, and iCCP:
4414 #if PNG_LIBPNG_VER >= 10504
4415 png_set_text_compression_level(png_ptr, level);
4417 png_set_text_compression_mem_level(png_ptr, level);
4419 png_set_text_compression_strategy(png_ptr,
4422 png_set_text_compression_window_bits(png_ptr,
4425 png_set_text_compression_method(png_ptr, method);
4428 Controlling row filtering
4430 If you want to control whether libpng uses filtering or not, which
4431 filters are used, and how it goes about picking row filters, you
4432 can call one of these functions. The selection and configuration
4433 of row filters can have a significant impact on the size and
4434 encoding speed and a somewhat lesser impact on the decoding speed
4435 of an image. Filtering is enabled by default for RGB and grayscale
4436 images (with and without alpha), but not for paletted images nor
4437 for any images with bit depths less than 8 bits/pixel.
4439 The 'method' parameter sets the main filtering method, which is
4440 currently only '0' in the PNG 1.2 specification. The 'filters'
4441 parameter sets which filter(s), if any, should be used for each
4442 scanline. Possible values are PNG_ALL_FILTERS, PNG_NO_FILTERS,
4443 or PNG_FAST_FILTERS to turn filtering on and off, or to turn on
4444 just the fast-decoding subset of filters, respectively.
4446 Individual filter types are PNG_FILTER_NONE, PNG_FILTER_SUB,
4447 PNG_FILTER_UP, PNG_FILTER_AVG, PNG_FILTER_PAETH, which can be bitwise
4448 ORed together with '|' to specify one or more filters to use.
4449 These filters are described in more detail in the PNG specification.
4450 If you intend to change the filter type during the course of writing
4451 the image, you should start with flags set for all of the filters
4452 you intend to use so that libpng can initialize its internal
4453 structures appropriately for all of the filter types. (Note that this
4454 means the first row must always be adaptively filtered, because libpng
4455 currently does not allocate the filter buffers until png_write_row()
4456 is called for the first time.)
4458 filters = PNG_NO_FILTERS;
4459 filters = PNG_ALL_FILTERS;
4460 filters = PNG_FAST_FILTERS;
4464 filters = PNG_FILTER_NONE | PNG_FILTER_SUB |
4465 PNG_FILTER_UP | PNG_FILTER_AVG |
4468 png_set_filter(png_ptr, PNG_FILTER_TYPE_BASE,
4471 The second parameter can also be
4472 PNG_INTRAPIXEL_DIFFERENCING if you are
4473 writing a PNG to be embedded in a MNG
4474 datastream. This parameter must be the
4475 same as the value of filter_method used
4478 Requesting debug printout
4480 The macro definition PNG_DEBUG can be used to request debugging
4481 printout. Set it to an integer value in the range 0 to 3. Higher
4482 numbers result in increasing amounts of debugging information. The
4483 information is printed to the "stderr" file, unless another file
4484 name is specified in the PNG_DEBUG_FILE macro definition.
4486 When PNG_DEBUG > 0, the following functions (macros) become available:
4488 png_debug(level, message)
4489 png_debug1(level, message, p1)
4490 png_debug2(level, message, p1, p2)
4492 in which "level" is compared to PNG_DEBUG to decide whether to print
4493 the message, "message" is the formatted string to be printed,
4494 and p1 and p2 are parameters that are to be embedded in the string
4495 according to printf-style formatting directives. For example,
4497 png_debug1(2, "foo=%d", foo);
4502 fprintf(PNG_DEBUG_FILE, "foo=%d\n", foo);
4504 When PNG_DEBUG is defined but is zero, the macros aren't defined, but you
4505 can still use PNG_DEBUG to control your own debugging:
4511 When PNG_DEBUG = 1, the macros are defined, but only png_debug statements
4512 having level = 0 will be printed. There aren't any such statements in
4513 this version of libpng, but if you insert some they will be printed.
4517 The MNG specification (available at http://www.libpng.org/pub/mng) allows
4518 certain extensions to PNG for PNG images that are embedded in MNG datastreams.
4519 Libpng can support some of these extensions. To enable them, use the
4520 png_permit_mng_features() function:
4522 feature_set = png_permit_mng_features(png_ptr, mask)
4524 mask is a png_uint_32 containing the bitwise OR of the
4525 features you want to enable. These include
4526 PNG_FLAG_MNG_EMPTY_PLTE
4527 PNG_FLAG_MNG_FILTER_64
4528 PNG_ALL_MNG_FEATURES
4530 feature_set is a png_uint_32 that is the bitwise AND of
4531 your mask with the set of MNG features that is
4532 supported by the version of libpng that you are using.
4534 It is an error to use this function when reading or writing a standalone
4535 PNG file with the PNG 8-byte signature. The PNG datastream must be wrapped
4536 in a MNG datastream. As a minimum, it must have the MNG 8-byte signature
4537 and the MHDR and MEND chunks. Libpng does not provide support for these
4538 or any other MNG chunks; your application must provide its own support for
4539 them. You may wish to consider using libmng (available at
4540 https://www.libmng.com/) instead.
4542 VIII. Changes to Libpng from version 0.88
4544 It should be noted that versions of libpng later than 0.96 are not
4545 distributed by the original libpng author, Guy Schalnat, nor by
4546 Andreas Dilger, who had taken over from Guy during 1996 and 1997, and
4547 distributed versions 0.89 through 0.96, but rather by another member
4548 of the original PNG Group, Glenn Randers-Pehrson. Guy and Andreas are
4549 still alive and well, but they have moved on to other things.
4551 The old libpng functions png_read_init(), png_write_init(),
4552 png_info_init(), png_read_destroy(), and png_write_destroy() have been
4553 moved to PNG_INTERNAL in version 0.95 to discourage their use. These
4554 functions will be removed from libpng version 1.4.0.
4556 The preferred method of creating and initializing the libpng structures is
4557 via the png_create_read_struct(), png_create_write_struct(), and
4558 png_create_info_struct() because they isolate the size of the structures
4559 from the application, allow version error checking, and also allow the
4560 use of custom error handling routines during the initialization, which
4561 the old functions do not. The functions png_read_destroy() and
4562 png_write_destroy() do not actually free the memory that libpng
4563 allocated for these structs, but just reset the data structures, so they
4564 can be used instead of png_destroy_read_struct() and
4565 png_destroy_write_struct() if you feel there is too much system overhead
4566 allocating and freeing the png_struct for each image read.
4568 Setting the error callbacks via png_set_message_fn() before
4569 png_read_init() as was suggested in libpng-0.88 is no longer supported
4570 because this caused applications that do not use custom error functions
4571 to fail if the png_ptr was not initialized to zero. It is still possible
4572 to set the error callbacks AFTER png_read_init(), or to change them with
4573 png_set_error_fn(), which is essentially the same function, but with a new
4574 name to force compilation errors with applications that try to use the old
4577 Support for the sCAL, iCCP, iTXt, and sPLT chunks was added at libpng-1.0.6;
4578 however, iTXt support was not enabled by default.
4580 Starting with version 1.0.7, you can find out which version of the library
4581 you are using at run-time:
4583 png_uint_32 libpng_vn = png_access_version_number();
4585 The number libpng_vn is constructed from the major version, minor
4586 version with leading zero, and release number with leading zero,
4587 (e.g., libpng_vn for version 1.0.7 is 10007).
4589 Note that this function does not take a png_ptr, so you can call it
4590 before you've created one.
4592 You can also check which version of png.h you used when compiling your
4595 png_uint_32 application_vn = PNG_LIBPNG_VER;
4597 IX. Changes to Libpng from version 1.0.x to 1.2.x
4599 Support for user memory management was enabled by default. To
4600 accomplish this, the functions png_create_read_struct_2(),
4601 png_create_write_struct_2(), png_set_mem_fn(), png_get_mem_ptr(),
4602 png_malloc_default(), and png_free_default() were added.
4604 Support for the iTXt chunk has been enabled by default as of
4607 Support for certain MNG features was enabled.
4609 Support for numbered error messages was added. However, we never got
4610 around to actually numbering the error messages. The function
4611 png_set_strip_error_numbers() was added (Note: the prototype for this
4612 function was inadvertently removed from png.h in PNG_NO_ASSEMBLER_CODE
4613 builds of libpng-1.2.15. It was restored in libpng-1.2.36).
4615 The png_malloc_warn() function was added at libpng-1.2.3. This issues
4616 a png_warning and returns NULL instead of aborting when it fails to
4617 acquire the requested memory allocation.
4619 Support for setting user limits on image width and height was enabled
4620 by default. The functions png_set_user_limits(), png_get_user_width_max(),
4621 and png_get_user_height_max() were added at libpng-1.2.6.
4623 The png_set_add_alpha() function was added at libpng-1.2.7.
4625 The function png_set_expand_gray_1_2_4_to_8() was added at libpng-1.2.9.
4626 Unlike png_set_gray_1_2_4_to_8(), the new function does not expand the
4627 tRNS chunk to alpha. The png_set_gray_1_2_4_to_8() function is
4630 A number of macro definitions in support of runtime selection of
4631 assembler code features (especially Intel MMX code support) were
4632 added at libpng-1.2.0:
4634 PNG_ASM_FLAG_MMX_SUPPORT_COMPILED
4635 PNG_ASM_FLAG_MMX_SUPPORT_IN_CPU
4636 PNG_ASM_FLAG_MMX_READ_COMBINE_ROW
4637 PNG_ASM_FLAG_MMX_READ_INTERLACE
4638 PNG_ASM_FLAG_MMX_READ_FILTER_SUB
4639 PNG_ASM_FLAG_MMX_READ_FILTER_UP
4640 PNG_ASM_FLAG_MMX_READ_FILTER_AVG
4641 PNG_ASM_FLAG_MMX_READ_FILTER_PAETH
4642 PNG_ASM_FLAGS_INITIALIZED
4648 We added the following functions in support of runtime
4649 selection of assembler code features:
4651 png_get_mmx_flagmask()
4652 png_set_mmx_thresholds()
4654 png_get_mmx_bitdepth_threshold()
4655 png_get_mmx_rowbytes_threshold()
4658 We replaced all of these functions with simple stubs in libpng-1.2.20,
4659 when the Intel assembler code was removed due to a licensing issue.
4661 These macros are deprecated:
4663 PNG_READ_TRANSFORMS_NOT_SUPPORTED
4664 PNG_PROGRESSIVE_READ_NOT_SUPPORTED
4665 PNG_NO_SEQUENTIAL_READ_SUPPORTED
4666 PNG_WRITE_TRANSFORMS_NOT_SUPPORTED
4667 PNG_READ_ANCILLARY_CHUNKS_NOT_SUPPORTED
4668 PNG_WRITE_ANCILLARY_CHUNKS_NOT_SUPPORTED
4670 They have been replaced, respectively, by:
4672 PNG_NO_READ_TRANSFORMS
4673 PNG_NO_PROGRESSIVE_READ
4674 PNG_NO_SEQUENTIAL_READ
4675 PNG_NO_WRITE_TRANSFORMS
4676 PNG_NO_READ_ANCILLARY_CHUNKS
4677 PNG_NO_WRITE_ANCILLARY_CHUNKS
4679 PNG_MAX_UINT was replaced with PNG_UINT_31_MAX. It has been
4680 deprecated since libpng-1.0.16 and libpng-1.2.6.
4683 png_check_sig(sig, num)
4685 !png_sig_cmp(sig, 0, num)
4686 It has been deprecated since libpng-0.90.
4689 png_set_gray_1_2_4_to_8()
4690 which also expands tRNS to alpha was replaced with
4691 png_set_expand_gray_1_2_4_to_8()
4692 which does not. It has been deprecated since libpng-1.0.18 and 1.2.9.
4694 X. Changes to Libpng from version 1.0.x/1.2.x to 1.4.x
4696 Private libpng prototypes and macro definitions were moved from
4697 png.h and pngconf.h into a new pngpriv.h header file.
4699 Functions png_set_benign_errors(), png_benign_error(), and
4700 png_chunk_benign_error() were added.
4702 Support for setting the maximum amount of memory that the application
4703 will allocate for reading chunks was added, as a security measure.
4704 The functions png_set_chunk_cache_max() and png_get_chunk_cache_max()
4705 were added to the library.
4707 We implemented support for I/O states by adding png_ptr member io_state
4708 and functions png_get_io_chunk_name() and png_get_io_state() in pngget.c
4710 We added PNG_TRANSFORM_GRAY_TO_RGB to the available high-level
4713 Checking for and reporting of errors in the IHDR chunk is more thorough.
4715 Support for global arrays was removed, to improve thread safety.
4717 Some obsolete/deprecated macros and functions have been removed.
4719 Typecasted NULL definitions such as
4720 #define png_voidp_NULL (png_voidp)NULL
4721 were eliminated. If you used these in your application, just use
4724 The png_struct and info_struct members "trans" and "trans_values" were
4725 changed to "trans_alpha" and "trans_color", respectively.
4727 The obsolete, unused pnggccrd.c and pngvcrd.c files and related makefiles
4730 The PNG_1_0_X and PNG_1_2_X macros were eliminated.
4732 The PNG_LEGACY_SUPPORTED macro was eliminated.
4734 Many WIN32_WCE #ifdefs were removed.
4736 The functions png_read_init(info_ptr), png_write_init(info_ptr),
4737 png_info_init(info_ptr), png_read_destroy(), and png_write_destroy()
4738 have been removed. They have been deprecated since libpng-0.95.
4740 The png_permit_empty_plte() was removed. It has been deprecated
4741 since libpng-1.0.9. Use png_permit_mng_features() instead.
4743 We removed the obsolete stub functions png_get_mmx_flagmask(),
4744 png_set_mmx_thresholds(), png_get_asm_flags(),
4745 png_get_mmx_bitdepth_threshold(), png_get_mmx_rowbytes_threshold(),
4746 png_set_asm_flags(), and png_mmx_supported()
4748 We removed the obsolete png_check_sig(), png_memcpy_check(), and
4749 png_memset_check() functions. Instead use !png_sig_cmp(), memcpy(),
4750 and memset(), respectively.
4752 The function png_set_gray_1_2_4_to_8() was removed. It has been
4753 deprecated since libpng-1.0.18 and 1.2.9, when it was replaced with
4754 png_set_expand_gray_1_2_4_to_8() because the former function also
4755 expanded any tRNS chunk to an alpha channel.
4757 Macros for png_get_uint_16, png_get_uint_32, and png_get_int_32
4758 were added and are used by default instead of the corresponding
4759 functions. Unfortunately,
4760 from libpng-1.4.0 until 1.4.4, the png_get_uint_16 macro (but not the
4761 function) incorrectly returned a value of type png_uint_32.
4763 We changed the prototype for png_malloc() from
4764 png_malloc(png_structp png_ptr, png_uint_32 size)
4766 png_malloc(png_structp png_ptr, png_alloc_size_t size)
4768 This also applies to the prototype for the user replacement malloc_fn().
4770 The png_calloc() function was added and is used in place of
4771 of "png_malloc(); memset();" except in the case in png_read_png()
4772 where the array consists of pointers; in this case a "for" loop is used
4773 after the png_malloc() to set the pointers to NULL, to give robust.
4774 behavior in case the application runs out of memory part-way through
4777 We changed the prototypes of png_get_compression_buffer_size() and
4778 png_set_compression_buffer_size() to work with png_size_t instead of
4781 Support for numbered error messages was removed by default, since we
4782 never got around to actually numbering the error messages. The function
4783 png_set_strip_error_numbers() was removed from the library by default.
4785 The png_zalloc() and png_zfree() functions are no longer exported.
4786 The png_zalloc() function no longer zeroes out the memory that it
4787 allocates. Applications that called png_zalloc(png_ptr, number, size)
4788 can call png_calloc(png_ptr, number*size) instead, and can call
4789 png_free() instead of png_zfree().
4791 Support for dithering was disabled by default in libpng-1.4.0, because
4792 it has not been well tested and doesn't actually "dither".
4794 removed, however, and could be enabled by building libpng with
4795 PNG_READ_DITHER_SUPPORTED defined. In libpng-1.4.2, this support
4796 was re-enabled, but the function was renamed png_set_quantize() to
4797 reflect more accurately what it actually does. At the same time,
4798 the PNG_DITHER_[RED,GREEN_BLUE]_BITS macros were also renamed to
4799 PNG_QUANTIZE_[RED,GREEN,BLUE]_BITS, and PNG_READ_DITHER_SUPPORTED
4800 was renamed to PNG_READ_QUANTIZE_SUPPORTED.
4802 We removed the trailing '.' from the warning and error messages.
4804 XI. Changes to Libpng from version 1.4.x to 1.5.x
4806 From libpng-1.4.0 until 1.4.4, the png_get_uint_16 macro (but not the
4807 function) incorrectly returned a value of type png_uint_32.
4808 The incorrect macro was removed from libpng-1.4.5.
4810 Checking for invalid palette index on write was added at libpng
4811 1.5.10. If a pixel contains an invalid (out-of-range) index libpng issues
4812 a benign error. This is enabled by default because this condition is an
4813 error according to the PNG specification, Clause 11.3.2, but the error can
4814 be ignored in each png_ptr with
4816 png_set_check_for_invalid_index(png_ptr, allowed);
4819 0: disable benign error (accept the
4820 invalid data without warning).
4821 1: enable benign error (treat the
4822 invalid data as an error or a
4825 If the error is ignored, or if png_benign_error() treats it as a warning,
4826 any invalid pixels are decoded as opaque black by the decoder and written
4827 as-is by the encoder.
4829 Retrieving the maximum palette index found was added at libpng-1.5.15.
4830 This statement must appear after png_read_png() or png_read_image() while
4831 reading, and after png_write_png() or png_write_image() while writing.
4833 int max_palette = png_get_palette_max(png_ptr, info_ptr);
4835 This will return the maximum palette index found in the image, or "-1" if
4836 the palette was not checked, or "0" if no palette was found. Note that this
4837 does not account for any palette index used by ancillary chunks such as the
4838 bKGD chunk; you must check those separately to determine the maximum
4839 palette index actually used.
4841 There are no substantial API changes between the non-deprecated parts of
4842 the 1.4.5 API and the 1.5.0 API; however, the ability to directly access
4843 members of the main libpng control structures, png_struct and png_info,
4844 deprecated in earlier versions of libpng, has been completely removed from
4845 libpng 1.5, and new private "pngstruct.h", "pnginfo.h", and "pngdebug.h"
4846 header files were created.
4848 We no longer include zlib.h in png.h. The include statement has been moved
4849 to pngstruct.h, where it is not accessible by applications. Applications that
4850 need access to information in zlib.h will need to add the '#include "zlib.h"'
4851 directive. It does not matter whether this is placed prior to or after
4852 the '"#include png.h"' directive.
4854 The png_sprintf(), png_strcpy(), and png_strncpy() macros are no longer used
4857 We moved the png_strlen(), png_memcpy(), png_memset(), and png_memcmp()
4858 macros into a private header file (pngpriv.h) that is not accessible to
4861 In png_get_iCCP, the type of "profile" was changed from png_charpp
4862 to png_bytepp, and in png_set_iCCP, from png_charp to png_const_bytep.
4864 There are changes of form in png.h, including new and changed macros to
4865 declare parts of the API. Some API functions with arguments that are
4866 pointers to data not modified within the function have been corrected to
4867 declare these arguments with PNG_CONST.
4869 Much of the internal use of C macros to control the library build has also
4870 changed and some of this is visible in the exported header files, in
4871 particular the use of macros to control data and API elements visible
4872 during application compilation may require significant revision to
4873 application code. (It is extremely rare for an application to do this.)
4875 Any program that compiled against libpng 1.4 and did not use deprecated
4876 features or access internal library structures should compile and work
4877 against libpng 1.5, except for the change in the prototype for
4878 png_get_iCCP() and png_set_iCCP() API functions mentioned above.
4880 libpng 1.5.0 adds PNG_ PASS macros to help in the reading and writing of
4881 interlaced images. The macros return the number of rows and columns in
4882 each pass and information that can be used to de-interlace and (if
4883 absolutely necessary) interlace an image.
4885 libpng 1.5.0 adds an API png_longjmp(png_ptr, value). This API calls
4886 the application-provided png_longjmp_ptr on the internal, but application
4887 initialized, longjmp buffer. It is provided as a convenience to avoid
4888 the need to use the png_jmpbuf macro, which had the unnecessary side
4889 effect of resetting the internal png_longjmp_ptr value.
4891 libpng 1.5.0 includes a complete fixed point API. By default this is
4892 present along with the corresponding floating point API. In general the
4893 fixed point API is faster and smaller than the floating point one because
4894 the PNG file format used fixed point, not floating point. This applies
4895 even if the library uses floating point in internal calculations. A new
4896 macro, PNG_FLOATING_ARITHMETIC_SUPPORTED, reveals whether the library
4897 uses floating point arithmetic (the default) or fixed point arithmetic
4898 internally for performance critical calculations such as gamma correction.
4899 In some cases, the gamma calculations may produce slightly different
4900 results. This has changed the results in png_rgb_to_gray and in alpha
4901 composition (png_set_background for example). This applies even if the
4902 original image was already linear (gamma == 1.0) and, therefore, it is
4903 not necessary to linearize the image. This is because libpng has *not*
4904 been changed to optimize that case correctly, yet.
4906 Fixed point support for the sCAL chunk comes with an important caveat;
4907 the sCAL specification uses a decimal encoding of floating point values
4908 and the accuracy of PNG fixed point values is insufficient for
4909 representation of these values. Consequently a "string" API
4910 (png_get_sCAL_s and png_set_sCAL_s) is the only reliable way of reading
4911 arbitrary sCAL chunks in the absence of either the floating point API or
4912 internal floating point calculations. Starting with libpng-1.5.0, both
4913 of these functions are present when PNG_sCAL_SUPPORTED is defined. Prior
4914 to libpng-1.5.0, their presence also depended upon PNG_FIXED_POINT_SUPPORTED
4915 being defined and PNG_FLOATING_POINT_SUPPORTED not being defined.
4917 Applications no longer need to include the optional distribution header
4918 file pngusr.h or define the corresponding macros during application
4919 build in order to see the correct variant of the libpng API. From 1.5.0
4920 application code can check for the corresponding _SUPPORTED macro:
4922 #ifdef PNG_INCH_CONVERSIONS_SUPPORTED
4923 /* code that uses the inch conversion APIs. */
4926 This macro will only be defined if the inch conversion functions have been
4927 compiled into libpng. The full set of macros, and whether or not support
4928 has been compiled in, are available in the header file pnglibconf.h.
4929 This header file is specific to the libpng build. Notice that prior to
4930 1.5.0 the _SUPPORTED macros would always have the default definition unless
4931 reset by pngusr.h or by explicit settings on the compiler command line.
4932 These settings may produce compiler warnings or errors in 1.5.0 because
4933 of macro redefinition.
4935 Applications can now choose whether to use these macros or to call the
4936 corresponding function by defining PNG_USE_READ_MACROS or
4937 PNG_NO_USE_READ_MACROS before including png.h. Notice that this is
4938 only supported from 1.5.0; defining PNG_NO_USE_READ_MACROS prior to 1.5.0
4939 will lead to a link failure.
4941 Prior to libpng-1.5.4, the zlib compressor used the same set of parameters
4942 when compressing the IDAT data and textual data such as zTXt and iCCP.
4943 In libpng-1.5.4 we reinitialized the zlib stream for each type of data.
4944 We added five png_set_text_*() functions for setting the parameters to
4945 use with textual data.
4947 Prior to libpng-1.5.4, the PNG_READ_16_TO_8_ACCURATE_SCALE_SUPPORTED
4948 option was off by default, and slightly inaccurate scaling occurred.
4949 This option can no longer be turned off, and the choice of accurate
4950 or inaccurate 16-to-8 scaling is by using the new png_set_scale_16_to_8()
4951 API for accurate scaling or the old png_set_strip_16_to_8() API for simple
4952 chopping. In libpng-1.5.4, the PNG_READ_16_TO_8_ACCURATE_SCALE_SUPPORTED
4953 macro became PNG_READ_SCALE_16_TO_8_SUPPORTED, and the PNG_READ_16_TO_8
4954 macro became PNG_READ_STRIP_16_TO_8_SUPPORTED, to enable the two
4955 png_set_*_16_to_8() functions separately.
4957 Prior to libpng-1.5.4, the png_set_user_limits() function could only be
4958 used to reduce the width and height limits from the value of
4959 PNG_USER_WIDTH_MAX and PNG_USER_HEIGHT_MAX, although this document said
4960 that it could be used to override them. Now this function will reduce or
4961 increase the limits.
4963 Starting in libpng-1.5.22, default user limits were established. These
4964 can be overridden by application calls to png_set_user_limits(),
4965 png_set_user_chunk_cache_max(), and/or png_set_user_malloc_max().
4967 max possible default
4968 png_user_width_max 0x7fffffff 1,000,000
4969 png_user_height_max 0x7fffffff 1,000,000
4970 png_user_chunk_cache_max 0 (unlimited) 1000
4971 png_user_chunk_malloc_max 0 (unlimited) 8,000,000
4973 The png_set_option() function (and the "options" member of the png struct) was
4974 added to libpng-1.5.15, with option PNG_ARM_NEON.
4976 The library now supports a complete fixed point implementation and can
4977 thus be used on systems that have no floating point support or very
4978 limited or slow support. Previously gamma correction, an essential part
4979 of complete PNG support, required reasonably fast floating point.
4981 As part of this the choice of internal implementation has been made
4982 independent of the choice of fixed versus floating point APIs and all the
4983 missing fixed point APIs have been implemented.
4985 The exact mechanism used to control attributes of API functions has
4986 changed, as described in the INSTALL file.
4988 A new test program, pngvalid, is provided in addition to pngtest.
4989 pngvalid validates the arithmetic accuracy of the gamma correction
4990 calculations and includes a number of validations of the file format.
4991 A subset of the full range of tests is run when "make check" is done
4992 (in the 'configure' build.) pngvalid also allows total allocated memory
4993 usage to be evaluated and performs additional memory overwrite validation.
4995 Many changes to individual feature macros have been made. The following
4996 are the changes most likely to be noticed by library builders who
4999 1) All feature macros now have consistent naming:
5001 #define PNG_NO_feature turns the feature off
5002 #define PNG_feature_SUPPORTED turns the feature on
5004 pnglibconf.h contains one line for each feature macro which is either:
5006 #define PNG_feature_SUPPORTED
5008 if the feature is supported or:
5010 /*#undef PNG_feature_SUPPORTED*/
5012 if it is not. Library code consistently checks for the 'SUPPORTED' macro.
5013 It does not, and libpng applications should not, check for the 'NO' macro
5014 which will not normally be defined even if the feature is not supported.
5015 The 'NO' macros are only used internally for setting or not setting the
5016 corresponding 'SUPPORTED' macros.
5018 Compatibility with the old names is provided as follows:
5020 PNG_INCH_CONVERSIONS turns on PNG_INCH_CONVERSIONS_SUPPORTED
5022 And the following definitions disable the corresponding feature:
5024 PNG_SETJMP_NOT_SUPPORTED disables SETJMP
5025 PNG_READ_TRANSFORMS_NOT_SUPPORTED disables READ_TRANSFORMS
5026 PNG_NO_READ_COMPOSITED_NODIV disables READ_COMPOSITE_NODIV
5027 PNG_WRITE_TRANSFORMS_NOT_SUPPORTED disables WRITE_TRANSFORMS
5028 PNG_READ_ANCILLARY_CHUNKS_NOT_SUPPORTED disables READ_ANCILLARY_CHUNKS
5029 PNG_WRITE_ANCILLARY_CHUNKS_NOT_SUPPORTED disables WRITE_ANCILLARY_CHUNKS
5031 Library builders should remove use of the above, inconsistent, names.
5033 2) Warning and error message formatting was previously conditional on
5034 the STDIO feature. The library has been changed to use the
5035 CONSOLE_IO feature instead. This means that if CONSOLE_IO is disabled
5036 the library no longer uses the printf(3) functions, even though the
5037 default read/write implementations use (FILE) style stdio.h functions.
5039 3) Three feature macros now control the fixed/floating point decisions:
5041 PNG_FLOATING_POINT_SUPPORTED enables the floating point APIs
5043 PNG_FIXED_POINT_SUPPORTED enables the fixed point APIs; however, in
5044 practice these are normally required internally anyway (because the PNG
5045 file format is fixed point), therefore in most cases PNG_NO_FIXED_POINT
5046 merely stops the function from being exported.
5048 PNG_FLOATING_ARITHMETIC_SUPPORTED chooses between the internal floating
5049 point implementation or the fixed point one. Typically the fixed point
5050 implementation is larger and slower than the floating point implementation
5051 on a system that supports floating point; however, it may be faster on a
5052 system which lacks floating point hardware and therefore uses a software
5055 4) Added PNG_{READ,WRITE}_INT_FUNCTIONS_SUPPORTED. This allows the
5056 functions to read and write ints to be disabled independently of
5057 PNG_USE_READ_MACROS, which allows libpng to be built with the functions
5058 even though the default is to use the macros - this allows applications
5059 to choose at app buildtime whether or not to use macros (previously
5060 impossible because the functions weren't in the default build.)
5062 XII. Changes to Libpng from version 1.5.x to 1.6.x
5064 A "simplified API" has been added (see documentation in png.h and a simple
5065 example in contrib/examples/pngtopng.c). The new publicly visible API
5066 includes the following:
5075 png_image_begin_read_from_file()
5076 png_image_begin_read_from_stdio()
5077 png_image_begin_read_from_memory()
5078 png_image_finish_read()
5081 png_image_write_to_file()
5082 png_image_write_to_memory()
5083 png_image_write_to_stdio()
5085 Starting with libpng-1.6.0, you can configure libpng to prefix all exported
5086 symbols, using the PNG_PREFIX macro.
5088 We no longer include string.h in png.h. The include statement has been moved
5089 to pngpriv.h, where it is not accessible by applications. Applications that
5090 need access to information in string.h must add an '#include <string.h>'
5091 directive. It does not matter whether this is placed prior to or after
5092 the '#include "png.h"' directive.
5094 The following API are now DEPRECATED:
5096 png_convert_to_rfc1123() which has been replaced
5097 with png_convert_to_rfc1123_buffer()
5098 png_malloc_default()
5102 The following have been removed:
5103 png_get_io_chunk_name(), which has been replaced
5104 with png_get_io_chunk_type(). The new
5105 function returns a 32-bit integer instead of
5107 The png_sizeof(), png_strlen(), png_memcpy(), png_memcmp(), and
5108 png_memset() macros are no longer used in the libpng sources and
5109 have been removed. These had already been made invisible to applications
5110 (i.e., defined in the private pngpriv.h header file) since libpng-1.5.0.
5112 The signatures of many exported functions were changed, such that
5113 png_structp became png_structrp or png_const_structrp
5114 png_infop became png_inforp or png_const_inforp
5115 where "rp" indicates a "restricted pointer".
5117 Dropped support for 16-bit platforms. The support for FAR/far types has
5118 been eliminated and the definition of png_alloc_size_t is now controlled
5119 by a flag so that 'small size_t' systems can select it if necessary.
5121 Error detection in some chunks has improved; in particular the iCCP chunk
5122 reader now does pretty complete validation of the basic format. Some bad
5123 profiles that were previously accepted are now accepted with a warning or
5124 rejected, depending upon the png_set_benign_errors() setting, in particular
5125 the very old broken Microsoft/HP 3144-byte sRGB profile. Starting with
5126 libpng-1.6.11, recognizing and checking sRGB profiles can be avoided by
5129 #if defined(PNG_SKIP_sRGB_CHECK_PROFILE) && \
5130 defined(PNG_SET_OPTION_SUPPORTED)
5131 png_set_option(png_ptr, PNG_SKIP_sRGB_CHECK_PROFILE,
5135 It's not a good idea to do this if you are using the "simplified API",
5136 which needs to be able to recognize sRGB profiles conveyed via the iCCP
5139 The PNG spec requirement that only grayscale profiles may appear in images
5140 with color type 0 or 4 and that even if the image only contains gray pixels,
5141 only RGB profiles may appear in images with color type 2, 3, or 6, is now
5142 enforced. The sRGB chunk is allowed to appear in images with any color type
5143 and is interpreted by libpng to convey a one-tracer-curve gray profile or a
5144 three-tracer-curve RGB profile as appropriate.
5146 Libpng 1.5.x erroneously used /MD for Debug DLL builds; if you used the debug
5147 builds in your app and you changed your app to use /MD you will need to
5148 change it back to /MDd for libpng 1.6.x.
5150 Prior to libpng-1.6.0 a warning would be issued if the iTXt chunk contained
5151 an empty language field or an empty translated keyword. Both of these
5152 are allowed by the PNG specification, so these warnings are no longer issued.
5154 The library now issues an error if the application attempts to set a
5155 transform after it calls png_read_update_info() or if it attempts to call
5156 both png_read_update_info() and png_start_read_image() or to call either
5157 of them more than once.
5159 The default condition for benign_errors is now to treat benign errors as
5160 warnings while reading and as errors while writing.
5162 The library now issues a warning if both background processing and RGB to
5163 gray are used when gamma correction happens. As with previous versions of
5164 the library the results are numerically very incorrect in this case.
5166 There are some minor arithmetic changes in some transforms such as
5167 png_set_background(), that might be detected by certain regression tests.
5169 Unknown chunk handling has been improved internally, without any API change.
5170 This adds more correct option control of the unknown handling, corrects
5171 a pre-existing bug where the per-chunk 'keep' setting is ignored, and makes
5172 it possible to skip IDAT chunks in the sequential reader.
5174 The machine-generated configure files are no longer included in branches
5175 libpng16 and later of the GIT repository. They continue to be included
5176 in the tarball releases, however.
5178 Libpng-1.6.0 through 1.6.2 used the CMF bytes at the beginning of the IDAT
5179 stream to set the size of the sliding window for reading instead of using the
5180 default 32-kbyte sliding window size. It was discovered that there are
5181 hundreds of PNG files in the wild that have incorrect CMF bytes that caused
5182 zlib to issue the "invalid distance too far back" error and reject the file.
5183 Libpng-1.6.3 and later calculate their own safe CMF from the image dimensions,
5184 provide a way to revert to the libpng-1.5.x behavior (ignoring the CMF bytes
5185 and using a 32-kbyte sliding window), by using
5187 png_set_option(png_ptr, PNG_MAXIMUM_INFLATE_WINDOW,
5190 and provide a tool (contrib/tools/pngfix) for rewriting a PNG file while
5191 optimizing the CMF bytes in its IDAT chunk correctly.
5193 Libpng-1.6.0 and libpng-1.6.1 wrote uncompressed iTXt chunks with the wrong
5194 length, which resulted in PNG files that cannot be read beyond the bad iTXt
5195 chunk. This error was fixed in libpng-1.6.3, and a tool (called
5196 contrib/tools/png-fix-itxt) has been added to the libpng distribution.
5198 Starting with libpng-1.6.17, the PNG_SAFE_LIMITS macro was eliminated
5199 and safe limits are used by default (users who need larger limits
5200 can still override them at compile time or run time, as described above).
5204 png_user_width_max 1,000,000 2,147,483,647
5205 png_user_height_max 1,000,000 2,147,483,647
5206 png_user_chunk_cache_max 128 unlimited
5207 png_user_chunk_malloc_max 8,000,000 unlimited
5209 Starting with libpng-1.6.18, a PNG_RELEASE_BUILD macro was added, which allows
5210 library builders to control compilation for an installed system (a release build).
5211 It can be set for testing debug or beta builds to ensure that they will compile
5212 when the build type is switched to RC or STABLE. In essence this overrides the
5213 PNG_LIBPNG_BUILD_BASE_TYPE definition which is not directly user controllable.
5215 Starting with libpng-1.6.19, attempting to set an over-length PLTE chunk
5216 is an error. Previously this requirement of the PNG specification was not
5217 enforced, and the palette was always limited to 256 entries. An over-length
5218 PLTE chunk found in an input PNG is silently truncated.
5220 Starting with libpng-1.6.31, the eXIf chunk is supported. Libpng does not
5221 attempt to decode the Exif profile; it simply returns a byte array
5222 containing the profile to the calling application which must do its own
5225 XIII. Detecting libpng
5227 The png_get_io_ptr() function has been present since libpng-0.88, has never
5228 changed, and is unaffected by conditional compilation macros. It is the
5229 best choice for use in configure scripts for detecting the presence of any
5230 libpng version since 0.88. In an autoconf "configure.in" you could use
5232 AC_CHECK_LIB(png, png_get_io_ptr, ...
5234 XV. Source code repository
5236 Since about February 2009, version 1.2.34, libpng has been under "git" source
5237 control. The git repository was built from old libpng-x.y.z.tar.gz files
5238 going back to version 0.70. You can access the git repository (read only)
5241 https://github.com/glennrp/libpng or
5242 https://git.code.sf.net/p/libpng/code.git
5244 or you can browse it with a web browser at
5246 https://github.com/glennrp/libpng or
5247 https://sourceforge.net/p/libpng/code/ci/libpng16/tree/
5249 Patches can be sent to glennrp at users.sourceforge.net or to
5250 png-mng-implement at lists.sourceforge.net or you can upload them to
5251 the libpng bug tracker at
5253 https://libpng.sourceforge.io/
5255 or as a "pull request" to
5257 https://github.com/glennrp/libpng/pulls
5259 We also accept patches built from the tar or zip distributions, and
5260 simple verbal discriptions of bug fixes, reported either to the
5261 SourceForge bug tracker, to the png-mng-implement at lists.sf.net
5262 mailing list, as github issues, or directly to glennrp.
5266 Our coding style is similar to the "Allman" style
5267 (See https://en.wikipedia.org/wiki/Indent_style#Allman_style), with curly
5268 braces on separate lines:
5275 else if (another condition)
5280 The braces can be omitted from simple one-line actions:
5285 We use 3-space indentation, except for continued statements which
5286 are usually indented the same as the first line of the statement
5287 plus four more spaces.
5289 For macro definitions we use 2-space indentation, always leaving the "#"
5290 in the first column.
5292 #ifndef PNG_NO_FEATURE
5293 # ifndef PNG_FEATURE_SUPPORTED
5294 # define PNG_FEATURE_SUPPORTED
5298 Comments appear with the leading "/*" at the same indentation as
5299 the statement that follows the comment:
5301 /* Single-line comment */
5304 /* This is a multiple-line
5309 Very short comments can be placed after the end of the statement
5310 to which they pertain:
5312 statement; /* comment */
5314 We don't use C++ style ("//") comments. We have, however,
5315 used them in the past in some now-abandoned MMX assembler
5318 Functions and their curly braces are not indented, and
5319 exported functions are marked with PNGAPI:
5321 /* This is a public function that is visible to
5322 * application programmers. It does thus-and-so.
5325 png_exported_function(png_ptr, png_info, foo)
5330 The return type and decorations are placed on a separate line
5331 ahead of the function name, as illustrated above.
5333 The prototypes for all exported functions appear in png.h,
5334 above the comment that says
5336 /* Maintainer: Put new public prototypes here ... */
5338 We mark all non-exported functions with "/* PRIVATE */"":
5341 png_non_exported_function(png_ptr, png_info, foo)
5346 The prototypes for non-exported functions (except for those in
5347 pngtest) appear in pngpriv.h above the comment that says
5349 /* Maintainer: Put new private prototypes here ^ */
5351 To avoid polluting the global namespace, the names of all exported
5352 functions and variables begin with "png_", and all publicly visible C
5353 preprocessor macros begin with "PNG". We request that applications that
5354 use libpng *not* begin any of their own symbols with either of these strings.
5356 We put a space after the "sizeof" operator and we omit the
5357 optional parentheses around its argument when the argument
5358 is an expression, not a type name, and we always enclose the
5359 sizeof operator, with its argument, in parentheses:
5361 (sizeof (png_uint_32))
5364 Prior to libpng-1.6.0 we used a "png_sizeof()" macro, formatted as
5365 though it were a function.
5367 Control keywords if, for, while, and switch are always followed by a space
5368 to distinguish them from function calls, which have no trailing space.
5370 We put a space after each comma and after each semicolon
5371 in "for" statements, and we put spaces before and after each
5372 C binary operator and after "for" or "while", and before
5373 "?". We don't put a space between a typecast and the expression
5374 being cast, nor do we put one between a function name and the
5375 left parenthesis that follows it:
5377 for (i = 2; i > 0; --i)
5378 y[i] = a(x) + (int)b;
5380 We prefer #ifdef and #ifndef to #if defined() and #if !defined()
5381 when there is only one macro being tested. We always use parentheses
5384 We express integer constants that are used as bit masks in hex format,
5385 with an even number of lower-case hex digits, and to make them unsigned
5386 (e.g., 0x00U, 0xffU, 0x0100U) and long if they are greater than 0x7fff
5389 We prefer to use underscores rather than camelCase in names, except
5390 for a few type names that we inherit from zlib.h.
5392 We prefer "if (something != 0)" and "if (something == 0)" over
5393 "if (something)" and if "(!something)", respectively, and for pointers
5394 we prefer "if (some_pointer != NULL)" or "if (some_pointer == NULL)".
5396 We do not use the TAB character for indentation in the C sources.
5398 Lines do not exceed 80 characters.
5400 Other rules can be inferred by inspecting the libpng source.
5402 XVI. Y2K Compliance in libpng
5404 Since the PNG Development group is an ad-hoc body, we can't make
5405 an official declaration.
5407 This is your unofficial assurance that libpng from version 0.71 and
5408 upward through 1.6.32 are Y2K compliant. It is my belief that earlier
5409 versions were also Y2K compliant.
5411 Libpng only has two year fields. One is a 2-byte unsigned integer
5412 that will hold years up to 65535. The other, which is deprecated,
5413 holds the date in text format, and will hold years up to 9999.
5416 "png_uint_16 year" in png_time_struct.
5419 "char time_buffer[29]" in png_struct. This is no longer used
5420 in libpng-1.6.x and will be removed from libpng-1.7.0.
5422 There are seven time-related functions:
5424 png_convert_to_rfc_1123_buffer() in png.c
5425 (formerly png_convert_to_rfc_1152() in error, and
5426 also formerly png_convert_to_rfc_1123())
5427 png_convert_from_struct_tm() in pngwrite.c, called
5429 png_convert_from_time_t() in pngwrite.c
5430 png_get_tIME() in pngget.c
5431 png_handle_tIME() in pngrutil.c, called in pngread.c
5432 png_set_tIME() in pngset.c
5433 png_write_tIME() in pngwutil.c, called in pngwrite.c
5435 All appear to handle dates properly in a Y2K environment. The
5436 png_convert_from_time_t() function calls gmtime() to convert from system
5437 clock time, which returns (year - 1900), which we properly convert to
5438 the full 4-digit year. There is a possibility that applications using
5439 libpng are not passing 4-digit years into the png_convert_to_rfc_1123()
5440 function, or that they are incorrectly passing only a 2-digit year
5441 instead of "year - 1900" into the png_convert_from_struct_tm() function,
5442 but this is not under our control. The libpng documentation has always
5443 stated that it works with 4-digit years, and the APIs have been
5446 The tIME chunk itself is also Y2K compliant. It uses a 2-byte unsigned
5447 integer to hold the year, and can hold years as large as 65535.
5449 zlib, upon which libpng depends, is also Y2K compliant. It contains
5450 no date-related code.
5453 Glenn Randers-Pehrson
5455 PNG Development Group