1 libpng-manual.txt - A description on how to use and modify libpng
3 libpng version 1.6.0 - February 14, 2013
4 Updated and distributed by Glenn Randers-Pehrson
5 <glennrp at users.sourceforge.net>
6 Copyright (c) 1998-2013 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 1.6.0 - February 14, 2013
15 Updated and distributed by Glenn Randers-Pehrson
16 Copyright (c) 1998-2013 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. There are five sections to this
54 file: introduction, structures, reading, writing, and modification and
55 configuration notes for various special platforms. In addition to this
56 file, example.c is a good starting point for using the library, as
57 it is heavily commented and should include everything most people
58 will need. We assume that libpng is already installed; see the
59 INSTALL file for instructions on how to install libpng.
61 For examples of libpng usage, see the files "example.c", "pngtest.c",
62 and the files in the "contrib" directory, all of which are included in
63 the libpng distribution.
65 Libpng was written as a companion to the PNG specification, as a way
66 of reducing the amount of time and effort it takes to support the PNG
67 file format in application programs.
69 The PNG specification (second edition), November 2003, is available as
70 a W3C Recommendation and as an ISO Standard (ISO/IEC 15948:2004 (E)) at
71 <http://www.w3.org/TR/2003/REC-PNG-20031110/
72 The W3C and ISO documents have identical technical content.
74 The PNG-1.2 specification is available at
75 <http://www.libpng.org/pub/png/documents/>. It is technically equivalent
76 to the PNG specification (second edition) but has some additional material.
78 The PNG-1.0 specification is available
79 as RFC 2083 <http://www.libpng.org/pub/png/documents/> and as a
80 W3C Recommendation <http://www.w3.org/TR/REC.png.html>.
82 Some additional chunks are described in the special-purpose public chunks
83 documents at <http://www.libpng.org/pub/png/documents/>.
86 about PNG, and the latest version of libpng, can be found at the PNG home
87 page, <http://www.libpng.org/pub/png/>.
89 Most users will not have to modify the library significantly; advanced
90 users may want to modify it more. All attempts were made to make it as
91 complete as possible, while keeping the code easy to understand.
92 Currently, this library only supports C. Support for other languages
95 Libpng has been designed to handle multiple sessions at one time,
96 to be easily modifiable, to be portable to the vast majority of
97 machines (ANSI, K&R, 16-, 32-, and 64-bit) available, and to be easy
98 to use. The ultimate goal of libpng is to promote the acceptance of
99 the PNG file format in whatever way possible. While there is still
100 work to be done (see the TODO file), libpng should cover the
101 majority of the needs of its users.
103 Libpng uses zlib for its compression and decompression of PNG files.
104 Further information about zlib, and the latest version of zlib, can
105 be found at the zlib home page, <http://www.info-zip.org/pub/infozip/zlib/>.
106 The zlib compression utility is a general purpose utility that is
107 useful for more than PNG files, and can be used without libpng.
108 See the documentation delivered with zlib for more details.
109 You can usually find the source files for the zlib utility wherever you
110 find the libpng source files.
112 Libpng is thread safe, provided the threads are using different
113 instances of the structures. Each thread should have its own
114 png_struct and png_info instances, and thus its own image.
115 Libpng does not protect itself against two threads using the
116 same instance of a structure.
120 There are two main structures that are important to libpng, png_struct
121 and png_info. Both are internal structures that are no longer exposed
122 in the libpng interface (as of libpng 1.5.0).
124 The png_info structure is designed to provide information about the
125 PNG file. At one time, the fields of png_info were intended to be
126 directly accessible to the user. However, this tended to cause problems
127 with applications using dynamically loaded libraries, and as a result
128 a set of interface functions for png_info (the png_get_*() and png_set_*()
129 functions) was developed, and direct access to the png_info fields was
132 The png_struct structure is the object used by the library to decode a
133 single image. As of 1.5.0 this structure is also not exposed.
135 Almost all libpng APIs require a pointer to a png_struct as the first argument.
136 Many (in particular the png_set and png_get APIs) also require a pointer
137 to png_info as the second argument. Some application visible macros
138 defined in png.h designed for basic data access (reading and writing
139 integers in the PNG format) don't take a png_info pointer, but it's almost
140 always safe to assume that a (png_struct*) has to be passed to call an API
143 You can have more than one png_info structure associated with an image,
144 as illustrated in pngtest.c, one for information valid prior to the
145 IDAT chunks and another (called "end_info" below) for things after them.
147 The png.h header file is an invaluable reference for programming with libpng.
148 And while I'm on the topic, make sure you include the libpng header file:
152 and also (as of libpng-1.5.0) the zlib header file, if you need it:
158 The png.h header file defines a number of integral types used by the
159 APIs. Most of these are fairly obvious; for example types corresponding
160 to integers of particular sizes and types for passing color values.
162 One exception is how non-integral numbers are handled. For application
163 convenience most APIs that take such numbers have C (double) arguments;
164 however, internally PNG, and libpng, use 32 bit signed integers and encode
165 the value by multiplying by 100,000. As of libpng 1.5.0 a convenience
166 macro PNG_FP_1 is defined in png.h along with a type (png_fixed_point)
167 which is simply (png_int_32).
169 All APIs that take (double) arguments also have a matching API that
170 takes the corresponding fixed point integer arguments. The fixed point
171 API has the same name as the floating point one with "_fixed" appended.
172 The actual range of values permitted in the APIs is frequently less than
173 the full range of (png_fixed_point) (-21474 to +21474). When APIs require
174 a non-negative argument the type is recorded as png_uint_32 above. Consult
175 the header file and the text below for more information.
177 Special care must be take with sCAL chunk handling because the chunk itself
178 uses non-integral values encoded as strings containing decimal floating point
179 numbers. See the comments in the header file.
183 The main header file function declarations are frequently protected by C
184 preprocessing directives of the form:
186 #ifdef PNG_feature_SUPPORTED
190 #ifdef PNG_feature_SUPPORTED
194 The library can be built without support for these APIs, although a
195 standard build will have all implemented APIs. Application programs
196 should check the feature macros before using an API for maximum
197 portability. From libpng 1.5.0 the feature macros set during the build
198 of libpng are recorded in the header file "pnglibconf.h" and this file
199 is always included by png.h.
201 If you don't need to change the library configuration from the default, skip to
202 the next section ("Reading").
204 Notice that some of the makefiles in the 'scripts' directory and (in 1.5.0) all
205 of the build project files in the 'projects' directory simply copy
206 scripts/pnglibconf.h.prebuilt to pnglibconf.h. This means that these build
207 systems do not permit easy auto-configuration of the library - they only
208 support the default configuration.
210 The easiest way to make minor changes to the libpng configuration when
211 auto-configuration is supported is to add definitions to the command line
212 using (typically) CPPFLAGS. For example:
214 CPPFLAGS=-DPNG_NO_FLOATING_ARITHMETIC
216 will change the internal libpng math implementation for gamma correction and
217 other arithmetic calculations to fixed point, avoiding the need for fast
218 floating point support. The result can be seen in the generated pnglibconf.h -
219 make sure it contains the changed feature macro setting.
221 If you need to make more extensive configuration changes - more than one or two
222 feature macro settings - you can either add -DPNG_USER_CONFIG to the build
223 command line and put a list of feature macro settings in pngusr.h or you can set
224 DFA_XTRA (a makefile variable) to a file containing the same information in the
225 form of 'option' settings.
227 A. Changing pnglibconf.h
229 A variety of methods exist to build libpng. Not all of these support
230 reconfiguration of pnglibconf.h. To reconfigure pnglibconf.h it must either be
231 rebuilt from scripts/pnglibconf.dfa using awk or it must be edited by hand.
233 Hand editing is achieved by copying scripts/pnglibconf.h.prebuilt to
234 pnglibconf.h and changing the lines defining the supported features, paying
235 very close attention to the 'option' information in scripts/pnglibconf.dfa
236 that describes those features and their requirements. This is easy to get
239 B. Configuration using DFA_XTRA
241 Rebuilding from pnglibconf.dfa is easy if a functioning 'awk', or a later
242 variant such as 'nawk' or 'gawk', is available. The configure build will
243 automatically find an appropriate awk and build pnglibconf.h.
244 The scripts/pnglibconf.mak file contains a set of make rules for doing the
245 same thing if configure is not used, and many of the makefiles in the scripts
246 directory use this approach.
248 When rebuilding simply write a new file containing changed options and set
249 DFA_XTRA to the name of this file. This causes the build to append the new file
250 to the end of scripts/pnglibconf.dfa. The pngusr.dfa file should contain lines
251 of the following forms:
255 This turns all optional features off. Include it at the start of pngusr.dfa to
256 make it easier to build a minimal configuration. You will need to turn at least
257 some features on afterward to enable either reading or writing code, or both.
262 Enable or disable a single feature. This will automatically enable other
263 features required by a feature that is turned on or disable other features that
264 require a feature which is turned off. Conflicting settings will cause an error
265 message to be emitted by awk.
267 setting feature default value
269 Changes the default value of setting 'feature' to 'value'. There are a small
270 number of settings listed at the top of pnglibconf.h, they are documented in the
271 source code. Most of these values have performance implications for the library
272 but most of them have no visible effect on the API. Some can also be overridden
275 This method of building a customized pnglibconf.h is illustrated in
276 contrib/pngminim/*. See the "$(PNGCONF):" target in the makefile and
277 pngusr.dfa in these directories.
279 C. Configuration using PNG_USR_CONFIG
281 If -DPNG_USR_CONFIG is added to the CFLAGS when pnglibconf.h is built the file
282 pngusr.h will automatically be included before the options in
283 scripts/pnglibconf.dfa are processed. Your pngusr.h file should contain only
284 macro definitions turning features on or off or setting settings.
286 Apart from the global setting "everything = off" all the options listed above
287 can be set using macros in pngusr.h:
289 #define PNG_feature_SUPPORTED
295 #define PNG_NO_feature
301 #define PNG_feature value
305 setting feature default value
307 Notice that in both cases, pngusr.dfa and pngusr.h, the contents of the
308 pngusr file you supply override the contents of scripts/pnglibconf.dfa
310 If confusing or incomprehensible behavior results it is possible to
311 examine the intermediate file pnglibconf.dfn to find the full set of
312 dependency information for each setting and option. Simply locate the
313 feature in the file and read the C comments that precede it.
315 This method is also illustrated in the contrib/pngminim/* makefiles and
320 We'll now walk you through the possible functions to call when reading
321 in a PNG file sequentially, briefly explaining the syntax and purpose
322 of each one. See example.c and png.h for more detail. While
323 progressive reading is covered in the next section, you will still
324 need some of the functions discussed in this section to read a PNG
329 You will want to do the I/O initialization(*) before you get into libpng,
330 so if it doesn't work, you don't have much to undo. Of course, you
331 will also want to insure that you are, in fact, dealing with a PNG
332 file. Libpng provides a simple check to see if a file is a PNG file.
333 To use it, pass in the first 1 to 8 bytes of the file to the function
334 png_sig_cmp(), and it will return 0 (false) if the bytes match the
335 corresponding bytes of the PNG signature, or nonzero (true) otherwise.
336 Of course, the more bytes you pass in, the greater the accuracy of the
339 If you are intending to keep the file pointer open for use in libpng,
340 you must ensure you don't read more than 8 bytes from the beginning
341 of the file, and you also have to make a call to png_set_sig_bytes_read()
342 with the number of bytes you read from the beginning. Libpng will
343 then only check the bytes (if any) that your program didn't read.
345 (*): If you are not using the standard I/O functions, you will need
346 to replace them with custom functions. See the discussion under
350 FILE *fp = fopen(file_name, "rb");
356 fread(header, 1, number, fp);
357 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. Note that it is impossible to "discard" data in a critical
473 Choices for (int) crit_action are
474 PNG_CRC_DEFAULT 0 error/quit
475 PNG_CRC_ERROR_QUIT 1 error/quit
476 PNG_CRC_WARN_USE 3 warn/use data
477 PNG_CRC_QUIET_USE 4 quiet/use data
478 PNG_CRC_NO_CHANGE 5 use the current value
480 Choices for (int) ancil_action are
481 PNG_CRC_DEFAULT 0 error/quit
482 PNG_CRC_ERROR_QUIT 1 error/quit
483 PNG_CRC_WARN_DISCARD 2 warn/discard data
484 PNG_CRC_WARN_USE 3 warn/use data
485 PNG_CRC_QUIET_USE 4 quiet/use data
486 PNG_CRC_NO_CHANGE 5 use the current value
488 Setting up callback code
490 You can set up a callback function to handle any unknown chunks in the
491 input stream. You must supply the function
493 read_chunk_callback(png_structp png_ptr,
494 png_unknown_chunkp chunk);
496 /* The unknown chunk structure contains your
497 chunk data, along with similar data for any other
504 /* Note that libpng has already taken care of
507 /* put your code here. Search for your chunk in the
508 unknown chunk structure, process it, and return one
511 return (-n); /* chunk had an error */
512 return (0); /* did not recognize */
513 return (n); /* success */
516 (You can give your function another name that you like instead of
517 "read_chunk_callback")
519 To inform libpng about your function, use
521 png_set_read_user_chunk_fn(png_ptr, user_chunk_ptr,
522 read_chunk_callback);
524 This names not only the callback function, but also a user pointer that
525 you can retrieve with
527 png_get_user_chunk_ptr(png_ptr);
529 If you call the png_set_read_user_chunk_fn() function, then all unknown
530 chunks will be saved when read, in case your callback function will need
531 one or more of them. This behavior can be changed with the
532 png_set_keep_unknown_chunks() function, described below.
534 At this point, you can set up a callback function that will be
535 called after each row has been read, which you can use to control
536 a progress meter or the like. It's demonstrated in pngtest.c.
537 You must supply a function
539 void read_row_callback(png_structp png_ptr,
540 png_uint_32 row, int pass);
542 /* put your code here */
545 (You can give it another name that you like instead of "read_row_callback")
547 To inform libpng about your function, use
549 png_set_read_status_fn(png_ptr, read_row_callback);
551 When this function is called the row has already been completely processed and
552 the 'row' and 'pass' refer to the next row to be handled. For the
553 non-interlaced case the row that was just handled is simply one less than the
554 passed in row number, and pass will always be 0. For the interlaced case the
555 same applies unless the row value is 0, in which case the row just handled was
556 the last one from one of the preceding passes. Because interlacing may skip a
557 pass you cannot be sure that the preceding pass is just 'pass-1', if you really
558 need to know what the last pass is record (row,pass) from the callback and use
559 the last recorded value each time.
561 As with the user transform you can find the output row using the
562 PNG_ROW_FROM_PASS_ROW macro.
564 Unknown-chunk handling
566 Now you get to set the way the library processes unknown chunks in the
567 input PNG stream. Both known and unknown chunks will be read. Normal
568 behavior is that known chunks will be parsed into information in
569 various info_ptr members while unknown chunks will be discarded. This
570 behavior can be wasteful if your application will never use some known
571 chunk types. To change this, you can call:
573 png_set_keep_unknown_chunks(png_ptr, keep,
574 chunk_list, num_chunks);
576 keep - 0: default unknown chunk handling
577 1: ignore; do not keep
578 2: keep only if safe-to-copy
579 3: keep even if unsafe-to-copy
581 You can use these definitions:
582 PNG_HANDLE_CHUNK_AS_DEFAULT 0
583 PNG_HANDLE_CHUNK_NEVER 1
584 PNG_HANDLE_CHUNK_IF_SAFE 2
585 PNG_HANDLE_CHUNK_ALWAYS 3
587 chunk_list - list of chunks affected (a byte string,
588 five bytes per chunk, NULL or '\0' if
589 num_chunks is positive; ignored if
592 num_chunks - number of chunks affected; if 0, all
593 unknown chunks are affected. If positive,
594 only the chunks in the list are affected,
595 and if negative all unknown chunks and
596 all known chunks except for the IHDR,
597 PLTE, tRNS, IDAT, and IEND chunks are
600 Unknown chunks declared in this way will be saved as raw data onto a
601 list of png_unknown_chunk structures. If a chunk that is normally
602 known to libpng is named in the list, it will be handled as unknown,
603 according to the "keep" directive. If a chunk is named in successive
604 instances of png_set_keep_unknown_chunks(), the final instance will
605 take precedence. The IHDR and IEND chunks should not be named in
606 chunk_list; if they are, libpng will process them normally anyway.
607 If you know that your application will never make use of some particular
608 chunks, use PNG_HANDLE_CHUNK_NEVER (or 1) as demonstrated below.
610 Here is an example of the usage of png_set_keep_unknown_chunks(),
611 where the private "vpAg" chunk will later be processed by a user chunk
614 png_byte vpAg[5]={118, 112, 65, 103, (png_byte) '\0'};
616 #if defined(PNG_UNKNOWN_CHUNKS_SUPPORTED)
617 png_byte unused_chunks[]=
619 104, 73, 83, 84, (png_byte) '\0', /* hIST */
620 105, 84, 88, 116, (png_byte) '\0', /* iTXt */
621 112, 67, 65, 76, (png_byte) '\0', /* pCAL */
622 115, 67, 65, 76, (png_byte) '\0', /* sCAL */
623 115, 80, 76, 84, (png_byte) '\0', /* sPLT */
624 116, 73, 77, 69, (png_byte) '\0', /* tIME */
630 #if defined(PNG_UNKNOWN_CHUNKS_SUPPORTED)
631 /* ignore all unknown chunks: */
632 png_set_keep_unknown_chunks(read_ptr, 1, NULL, 0);
634 /* except for vpAg: */
635 png_set_keep_unknown_chunks(read_ptr, 2, vpAg, 1);
637 /* also ignore unused known chunks: */
638 png_set_keep_unknown_chunks(read_ptr, 1, unused_chunks,
639 (int)sizeof(unused_chunks)/5);
644 The PNG specification allows the width and height of an image to be as
645 large as 2^31-1 (0x7fffffff), or about 2.147 billion rows and columns.
646 Since very few applications really need to process such large images,
647 we have imposed an arbitrary 1-million limit on rows and columns.
648 Larger images will be rejected immediately with a png_error() call. If
649 you wish to change this limit, you can use
651 png_set_user_limits(png_ptr, width_max, height_max);
653 to set your own limits, or use width_max = height_max = 0x7fffffffL
654 to allow all valid dimensions (libpng may reject some very large images
655 anyway because of potential buffer overflow conditions).
657 You should put this statement after you create the PNG structure and
658 before calling png_read_info(), png_read_png(), or png_process_data().
660 When writing a PNG datastream, put this statement before calling
661 png_write_info() or png_write_png().
663 If you need to retrieve the limits that are being applied, use
665 width_max = png_get_user_width_max(png_ptr);
666 height_max = png_get_user_height_max(png_ptr);
668 The PNG specification sets no limit on the number of ancillary chunks
669 allowed in a PNG datastream. You can impose a limit on the total number
670 of sPLT, tEXt, iTXt, zTXt, and unknown chunks that will be stored, with
672 png_set_chunk_cache_max(png_ptr, user_chunk_cache_max);
674 where 0x7fffffffL means unlimited. You can retrieve this limit with
676 chunk_cache_max = png_get_chunk_cache_max(png_ptr);
678 You can also set a limit on the amount of memory that a compressed chunk
679 other than IDAT can occupy, with
681 png_set_chunk_malloc_max(png_ptr, user_chunk_malloc_max);
683 and you can retrieve the limit with
685 chunk_malloc_max = png_get_chunk_malloc_max(png_ptr);
687 Any chunks that would cause either of these limits to be exceeded will
690 Information about your system
692 If you intend to display the PNG or to incorporate it in other image data you
693 need to tell libpng information about your display or drawing surface so that
694 libpng can convert the values in the image to match the display.
696 From libpng-1.5.4 this information can be set before reading the PNG file
697 header. In earlier versions png_set_gamma() existed but behaved incorrectly if
698 called before the PNG file header had been read and png_set_alpha_mode() did not
701 If you need to support versions prior to libpng-1.5.4 test the version number
702 as illustrated below using "PNG_LIBPNG_VER >= 10504" and follow the procedures
703 described in the appropriate manual page.
705 You give libpng the encoding expected by your system expressed as a 'gamma'
706 value. You can also specify a default encoding for the PNG file in
707 case the required information is missing from the file. By default libpng
708 assumes that the PNG data matches your system, to keep this default call:
710 png_set_gamma(png_ptr, screen_gamma, 1/screen_gamma/*file gamma*/);
712 or you can use the fixed point equivalent:
714 png_set_gamma_fixed(png_ptr, PNG_FP_1*screen_gamma,
715 PNG_FP_1/screen_gamma);
717 If you don't know the gamma for your system it is probably 2.2 - a good
718 approximation to the IEC standard for display systems (sRGB). If images are
719 too contrasty or washed out you got the value wrong - check your system
722 Many systems permit the system gamma to be changed via a lookup table in the
723 display driver, a few systems, including older Macs, change the response by
724 default. As of 1.5.4 three special values are available to handle common
727 PNG_DEFAULT_sRGB: Indicates that the system conforms to the
728 IEC 61966-2-1 standard. This matches almost
730 PNG_GAMMA_MAC_18: Indicates that the system is an older
731 (pre Mac OS 10.6) Apple Macintosh system with
732 the default settings.
733 PNG_GAMMA_LINEAR: Just the fixed point value for 1.0 - indicates
734 that the system expects data with no gamma
737 You would use the linear (unencoded) value if you need to process the pixel
738 values further because this avoids the need to decode and reencode each
739 component value whenever arithmetic is performed. A lot of graphics software
740 uses linear values for this reason, often with higher precision component values
741 to preserve overall accuracy.
743 The second thing you may need to tell libpng about is how your system handles
744 alpha channel information. Some, but not all, PNG files contain an alpha
745 channel. To display these files correctly you need to compose the data onto a
746 suitable background, as described in the PNG specification.
748 Libpng only supports composing onto a single color (using png_set_background;
749 see below). Otherwise you must do the composition yourself and, in this case,
750 you may need to call png_set_alpha_mode:
752 #if PNG_LIBPNG_VER >= 10504
753 png_set_alpha_mode(png_ptr, mode, screen_gamma);
755 png_set_gamma(png_ptr, screen_gamma, 1.0/screen_gamma);
758 The screen_gamma value is the same as the argument to png_set_gamma; however,
759 how it affects the output depends on the mode. png_set_alpha_mode() sets the
760 file gamma default to 1/screen_gamma, so normally you don't need to call
761 png_set_gamma. If you need different defaults call png_set_gamma() before
762 png_set_alpha_mode() - if you call it after it will override the settings made
763 by png_set_alpha_mode().
765 The mode is as follows:
767 PNG_ALPHA_PNG: The data is encoded according to the PNG specification. Red,
768 green and blue, or gray, components are gamma encoded color
769 values and are not premultiplied by the alpha value. The
770 alpha value is a linear measure of the contribution of the
771 pixel to the corresponding final output pixel.
773 You should normally use this format if you intend to perform
774 color correction on the color values; most, maybe all, color
775 correction software has no handling for the alpha channel and,
776 anyway, the math to handle pre-multiplied component values is
777 unnecessarily complex.
779 Before you do any arithmetic on the component values you need
780 to remove the gamma encoding and multiply out the alpha
781 channel. See the PNG specification for more detail. It is
782 important to note that when an image with an alpha channel is
783 scaled, linear encoded, pre-multiplied component values must
786 The remaining modes assume you don't need to do any further color correction or
787 that if you do, your color correction software knows all about alpha (it
790 PNG_ALPHA_STANDARD: The data libpng produces
791 is encoded in the standard way
792 assumed by most correctly written graphics software.
793 The gamma encoding will be removed by libpng and the
794 linear component values will be pre-multiplied by the
797 With this format the final image must be re-encoded to
798 match the display gamma before the image is displayed.
799 If your system doesn't do that, yet still seems to
800 perform arithmetic on the pixels without decoding them,
801 it is broken - check out the modes below.
803 With PNG_ALPHA_STANDARD libpng always produces linear
804 component values, whatever screen_gamma you supply. The
805 screen_gamma value is, however, used as a default for
806 the file gamma if the PNG file has no gamma information.
808 If you call png_set_gamma() after png_set_alpha_mode() you
809 will override the linear encoding. Instead the
810 pre-multiplied pixel values will be gamma encoded but
811 the alpha channel will still be linear. This may
812 actually match the requirements of some broken software,
815 While linear 8-bit data is often used it has
816 insufficient precision for any image with a reasonable
817 dynamic range. To avoid problems, and if your software
818 supports it, use png_set_expand_16() to force all
819 components to 16 bits.
821 PNG_ALPHA_OPTIMIZED: This mode is the same
822 as PNG_ALPHA_STANDARD except that
823 completely opaque pixels are gamma encoded according to
824 the screen_gamma value. Pixels with alpha less than 1.0
825 will still have linear components.
827 Use this format if you have control over your
828 compositing software and so don't do other arithmetic
829 (such as scaling) on the data you get from libpng. Your
830 compositing software can simply copy opaque pixels to
831 the output but still has linear values for the
834 In normal compositing, where the alpha channel encodes
835 partial pixel coverage (as opposed to broad area
836 translucency), the inaccuracies of the 8-bit
837 representation of non-opaque pixels are irrelevant.
839 You can also try this format if your software is broken;
840 it might look better.
842 PNG_ALPHA_BROKEN: This is PNG_ALPHA_STANDARD;
843 however, all component values,
844 including the alpha channel are gamma encoded. This is
845 an appropriate format to try if your software, or more
846 likely hardware, is totally broken, i.e., if it performs
847 linear arithmetic directly on gamma encoded values.
849 In most cases of broken software or hardware the bug in the final display
850 manifests as a subtle halo around composited parts of the image. You may not
851 even perceive this as a halo; the composited part of the image may simply appear
852 separate from the background, as though it had been cut out of paper and pasted
855 If you don't have to deal with bugs in software or hardware, or if you can fix
856 them, there are three recommended ways of using png_set_alpha_mode():
858 png_set_alpha_mode(png_ptr, PNG_ALPHA_PNG,
861 You can do color correction on the result (libpng does not currently
862 support color correction internally). When you handle the alpha channel
863 you need to undo the gamma encoding and multiply out the alpha.
865 png_set_alpha_mode(png_ptr, PNG_ALPHA_STANDARD,
867 png_set_expand_16(png_ptr);
869 If you are using the high level interface, don't call png_set_expand_16();
870 instead pass PNG_TRANSFORM_EXPAND_16 to the interface.
872 With this mode you can't do color correction, but you can do arithmetic,
873 including composition and scaling, on the data without further processing.
875 png_set_alpha_mode(png_ptr, PNG_ALPHA_OPTIMIZED,
878 You can avoid the expansion to 16-bit components with this mode, but you
879 lose the ability to scale the image or perform other linear arithmetic.
880 All you can do is compose the result onto a matching output. Since this
881 mode is libpng-specific you also need to write your own composition
884 If you don't need, or can't handle, the alpha channel you can call
885 png_set_background() to remove it by compositing against a fixed color. Don't
886 call png_set_strip_alpha() to do this - it will leave spurious pixel values in
887 transparent parts of this image.
889 png_set_background(png_ptr, &background_color,
890 PNG_BACKGROUND_GAMMA_SCREEN, 0, 1);
892 The background_color is an RGB or grayscale value according to the data format
893 libpng will produce for you. Because you don't yet know the format of the PNG
894 file, if you call png_set_background at this point you must arrange for the
895 format produced by libpng to always have 8-bit or 16-bit components and then
896 store the color as an 8-bit or 16-bit color as appropriate. The color contains
897 separate gray and RGB component values, so you can let libpng produce gray or
898 RGB output according to the input format, but low bit depth grayscale images
899 must always be converted to at least 8-bit format. (Even though low bit depth
900 grayscale images can't have an alpha channel they can have a transparent
903 You set the transforms you need later, either as flags to the high level
904 interface or libpng API calls for the low level interface. For reference the
905 settings and API calls required are:
908 PNG_TRANSFORM_SCALE_16 | PNG_EXPAND
909 png_set_expand(png_ptr); png_set_scale_16(png_ptr);
911 If you must get exactly the same inaccurate results
912 produced by default in versions prior to libpng-1.5.4,
913 use PNG_TRANSFORM_STRIP_16 and png_set_strip_16(png_ptr)
917 PNG_TRANSFORM_EXPAND_16
918 png_set_expand_16(png_ptr);
920 In either case palette image data will be expanded to RGB. If you just want
921 color data you can add PNG_TRANSFORM_GRAY_TO_RGB or png_set_gray_to_rgb(png_ptr)
924 Calling png_set_background before the PNG file header is read will not work
925 prior to libpng-1.5.4. Because the failure may result in unexpected warnings or
926 errors it is therefore much safer to call png_set_background after the head has
927 been read. Unfortunately this means that prior to libpng-1.5.4 it cannot be
928 used with the high level interface.
930 The high-level read interface
932 At this point there are two ways to proceed; through the high-level
933 read interface, or through a sequence of low-level read operations.
934 You can use the high-level interface if (a) you are willing to read
935 the entire image into memory, and (b) the input transformations
936 you want to do are limited to the following set:
938 PNG_TRANSFORM_IDENTITY No transformation
939 PNG_TRANSFORM_SCALE_16 Strip 16-bit samples to
941 PNG_TRANSFORM_STRIP_16 Chop 16-bit samples to
942 8-bit less accurately
943 PNG_TRANSFORM_STRIP_ALPHA Discard the alpha channel
944 PNG_TRANSFORM_PACKING Expand 1, 2 and 4-bit
946 PNG_TRANSFORM_PACKSWAP Change order of packed
948 PNG_TRANSFORM_EXPAND Perform set_expand()
949 PNG_TRANSFORM_INVERT_MONO Invert monochrome images
950 PNG_TRANSFORM_SHIFT Normalize pixels to the
952 PNG_TRANSFORM_BGR Flip RGB to BGR, RGBA
954 PNG_TRANSFORM_SWAP_ALPHA Flip RGBA to ARGB or GA
956 PNG_TRANSFORM_INVERT_ALPHA Change alpha from opacity
958 PNG_TRANSFORM_SWAP_ENDIAN Byte-swap 16-bit samples
959 PNG_TRANSFORM_GRAY_TO_RGB Expand grayscale samples
960 to RGB (or GA to RGBA)
961 PNG_TRANSFORM_EXPAND_16 Expand samples to 16 bits
963 (This excludes setting a background color, doing gamma transformation,
964 quantizing, and setting filler.) If this is the case, simply do this:
966 png_read_png(png_ptr, info_ptr, png_transforms, NULL)
968 where png_transforms is an integer containing the bitwise OR of some
969 set of transformation flags. This call is equivalent to png_read_info(),
970 followed the set of transformations indicated by the transform mask,
971 then png_read_image(), and finally png_read_end().
973 (The final parameter of this call is not yet used. Someday it might point
974 to transformation parameters required by some future input transform.)
976 You must use png_transforms and not call any png_set_transform() functions
977 when you use png_read_png().
979 After you have called png_read_png(), you can retrieve the image data
982 row_pointers = png_get_rows(png_ptr, info_ptr);
984 where row_pointers is an array of pointers to the pixel data for each row:
986 png_bytep row_pointers[height];
988 If you know your image size and pixel size ahead of time, you can allocate
989 row_pointers prior to calling png_read_png() with
991 if (height > PNG_UINT_32_MAX/png_sizeof(png_byte))
993 "Image is too tall to process in memory");
995 if (width > PNG_UINT_32_MAX/pixel_size)
997 "Image is too wide to process in memory");
999 row_pointers = png_malloc(png_ptr,
1000 height*png_sizeof(png_bytep));
1002 for (int i=0; i<height, i++)
1003 row_pointers[i]=NULL; /* security precaution */
1005 for (int i=0; i<height, i++)
1006 row_pointers[i]=png_malloc(png_ptr,
1009 png_set_rows(png_ptr, info_ptr, &row_pointers);
1011 Alternatively you could allocate your image in one big block and define
1012 row_pointers[i] to point into the proper places in your block.
1014 If you use png_set_rows(), the application is responsible for freeing
1015 row_pointers (and row_pointers[i], if they were separately allocated).
1017 If you don't allocate row_pointers ahead of time, png_read_png() will
1018 do it, and it'll be free'ed by libpng when you call png_destroy_*().
1020 The low-level read interface
1022 If you are going the low-level route, you are now ready to read all
1023 the file information up to the actual image data. You do this with a
1024 call to png_read_info().
1026 png_read_info(png_ptr, info_ptr);
1028 This will process all chunks up to but not including the image data.
1030 This also copies some of the data from the PNG file into the decode structure
1031 for use in later transformations. Important information copied in is:
1033 1) The PNG file gamma from the gAMA chunk. This overwrites the default value
1034 provided by an earlier call to png_set_gamma or png_set_alpha_mode.
1036 2) Prior to libpng-1.5.4 the background color from a bKGd chunk. This
1037 damages the information provided by an earlier call to png_set_background
1038 resulting in unexpected behavior. Libpng-1.5.4 no longer does this.
1040 3) The number of significant bits in each component value. Libpng uses this to
1041 optimize gamma handling by reducing the internal lookup table sizes.
1043 4) The transparent color information from a tRNS chunk. This can be modified by
1044 a later call to png_set_tRNS.
1046 Querying the info structure
1048 Functions are used to get the information from the info_ptr once it
1049 has been read. Note that these fields may not be completely filled
1050 in until png_read_end() has read the chunk data following the image.
1052 png_get_IHDR(png_ptr, info_ptr, &width, &height,
1053 &bit_depth, &color_type, &interlace_type,
1054 &compression_type, &filter_method);
1056 width - holds the width of the image
1057 in pixels (up to 2^31).
1059 height - holds the height of the image
1060 in pixels (up to 2^31).
1062 bit_depth - holds the bit depth of one of the
1063 image channels. (valid values are
1064 1, 2, 4, 8, 16 and depend also on
1065 the color_type. See also
1066 significant bits (sBIT) below).
1068 color_type - describes which color/alpha channels
1071 (bit depths 1, 2, 4, 8, 16)
1072 PNG_COLOR_TYPE_GRAY_ALPHA
1074 PNG_COLOR_TYPE_PALETTE
1075 (bit depths 1, 2, 4, 8)
1078 PNG_COLOR_TYPE_RGB_ALPHA
1081 PNG_COLOR_MASK_PALETTE
1082 PNG_COLOR_MASK_COLOR
1083 PNG_COLOR_MASK_ALPHA
1085 interlace_type - (PNG_INTERLACE_NONE or
1086 PNG_INTERLACE_ADAM7)
1088 compression_type - (must be PNG_COMPRESSION_TYPE_BASE
1091 filter_method - (must be PNG_FILTER_TYPE_BASE
1092 for PNG 1.0, and can also be
1093 PNG_INTRAPIXEL_DIFFERENCING if
1094 the PNG datastream is embedded in
1095 a MNG-1.0 datastream)
1097 Any or all of interlace_type, compression_type, or
1098 filter_method can be NULL if you are
1099 not interested in their values.
1101 Note that png_get_IHDR() returns 32-bit data into
1102 the application's width and height variables.
1103 This is an unsafe situation if these are 16-bit
1104 variables. In such situations, the
1105 png_get_image_width() and png_get_image_height()
1106 functions described below are safer.
1108 width = png_get_image_width(png_ptr,
1111 height = png_get_image_height(png_ptr,
1114 bit_depth = png_get_bit_depth(png_ptr,
1117 color_type = png_get_color_type(png_ptr,
1120 interlace_type = png_get_interlace_type(png_ptr,
1123 compression_type = png_get_compression_type(png_ptr,
1126 filter_method = png_get_filter_type(png_ptr,
1129 channels = png_get_channels(png_ptr, info_ptr);
1131 channels - number of channels of info for the
1132 color type (valid values are 1 (GRAY,
1133 PALETTE), 2 (GRAY_ALPHA), 3 (RGB),
1134 4 (RGB_ALPHA or RGB + filler byte))
1136 rowbytes = png_get_rowbytes(png_ptr, info_ptr);
1138 rowbytes - number of bytes needed to hold a row
1140 signature = png_get_signature(png_ptr, info_ptr);
1142 signature - holds the signature read from the
1143 file (if any). The data is kept in
1144 the same offset it would be if the
1145 whole signature were read (i.e. if an
1146 application had already read in 4
1147 bytes of signature before starting
1148 libpng, the remaining 4 bytes would
1149 be in signature[4] through signature[7]
1150 (see png_set_sig_bytes())).
1152 These are also important, but their validity depends on whether the chunk
1153 has been read. The png_get_valid(png_ptr, info_ptr, PNG_INFO_<chunk>) and
1154 png_get_<chunk>(png_ptr, info_ptr, ...) functions return non-zero if the
1155 data has been read, or zero if it is missing. The parameters to the
1156 png_get_<chunk> are set directly if they are simple data types, or a
1157 pointer into the info_ptr is returned for any complex types.
1159 The colorspace data from gAMA, cHRM, sRGB, iCCP, and sBIT chunks
1160 is simply returned to give the application information about how the
1161 image was encoded. Libpng itself only does transformations using the file
1162 gamma when combining semitransparent pixels with the background color, and,
1163 since libpng-1.6.0, when converting between 8-bit sRGB and 16-bit linear pixels
1164 within the simplified API. Libpng also uses the file gamma when converting
1165 RGB to gray, beginning with libpng-1.0.5, if the application calls
1166 png_set_rgb_to_gray()).
1168 png_get_PLTE(png_ptr, info_ptr, &palette,
1171 palette - the palette for the file
1172 (array of png_color)
1174 num_palette - number of entries in the palette
1176 png_get_gAMA(png_ptr, info_ptr, &file_gamma);
1177 png_get_gAMA_fixed(png_ptr, info_ptr, &int_file_gamma);
1179 file_gamma - the gamma at which the file is
1180 written (PNG_INFO_gAMA)
1182 int_file_gamma - 100,000 times the gamma at which the
1185 png_get_cHRM(png_ptr, info_ptr, &white_x, &white_y, &red_x,
1186 &red_y, &green_x, &green_y, &blue_x, &blue_y)
1187 png_get_cHRM_XYZ(png_ptr, info_ptr, &red_X, &red_Y, &red_Z,
1188 &green_X, &green_Y, &green_Z, &blue_X, &blue_Y,
1190 png_get_cHRM_fixed(png_ptr, info_ptr, &int_white_x,
1191 &int_white_y, &int_red_x, &int_red_y,
1192 &int_green_x, &int_green_y, &int_blue_x,
1194 png_get_cHRM_XYZ_fixed(png_ptr, info_ptr, &int_red_X, &int_red_Y,
1195 &int_red_Z, &int_green_X, &int_green_Y,
1196 &int_green_Z, &int_blue_X, &int_blue_Y,
1199 {white,red,green,blue}_{x,y}
1200 A color space encoding specified using the
1201 chromaticities of the end points and the
1202 white point. (PNG_INFO_cHRM)
1204 {red,green,blue}_{X,Y,Z}
1205 A color space encoding specified using the
1206 encoding end points - the CIE tristimulus
1207 specification of the intended color of the red,
1208 green and blue channels in the PNG RGB data.
1209 The white point is simply the sum of the three
1210 end points. (PNG_INFO_cHRM)
1212 png_get_sRGB(png_ptr, info_ptr, &srgb_intent);
1214 file_srgb_intent - the rendering intent (PNG_INFO_sRGB)
1215 The presence of the sRGB chunk
1216 means that the pixel data is in the
1217 sRGB color space. This chunk also
1218 implies specific values of gAMA and
1221 png_get_iCCP(png_ptr, info_ptr, &name,
1222 &compression_type, &profile, &proflen);
1224 name - The profile name.
1226 compression_type - The compression type; always
1227 PNG_COMPRESSION_TYPE_BASE for PNG 1.0.
1228 You may give NULL to this argument to
1231 profile - International Color Consortium color
1232 profile data. May contain NULs.
1234 proflen - length of profile data in bytes.
1236 png_get_sBIT(png_ptr, info_ptr, &sig_bit);
1238 sig_bit - the number of significant bits for
1239 (PNG_INFO_sBIT) each of the gray,
1240 red, green, and blue channels,
1241 whichever are appropriate for the
1242 given color type (png_color_16)
1244 png_get_tRNS(png_ptr, info_ptr, &trans_alpha,
1245 &num_trans, &trans_color);
1247 trans_alpha - array of alpha (transparency)
1248 entries for palette (PNG_INFO_tRNS)
1250 num_trans - number of transparent entries
1253 trans_color - graylevel or color sample values of
1254 the single transparent color for
1255 non-paletted images (PNG_INFO_tRNS)
1257 png_get_hIST(png_ptr, info_ptr, &hist);
1260 hist - histogram of palette (array of
1263 png_get_tIME(png_ptr, info_ptr, &mod_time);
1265 mod_time - time image was last modified
1268 png_get_bKGD(png_ptr, info_ptr, &background);
1270 background - background color (of type
1271 png_color_16p) (PNG_VALID_bKGD)
1272 valid 16-bit red, green and blue
1273 values, regardless of color_type
1275 num_comments = png_get_text(png_ptr, info_ptr,
1276 &text_ptr, &num_text);
1278 num_comments - number of comments
1280 text_ptr - array of png_text holding image
1283 text_ptr[i].compression - type of compression used
1284 on "text" PNG_TEXT_COMPRESSION_NONE
1285 PNG_TEXT_COMPRESSION_zTXt
1286 PNG_ITXT_COMPRESSION_NONE
1287 PNG_ITXT_COMPRESSION_zTXt
1289 text_ptr[i].key - keyword for comment. Must contain
1292 text_ptr[i].text - text comments for current
1293 keyword. Can be empty.
1295 text_ptr[i].text_length - length of text string,
1296 after decompression, 0 for iTXt
1298 text_ptr[i].itxt_length - length of itxt string,
1299 after decompression, 0 for tEXt/zTXt
1301 text_ptr[i].lang - language of comment (empty
1302 string for unknown).
1304 text_ptr[i].lang_key - keyword in UTF-8
1305 (empty string for unknown).
1307 Note that the itxt_length, lang, and lang_key
1308 members of the text_ptr structure only exist when the
1309 library is built with iTXt chunk support. Prior to
1310 libpng-1.4.0 the library was built by default without
1311 iTXt support. Also note that when iTXt is supported,
1312 they contain NULL pointers when the "compression"
1313 field contains PNG_TEXT_COMPRESSION_NONE or
1314 PNG_TEXT_COMPRESSION_zTXt.
1316 num_text - number of comments (same as
1317 num_comments; you can put NULL here
1318 to avoid the duplication)
1320 Note while png_set_text() will accept text, language,
1321 and translated keywords that can be NULL pointers, the
1322 structure returned by png_get_text will always contain
1323 regular zero-terminated C strings. They might be
1324 empty strings but they will never be NULL pointers.
1326 num_spalettes = png_get_sPLT(png_ptr, info_ptr,
1329 num_spalettes - number of sPLT chunks read.
1331 palette_ptr - array of palette structures holding
1332 contents of one or more sPLT chunks
1335 png_get_oFFs(png_ptr, info_ptr, &offset_x, &offset_y,
1338 offset_x - positive offset from the left edge
1339 of the screen (can be negative)
1341 offset_y - positive offset from the top edge
1342 of the screen (can be negative)
1344 unit_type - PNG_OFFSET_PIXEL, PNG_OFFSET_MICROMETER
1346 png_get_pHYs(png_ptr, info_ptr, &res_x, &res_y,
1349 res_x - pixels/unit physical resolution in
1352 res_y - pixels/unit physical resolution in
1355 unit_type - PNG_RESOLUTION_UNKNOWN,
1356 PNG_RESOLUTION_METER
1358 png_get_sCAL(png_ptr, info_ptr, &unit, &width,
1361 unit - physical scale units (an integer)
1363 width - width of a pixel in physical scale units
1365 height - height of a pixel in physical scale units
1366 (width and height are doubles)
1368 png_get_sCAL_s(png_ptr, info_ptr, &unit, &width,
1371 unit - physical scale units (an integer)
1373 width - width of a pixel in physical scale units
1374 (expressed as a string)
1376 height - height of a pixel in physical scale units
1377 (width and height are strings like "2.54")
1379 num_unknown_chunks = png_get_unknown_chunks(png_ptr,
1380 info_ptr, &unknowns)
1382 unknowns - array of png_unknown_chunk
1383 structures holding unknown chunks
1385 unknowns[i].name - name of unknown chunk
1387 unknowns[i].data - data of unknown chunk
1389 unknowns[i].size - size of unknown chunk's data
1391 unknowns[i].location - position of chunk in file
1393 The value of "i" corresponds to the order in which the
1394 chunks were read from the PNG file or inserted with the
1395 png_set_unknown_chunks() function.
1397 The value of "location" is a bitwise "or" of
1399 PNG_HAVE_IHDR (0x01)
1400 PNG_HAVE_PLTE (0x02)
1401 PNG_AFTER_IDAT (0x08)
1403 The data from the pHYs chunk can be retrieved in several convenient
1406 res_x = png_get_x_pixels_per_meter(png_ptr,
1409 res_y = png_get_y_pixels_per_meter(png_ptr,
1412 res_x_and_y = png_get_pixels_per_meter(png_ptr,
1415 res_x = png_get_x_pixels_per_inch(png_ptr,
1418 res_y = png_get_y_pixels_per_inch(png_ptr,
1421 res_x_and_y = png_get_pixels_per_inch(png_ptr,
1424 aspect_ratio = png_get_pixel_aspect_ratio(png_ptr,
1427 Each of these returns 0 [signifying "unknown"] if
1428 the data is not present or if res_x is 0;
1429 res_x_and_y is 0 if res_x != res_y
1431 Note that because of the way the resolutions are
1432 stored internally, the inch conversions won't
1433 come out to exactly even number. For example,
1434 72 dpi is stored as 0.28346 pixels/meter, and
1435 when this is retrieved it is 71.9988 dpi, so
1436 be sure to round the returned value appropriately
1437 if you want to display a reasonable-looking result.
1439 The data from the oFFs chunk can be retrieved in several convenient
1442 x_offset = png_get_x_offset_microns(png_ptr, info_ptr);
1444 y_offset = png_get_y_offset_microns(png_ptr, info_ptr);
1446 x_offset = png_get_x_offset_inches(png_ptr, info_ptr);
1448 y_offset = png_get_y_offset_inches(png_ptr, info_ptr);
1450 Each of these returns 0 [signifying "unknown" if both
1451 x and y are 0] if the data is not present or if the
1452 chunk is present but the unit is the pixel. The
1453 remark about inexact inch conversions applies here
1454 as well, because a value in inches can't always be
1455 converted to microns and back without some loss
1458 For more information, see the
1459 PNG specification for chunk contents. Be careful with trusting
1460 rowbytes, as some of the transformations could increase the space
1461 needed to hold a row (expand, filler, gray_to_rgb, etc.).
1462 See png_read_update_info(), below.
1464 A quick word about text_ptr and num_text. PNG stores comments in
1465 keyword/text pairs, one pair per chunk, with no limit on the number
1466 of text chunks, and a 2^31 byte limit on their size. While there are
1467 suggested keywords, there is no requirement to restrict the use to these
1468 strings. It is strongly suggested that keywords and text be sensible
1469 to humans (that's the point), so don't use abbreviations. Non-printing
1470 symbols are not allowed. See the PNG specification for more details.
1471 There is also no requirement to have text after the keyword.
1473 Keywords should be limited to 79 Latin-1 characters without leading or
1474 trailing spaces, but non-consecutive spaces are allowed within the
1475 keyword. It is possible to have the same keyword any number of times.
1476 The text_ptr is an array of png_text structures, each holding a
1477 pointer to a language string, a pointer to a keyword and a pointer to
1478 a text string. The text string, language code, and translated
1479 keyword may be empty or NULL pointers. The keyword/text
1480 pairs are put into the array in the order that they are received.
1481 However, some or all of the text chunks may be after the image, so, to
1482 make sure you have read all the text chunks, don't mess with these
1483 until after you read the stuff after the image. This will be
1484 mentioned again below in the discussion that goes with png_read_end().
1486 Input transformations
1488 After you've read the header information, you can set up the library
1489 to handle any special transformations of the image data. The various
1490 ways to transform the data will be described in the order that they
1491 should occur. This is important, as some of these change the color
1492 type and/or bit depth of the data, and some others only work on
1493 certain color types and bit depths.
1495 Transformations you request are ignored if they don't have any meaning for a
1496 particular input data format. However some transformations can have an effect
1497 as a result of a previous transformation. If you specify a contradictory set of
1498 transformations, for example both adding and removing the alpha channel, you
1499 cannot predict the final result.
1501 The color used for the transparency values should be supplied in the same
1502 format/depth as the current image data. It is stored in the same format/depth
1503 as the image data in a tRNS chunk, so this is what libpng expects for this data.
1505 The color used for the background value depends on the need_expand argument as
1508 Data will be decoded into the supplied row buffers packed into bytes
1509 unless the library has been told to transform it into another format.
1510 For example, 4 bit/pixel paletted or grayscale data will be returned
1511 2 pixels/byte with the leftmost pixel in the high-order bits of the
1512 byte, unless png_set_packing() is called. 8-bit RGB data will be stored
1513 in RGB RGB RGB format unless png_set_filler() or png_set_add_alpha()
1514 is called to insert filler bytes, either before or after each RGB triplet.
1515 16-bit RGB data will be returned RRGGBB RRGGBB, with the most significant
1516 byte of the color value first, unless png_set_scale_16() is called to
1517 transform it to regular RGB RGB triplets, or png_set_filler() or
1518 png_set_add alpha() is called to insert filler bytes, either before or
1519 after each RRGGBB triplet. Similarly, 8-bit or 16-bit grayscale data can
1520 be modified with png_set_filler(), png_set_add_alpha(), png_set_strip_16(),
1521 or png_set_scale_16().
1523 The following code transforms grayscale images of less than 8 to 8 bits,
1524 changes paletted images to RGB, and adds a full alpha channel if there is
1525 transparency information in a tRNS chunk. This is most useful on
1526 grayscale images with bit depths of 2 or 4 or if there is a multiple-image
1527 viewing application that wishes to treat all images in the same way.
1529 if (color_type == PNG_COLOR_TYPE_PALETTE)
1530 png_set_palette_to_rgb(png_ptr);
1532 if (png_get_valid(png_ptr, info_ptr,
1533 PNG_INFO_tRNS)) png_set_tRNS_to_alpha(png_ptr);
1535 if (color_type == PNG_COLOR_TYPE_GRAY &&
1536 bit_depth < 8) png_set_expand_gray_1_2_4_to_8(png_ptr);
1538 The first two functions are actually aliases for png_set_expand(), added
1539 in libpng version 1.0.4, with the function names expanded to improve code
1540 readability. In some future version they may actually do different
1543 As of libpng version 1.2.9, png_set_expand_gray_1_2_4_to_8() was
1544 added. It expands the sample depth without changing tRNS to alpha.
1546 As of libpng version 1.5.2, png_set_expand_16() was added. It behaves as
1547 png_set_expand(); however, the resultant channels have 16 bits rather than 8.
1548 Use this when the output color or gray channels are made linear to avoid fairly
1549 severe accuracy loss.
1552 png_set_expand_16(png_ptr);
1554 PNG can have files with 16 bits per channel. If you only can handle
1555 8 bits per channel, this will strip the pixels down to 8-bit.
1557 if (bit_depth == 16)
1558 #if PNG_LIBPNG_VER >= 10504
1559 png_set_scale_16(png_ptr);
1561 png_set_strip_16(png_ptr);
1564 (The more accurate "png_set_scale_16()" API became available in libpng version
1567 If you need to process the alpha channel on the image separately from the image
1568 data (for example if you convert it to a bitmap mask) it is possible to have
1569 libpng strip the channel leaving just RGB or gray data:
1571 if (color_type & PNG_COLOR_MASK_ALPHA)
1572 png_set_strip_alpha(png_ptr);
1574 If you strip the alpha channel you need to find some other way of dealing with
1575 the information. If, instead, you want to convert the image to an opaque
1576 version with no alpha channel use png_set_background; see below.
1578 As of libpng version 1.5.2, almost all useful expansions are supported, the
1579 major ommissions are conversion of grayscale to indexed images (which can be
1580 done trivially in the application) and conversion of indexed to grayscale (which
1581 can be done by a trivial manipulation of the palette.)
1583 In the following table, the 01 means grayscale with depth<8, 31 means
1584 indexed with depth<8, other numerals represent the color type, "T" means
1585 the tRNS chunk is present, A means an alpha channel is present, and O
1586 means tRNS or alpha is present but all pixels in the image are opaque.
1588 FROM 01 31 0 0T 0O 2 2T 2O 3 3T 3O 4A 4O 6A 6O
1590 01 - [G] - - - - - - - - - - - - -
1591 31 [Q] Q [Q] [Q] [Q] Q Q Q Q Q Q [Q] [Q] Q Q
1592 0 1 G + . . G G G G G G B B GB GB
1593 0T lt Gt t + . Gt G G Gt G G Bt Bt GBt GBt
1594 0O lt Gt t . + Gt Gt G Gt Gt G Bt Bt GBt GBt
1595 2 C P C C C + . . C - - CB CB B B
1596 2T Ct - Ct C C t + t - - - CBt CBt Bt Bt
1597 2O Ct - Ct C C t t + - - - CBt CBt Bt Bt
1598 3 [Q] p [Q] [Q] [Q] Q Q Q + . . [Q] [Q] Q Q
1599 3T [Qt] p [Qt][Q] [Q] Qt Qt Qt t + t [Qt][Qt] Qt Qt
1600 3O [Qt] p [Qt][Q] [Q] Qt Qt Qt t t + [Qt][Qt] Qt Qt
1601 4A lA G A T T GA GT GT GA GT GT + BA G GBA
1602 4O lA GBA A T T GA GT GT GA GT GT BA + GBA G
1603 6A CA PA CA C C A T tT PA P P C CBA + BA
1604 6O CA PBA CA C C A tT T PA P P CBA C BA +
1607 "+" identifies entries where 'from' and 'to' are the same.
1608 "-" means the transformation is not supported.
1609 "." means nothing is necessary (a tRNS chunk can just be ignored).
1610 "t" means the transformation is obtained by png_set_tRNS.
1611 "A" means the transformation is obtained by png_set_add_alpha().
1612 "X" means the transformation is obtained by png_set_expand().
1613 "1" means the transformation is obtained by
1614 png_set_expand_gray_1_2_4_to_8() (and by png_set_expand()
1615 if there is no transparency in the original or the final
1617 "C" means the transformation is obtained by png_set_gray_to_rgb().
1618 "G" means the transformation is obtained by png_set_rgb_to_gray().
1619 "P" means the transformation is obtained by
1620 png_set_expand_palette_to_rgb().
1621 "p" means the transformation is obtained by png_set_packing().
1622 "Q" means the transformation is obtained by png_set_quantize().
1623 "T" means the transformation is obtained by
1624 png_set_tRNS_to_alpha().
1625 "B" means the transformation is obtained by
1626 png_set_background(), or png_strip_alpha().
1628 When an entry has multiple transforms listed all are required to cause the
1629 right overall transformation. When two transforms are separated by a comma
1630 either will do the job. When transforms are enclosed in [] the transform should
1631 do the job but this is currently unimplemented - a different format will result
1632 if the suggested transformations are used.
1634 In PNG files, the alpha channel in an image
1635 is the level of opacity. If you need the alpha channel in an image to
1636 be the level of transparency instead of opacity, you can invert the
1637 alpha channel (or the tRNS chunk data) after it's read, so that 0 is
1638 fully opaque and 255 (in 8-bit or paletted images) or 65535 (in 16-bit
1639 images) is fully transparent, with
1641 png_set_invert_alpha(png_ptr);
1643 PNG files pack pixels of bit depths 1, 2, and 4 into bytes as small as
1644 they can, resulting in, for example, 8 pixels per byte for 1 bit
1645 files. This code expands to 1 pixel per byte without changing the
1646 values of the pixels:
1649 png_set_packing(png_ptr);
1651 PNG files have possible bit depths of 1, 2, 4, 8, and 16. All pixels
1652 stored in a PNG image have been "scaled" or "shifted" up to the next
1653 higher possible bit depth (e.g. from 5 bits/sample in the range [0,31]
1654 to 8 bits/sample in the range [0, 255]). However, it is also possible
1655 to convert the PNG pixel data back to the original bit depth of the
1656 image. This call reduces the pixels back down to the original bit depth:
1658 png_color_8p sig_bit;
1660 if (png_get_sBIT(png_ptr, info_ptr, &sig_bit))
1661 png_set_shift(png_ptr, sig_bit);
1663 PNG files store 3-color pixels in red, green, blue order. This code
1664 changes the storage of the pixels to blue, green, red:
1666 if (color_type == PNG_COLOR_TYPE_RGB ||
1667 color_type == PNG_COLOR_TYPE_RGB_ALPHA)
1668 png_set_bgr(png_ptr);
1670 PNG files store RGB pixels packed into 3 or 6 bytes. This code expands them
1671 into 4 or 8 bytes for windowing systems that need them in this format:
1673 if (color_type == PNG_COLOR_TYPE_RGB)
1674 png_set_filler(png_ptr, filler, PNG_FILLER_BEFORE);
1676 where "filler" is the 8 or 16-bit number to fill with, and the location is
1677 either PNG_FILLER_BEFORE or PNG_FILLER_AFTER, depending upon whether
1678 you want the filler before the RGB or after. This transformation
1679 does not affect images that already have full alpha channels. To add an
1680 opaque alpha channel, use filler=0xff or 0xffff and PNG_FILLER_AFTER which
1681 will generate RGBA pixels.
1683 Note that png_set_filler() does not change the color type. If you want
1684 to do that, you can add a true alpha channel with
1686 if (color_type == PNG_COLOR_TYPE_RGB ||
1687 color_type == PNG_COLOR_TYPE_GRAY)
1688 png_set_add_alpha(png_ptr, filler, PNG_FILLER_AFTER);
1690 where "filler" contains the alpha value to assign to each pixel.
1691 This function was added in libpng-1.2.7.
1693 If you are reading an image with an alpha channel, and you need the
1694 data as ARGB instead of the normal PNG format RGBA:
1696 if (color_type == PNG_COLOR_TYPE_RGB_ALPHA)
1697 png_set_swap_alpha(png_ptr);
1699 For some uses, you may want a grayscale image to be represented as
1700 RGB. This code will do that conversion:
1702 if (color_type == PNG_COLOR_TYPE_GRAY ||
1703 color_type == PNG_COLOR_TYPE_GRAY_ALPHA)
1704 png_set_gray_to_rgb(png_ptr);
1706 Conversely, you can convert an RGB or RGBA image to grayscale or grayscale
1709 if (color_type == PNG_COLOR_TYPE_RGB ||
1710 color_type == PNG_COLOR_TYPE_RGB_ALPHA)
1711 png_set_rgb_to_gray(png_ptr, error_action,
1712 double red_weight, double green_weight);
1714 error_action = 1: silently do the conversion
1716 error_action = 2: issue a warning if the original
1717 image has any pixel where
1718 red != green or red != blue
1720 error_action = 3: issue an error and abort the
1721 conversion if the original
1722 image has any pixel where
1723 red != green or red != blue
1725 red_weight: weight of red component
1727 green_weight: weight of green component
1728 If either weight is negative, default
1731 In the corresponding fixed point API the red_weight and green_weight values are
1732 simply scaled by 100,000:
1734 png_set_rgb_to_gray(png_ptr, error_action,
1735 png_fixed_point red_weight,
1736 png_fixed_point green_weight);
1738 If you have set error_action = 1 or 2, you can
1739 later check whether the image really was gray, after processing
1740 the image rows, with the png_get_rgb_to_gray_status(png_ptr) function.
1741 It will return a png_byte that is zero if the image was gray or
1742 1 if there were any non-gray pixels. Background and sBIT data
1743 will be silently converted to grayscale, using the green channel
1744 data for sBIT, regardless of the error_action setting.
1746 The default values come from the PNG file cHRM chunk if present; otherwise, the
1747 defaults correspond to the ITU-R recommendation 709, and also the sRGB color
1748 space, as recommended in the Charles Poynton's Colour FAQ,
1749 <http://www.poynton.com/>, in section 9:
1751 <http://www.poynton.com/notes/colour_and_gamma/ColorFAQ.html#RTFToC9>
1753 Y = 0.2126 * R + 0.7152 * G + 0.0722 * B
1755 Previous versions of this document, 1998 through 2002, recommended a slightly
1758 Y = 0.212671 * R + 0.715160 * G + 0.072169 * B
1760 Libpng uses an integer approximation:
1762 Y = (6968 * R + 23434 * G + 2366 * B)/32768
1764 The calculation is done in a linear colorspace, if the image gamma
1767 The png_set_background() function has been described already; it tells libpng to
1768 composite images with alpha or simple transparency against the supplied
1769 background color. For compatibility with versions of libpng earlier than
1770 libpng-1.5.4 it is recommended that you call the function after reading the file
1771 header, even if you don't want to use the color in a bKGD chunk, if one exists.
1773 If the PNG file contains a bKGD chunk (PNG_INFO_bKGD valid),
1774 you may use this color, or supply another color more suitable for
1775 the current display (e.g., the background color from a web page). You
1776 need to tell libpng how the color is represented, both the format of the
1777 component values in the color (the number of bits) and the gamma encoding of the
1778 color. The function takes two arguments, background_gamma_mode and need_expand
1779 to convey this information; however, only two combinations are likely to be
1782 png_color_16 my_background;
1783 png_color_16p image_background;
1785 if (png_get_bKGD(png_ptr, info_ptr, &image_background))
1786 png_set_background(png_ptr, image_background,
1787 PNG_BACKGROUND_GAMMA_FILE, 1/*needs to be expanded*/, 1);
1789 png_set_background(png_ptr, &my_background,
1790 PNG_BACKGROUND_GAMMA_SCREEN, 0/*do not expand*/, 1);
1792 The second call was described above - my_background is in the format of the
1793 final, display, output produced by libpng. Because you now know the format of
1794 the PNG it is possible to avoid the need to choose either 8-bit or 16-bit
1795 output and to retain palette images (the palette colors will be modified
1796 appropriately and the tRNS chunk removed.) However, if you are doing this,
1797 take great care not to ask for transformations without checking first that
1800 In the first call the background color has the original bit depth and color type
1801 of the PNG file. So, for palette images the color is supplied as a palette
1802 index and for low bit greyscale images the color is a reduced bit value in
1803 image_background->gray.
1805 If you didn't call png_set_gamma() before reading the file header, for example
1806 if you need your code to remain compatible with older versions of libpng prior
1807 to libpng-1.5.4, this is the place to call it.
1809 Do not call it if you called png_set_alpha_mode(); doing so will damage the
1810 settings put in place by png_set_alpha_mode(). (If png_set_alpha_mode() is
1811 supported then you can certainly do png_set_gamma() before reading the PNG
1814 This API unconditionally sets the screen and file gamma values, so it will
1815 override the value in the PNG file unless it is called before the PNG file
1816 reading starts. For this reason you must always call it with the PNG file
1817 value when you call it in this position:
1819 if (png_get_gAMA(png_ptr, info_ptr, &file_gamma))
1820 png_set_gamma(png_ptr, screen_gamma, file_gamma);
1823 png_set_gamma(png_ptr, screen_gamma, 0.45455);
1825 If you need to reduce an RGB file to a paletted file, or if a paletted
1826 file has more entries then will fit on your screen, png_set_quantize()
1827 will do that. Note that this is a simple match quantization that merely
1828 finds the closest color available. This should work fairly well with
1829 optimized palettes, but fairly badly with linear color cubes. If you
1830 pass a palette that is larger than maximum_colors, the file will
1831 reduce the number of colors in the palette so it will fit into
1832 maximum_colors. If there is a histogram, libpng will use it to make
1833 more intelligent choices when reducing the palette. If there is no
1834 histogram, it may not do as good a job.
1836 if (color_type & PNG_COLOR_MASK_COLOR)
1838 if (png_get_valid(png_ptr, info_ptr,
1841 png_uint_16p histogram = NULL;
1843 png_get_hIST(png_ptr, info_ptr,
1845 png_set_quantize(png_ptr, palette, num_palette,
1846 max_screen_colors, histogram, 1);
1851 png_color std_color_cube[MAX_SCREEN_COLORS] =
1854 png_set_quantize(png_ptr, std_color_cube,
1855 MAX_SCREEN_COLORS, MAX_SCREEN_COLORS,
1860 PNG files describe monochrome as black being zero and white being one.
1861 The following code will reverse this (make black be one and white be
1864 if (bit_depth == 1 && color_type == PNG_COLOR_TYPE_GRAY)
1865 png_set_invert_mono(png_ptr);
1867 This function can also be used to invert grayscale and gray-alpha images:
1869 if (color_type == PNG_COLOR_TYPE_GRAY ||
1870 color_type == PNG_COLOR_TYPE_GRAY_ALPHA)
1871 png_set_invert_mono(png_ptr);
1873 PNG files store 16-bit pixels in network byte order (big-endian,
1874 ie. most significant bits first). This code changes the storage to the
1875 other way (little-endian, i.e. least significant bits first, the
1876 way PCs store them):
1878 if (bit_depth == 16)
1879 png_set_swap(png_ptr);
1881 If you are using packed-pixel images (1, 2, or 4 bits/pixel), and you
1882 need to change the order the pixels are packed into bytes, you can use:
1885 png_set_packswap(png_ptr);
1887 Finally, you can write your own transformation function if none of
1888 the existing ones meets your needs. This is done by setting a callback
1891 png_set_read_user_transform_fn(png_ptr,
1894 You must supply the function
1896 void read_transform_fn(png_structp png_ptr, png_row_infop
1897 row_info, png_bytep data)
1899 See pngtest.c for a working example. Your function will be called
1900 after all of the other transformations have been processed. Take care with
1901 interlaced images if you do the interlace yourself - the width of the row is the
1902 width in 'row_info', not the overall image width.
1904 If supported, libpng provides two information routines that you can use to find
1905 where you are in processing the image:
1907 png_get_current_pass_number(png_structp png_ptr);
1908 png_get_current_row_number(png_structp png_ptr);
1910 Don't try using these outside a transform callback - firstly they are only
1911 supported if user transforms are supported, secondly they may well return
1912 unexpected results unless the row is actually being processed at the moment they
1916 images the value returned is the row in the input sub-image image. Use
1917 PNG_ROW_FROM_PASS_ROW(row, pass) and PNG_COL_FROM_PASS_COL(col, pass) to
1918 find the output pixel (x,y) given an interlaced sub-image pixel (row,col,pass).
1920 The discussion of interlace handling above contains more information on how to
1923 You can also set up a pointer to a user structure for use by your
1924 callback function, and you can inform libpng that your transform
1925 function will change the number of channels or bit depth with the
1928 png_set_user_transform_info(png_ptr, user_ptr,
1929 user_depth, user_channels);
1931 The user's application, not libpng, is responsible for allocating and
1932 freeing any memory required for the user structure.
1934 You can retrieve the pointer via the function
1935 png_get_user_transform_ptr(). For example:
1937 voidp read_user_transform_ptr =
1938 png_get_user_transform_ptr(png_ptr);
1940 The last thing to handle is interlacing; this is covered in detail below,
1941 but you must call the function here if you want libpng to handle expansion
1942 of the interlaced image.
1944 number_of_passes = png_set_interlace_handling(png_ptr);
1946 After setting the transformations, libpng can update your png_info
1947 structure to reflect any transformations you've requested with this
1950 png_read_update_info(png_ptr, info_ptr);
1952 This is most useful to update the info structure's rowbytes
1953 field so you can use it to allocate your image memory. This function
1954 will also update your palette with the correct screen_gamma and
1955 background if these have been given with the calls above. You may
1956 only call png_read_update_info() once with a particular info_ptr.
1958 After you call png_read_update_info(), you can allocate any
1959 memory you need to hold the image. The row data is simply
1960 raw byte data for all forms of images. As the actual allocation
1961 varies among applications, no example will be given. If you
1962 are allocating one large chunk, you will need to build an
1963 array of pointers to each row, as it will be needed for some
1964 of the functions below.
1966 Remember: Before you call png_read_update_info(), the png_get_*()
1967 functions return the values corresponding to the original PNG image.
1968 After you call png_read_update_info the values refer to the image
1969 that libpng will output. Consequently you must call all the png_set_
1970 functions before you call png_read_update_info(). This is particularly
1971 important for png_set_interlace_handling() - if you are going to call
1972 png_read_update_info() you must call png_set_interlace_handling() before
1973 it unless you want to receive interlaced output.
1977 After you've allocated memory, you can read the image data.
1978 The simplest way to do this is in one function call. If you are
1979 allocating enough memory to hold the whole image, you can just
1980 call png_read_image() and libpng will read in all the image data
1981 and put it in the memory area supplied. You will need to pass in
1982 an array of pointers to each row.
1984 This function automatically handles interlacing, so you don't
1985 need to call png_set_interlace_handling() (unless you call
1986 png_read_update_info()) or call this function multiple times, or any
1987 of that other stuff necessary with png_read_rows().
1989 png_read_image(png_ptr, row_pointers);
1991 where row_pointers is:
1993 png_bytep row_pointers[height];
1995 You can point to void or char or whatever you use for pixels.
1997 If you don't want to read in the whole image at once, you can
1998 use png_read_rows() instead. If there is no interlacing (check
1999 interlace_type == PNG_INTERLACE_NONE), this is simple:
2001 png_read_rows(png_ptr, row_pointers, NULL,
2004 where row_pointers is the same as in the png_read_image() call.
2006 If you are doing this just one row at a time, you can do this with
2007 a single row_pointer instead of an array of row_pointers:
2009 png_bytep row_pointer = row;
2010 png_read_row(png_ptr, row_pointer, NULL);
2012 If the file is interlaced (interlace_type != 0 in the IHDR chunk), things
2013 get somewhat harder. The only current (PNG Specification version 1.2)
2014 interlacing type for PNG is (interlace_type == PNG_INTERLACE_ADAM7);
2015 a somewhat complicated 2D interlace scheme, known as Adam7, that
2016 breaks down an image into seven smaller images of varying size, based
2017 on an 8x8 grid. This number is defined (from libpng 1.5) as
2018 PNG_INTERLACE_ADAM7_PASSES in png.h
2020 libpng can fill out those images or it can give them to you "as is".
2021 It is almost always better to have libpng handle the interlacing for you.
2022 If you want the images filled out, there are two ways to do that. The one
2023 mentioned in the PNG specification is to expand each pixel to cover
2024 those pixels that have not been read yet (the "rectangle" method).
2025 This results in a blocky image for the first pass, which gradually
2026 smooths out as more pixels are read. The other method is the "sparkle"
2027 method, where pixels are drawn only in their final locations, with the
2028 rest of the image remaining whatever colors they were initialized to
2029 before the start of the read. The first method usually looks better,
2030 but tends to be slower, as there are more pixels to put in the rows.
2032 If, as is likely, you want libpng to expand the images, call this before
2033 calling png_start_read_image() or png_read_update_info():
2035 if (interlace_type == PNG_INTERLACE_ADAM7)
2037 = png_set_interlace_handling(png_ptr);
2039 This will return the number of passes needed. Currently, this is seven,
2040 but may change if another interlace type is added. This function can be
2041 called even if the file is not interlaced, where it will return one pass.
2042 You then need to read the whole image 'number_of_passes' times. Each time
2043 will distribute the pixels from the current pass to the correct place in
2044 the output image, so you need to supply the same rows to png_read_rows in
2047 If you are not going to display the image after each pass, but are
2048 going to wait until the entire image is read in, use the sparkle
2049 effect. This effect is faster and the end result of either method
2050 is exactly the same. If you are planning on displaying the image
2051 after each pass, the "rectangle" effect is generally considered the
2054 If you only want the "sparkle" effect, just call png_read_rows() as
2055 normal, with the third parameter NULL. Make sure you make pass over
2056 the image number_of_passes times, and you don't change the data in the
2057 rows between calls. You can change the locations of the data, just
2058 not the data. Each pass only writes the pixels appropriate for that
2059 pass, and assumes the data from previous passes is still valid.
2061 png_read_rows(png_ptr, row_pointers, NULL,
2064 If you only want the first effect (the rectangles), do the same as
2065 before except pass the row buffer in the third parameter, and leave
2066 the second parameter NULL.
2068 png_read_rows(png_ptr, NULL, row_pointers,
2071 If you don't want libpng to handle the interlacing details, just call
2072 png_read_rows() PNG_INTERLACE_ADAM7_PASSES times to read in all the images.
2073 Each of the images is a valid image by itself; however, you will almost
2074 certainly need to distribute the pixels from each sub-image to the
2075 correct place. This is where everything gets very tricky.
2077 If you want to retrieve the separate images you must pass the correct
2078 number of rows to each successive call of png_read_rows(). The calculation
2079 gets pretty complicated for small images, where some sub-images may
2080 not even exist because either their width or height ends up zero.
2081 libpng provides two macros to help you in 1.5 and later versions:
2083 png_uint_32 width = PNG_PASS_COLS(image_width, pass_number);
2084 png_uint_32 height = PNG_PASS_ROWS(image_height, pass_number);
2086 Respectively these tell you the width and height of the sub-image
2087 corresponding to the numbered pass. 'pass' is in in the range 0 to 6 -
2088 this can be confusing because the specification refers to the same passes
2089 as 1 to 7! Be careful, you must check both the width and height before
2090 calling png_read_rows() and not call it for that pass if either is zero.
2092 You can, of course, read each sub-image row by row. If you want to
2093 produce optimal code to make a pixel-by-pixel transformation of an
2094 interlaced image this is the best approach; read each row of each pass,
2095 transform it, and write it out to a new interlaced image.
2097 If you want to de-interlace the image yourself libpng provides further
2098 macros to help that tell you where to place the pixels in the output image.
2099 Because the interlacing scheme is rectangular - sub-image pixels are always
2100 arranged on a rectangular grid - all you need to know for each pass is the
2101 starting column and row in the output image of the first pixel plus the
2102 spacing between each pixel. As of libpng 1.5 there are four macros to
2103 retrieve this information:
2105 png_uint_32 x = PNG_PASS_START_COL(pass);
2106 png_uint_32 y = PNG_PASS_START_ROW(pass);
2107 png_uint_32 xStep = 1U << PNG_PASS_COL_SHIFT(pass);
2108 png_uint_32 yStep = 1U << PNG_PASS_ROW_SHIFT(pass);
2110 These allow you to write the obvious loop:
2112 png_uint_32 input_y = 0;
2113 png_uint_32 output_y = PNG_PASS_START_ROW(pass);
2115 while (output_y < output_image_height)
2117 png_uint_32 input_x = 0;
2118 png_uint_32 output_x = PNG_PASS_START_COL(pass);
2120 while (output_x < output_image_width)
2122 image[output_y][output_x] =
2123 subimage[pass][input_y][input_x++];
2132 Notice that the steps between successive output rows and columns are
2133 returned as shifts. This is possible because the pixels in the subimages
2134 are always a power of 2 apart - 1, 2, 4 or 8 pixels - in the original
2135 image. In practice you may need to directly calculate the output coordinate
2136 given an input coordinate. libpng provides two further macros for this
2139 png_uint_32 output_x = PNG_COL_FROM_PASS_COL(input_x, pass);
2140 png_uint_32 output_y = PNG_ROW_FROM_PASS_ROW(input_y, pass);
2142 Finally a pair of macros are provided to tell you if a particular image
2143 row or column appears in a given pass:
2145 int col_in_pass = PNG_COL_IN_INTERLACE_PASS(output_x, pass);
2146 int row_in_pass = PNG_ROW_IN_INTERLACE_PASS(output_y, pass);
2148 Bear in mind that you will probably also need to check the width and height
2149 of the pass in addition to the above to be sure the pass even exists!
2151 With any luck you are convinced by now that you don't want to do your own
2152 interlace handling. In reality normally the only good reason for doing this
2153 is if you are processing PNG files on a pixel-by-pixel basis and don't want
2154 to load the whole file into memory when it is interlaced.
2156 libpng includes a test program, pngvalid, that illustrates reading and
2157 writing of interlaced images. If you can't get interlacing to work in your
2158 code and don't want to leave it to libpng (the recommended approach), see
2159 how pngvalid.c does it.
2161 Finishing a sequential read
2163 After you are finished reading the image through the
2164 low-level interface, you can finish reading the file. If you are
2165 interested in comments or time, which may be stored either before or
2166 after the image data, you should pass the separate png_info struct if
2167 you want to keep the comments from before and after the image
2170 png_infop end_info = png_create_info_struct(png_ptr);
2174 png_destroy_read_struct(&png_ptr, &info_ptr,
2179 png_read_end(png_ptr, end_info);
2181 If you are not interested, you should still call png_read_end()
2182 but you can pass NULL, avoiding the need to create an end_info structure.
2184 png_read_end(png_ptr, (png_infop)NULL);
2186 If you don't call png_read_end(), then your file pointer will be
2187 left pointing to the first chunk after the last IDAT, which is probably
2188 not what you want if you expect to read something beyond the end of
2191 When you are done, you can free all memory allocated by libpng like this:
2193 png_destroy_read_struct(&png_ptr, &info_ptr,
2196 or, if you didn't create an end_info structure,
2198 png_destroy_read_struct(&png_ptr, &info_ptr,
2201 It is also possible to individually free the info_ptr members that
2202 point to libpng-allocated storage with the following function:
2204 png_free_data(png_ptr, info_ptr, mask, seq)
2206 mask - identifies data to be freed, a mask
2207 containing the bitwise OR of one or
2209 PNG_FREE_PLTE, PNG_FREE_TRNS,
2210 PNG_FREE_HIST, PNG_FREE_ICCP,
2211 PNG_FREE_PCAL, PNG_FREE_ROWS,
2212 PNG_FREE_SCAL, PNG_FREE_SPLT,
2213 PNG_FREE_TEXT, PNG_FREE_UNKN,
2214 or simply PNG_FREE_ALL
2216 seq - sequence number of item to be freed
2219 This function may be safely called when the relevant storage has
2220 already been freed, or has not yet been allocated, or was allocated
2221 by the user and not by libpng, and will in those cases do nothing.
2222 The "seq" parameter is ignored if only one item of the selected data
2223 type, such as PLTE, is allowed. If "seq" is not -1, and multiple items
2224 are allowed for the data type identified in the mask, such as text or
2225 sPLT, only the n'th item in the structure is freed, where n is "seq".
2227 The default behavior is only to free data that was allocated internally
2228 by libpng. This can be changed, so that libpng will not free the data,
2229 or so that it will free data that was allocated by the user with png_malloc()
2230 or png_calloc() and passed in via a png_set_*() function, with
2232 png_data_freer(png_ptr, info_ptr, freer, mask)
2235 PNG_DESTROY_WILL_FREE_DATA
2236 PNG_SET_WILL_FREE_DATA
2237 PNG_USER_WILL_FREE_DATA
2239 mask - which data elements are affected
2240 same choices as in png_free_data()
2242 This function only affects data that has already been allocated.
2243 You can call this function after reading the PNG data but before calling
2244 any png_set_*() functions, to control whether the user or the png_set_*()
2245 function is responsible for freeing any existing data that might be present,
2246 and again after the png_set_*() functions to control whether the user
2247 or png_destroy_*() is supposed to free the data. When the user assumes
2248 responsibility for libpng-allocated data, the application must use
2249 png_free() to free it, and when the user transfers responsibility to libpng
2250 for data that the user has allocated, the user must have used png_malloc()
2251 or png_calloc() to allocate it.
2253 If you allocated your row_pointers in a single block, as suggested above in
2254 the description of the high level read interface, you must not transfer
2255 responsibility for freeing it to the png_set_rows or png_read_destroy function,
2256 because they would also try to free the individual row_pointers[i].
2258 If you allocated text_ptr.text, text_ptr.lang, and text_ptr.translated_keyword
2259 separately, do not transfer responsibility for freeing text_ptr to libpng,
2260 because when libpng fills a png_text structure it combines these members with
2261 the key member, and png_free_data() will free only text_ptr.key. Similarly,
2262 if you transfer responsibility for free'ing text_ptr from libpng to your
2263 application, your application must not separately free those members.
2265 The png_free_data() function will turn off the "valid" flag for anything
2266 it frees. If you need to turn the flag off for a chunk that was freed by
2267 your application instead of by libpng, you can use
2269 png_set_invalid(png_ptr, info_ptr, mask);
2271 mask - identifies the chunks to be made invalid,
2272 containing the bitwise OR of one or
2274 PNG_INFO_gAMA, PNG_INFO_sBIT,
2275 PNG_INFO_cHRM, PNG_INFO_PLTE,
2276 PNG_INFO_tRNS, PNG_INFO_bKGD,
2277 PNG_INFO_hIST, PNG_INFO_pHYs,
2278 PNG_INFO_oFFs, PNG_INFO_tIME,
2279 PNG_INFO_pCAL, PNG_INFO_sRGB,
2280 PNG_INFO_iCCP, PNG_INFO_sPLT,
2281 PNG_INFO_sCAL, PNG_INFO_IDAT
2283 For a more compact example of reading a PNG image, see the file example.c.
2285 Reading PNG files progressively
2287 The progressive reader is slightly different then the non-progressive
2288 reader. Instead of calling png_read_info(), png_read_rows(), and
2289 png_read_end(), you make one call to png_process_data(), which calls
2290 callbacks when it has the info, a row, or the end of the image. You
2291 set up these callbacks with png_set_progressive_read_fn(). You don't
2292 have to worry about the input/output functions of libpng, as you are
2293 giving the library the data directly in png_process_data(). I will
2294 assume that you have read the section on reading PNG files above,
2295 so I will only highlight the differences (although I will show
2298 png_structp png_ptr;
2301 /* An example code fragment of how you would
2302 initialize the progressive reader in your
2305 initialize_png_reader()
2307 png_ptr = png_create_read_struct
2308 (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
2309 user_error_fn, user_warning_fn);
2314 info_ptr = png_create_info_struct(png_ptr);
2318 png_destroy_read_struct(&png_ptr,
2319 (png_infopp)NULL, (png_infopp)NULL);
2323 if (setjmp(png_jmpbuf(png_ptr)))
2325 png_destroy_read_struct(&png_ptr, &info_ptr,
2330 /* This one's new. You can provide functions
2331 to be called when the header info is valid,
2332 when each row is completed, and when the image
2333 is finished. If you aren't using all functions,
2334 you can specify NULL parameters. Even when all
2335 three functions are NULL, you need to call
2336 png_set_progressive_read_fn(). You can use
2337 any struct as the user_ptr (cast to a void pointer
2338 for the function call), and retrieve the pointer
2339 from inside the callbacks using the function
2341 png_get_progressive_ptr(png_ptr);
2343 which will return a void pointer, which you have
2344 to cast appropriately.
2346 png_set_progressive_read_fn(png_ptr, (void *)user_ptr,
2347 info_callback, row_callback, end_callback);
2352 /* A code fragment that you call as you receive blocks
2355 process_data(png_bytep buffer, png_uint_32 length)
2357 if (setjmp(png_jmpbuf(png_ptr)))
2359 png_destroy_read_struct(&png_ptr, &info_ptr,
2364 /* This one's new also. Simply give it a chunk
2365 of data from the file stream (in order, of
2366 course). On machines with segmented memory
2367 models machines, don't give it any more than
2368 64K. The library seems to run fine with sizes
2369 of 4K. Although you can give it much less if
2370 necessary (I assume you can give it chunks of
2371 1 byte, I haven't tried less then 256 bytes
2372 yet). When this function returns, you may
2373 want to display any rows that were generated
2374 in the row callback if you don't already do
2377 png_process_data(png_ptr, info_ptr, buffer, length);
2379 /* At this point you can call png_process_data_skip if
2380 you want to handle data the library will skip yourself;
2381 it simply returns the number of bytes to skip (and stops
2382 libpng skipping that number of bytes on the next
2383 png_process_data call).
2387 /* This function is called (as set by
2388 png_set_progressive_read_fn() above) when enough data
2389 has been supplied so all of the header has been
2393 info_callback(png_structp png_ptr, png_infop info)
2395 /* Do any setup here, including setting any of
2396 the transformations mentioned in the Reading
2397 PNG files section. For now, you _must_ call
2398 either png_start_read_image() or
2399 png_read_update_info() after all the
2400 transformations are set (even if you don't set
2401 any). You may start getting rows before
2402 png_process_data() returns, so this is your
2403 last chance to prepare for that.
2405 This is where you turn on interlace handling,
2406 assuming you don't want to do it yourself.
2408 If you need to you can stop the processing of
2409 your original input data at this point by calling
2410 png_process_data_pause. This returns the number
2411 of unprocessed bytes from the last png_process_data
2412 call - it is up to you to ensure that the next call
2413 sees these bytes again. If you don't want to bother
2414 with this you can get libpng to cache the unread
2415 bytes by setting the 'save' parameter (see png.h) but
2416 then libpng will have to copy the data internally.
2420 /* This function is called when each row of image
2423 row_callback(png_structp png_ptr, png_bytep new_row,
2424 png_uint_32 row_num, int pass)
2426 /* If the image is interlaced, and you turned
2427 on the interlace handler, this function will
2428 be called for every row in every pass. Some
2429 of these rows will not be changed from the
2430 previous pass. When the row is not changed,
2431 the new_row variable will be NULL. The rows
2432 and passes are called in order, so you don't
2433 really need the row_num and pass, but I'm
2434 supplying them because it may make your life
2437 If you did not turn on interlace handling then
2438 the callback is called for each row of each
2439 sub-image when the image is interlaced. In this
2440 case 'row_num' is the row in the sub-image, not
2441 the row in the output image as it is in all other
2444 For the non-NULL rows of interlaced images when
2445 you have switched on libpng interlace handling,
2446 you must call png_progressive_combine_row()
2447 passing in the row and the old row. You can
2448 call this function for NULL rows (it will just
2449 return) and for non-interlaced images (it just
2450 does the memcpy for you) if it will make the
2451 code easier. Thus, you can just do this for
2452 all cases if you switch on interlace handling;
2455 png_progressive_combine_row(png_ptr, old_row,
2458 /* where old_row is what was displayed for
2459 previously for the row. Note that the first
2460 pass (pass == 0, really) will completely cover
2461 the old row, so the rows do not have to be
2462 initialized. After the first pass (and only
2463 for interlaced images), you will have to pass
2464 the current row, and the function will combine
2465 the old row and the new row.
2467 You can also call png_process_data_pause in this
2468 callback - see above.
2473 end_callback(png_structp png_ptr, png_infop info)
2475 /* This function is called after the whole image
2476 has been read, including any chunks after the
2477 image (up to and including the IEND). You
2478 will usually have the same info chunk as you
2479 had in the header, although some data may have
2480 been added to the comments and time fields.
2482 Most people won't do much here, perhaps setting
2483 a flag that marks the image as finished.
2491 Much of this is very similar to reading. However, everything of
2492 importance is repeated here, so you won't have to constantly look
2493 back up in the reading section to understand writing.
2497 You will want to do the I/O initialization before you get into libpng,
2498 so if it doesn't work, you don't have anything to undo. If you are not
2499 using the standard I/O functions, you will need to replace them with
2500 custom writing functions. See the discussion under Customizing libpng.
2502 FILE *fp = fopen(file_name, "wb");
2507 Next, png_struct and png_info need to be allocated and initialized.
2508 As these can be both relatively large, you may not want to store these
2509 on the stack, unless you have stack space to spare. Of course, you
2510 will want to check if they return NULL. If you are also reading,
2511 you won't want to name your read structure and your write structure
2512 both "png_ptr"; you can call them anything you like, such as
2513 "read_ptr" and "write_ptr". Look at pngtest.c, for example.
2515 png_structp png_ptr = png_create_write_struct
2516 (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
2517 user_error_fn, user_warning_fn);
2522 png_infop info_ptr = png_create_info_struct(png_ptr);
2525 png_destroy_write_struct(&png_ptr,
2530 If you want to use your own memory allocation routines,
2531 define PNG_USER_MEM_SUPPORTED and use
2532 png_create_write_struct_2() instead of png_create_write_struct():
2534 png_structp png_ptr = png_create_write_struct_2
2535 (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
2536 user_error_fn, user_warning_fn, (png_voidp)
2537 user_mem_ptr, user_malloc_fn, user_free_fn);
2539 After you have these structures, you will need to set up the
2540 error handling. When libpng encounters an error, it expects to
2541 longjmp() back to your routine. Therefore, you will need to call
2542 setjmp() and pass the png_jmpbuf(png_ptr). If you
2543 write the file from different routines, you will need to update
2544 the png_jmpbuf(png_ptr) every time you enter a new routine that will
2545 call a png_*() function. See your documentation of setjmp/longjmp
2546 for your compiler for more information on setjmp/longjmp. See
2547 the discussion on libpng error handling in the Customizing Libpng
2548 section below for more information on the libpng error handling.
2550 if (setjmp(png_jmpbuf(png_ptr)))
2552 png_destroy_write_struct(&png_ptr, &info_ptr);
2559 If you would rather avoid the complexity of setjmp/longjmp issues,
2560 you can compile libpng with PNG_NO_SETJMP, in which case
2561 errors will result in a call to PNG_ABORT() which defaults to abort().
2563 You can #define PNG_ABORT() to a function that does something
2564 more useful than abort(), as long as your function does not
2567 Checking for invalid palette index on write was added at libpng
2568 1.5.10. If a pixel contains an invalid (out-of-range) index libpng issues
2569 a benign error. This is enabled by default because this condition is an
2570 error according to the PNG specification, Clause 11.3.2, but the error can
2571 be ignored in each png_ptr with
2573 png_set_check_for_invalid_index(png_ptr, 0);
2575 If the error is ignored, or if png_benign_error() treats it as a warning,
2576 any invalid pixels are written as-is by the encoder, resulting in an
2577 invalid PNG datastream as output. In this case the application is
2578 responsible for ensuring that the pixel indexes are in range when it writes
2579 a PLTE chunk with fewer entries than the bit depth would allow.
2581 Now you need to set up the output code. The default for libpng is to
2582 use the C function fwrite(). If you use this, you will need to pass a
2583 valid FILE * in the function png_init_io(). Be sure that the file is
2584 opened in binary mode. Again, if you wish to handle writing data in
2585 another way, see the discussion on libpng I/O handling in the Customizing
2586 Libpng section below.
2588 png_init_io(png_ptr, fp);
2590 If you are embedding your PNG into a datastream such as MNG, and don't
2591 want libpng to write the 8-byte signature, or if you have already
2592 written the signature in your application, use
2594 png_set_sig_bytes(png_ptr, 8);
2596 to inform libpng that it should not write a signature.
2600 At this point, you can set up a callback function that will be
2601 called after each row has been written, which you can use to control
2602 a progress meter or the like. It's demonstrated in pngtest.c.
2603 You must supply a function
2605 void write_row_callback(png_structp png_ptr, png_uint_32 row,
2608 /* put your code here */
2611 (You can give it another name that you like instead of "write_row_callback")
2613 To inform libpng about your function, use
2615 png_set_write_status_fn(png_ptr, write_row_callback);
2617 When this function is called the row has already been completely processed and
2618 it has also been written out. The 'row' and 'pass' refer to the next row to be
2620 non-interlaced case the row that was just handled is simply one less than the
2621 passed in row number, and pass will always be 0. For the interlaced case the
2622 same applies unless the row value is 0, in which case the row just handled was
2623 the last one from one of the preceding passes. Because interlacing may skip a
2624 pass you cannot be sure that the preceding pass is just 'pass-1', if you really
2625 need to know what the last pass is record (row,pass) from the callback and use
2626 the last recorded value each time.
2628 As with the user transform you can find the output row using the
2629 PNG_ROW_FROM_PASS_ROW macro.
2631 You now have the option of modifying how the compression library will
2632 run. The following functions are mainly for testing, but may be useful
2633 in some cases, like if you need to write PNG files extremely fast and
2634 are willing to give up some compression, or if you want to get the
2635 maximum possible compression at the expense of slower writing. If you
2636 have no special needs in this area, let the library do what it wants by
2637 not calling this function at all, as it has been tuned to deliver a good
2638 speed/compression ratio. The second parameter to png_set_filter() is
2639 the filter method, for which the only valid values are 0 (as of the
2640 July 1999 PNG specification, version 1.2) or 64 (if you are writing
2641 a PNG datastream that is to be embedded in a MNG datastream). The third
2642 parameter is a flag that indicates which filter type(s) are to be tested
2643 for each scanline. See the PNG specification for details on the specific
2647 /* turn on or off filtering, and/or choose
2648 specific filters. You can use either a single
2649 PNG_FILTER_VALUE_NAME or the bitwise OR of one
2650 or more PNG_FILTER_NAME masks.
2652 png_set_filter(png_ptr, 0,
2653 PNG_FILTER_NONE | PNG_FILTER_VALUE_NONE |
2654 PNG_FILTER_SUB | PNG_FILTER_VALUE_SUB |
2655 PNG_FILTER_UP | PNG_FILTER_VALUE_UP |
2656 PNG_FILTER_AVG | PNG_FILTER_VALUE_AVG |
2657 PNG_FILTER_PAETH | PNG_FILTER_VALUE_PAETH|
2660 If an application wants to start and stop using particular filters during
2661 compression, it should start out with all of the filters (to ensure that
2662 the previous row of pixels will be stored in case it's needed later),
2663 and then add and remove them after the start of compression.
2665 If you are writing a PNG datastream that is to be embedded in a MNG
2666 datastream, the second parameter can be either 0 or 64.
2668 The png_set_compression_*() functions interface to the zlib compression
2669 library, and should mostly be ignored unless you really know what you are
2670 doing. The only generally useful call is png_set_compression_level()
2671 which changes how much time zlib spends on trying to compress the image
2672 data. See the Compression Library (zlib.h and algorithm.txt, distributed
2673 with zlib) for details on the compression levels.
2677 /* Set the zlib compression level */
2678 png_set_compression_level(png_ptr,
2679 Z_BEST_COMPRESSION);
2681 /* Set other zlib parameters for compressing IDAT */
2682 png_set_compression_mem_level(png_ptr, 8);
2683 png_set_compression_strategy(png_ptr,
2684 Z_DEFAULT_STRATEGY);
2685 png_set_compression_window_bits(png_ptr, 15);
2686 png_set_compression_method(png_ptr, 8);
2687 png_set_compression_buffer_size(png_ptr, 8192)
2689 /* Set zlib parameters for text compression
2690 * If you don't call these, the parameters
2691 * fall back on those defined for IDAT chunks
2693 png_set_text_compression_mem_level(png_ptr, 8);
2694 png_set_text_compression_strategy(png_ptr,
2695 Z_DEFAULT_STRATEGY);
2696 png_set_text_compression_window_bits(png_ptr, 15);
2697 png_set_text_compression_method(png_ptr, 8);
2699 Setting the contents of info for output
2701 You now need to fill in the png_info structure with all the data you
2702 wish to write before the actual image. Note that the only thing you
2703 are allowed to write after the image is the text chunks and the time
2704 chunk (as of PNG Specification 1.2, anyway). See png_write_end() and
2705 the latest PNG specification for more information on that. If you
2706 wish to write them before the image, fill them in now, and flag that
2707 data as being valid. If you want to wait until after the data, don't
2708 fill them until png_write_end(). For all the fields in png_info and
2709 their data types, see png.h. For explanations of what the fields
2710 contain, see the PNG specification.
2712 Some of the more important parts of the png_info are:
2714 png_set_IHDR(png_ptr, info_ptr, width, height,
2715 bit_depth, color_type, interlace_type,
2716 compression_type, filter_method)
2718 width - holds the width of the image
2719 in pixels (up to 2^31).
2721 height - holds the height of the image
2722 in pixels (up to 2^31).
2724 bit_depth - holds the bit depth of one of the
2726 (valid values are 1, 2, 4, 8, 16
2727 and depend also on the
2728 color_type. See also significant
2731 color_type - describes which color/alpha
2732 channels are present.
2734 (bit depths 1, 2, 4, 8, 16)
2735 PNG_COLOR_TYPE_GRAY_ALPHA
2737 PNG_COLOR_TYPE_PALETTE
2738 (bit depths 1, 2, 4, 8)
2741 PNG_COLOR_TYPE_RGB_ALPHA
2744 PNG_COLOR_MASK_PALETTE
2745 PNG_COLOR_MASK_COLOR
2746 PNG_COLOR_MASK_ALPHA
2748 interlace_type - PNG_INTERLACE_NONE or
2751 compression_type - (must be
2752 PNG_COMPRESSION_TYPE_DEFAULT)
2754 filter_method - (must be PNG_FILTER_TYPE_DEFAULT
2755 or, if you are writing a PNG to
2756 be embedded in a MNG datastream,
2758 PNG_INTRAPIXEL_DIFFERENCING)
2760 If you call png_set_IHDR(), the call must appear before any of the
2761 other png_set_*() functions, because they might require access to some of
2762 the IHDR settings. The remaining png_set_*() functions can be called
2765 If you wish, you can reset the compression_type, interlace_type, or
2766 filter_method later by calling png_set_IHDR() again; if you do this, the
2767 width, height, bit_depth, and color_type must be the same in each call.
2769 png_set_PLTE(png_ptr, info_ptr, palette,
2772 palette - the palette for the file
2773 (array of png_color)
2774 num_palette - number of entries in the palette
2776 png_set_gAMA(png_ptr, info_ptr, file_gamma);
2777 png_set_gAMA_fixed(png_ptr, info_ptr, int_file_gamma);
2779 file_gamma - the gamma at which the image was
2780 created (PNG_INFO_gAMA)
2782 int_file_gamma - 100,000 times the gamma at which
2783 the image was created
2785 png_set_cHRM(png_ptr, info_ptr, white_x, white_y, red_x, red_y,
2786 green_x, green_y, blue_x, blue_y)
2787 png_set_cHRM_XYZ(png_ptr, info_ptr, red_X, red_Y, red_Z, green_X,
2788 green_Y, green_Z, blue_X, blue_Y, blue_Z)
2789 png_set_cHRM_fixed(png_ptr, info_ptr, int_white_x, int_white_y,
2790 int_red_x, int_red_y, int_green_x, int_green_y,
2791 int_blue_x, int_blue_y)
2792 png_set_cHRM_XYZ_fixed(png_ptr, info_ptr, int_red_X, int_red_Y,
2793 int_red_Z, int_green_X, int_green_Y, int_green_Z,
2794 int_blue_X, int_blue_Y, int_blue_Z)
2796 {white,red,green,blue}_{x,y}
2797 A color space encoding specified using the chromaticities
2798 of the end points and the white point.
2800 {red,green,blue}_{X,Y,Z}
2801 A color space encoding specified using the encoding end
2802 points - the CIE tristimulus specification of the intended
2803 color of the red, green and blue channels in the PNG RGB
2804 data. The white point is simply the sum of the three end
2807 png_set_sRGB(png_ptr, info_ptr, srgb_intent);
2809 srgb_intent - the rendering intent
2810 (PNG_INFO_sRGB) The presence of
2811 the sRGB chunk means that the pixel
2812 data is in the sRGB color space.
2813 This chunk also implies specific
2814 values of gAMA and cHRM. Rendering
2815 intent is the CSS-1 property that
2816 has been defined by the International
2818 (http://www.color.org).
2820 PNG_sRGB_INTENT_SATURATION,
2821 PNG_sRGB_INTENT_PERCEPTUAL,
2822 PNG_sRGB_INTENT_ABSOLUTE, or
2823 PNG_sRGB_INTENT_RELATIVE.
2826 png_set_sRGB_gAMA_and_cHRM(png_ptr, info_ptr,
2829 srgb_intent - the rendering intent
2830 (PNG_INFO_sRGB) The presence of the
2831 sRGB chunk means that the pixel
2832 data is in the sRGB color space.
2833 This function also causes gAMA and
2834 cHRM chunks with the specific values
2835 that are consistent with sRGB to be
2838 png_set_iCCP(png_ptr, info_ptr, name, compression_type,
2841 name - The profile name.
2843 compression_type - The compression type; always
2844 PNG_COMPRESSION_TYPE_BASE for PNG 1.0.
2845 You may give NULL to this argument to
2848 profile - International Color Consortium color
2849 profile data. May contain NULs.
2851 proflen - length of profile data in bytes.
2853 png_set_sBIT(png_ptr, info_ptr, sig_bit);
2855 sig_bit - the number of significant bits for
2856 (PNG_INFO_sBIT) each of the gray, red,
2857 green, and blue channels, whichever are
2858 appropriate for the given color type
2861 png_set_tRNS(png_ptr, info_ptr, trans_alpha,
2862 num_trans, trans_color);
2864 trans_alpha - array of alpha (transparency)
2865 entries for palette (PNG_INFO_tRNS)
2867 num_trans - number of transparent entries
2870 trans_color - graylevel or color sample values
2871 (in order red, green, blue) of the
2872 single transparent color for
2873 non-paletted images (PNG_INFO_tRNS)
2875 png_set_hIST(png_ptr, info_ptr, hist);
2877 hist - histogram of palette (array of
2878 png_uint_16) (PNG_INFO_hIST)
2880 png_set_tIME(png_ptr, info_ptr, mod_time);
2882 mod_time - time image was last modified
2885 png_set_bKGD(png_ptr, info_ptr, background);
2887 background - background color (of type
2888 png_color_16p) (PNG_VALID_bKGD)
2890 png_set_text(png_ptr, info_ptr, text_ptr, num_text);
2892 text_ptr - array of png_text holding image
2895 text_ptr[i].compression - type of compression used
2896 on "text" PNG_TEXT_COMPRESSION_NONE
2897 PNG_TEXT_COMPRESSION_zTXt
2898 PNG_ITXT_COMPRESSION_NONE
2899 PNG_ITXT_COMPRESSION_zTXt
2900 text_ptr[i].key - keyword for comment. Must contain
2902 text_ptr[i].text - text comments for current
2903 keyword. Can be NULL or empty.
2904 text_ptr[i].text_length - length of text string,
2905 after decompression, 0 for iTXt
2906 text_ptr[i].itxt_length - length of itxt string,
2907 after decompression, 0 for tEXt/zTXt
2908 text_ptr[i].lang - language of comment (NULL or
2910 text_ptr[i].translated_keyword - keyword in UTF-8 (NULL
2911 or empty for unknown).
2913 Note that the itxt_length, lang, and lang_key
2914 members of the text_ptr structure only exist when the
2915 library is built with iTXt chunk support. Prior to
2916 libpng-1.4.0 the library was built by default without
2917 iTXt support. Also note that when iTXt is supported,
2918 they contain NULL pointers when the "compression"
2919 field contains PNG_TEXT_COMPRESSION_NONE or
2920 PNG_TEXT_COMPRESSION_zTXt.
2922 num_text - number of comments
2924 png_set_sPLT(png_ptr, info_ptr, &palette_ptr,
2927 palette_ptr - array of png_sPLT_struct structures
2928 to be added to the list of palettes
2929 in the info structure.
2930 num_spalettes - number of palette structures to be
2933 png_set_oFFs(png_ptr, info_ptr, offset_x, offset_y,
2936 offset_x - positive offset from the left
2939 offset_y - positive offset from the top
2942 unit_type - PNG_OFFSET_PIXEL, PNG_OFFSET_MICROMETER
2944 png_set_pHYs(png_ptr, info_ptr, res_x, res_y,
2947 res_x - pixels/unit physical resolution
2950 res_y - pixels/unit physical resolution
2953 unit_type - PNG_RESOLUTION_UNKNOWN,
2954 PNG_RESOLUTION_METER
2956 png_set_sCAL(png_ptr, info_ptr, unit, width, height)
2958 unit - physical scale units (an integer)
2960 width - width of a pixel in physical scale units
2962 height - height of a pixel in physical scale units
2963 (width and height are doubles)
2965 png_set_sCAL_s(png_ptr, info_ptr, unit, width, height)
2967 unit - physical scale units (an integer)
2969 width - width of a pixel in physical scale units
2970 expressed as a string
2972 height - height of a pixel in physical scale units
2973 (width and height are strings like "2.54")
2975 png_set_unknown_chunks(png_ptr, info_ptr, &unknowns,
2978 unknowns - array of png_unknown_chunk
2979 structures holding unknown chunks
2980 unknowns[i].name - name of unknown chunk
2981 unknowns[i].data - data of unknown chunk
2982 unknowns[i].size - size of unknown chunk's data
2983 unknowns[i].location - position to write chunk in file
2984 0: do not write chunk
2985 PNG_HAVE_IHDR: before PLTE
2986 PNG_HAVE_PLTE: before IDAT
2987 PNG_AFTER_IDAT: after IDAT
2989 The "location" member is set automatically according to
2990 what part of the output file has already been written.
2991 You can change its value after calling png_set_unknown_chunks()
2992 as demonstrated in pngtest.c. Within each of the "locations",
2993 the chunks are sequenced according to their position in the
2994 structure (that is, the value of "i", which is the order in which
2995 the chunk was either read from the input file or defined with
2996 png_set_unknown_chunks).
2998 A quick word about text and num_text. text is an array of png_text
2999 structures. num_text is the number of valid structures in the array.
3000 Each png_text structure holds a language code, a keyword, a text value,
3001 and a compression type.
3003 The compression types have the same valid numbers as the compression
3004 types of the image data. Currently, the only valid number is zero.
3005 However, you can store text either compressed or uncompressed, unlike
3006 images, which always have to be compressed. So if you don't want the
3007 text compressed, set the compression type to PNG_TEXT_COMPRESSION_NONE.
3008 Because tEXt and zTXt chunks don't have a language field, if you
3009 specify PNG_TEXT_COMPRESSION_NONE or PNG_TEXT_COMPRESSION_zTXt
3010 any language code or translated keyword will not be written out.
3012 Until text gets around a few hundred bytes, it is not worth compressing it.
3013 After the text has been written out to the file, the compression type
3014 is set to PNG_TEXT_COMPRESSION_NONE_WR or PNG_TEXT_COMPRESSION_zTXt_WR,
3015 so that it isn't written out again at the end (in case you are calling
3016 png_write_end() with the same struct).
3018 The keywords that are given in the PNG Specification are:
3020 Title Short (one line) title or
3023 Author Name of image's creator
3025 Description Description of image (possibly long)
3027 Copyright Copyright notice
3029 Creation Time Time of original image creation
3030 (usually RFC 1123 format, see below)
3032 Software Software used to create the image
3034 Disclaimer Legal disclaimer
3036 Warning Warning of nature of content
3038 Source Device used to create the image
3040 Comment Miscellaneous comment; conversion
3041 from other image format
3043 The keyword-text pairs work like this. Keywords should be short
3044 simple descriptions of what the comment is about. Some typical
3045 keywords are found in the PNG specification, as is some recommendations
3046 on keywords. You can repeat keywords in a file. You can even write
3047 some text before the image and some after. For example, you may want
3048 to put a description of the image before the image, but leave the
3049 disclaimer until after, so viewers working over modem connections
3050 don't have to wait for the disclaimer to go over the modem before
3051 they start seeing the image. Finally, keywords should be full
3052 words, not abbreviations. Keywords and text are in the ISO 8859-1
3053 (Latin-1) character set (a superset of regular ASCII) and can not
3054 contain NUL characters, and should not contain control or other
3055 unprintable characters. To make the comments widely readable, stick
3056 with basic ASCII, and avoid machine specific character set extensions
3057 like the IBM-PC character set. The keyword must be present, but
3058 you can leave off the text string on non-compressed pairs.
3059 Compressed pairs must have a text string, as only the text string
3060 is compressed anyway, so the compression would be meaningless.
3062 PNG supports modification time via the png_time structure. Two
3063 conversion routines are provided, png_convert_from_time_t() for
3064 time_t and png_convert_from_struct_tm() for struct tm. The
3065 time_t routine uses gmtime(). You don't have to use either of
3066 these, but if you wish to fill in the png_time structure directly,
3067 you should provide the time in universal time (GMT) if possible
3068 instead of your local time. Note that the year number is the full
3069 year (e.g. 1998, rather than 98 - PNG is year 2000 compliant!), and
3070 that months start with 1.
3072 If you want to store the time of the original image creation, you should
3073 use a plain tEXt chunk with the "Creation Time" keyword. This is
3074 necessary because the "creation time" of a PNG image is somewhat vague,
3075 depending on whether you mean the PNG file, the time the image was
3076 created in a non-PNG format, a still photo from which the image was
3077 scanned, or possibly the subject matter itself. In order to facilitate
3078 machine-readable dates, it is recommended that the "Creation Time"
3079 tEXt chunk use RFC 1123 format dates (e.g. "22 May 1997 18:07:10 GMT"),
3080 although this isn't a requirement. Unlike the tIME chunk, the
3081 "Creation Time" tEXt chunk is not expected to be automatically changed
3082 by the software. To facilitate the use of RFC 1123 dates, a function
3083 png_convert_to_rfc1123_buffer(png_ptr, buffer, png_timep) is provided to
3084 convert from PNG time to an RFC 1123 format string. The caller must provide
3085 a writeable buffer of at least 29 bytes.
3087 Writing unknown chunks
3089 You can use the png_set_unknown_chunks function to queue up chunks
3090 for writing. You give it a chunk name, raw data, and a size; that's
3091 all there is to it. The chunks will be written by the next following
3092 png_write_info_before_PLTE, png_write_info, or png_write_end function.
3093 Any chunks previously read into the info structure's unknown-chunk
3094 list will also be written out in a sequence that satisfies the PNG
3095 specification's ordering rules.
3097 The high-level write interface
3099 At this point there are two ways to proceed; through the high-level
3100 write interface, or through a sequence of low-level write operations.
3101 You can use the high-level interface if your image data is present
3102 in the info structure. All defined output
3103 transformations are permitted, enabled by the following masks.
3105 PNG_TRANSFORM_IDENTITY No transformation
3106 PNG_TRANSFORM_PACKING Pack 1, 2 and 4-bit samples
3107 PNG_TRANSFORM_PACKSWAP Change order of packed
3109 PNG_TRANSFORM_INVERT_MONO Invert monochrome images
3110 PNG_TRANSFORM_SHIFT Normalize pixels to the
3112 PNG_TRANSFORM_BGR Flip RGB to BGR, RGBA
3114 PNG_TRANSFORM_SWAP_ALPHA Flip RGBA to ARGB or GA
3116 PNG_TRANSFORM_INVERT_ALPHA Change alpha from opacity
3118 PNG_TRANSFORM_SWAP_ENDIAN Byte-swap 16-bit samples
3119 PNG_TRANSFORM_STRIP_FILLER Strip out filler
3121 PNG_TRANSFORM_STRIP_FILLER_BEFORE Strip out leading
3123 PNG_TRANSFORM_STRIP_FILLER_AFTER Strip out trailing
3126 If you have valid image data in the info structure (you can use
3127 png_set_rows() to put image data in the info structure), simply do this:
3129 png_write_png(png_ptr, info_ptr, png_transforms, NULL)
3131 where png_transforms is an integer containing the bitwise OR of some set of
3132 transformation flags. This call is equivalent to png_write_info(),
3133 followed the set of transformations indicated by the transform mask,
3134 then png_write_image(), and finally png_write_end().
3136 (The final parameter of this call is not yet used. Someday it might point
3137 to transformation parameters required by some future output transform.)
3139 You must use png_transforms and not call any png_set_transform() functions
3140 when you use png_write_png().
3142 The low-level write interface
3144 If you are going the low-level route instead, you are now ready to
3145 write all the file information up to the actual image data. You do
3146 this with a call to png_write_info().
3148 png_write_info(png_ptr, info_ptr);
3150 Note that there is one transformation you may need to do before
3151 png_write_info(). In PNG files, the alpha channel in an image is the
3152 level of opacity. If your data is supplied as a level of transparency,
3153 you can invert the alpha channel before you write it, so that 0 is
3154 fully transparent and 255 (in 8-bit or paletted images) or 65535
3155 (in 16-bit images) is fully opaque, with
3157 png_set_invert_alpha(png_ptr);
3159 This must appear before png_write_info() instead of later with the
3160 other transformations because in the case of paletted images the tRNS
3161 chunk data has to be inverted before the tRNS chunk is written. If
3162 your image is not a paletted image, the tRNS data (which in such cases
3163 represents a single color to be rendered as transparent) won't need to
3164 be changed, and you can safely do this transformation after your
3165 png_write_info() call.
3167 If you need to write a private chunk that you want to appear before
3168 the PLTE chunk when PLTE is present, you can write the PNG info in
3169 two steps, and insert code to write your own chunk between them:
3171 png_write_info_before_PLTE(png_ptr, info_ptr);
3172 png_set_unknown_chunks(png_ptr, info_ptr, ...);
3173 png_write_info(png_ptr, info_ptr);
3175 After you've written the file information, you can set up the library
3176 to handle any special transformations of the image data. The various
3177 ways to transform the data will be described in the order that they
3178 should occur. This is important, as some of these change the color
3179 type and/or bit depth of the data, and some others only work on
3180 certain color types and bit depths. Even though each transformation
3181 checks to see if it has data that it can do something with, you should
3182 make sure to only enable a transformation if it will be valid for the
3183 data. For example, don't swap red and blue on grayscale data.
3185 PNG files store RGB pixels packed into 3 or 6 bytes. This code tells
3186 the library to strip input data that has 4 or 8 bytes per pixel down
3187 to 3 or 6 bytes (or strip 2 or 4-byte grayscale+filler data to 1 or 2
3190 png_set_filler(png_ptr, 0, PNG_FILLER_BEFORE);
3192 where the 0 is unused, and the location is either PNG_FILLER_BEFORE or
3193 PNG_FILLER_AFTER, depending upon whether the filler byte in the pixel
3194 is stored XRGB or RGBX.
3196 PNG files pack pixels of bit depths 1, 2, and 4 into bytes as small as
3197 they can, resulting in, for example, 8 pixels per byte for 1 bit files.
3198 If the data is supplied at 1 pixel per byte, use this code, which will
3199 correctly pack the pixels into a single byte:
3201 png_set_packing(png_ptr);
3203 PNG files reduce possible bit depths to 1, 2, 4, 8, and 16. If your
3204 data is of another bit depth, you can write an sBIT chunk into the
3205 file so that decoders can recover the original data if desired.
3207 /* Set the true bit depth of the image data */
3208 if (color_type & PNG_COLOR_MASK_COLOR)
3210 sig_bit.red = true_bit_depth;
3211 sig_bit.green = true_bit_depth;
3212 sig_bit.blue = true_bit_depth;
3217 sig_bit.gray = true_bit_depth;
3220 if (color_type & PNG_COLOR_MASK_ALPHA)
3222 sig_bit.alpha = true_bit_depth;
3225 png_set_sBIT(png_ptr, info_ptr, &sig_bit);
3227 If the data is stored in the row buffer in a bit depth other than
3228 one supported by PNG (e.g. 3 bit data in the range 0-7 for a 4-bit PNG),
3229 this will scale the values to appear to be the correct bit depth as
3232 png_set_shift(png_ptr, &sig_bit);
3234 PNG files store 16-bit pixels in network byte order (big-endian,
3235 ie. most significant bits first). This code would be used if they are
3236 supplied the other way (little-endian, i.e. least significant bits
3237 first, the way PCs store them):
3240 png_set_swap(png_ptr);
3242 If you are using packed-pixel images (1, 2, or 4 bits/pixel), and you
3243 need to change the order the pixels are packed into bytes, you can use:
3246 png_set_packswap(png_ptr);
3248 PNG files store 3 color pixels in red, green, blue order. This code
3249 would be used if they are supplied as blue, green, red:
3251 png_set_bgr(png_ptr);
3253 PNG files describe monochrome as black being zero and white being
3254 one. This code would be used if the pixels are supplied with this reversed
3255 (black being one and white being zero):
3257 png_set_invert_mono(png_ptr);
3259 Finally, you can write your own transformation function if none of
3260 the existing ones meets your needs. This is done by setting a callback
3263 png_set_write_user_transform_fn(png_ptr,
3264 write_transform_fn);
3266 You must supply the function
3268 void write_transform_fn(png_structp png_ptr, png_row_infop
3269 row_info, png_bytep data)
3271 See pngtest.c for a working example. Your function will be called
3272 before any of the other transformations are processed. If supported
3273 libpng also supplies an information routine that may be called from
3276 png_get_current_row_number(png_ptr);
3277 png_get_current_pass_number(png_ptr);
3279 This returns the current row passed to the transform. With interlaced
3280 images the value returned is the row in the input sub-image image. Use
3281 PNG_ROW_FROM_PASS_ROW(row, pass) and PNG_COL_FROM_PASS_COL(col, pass) to
3282 find the output pixel (x,y) given an interlaced sub-image pixel (row,col,pass).
3284 The discussion of interlace handling above contains more information on how to
3287 You can also set up a pointer to a user structure for use by your
3290 png_set_user_transform_info(png_ptr, user_ptr, 0, 0);
3292 The user_channels and user_depth parameters of this function are ignored
3293 when writing; you can set them to zero as shown.
3295 You can retrieve the pointer via the function png_get_user_transform_ptr().
3298 voidp write_user_transform_ptr =
3299 png_get_user_transform_ptr(png_ptr);
3301 It is possible to have libpng flush any pending output, either manually,
3302 or automatically after a certain number of lines have been written. To
3303 flush the output stream a single time call:
3305 png_write_flush(png_ptr);
3307 and to have libpng flush the output stream periodically after a certain
3308 number of scanlines have been written, call:
3310 png_set_flush(png_ptr, nrows);
3312 Note that the distance between rows is from the last time png_write_flush()
3313 was called, or the first row of the image if it has never been called.
3314 So if you write 50 lines, and then png_set_flush 25, it will flush the
3315 output on the next scanline, and every 25 lines thereafter, unless
3316 png_write_flush() is called before 25 more lines have been written.
3317 If nrows is too small (less than about 10 lines for a 640 pixel wide
3318 RGB image) the image compression may decrease noticeably (although this
3319 may be acceptable for real-time applications). Infrequent flushing will
3320 only degrade the compression performance by a few percent over images
3321 that do not use flushing.
3323 Writing the image data
3325 That's it for the transformations. Now you can write the image data.
3326 The simplest way to do this is in one function call. If you have the
3327 whole image in memory, you can just call png_write_image() and libpng
3328 will write the image. You will need to pass in an array of pointers to
3329 each row. This function automatically handles interlacing, so you don't
3330 need to call png_set_interlace_handling() or call this function multiple
3331 times, or any of that other stuff necessary with png_write_rows().
3333 png_write_image(png_ptr, row_pointers);
3335 where row_pointers is:
3337 png_byte *row_pointers[height];
3339 You can point to void or char or whatever you use for pixels.
3341 If you don't want to write the whole image at once, you can
3342 use png_write_rows() instead. If the file is not interlaced,
3345 png_write_rows(png_ptr, row_pointers,
3348 row_pointers is the same as in the png_write_image() call.
3350 If you are just writing one row at a time, you can do this with
3351 a single row_pointer instead of an array of row_pointers:
3353 png_bytep row_pointer = row;
3355 png_write_row(png_ptr, row_pointer);
3357 When the file is interlaced, things can get a good deal more complicated.
3358 The only currently (as of the PNG Specification version 1.2, dated July
3359 1999) defined interlacing scheme for PNG files is the "Adam7" interlace
3360 scheme, that breaks down an image into seven smaller images of varying
3361 size. libpng will build these images for you, or you can do them
3362 yourself. If you want to build them yourself, see the PNG specification
3363 for details of which pixels to write when.
3365 If you don't want libpng to handle the interlacing details, just
3366 use png_set_interlace_handling() and call png_write_rows() the
3367 correct number of times to write all the sub-images
3368 (png_set_interlace_handling() returns the number of sub-images.)
3370 If you want libpng to build the sub-images, call this before you start
3373 number_of_passes = png_set_interlace_handling(png_ptr);
3375 This will return the number of passes needed. Currently, this is seven,
3376 but may change if another interlace type is added.
3378 Then write the complete image number_of_passes times.
3380 png_write_rows(png_ptr, row_pointers, number_of_rows);
3382 Think carefully before you write an interlaced image. Typically code that
3383 reads such images reads all the image data into memory, uncompressed, before
3384 doing any processing. Only code that can display an image on the fly can
3385 take advantage of the interlacing and even then the image has to be exactly
3386 the correct size for the output device, because scaling an image requires
3387 adjacent pixels and these are not available until all the passes have been
3390 If you do write an interlaced image you will hardly ever need to handle
3391 the interlacing yourself. Call png_set_interlace_handling() and use the
3392 approach described above.
3394 The only time it is conceivable that you will really need to write an
3395 interlaced image pass-by-pass is when you have read one pass by pass and
3396 made some pixel-by-pixel transformation to it, as described in the read
3397 code above. In this case use the PNG_PASS_ROWS and PNG_PASS_COLS macros
3398 to determine the size of each sub-image in turn and simply write the rows
3399 you obtained from the read code.
3401 Finishing a sequential write
3403 After you are finished writing the image, you should finish writing
3404 the file. If you are interested in writing comments or time, you should
3405 pass an appropriately filled png_info pointer. If you are not interested,
3408 png_write_end(png_ptr, info_ptr);
3410 When you are done, you can free all memory used by libpng like this:
3412 png_destroy_write_struct(&png_ptr, &info_ptr);
3414 It is also possible to individually free the info_ptr members that
3415 point to libpng-allocated storage with the following function:
3417 png_free_data(png_ptr, info_ptr, mask, seq)
3419 mask - identifies data to be freed, a mask
3420 containing the bitwise OR of one or
3422 PNG_FREE_PLTE, PNG_FREE_TRNS,
3423 PNG_FREE_HIST, PNG_FREE_ICCP,
3424 PNG_FREE_PCAL, PNG_FREE_ROWS,
3425 PNG_FREE_SCAL, PNG_FREE_SPLT,
3426 PNG_FREE_TEXT, PNG_FREE_UNKN,
3427 or simply PNG_FREE_ALL
3429 seq - sequence number of item to be freed
3432 This function may be safely called when the relevant storage has
3433 already been freed, or has not yet been allocated, or was allocated
3434 by the user and not by libpng, and will in those cases do nothing.
3435 The "seq" parameter is ignored if only one item of the selected data
3436 type, such as PLTE, is allowed. If "seq" is not -1, and multiple items
3437 are allowed for the data type identified in the mask, such as text or
3438 sPLT, only the n'th item in the structure is freed, where n is "seq".
3440 If you allocated data such as a palette that you passed in to libpng
3441 with png_set_*, you must not free it until just before the call to
3442 png_destroy_write_struct().
3444 The default behavior is only to free data that was allocated internally
3445 by libpng. This can be changed, so that libpng will not free the data,
3446 or so that it will free data that was allocated by the user with png_malloc()
3447 or png_calloc() and passed in via a png_set_*() function, with
3449 png_data_freer(png_ptr, info_ptr, freer, mask)
3452 PNG_DESTROY_WILL_FREE_DATA
3453 PNG_SET_WILL_FREE_DATA
3454 PNG_USER_WILL_FREE_DATA
3456 mask - which data elements are affected
3457 same choices as in png_free_data()
3459 For example, to transfer responsibility for some data from a read structure
3460 to a write structure, you could use
3462 png_data_freer(read_ptr, read_info_ptr,
3463 PNG_USER_WILL_FREE_DATA,
3464 PNG_FREE_PLTE|PNG_FREE_tRNS|PNG_FREE_hIST)
3466 png_data_freer(write_ptr, write_info_ptr,
3467 PNG_DESTROY_WILL_FREE_DATA,
3468 PNG_FREE_PLTE|PNG_FREE_tRNS|PNG_FREE_hIST)
3470 thereby briefly reassigning responsibility for freeing to the user but
3471 immediately afterwards reassigning it once more to the write_destroy
3472 function. Having done this, it would then be safe to destroy the read
3473 structure and continue to use the PLTE, tRNS, and hIST data in the write
3476 This function only affects data that has already been allocated.
3477 You can call this function before calling after the png_set_*() functions
3478 to control whether the user or png_destroy_*() is supposed to free the data.
3479 When the user assumes responsibility for libpng-allocated data, the
3480 application must use
3481 png_free() to free it, and when the user transfers responsibility to libpng
3482 for data that the user has allocated, the user must have used png_malloc()
3483 or png_calloc() to allocate it.
3485 If you allocated text_ptr.text, text_ptr.lang, and text_ptr.translated_keyword
3486 separately, do not transfer responsibility for freeing text_ptr to libpng,
3487 because when libpng fills a png_text structure it combines these members with
3488 the key member, and png_free_data() will free only text_ptr.key. Similarly,
3489 if you transfer responsibility for free'ing text_ptr from libpng to your
3490 application, your application must not separately free those members.
3491 For a more compact example of writing a PNG image, see the file example.c.
3495 The simplified API, which became available in libpng-1.6.0, hides the details
3496 of both libpng and the PNG file format itself.
3497 It allows PNG files to be read into a very limited number of
3498 in-memory bitmap formats or to be written from the same formats. If these
3499 formats do not accomodate your needs then you can, and should, use the more
3500 sophisticated APIs above - these support a wide variety of in-memory formats
3501 and a wide variety of sophisticated transformations to those formats as well
3502 as a wide variety of APIs to manipulate ancilliary information.
3504 To read a PNG file using the simplified API:
3506 1) Declare a 'png_image' structure (see below) on the
3507 stack and memset() it to all zero.
3509 2) Call the appropriate png_image_begin_read... function.
3511 3) Set the png_image 'format' member to the required
3512 format and allocate a buffer for the image.
3514 4) Call png_image_finish_read to read the image into
3517 There are no restrictions on the format of the PNG input itself; all valid
3518 color types, bit depths, and interlace methods are acceptable, and the
3519 input image is transformed as necessary to the requested in-memory format
3520 during the png_image_finish_read() step.
3522 To write a PNG file using the simplified API:
3524 1) Declare a 'png_image' structure on the stack and memset()
3527 2) Initialize the members of the structure that describe the
3528 image, setting the 'format' member to the format of the
3531 3) Call the appropriate png_image_write... function with a
3532 pointer to the image to write the PNG data.
3534 png_image is a structure that describes the in-memory format of an image
3535 when it is being read or define the in-memory format of an image that you
3536 need to write. The "png_image" structure contains the following members:
3538 png_uint_32 version Set to PNG_IMAGE_VERSION
3539 png_uint_32 width Image width in pixels (columns)
3540 png_uint_32 height Image height in pixels (rows)
3541 png_uint_32 format Image format as defined below
3542 png_uint_32 flags A bit mask containing informational flags
3543 png_controlp opaque Initialize to NULL, free with png_image_free
3544 png_uint_32 colormap_entries; Number of entries in the color-map
3545 png_uint_32 warning_or_error;
3548 In the event of an error or warning the following field warning_or_error
3549 field will be set to a non-zero value and the 'message' field will contain
3550 a '\0' terminated string with the libpng error or warning message. If both
3551 warnings and an error were encountered, only the error is recorded. If there
3552 are multiple warnings, only the first one is recorded.
3554 The upper 30 bits of this value are reserved; the low two bits contain
3555 a two bit code such that a value more than 1 indicates a failure in the API
3558 0 - no warning or error
3561 3 - error preceded by warning
3563 The pixels (samples) of the image have one to four channels whose components
3564 have original values in the range 0 to 1.0:
3566 1: A single gray or luminance channel (G).
3567 2: A gray/luminance channel and an alpha channel (GA).
3568 3: Three red, green, blue color channels (RGB).
3569 4: Three color channels and an alpha channel (RGBA).
3571 The channels are encoded in one of two ways:
3573 a) As a small integer, value 0..255, contained in a single byte. For the
3574 alpha channel the original value is simply value/255. For the color or
3575 luminance channels the value is encoded according to the sRGB specification
3576 and matches the 8-bit format expected by typical display devices.
3578 The color/gray channels are not scaled (pre-multiplied) by the alpha
3579 channel and are suitable for passing to color management software.
3581 b) As a value in the range 0..65535, contained in a 2-byte integer. All
3582 channels can be converted to the original value by dividing by 65535; all
3583 channels are linear. Color channels use the RGB encoding (RGB end-points) of
3584 the sRGB specification. This encoding is identified by the
3585 PNG_FORMAT_FLAG_LINEAR flag below.
3587 When an alpha channel is present it is expected to denote pixel coverage
3588 of the color or luminance channels and is returned as an associated alpha
3589 channel: the color/gray channels are scaled (pre-multiplied) by the alpha
3592 When a color-mapped image is used as a result of calling
3593 png_image_read_colormap or png_image_write_colormap the channels are encoded
3594 in the color-map and the descriptions above apply to the color-map entries.
3595 The image data is encoded as small integers, value 0..255, that index the
3596 entries in the color-map. One integer (one byte) is stored for each pixel.
3600 The #defines to be used in png_image::format. Each #define identifies a
3601 particular layout of channel data and, if present, alpha values. There are
3602 separate defines for each of the two channel encodings.
3604 A format is built up using single bit flag values. Not all combinations are
3605 valid: use the bit flag values below for testing a format returned by the
3606 read APIs, but set formats from the derived values.
3608 When reading or writing color-mapped images the format should be set to the
3609 format of the entries in the color-map then png_image_{read,write}_colormap
3610 called to read or write the color-map and set the format correctly for the
3611 image data. Do not set the PNG_FORMAT_FLAG_COLORMAP bit directly!
3613 NOTE: libpng can be built with particular features disabled, if you see
3614 compiler errors because the definition of one of the following flags has been
3615 compiled out it is because libpng does not have the required support. It is
3616 possible, however, for the libpng configuration to enable the format on just
3617 read or just write; in that case you may see an error at run time. You can
3618 guard against this by checking for the definition of:
3620 PNG_SIMPLIFIED_{READ,WRITE}_{BGR,AFIRST}_SUPPORTED
3622 PNG_FORMAT_FLAG_ALPHA 0x01 format with an alpha channel
3623 PNG_FORMAT_FLAG_COLOR 0x02 color format: otherwise grayscale
3624 PNG_FORMAT_FLAG_LINEAR 0x04 png_uint_16 channels else png_byte
3625 PNG_FORMAT_FLAG_COLORMAP 0x08 libpng use only
3626 PNG_FORMAT_FLAG_BGR 0x10 BGR colors, else order is RGB
3627 PNG_FORMAT_FLAG_AFIRST 0x20 alpha channel comes first
3629 Supported formats are as follows. Future versions of libpng may support more
3630 formats; for compatibility with older versions simply check if the format
3631 macro is defined using #ifdef. These defines describe the in-memory layout
3632 of the components of the pixels of the image.
3634 First the single byte formats:
3637 PNG_FORMAT_GA PNG_FORMAT_FLAG_ALPHA
3638 PNG_FORMAT_AG (PNG_FORMAT_GA|PNG_FORMAT_FLAG_AFIRST)
3639 PNG_FORMAT_RGB PNG_FORMAT_FLAG_COLOR
3640 PNG_FORMAT_BGR (PNG_FORMAT_FLAG_COLOR|PNG_FORMAT_FLAG_BGR)
3641 PNG_FORMAT_RGBA (PNG_FORMAT_RGB|PNG_FORMAT_FLAG_ALPHA)
3642 PNG_FORMAT_ARGB (PNG_FORMAT_RGBA|PNG_FORMAT_FLAG_AFIRST)
3643 PNG_FORMAT_BGRA (PNG_FORMAT_BGR|PNG_FORMAT_FLAG_ALPHA)
3644 PNG_FORMAT_ABGR (PNG_FORMAT_BGRA|PNG_FORMAT_FLAG_AFIRST)
3646 Then the linear 2-byte formats. When naming these "Y" is used to
3647 indicate a luminance (gray) channel. The component order within the pixel
3648 is always the same - there is no provision for swapping the order of the
3649 components in the linear format.
3651 PNG_FORMAT_LINEAR_Y PNG_FORMAT_FLAG_LINEAR
3652 PNG_FORMAT_LINEAR_Y_ALPHA
3653 (PNG_FORMAT_FLAG_LINEAR|PNG_FORMAT_FLAG_ALPHA)
3654 PNG_FORMAT_LINEAR_RGB
3655 (PNG_FORMAT_FLAG_LINEAR|PNG_FORMAT_FLAG_COLOR)
3656 PNG_FORMAT_LINEAR_RGB_ALPHA
3657 (PNG_FORMAT_FLAG_LINEAR|PNG_FORMAT_FLAG_COLOR|
3658 PNG_FORMAT_FLAG_ALPHA)
3660 Color-mapped formats are obtained by calling png_image_{read,write}_colormap,
3661 as appropriate after setting png_image::format to the format of the color-map
3662 to be read or written. Applications may check the value of
3663 PNG_FORMAT_FLAG_COLORMAP to see if they have called the colormap API. The
3664 format of the color-map may be extracted using the following macro.
3666 PNG_FORMAT_OF_COLORMAP(fmt) ((fmt) & ~PNG_FORMAT_FLAG_COLORMAP)
3670 These are convenience macros to derive information from a png_image
3671 structure. The PNG_IMAGE_SAMPLE_ macros return values appropriate to the
3672 actual image sample values - either the entries in the color-map or the
3673 pixels in the image. The PNG_IMAGE_PIXEL_ macros return corresponding values
3674 for the pixels and will always return 1 after a call to
3675 png_image_{read,write}_colormap. The remaining macros return information
3676 about the rows in the image and the complete image.
3678 NOTE: All the macros that take a png_image::format parameter are compile time
3679 constants if the format parameter is, itself, a constant. Therefore these
3680 macros can be used in array declarations and case labels where required.
3681 Similarly the macros are also pre-processor constants (sizeof is not used) so
3682 they can be used in #if tests.
3684 First the information about the samples.
3686 PNG_IMAGE_SAMPLE_CHANNELS(fmt)
3687 Returns the total number of channels in a given format: 1..4
3689 PNG_IMAGE_SAMPLE_COMPONENT_SIZE(fmt)
3690 Returns the size in bytes of a single component of a pixel or color-map
3691 entry (as appropriate) in the image.
3693 PNG_IMAGE_SAMPLE_SIZE(fmt)
3694 This is the size of the sample data for one sample. If the image is
3695 color-mapped it is the size of one color-map entry (and image pixels are
3696 one byte in size), otherwise it is the size of one image pixel.
3698 PNG_IMAGE_COLORMAP_SIZE(fmt)
3699 The size of the color-map required by the format; this is the size of the
3700 color-map buffer passed to the png_image_{read,write}_colormap APIs, it is
3701 a fixed number determined by the format so can easily be allocated on the
3704 #define PNG_IMAGE_MAXIMUM_COLORMAP_COMPONENTS(fmt)\
3705 (PNG_IMAGE_SAMPLE_CHANNELS(fmt) * 256)
3706 /* The maximum size of the color-map required by the format expressed in a
3707 * count of components. This can be used to compile-time allocate a
3710 * png_uint_16 colormap[PNG_IMAGE_MAXIMUM_COLORMAP_COMPONENTS(linear_fmt)];
3712 * png_byte colormap[PNG_IMAGE_MAXIMUM_COLORMAP_COMPONENTS(sRGB_fmt)];
3714 * Alternatively use the PNG_IMAGE_COLORMAP_SIZE macro below to use the
3715 * information from one of the png_image_begin_read_ APIs and dynamically
3716 * allocate the required memory.
3720 Corresponding information about the pixels
3722 PNG_IMAGE_PIXEL_(test,fmt)
3724 PNG_IMAGE_PIXEL_CHANNELS(fmt)
3725 The number of separate channels (components) in a pixel; 1 for a
3728 PNG_IMAGE_PIXEL_COMPONENT_SIZE(fmt)\
3729 The size, in bytes, of each component in a pixel; 1 for a color-mapped
3732 PNG_IMAGE_PIXEL_SIZE(fmt)
3733 The size, in bytes, of a complete pixel; 1 for a color-mapped image.
3735 Information about the whole row, or whole image
3737 PNG_IMAGE_ROW_STRIDE(image)
3738 Returns the total number of components in a single row of the image; this
3739 is the minimum 'row stride', the minimum count of components between each
3740 row. For a color-mapped image this is the minimum number of bytes in a
3743 PNG_IMAGE_BUFFER_SIZE(image, row_stride)
3744 Returns the size, in bytes, of an image buffer given a png_image and a row
3745 stride - the number of components to leave space for in each row.
3747 PNG_IMAGE_FLAG_COLORSPACE_NOT_sRGB == 0x01
3748 This indicates the the RGB values of the in-memory bitmap do not
3749 correspond to the red, green and blue end-points defined by sRGB.
3751 PNG_IMAGE_FLAG_COLORMAP == 0x02
3752 The PNG is color-mapped. If this flag is set png_image_read_colormap
3753 can be used without further loss of image information. If it is not set
3754 png_image_read_colormap will cause significant loss if the image has any
3758 The png_image passed to the read APIs must have been initialized by setting
3759 the png_controlp field 'opaque' to NULL (or, better, memset the whole thing.)
3761 int png_image_begin_read_from_file( png_imagep image,
3762 const char *file_name)
3764 The named file is opened for read and the image header
3765 is filled in from the PNG header in the file.
3767 int png_image_begin_read_from_stdio (png_imagep image,
3770 The PNG header is read from the stdio FILE object.
3772 int png_image_begin_read_from_memory(png_imagep image,
3773 png_const_voidp memory, png_size_t size)
3775 The PNG header is read from the given memory buffer.
3777 int png_image_finish_read(png_imagep image,
3778 png_colorp background, void *buffer,
3779 png_int_32 row_stride, void *colormap));
3781 Finish reading the image into the supplied buffer and
3782 clean up the png_image structure.
3784 row_stride is the step, in png_byte or png_uint_16 units
3785 as appropriate, between adjacent rows. A positive stride
3786 indicates that the top-most row is first in the buffer -
3787 the normal top-down arrangement. A negative stride
3788 indicates that the bottom-most row is first in the buffer.
3790 background need only be supplied if an alpha channel must
3791 be removed from a png_byte format and the removal is to be
3792 done by compositing on a solid color; otherwise it may be
3793 NULL and any composition will be done directly onto the
3794 buffer. The value is an sRGB color to use for the
3795 background, for grayscale output the green channel is used.
3797 For linear output removing the alpha channel is always done
3798 by compositing on black.
3800 void png_image_free(png_imagep image)
3802 Free any data allocated by libpng in image->opaque,
3803 setting the pointer to NULL. May be called at any time
3804 after the structure is initialized.
3806 When the simplified API needs to convert between sRGB and linear colorspaces,
3807 the actual sRGB transfer curve defined in the sRGB specification (see the
3808 article at http://en.wikipedia.org/wiki/SRGB) is used, not the gamma=1/2.2
3809 approximation used elsewhere in libpng.
3813 For write you must initialize a png_image structure to describe the image to
3816 version: must be set to PNG_IMAGE_VERSION
3817 opaque: must be initialized to NULL
3818 width: image width in pixels
3819 height: image height in rows
3820 format: the format of the data you wish to write
3821 flags: set to 0 unless one of the defined flags applies; set
3822 PNG_IMAGE_FLAG_COLORSPACE_NOT_sRGB for color format images
3823 where the RGB values do not correspond to the colors in sRGB.
3824 colormap_entries: set to the number of entries in the color-map (0 to 256)
3826 int png_image_write_to_file, (png_imagep image,
3827 const char *file, int convert_to_8bit, const void *buffer,
3828 png_int_32 row_stride, const void *colormap));
3830 Write the image to the named file.
3832 int png_image_write_to_stdio(png_imagep image, FILE *file,
3833 int convert_to_8_bit, const void *buffer,
3834 png_int_32 row_stride, const void *colormap)
3836 Write the image to the given (FILE*).
3838 With all write APIs if image is in one of the linear formats with
3839 (png_uint_16) data then setting convert_to_8_bit will cause the output to be
3840 a (png_byte) PNG gamma encoded according to the sRGB specification, otherwise
3841 a 16-bit linear encoded PNG file is written.
3843 With all APIs row_stride is handled as in the read APIs - it is the spacing
3844 from one row to the next in component sized units (float) and if negative
3845 indicates a bottom-up row layout in the buffer.
3847 Note that the write API does not support interlacing, sub-8-bit pixels,
3848 and indexed (paletted) images.
3850 VI. Modifying/Customizing libpng
3852 There are two issues here. The first is changing how libpng does
3853 standard things like memory allocation, input/output, and error handling.
3854 The second deals with more complicated things like adding new chunks,
3855 adding new transformations, and generally changing how libpng works.
3856 Both of those are compile-time issues; that is, they are generally
3857 determined at the time the code is written, and there is rarely a need
3858 to provide the user with a means of changing them.
3860 Memory allocation, input/output, and error handling
3862 All of the memory allocation, input/output, and error handling in libpng
3863 goes through callbacks that are user-settable. The default routines are
3864 in pngmem.c, pngrio.c, pngwio.c, and pngerror.c, respectively. To change
3865 these functions, call the appropriate png_set_*_fn() function.
3867 Memory allocation is done through the functions png_malloc(), png_calloc(),
3868 and png_free(). The png_malloc() and png_free() functions currently just
3869 call the standard C functions and png_calloc() calls png_malloc() and then
3870 clears the newly allocated memory to zero; note that png_calloc(png_ptr, size)
3871 is not the same as the calloc(number, size) function provided by stdlib.h.
3872 There is limited support for certain systems with segmented memory
3873 architectures and the types of pointers declared by png.h match this; you
3874 will have to use appropriate pointers in your application. Since it is
3875 unlikely that the method of handling memory allocation on a platform
3876 will change between applications, these functions must be modified in
3877 the library at compile time. If you prefer to use a different method
3878 of allocating and freeing data, you can use png_create_read_struct_2() or
3879 png_create_write_struct_2() to register your own functions as described
3880 above. These functions also provide a void pointer that can be retrieved
3883 mem_ptr=png_get_mem_ptr(png_ptr);
3885 Your replacement memory functions must have prototypes as follows:
3887 png_voidp malloc_fn(png_structp png_ptr,
3888 png_alloc_size_t size);
3890 void free_fn(png_structp png_ptr, png_voidp ptr);
3892 Your malloc_fn() must return NULL in case of failure. The png_malloc()
3893 function will normally call png_error() if it receives a NULL from the
3894 system memory allocator or from your replacement malloc_fn().
3896 Your free_fn() will never be called with a NULL ptr, since libpng's
3897 png_free() checks for NULL before calling free_fn().
3899 Input/Output in libpng is done through png_read() and png_write(),
3900 which currently just call fread() and fwrite(). The FILE * is stored in
3901 png_struct and is initialized via png_init_io(). If you wish to change
3902 the method of I/O, the library supplies callbacks that you can set
3903 through the function png_set_read_fn() and png_set_write_fn() at run
3904 time, instead of calling the png_init_io() function. These functions
3905 also provide a void pointer that can be retrieved via the function
3906 png_get_io_ptr(). For example:
3908 png_set_read_fn(png_structp read_ptr,
3909 voidp read_io_ptr, png_rw_ptr read_data_fn)
3911 png_set_write_fn(png_structp write_ptr,
3912 voidp write_io_ptr, png_rw_ptr write_data_fn,
3913 png_flush_ptr output_flush_fn);
3915 voidp read_io_ptr = png_get_io_ptr(read_ptr);
3916 voidp write_io_ptr = png_get_io_ptr(write_ptr);
3918 The replacement I/O functions must have prototypes as follows:
3920 void user_read_data(png_structp png_ptr,
3921 png_bytep data, png_size_t length);
3923 void user_write_data(png_structp png_ptr,
3924 png_bytep data, png_size_t length);
3926 void user_flush_data(png_structp png_ptr);
3928 The user_read_data() function is responsible for detecting and
3929 handling end-of-data errors.
3931 Supplying NULL for the read, write, or flush functions sets them back
3932 to using the default C stream functions, which expect the io_ptr to
3933 point to a standard *FILE structure. It is probably a mistake
3934 to use NULL for one of write_data_fn and output_flush_fn but not both
3935 of them, unless you have built libpng with PNG_NO_WRITE_FLUSH defined.
3936 It is an error to read from a write stream, and vice versa.
3938 Error handling in libpng is done through png_error() and png_warning().
3939 Errors handled through png_error() are fatal, meaning that png_error()
3940 should never return to its caller. Currently, this is handled via
3941 setjmp() and longjmp() (unless you have compiled libpng with
3942 PNG_NO_SETJMP, in which case it is handled via PNG_ABORT()),
3943 but you could change this to do things like exit() if you should wish,
3944 as long as your function does not return.
3946 On non-fatal errors, png_warning() is called
3947 to print a warning message, and then control returns to the calling code.
3948 By default png_error() and png_warning() print a message on stderr via
3949 fprintf() unless the library is compiled with PNG_NO_CONSOLE_IO defined
3950 (because you don't want the messages) or PNG_NO_STDIO defined (because
3951 fprintf() isn't available). If you wish to change the behavior of the error
3952 functions, you will need to set up your own message callbacks. These
3953 functions are normally supplied at the time that the png_struct is created.
3954 It is also possible to redirect errors and warnings to your own replacement
3955 functions after png_create_*_struct() has been called by calling:
3957 png_set_error_fn(png_structp png_ptr,
3958 png_voidp error_ptr, png_error_ptr error_fn,
3959 png_error_ptr warning_fn);
3961 png_voidp error_ptr = png_get_error_ptr(png_ptr);
3963 If NULL is supplied for either error_fn or warning_fn, then the libpng
3964 default function will be used, calling fprintf() and/or longjmp() if a
3965 problem is encountered. The replacement error functions should have
3966 parameters as follows:
3968 void user_error_fn(png_structp png_ptr,
3969 png_const_charp error_msg);
3971 void user_warning_fn(png_structp png_ptr,
3972 png_const_charp warning_msg);
3974 The motivation behind using setjmp() and longjmp() is the C++ throw and
3975 catch exception handling methods. This makes the code much easier to write,
3976 as there is no need to check every return code of every function call.
3977 However, there are some uncertainties about the status of local variables
3978 after a longjmp, so the user may want to be careful about doing anything
3979 after setjmp returns non-zero besides returning itself. Consult your
3980 compiler documentation for more details. For an alternative approach, you
3981 may wish to use the "cexcept" facility (see http://cexcept.sourceforge.net),
3982 which is illustrated in pngvalid.c and in contrib/visupng.
3984 Beginning in libpng-1.4.0, the png_set_benign_errors() API became available.
3985 You can use this to handle certain errors (normally handled as errors)
3988 png_set_benign_errors (png_ptr, int allowed);
3990 allowed: 0: (default) treat png_benign_error() an error.
3991 1: treat png_benign_error() as a warning.
3995 If you need to read or write custom chunks, you may need to get deeper
3996 into the libpng code. The library now has mechanisms for storing
3997 and writing chunks of unknown type; you can even declare callbacks
3998 for custom chunks. However, this may not be good enough if the
3999 library code itself needs to know about interactions between your
4000 chunk and existing `intrinsic' chunks.
4002 If you need to write a new intrinsic chunk, first read the PNG
4003 specification. Acquire a first level of understanding of how it works.
4004 Pay particular attention to the sections that describe chunk names,
4005 and look at how other chunks were designed, so you can do things
4006 similarly. Second, check out the sections of libpng that read and
4007 write chunks. Try to find a chunk that is similar to yours and use
4008 it as a template. More details can be found in the comments inside
4009 the code. It is best to handle private or unknown chunks in a generic method,
4010 via callback functions, instead of by modifying libpng functions. This
4011 is illustrated in pngtest.c, which uses a callback function to handle a
4012 private "vpAg" chunk and the new "sTER" chunk, which are both unknown to
4015 If you wish to write your own transformation for the data, look through
4016 the part of the code that does the transformations, and check out some of
4017 the simpler ones to get an idea of how they work. Try to find a similar
4018 transformation to the one you want to add and copy off of it. More details
4019 can be found in the comments inside the code itself.
4021 Configuring for 16-bit platforms
4023 You will want to look into zconf.h to tell zlib (and thus libpng) that
4024 it cannot allocate more then 64K at a time. Even if you can, the memory
4025 won't be accessible. So limit zlib and libpng to 64K by defining MAXSEG_64K.
4029 For DOS users who only have access to the lower 640K, you will
4030 have to limit zlib's memory usage via a png_set_compression_mem_level()
4031 call. See zlib.h or zconf.h in the zlib library for more information.
4033 Configuring for Medium Model
4035 Libpng's support for medium model has been tested on most of the popular
4036 compilers. Make sure MAXSEG_64K gets defined, USE_FAR_KEYWORD gets
4037 defined, and FAR gets defined to far in pngconf.h, and you should be
4038 all set. Everything in the library (except for zlib's structure) is
4039 expecting far data. You must use the typedefs with the p or pp on
4040 the end for pointers (or at least look at them and be careful). Make
4041 note that the rows of data are defined as png_bytepp, which is
4042 an "unsigned char far * far *".
4044 Configuring for gui/windowing platforms:
4046 You will need to write new error and warning functions that use the GUI
4047 interface, as described previously, and set them to be the error and
4048 warning functions at the time that png_create_*_struct() is called,
4049 in order to have them available during the structure initialization.
4050 They can be changed later via png_set_error_fn(). On some compilers,
4051 you may also have to change the memory allocators (png_malloc, etc.).
4053 Configuring for compiler xxx:
4055 All includes for libpng are in pngconf.h. If you need to add, change
4056 or delete an include, this is the place to do it.
4057 The includes that are not needed outside libpng are placed in pngpriv.h,
4058 which is only used by the routines inside libpng itself.
4059 The files in libpng proper only include pngpriv.h and png.h, which
4060 in turn includes pngconf.h and, as of libpng-1.5.0, pnglibconf.h.
4061 As of libpng-1.5.0, pngpriv.h also includes three other private header
4062 files, pngstruct.h, pnginfo.h, and pngdebug.h, which contain material
4063 that previously appeared in the public headers.
4067 There are special functions to configure the compression. Perhaps the
4068 most useful one changes the compression level, which currently uses
4069 input compression values in the range 0 - 9. The library normally
4070 uses the default compression level (Z_DEFAULT_COMPRESSION = 6). Tests
4071 have shown that for a large majority of images, compression values in
4072 the range 3-6 compress nearly as well as higher levels, and do so much
4073 faster. For online applications it may be desirable to have maximum speed
4074 (Z_BEST_SPEED = 1). With versions of zlib after v0.99, you can also
4075 specify no compression (Z_NO_COMPRESSION = 0), but this would create
4076 files larger than just storing the raw bitmap. You can specify the
4077 compression level by calling:
4080 png_set_compression_level(png_ptr, level);
4082 Another useful one is to reduce the memory level used by the library.
4083 The memory level defaults to 8, but it can be lowered if you are
4084 short on memory (running DOS, for example, where you only have 640K).
4085 Note that the memory level does have an effect on compression; among
4086 other things, lower levels will result in sections of incompressible
4087 data being emitted in smaller stored blocks, with a correspondingly
4088 larger relative overhead of up to 15% in the worst case.
4091 png_set_compression_mem_level(png_ptr, level);
4093 The other functions are for configuring zlib. They are not recommended
4094 for normal use and may result in writing an invalid PNG file. See
4095 zlib.h for more information on what these mean.
4098 png_set_compression_strategy(png_ptr,
4101 png_set_compression_window_bits(png_ptr,
4104 png_set_compression_method(png_ptr, method);
4106 png_set_compression_buffer_size(png_ptr, size);
4108 As of libpng version 1.5.4, additional APIs became
4109 available to set these separately for non-IDAT
4110 compressed chunks such as zTXt, iTXt, and iCCP:
4113 #if PNG_LIBPNG_VER >= 10504
4114 png_set_text_compression_level(png_ptr, level);
4116 png_set_text_compression_mem_level(png_ptr, level);
4118 png_set_text_compression_strategy(png_ptr,
4121 png_set_text_compression_window_bits(png_ptr,
4124 png_set_text_compression_method(png_ptr, method);
4127 Controlling row filtering
4129 If you want to control whether libpng uses filtering or not, which
4130 filters are used, and how it goes about picking row filters, you
4131 can call one of these functions. The selection and configuration
4132 of row filters can have a significant impact on the size and
4133 encoding speed and a somewhat lesser impact on the decoding speed
4134 of an image. Filtering is enabled by default for RGB and grayscale
4135 images (with and without alpha), but not for paletted images nor
4136 for any images with bit depths less than 8 bits/pixel.
4138 The 'method' parameter sets the main filtering method, which is
4139 currently only '0' in the PNG 1.2 specification. The 'filters'
4140 parameter sets which filter(s), if any, should be used for each
4141 scanline. Possible values are PNG_ALL_FILTERS and PNG_NO_FILTERS
4142 to turn filtering on and off, respectively.
4144 Individual filter types are PNG_FILTER_NONE, PNG_FILTER_SUB,
4145 PNG_FILTER_UP, PNG_FILTER_AVG, PNG_FILTER_PAETH, which can be bitwise
4146 ORed together with '|' to specify one or more filters to use.
4147 These filters are described in more detail in the PNG specification.
4148 If you intend to change the filter type during the course of writing
4149 the image, you should start with flags set for all of the filters
4150 you intend to use so that libpng can initialize its internal
4151 structures appropriately for all of the filter types. (Note that this
4152 means the first row must always be adaptively filtered, because libpng
4153 currently does not allocate the filter buffers until png_write_row()
4154 is called for the first time.)
4156 filters = PNG_FILTER_NONE | PNG_FILTER_SUB
4157 PNG_FILTER_UP | PNG_FILTER_AVG |
4158 PNG_FILTER_PAETH | PNG_ALL_FILTERS;
4160 png_set_filter(png_ptr, PNG_FILTER_TYPE_BASE,
4162 The second parameter can also be
4163 PNG_INTRAPIXEL_DIFFERENCING if you are
4164 writing a PNG to be embedded in a MNG
4165 datastream. This parameter must be the
4166 same as the value of filter_method used
4169 It is also possible to influence how libpng chooses from among the
4170 available filters. This is done in one or both of two ways - by
4171 telling it how important it is to keep the same filter for successive
4172 rows, and by telling it the relative computational costs of the filters.
4174 double weights[3] = {1.5, 1.3, 1.1},
4175 costs[PNG_FILTER_VALUE_LAST] =
4176 {1.0, 1.3, 1.3, 1.5, 1.7};
4178 png_set_filter_heuristics(png_ptr,
4179 PNG_FILTER_HEURISTIC_WEIGHTED, 3,
4182 The weights are multiplying factors that indicate to libpng that the
4183 row filter should be the same for successive rows unless another row filter
4184 is that many times better than the previous filter. In the above example,
4185 if the previous 3 filters were SUB, SUB, NONE, the SUB filter could have a
4186 "sum of absolute differences" 1.5 x 1.3 times higher than other filters
4187 and still be chosen, while the NONE filter could have a sum 1.1 times
4188 higher than other filters and still be chosen. Unspecified weights are
4189 taken to be 1.0, and the specified weights should probably be declining
4190 like those above in order to emphasize recent filters over older filters.
4192 The filter costs specify for each filter type a relative decoding cost
4193 to be considered when selecting row filters. This means that filters
4194 with higher costs are less likely to be chosen over filters with lower
4195 costs, unless their "sum of absolute differences" is that much smaller.
4196 The costs do not necessarily reflect the exact computational speeds of
4197 the various filters, since this would unduly influence the final image
4200 Note that the numbers above were invented purely for this example and
4201 are given only to help explain the function usage. Little testing has
4202 been done to find optimum values for either the costs or the weights.
4204 Removing unwanted object code
4206 There are a bunch of #define's in pngconf.h that control what parts of
4207 libpng are compiled. All the defines end in _SUPPORTED. If you are
4208 never going to use a capability, you can change the #define to #undef
4209 before recompiling libpng and save yourself code and data space, or
4210 you can turn off individual capabilities with defines that begin with
4213 In libpng-1.5.0 and later, the #define's are in pnglibconf.h instead.
4215 You can also turn all of the transforms and ancillary chunk capabilities
4216 off en masse with compiler directives that define
4217 PNG_NO_READ[or WRITE]_TRANSFORMS, or PNG_NO_READ[or WRITE]_ANCILLARY_CHUNKS,
4219 along with directives to turn on any of the capabilities that you do
4220 want. The PNG_NO_READ[or WRITE]_TRANSFORMS directives disable the extra
4221 transformations but still leave the library fully capable of reading
4222 and writing PNG files with all known public chunks. Use of the
4223 PNG_NO_READ[or WRITE]_ANCILLARY_CHUNKS directive produces a library
4224 that is incapable of reading or writing ancillary chunks. If you are
4225 not using the progressive reading capability, you can turn that off
4226 with PNG_NO_PROGRESSIVE_READ (don't confuse this with the INTERLACING
4227 capability, which you'll still have).
4229 All the reading and writing specific code are in separate files, so the
4230 linker should only grab the files it needs. However, if you want to
4231 make sure, or if you are building a stand alone library, all the
4232 reading files start with "pngr" and all the writing files start with "pngw".
4233 The files that don't match either (like png.c, pngtrans.c, etc.)
4234 are used for both reading and writing, and always need to be included.
4235 The progressive reader is in pngpread.c
4237 If you are creating or distributing a dynamically linked library (a .so
4238 or DLL file), you should not remove or disable any parts of the library,
4239 as this will cause applications linked with different versions of the
4240 library to fail if they call functions not available in your library.
4241 The size of the library itself should not be an issue, because only
4242 those sections that are actually used will be loaded into memory.
4244 Requesting debug printout
4246 The macro definition PNG_DEBUG can be used to request debugging
4247 printout. Set it to an integer value in the range 0 to 3. Higher
4248 numbers result in increasing amounts of debugging information. The
4249 information is printed to the "stderr" file, unless another file
4250 name is specified in the PNG_DEBUG_FILE macro definition.
4252 When PNG_DEBUG > 0, the following functions (macros) become available:
4254 png_debug(level, message)
4255 png_debug1(level, message, p1)
4256 png_debug2(level, message, p1, p2)
4258 in which "level" is compared to PNG_DEBUG to decide whether to print
4259 the message, "message" is the formatted string to be printed,
4260 and p1 and p2 are parameters that are to be embedded in the string
4261 according to printf-style formatting directives. For example,
4263 png_debug1(2, "foo=%d\n", foo);
4268 fprintf(PNG_DEBUG_FILE, "foo=%d\n", foo);
4270 When PNG_DEBUG is defined but is zero, the macros aren't defined, but you
4271 can still use PNG_DEBUG to control your own debugging:
4277 When PNG_DEBUG = 1, the macros are defined, but only png_debug statements
4278 having level = 0 will be printed. There aren't any such statements in
4279 this version of libpng, but if you insert some they will be printed.
4281 Prepending a prefix to exported symbols
4283 Starting with libpng-1.6.0, you can configure libpng (when using the
4284 "configure" script) to prefix all exported symbols by means of the
4285 configuration option "--with-libpng-prefix=FOO_", where FOO_ can be any
4286 string beginning with a letter and containing only uppercase
4287 and lowercase letters, digits, and the underscore (i.e., a C language
4288 identifier). This creates a set of macros in pnglibconf.h, so this is
4289 transparent to applications; their function calls get transformed by
4290 the macros to use the modified names.
4294 The MNG specification (available at http://www.libpng.org/pub/mng) allows
4295 certain extensions to PNG for PNG images that are embedded in MNG datastreams.
4296 Libpng can support some of these extensions. To enable them, use the
4297 png_permit_mng_features() function:
4299 feature_set = png_permit_mng_features(png_ptr, mask)
4301 mask is a png_uint_32 containing the bitwise OR of the
4302 features you want to enable. These include
4303 PNG_FLAG_MNG_EMPTY_PLTE
4304 PNG_FLAG_MNG_FILTER_64
4305 PNG_ALL_MNG_FEATURES
4307 feature_set is a png_uint_32 that is the bitwise AND of
4308 your mask with the set of MNG features that is
4309 supported by the version of libpng that you are using.
4311 It is an error to use this function when reading or writing a standalone
4312 PNG file with the PNG 8-byte signature. The PNG datastream must be wrapped
4313 in a MNG datastream. As a minimum, it must have the MNG 8-byte signature
4314 and the MHDR and MEND chunks. Libpng does not provide support for these
4315 or any other MNG chunks; your application must provide its own support for
4316 them. You may wish to consider using libmng (available at
4317 http://www.libmng.com) instead.
4319 VIII. Changes to Libpng from version 0.88
4321 It should be noted that versions of libpng later than 0.96 are not
4322 distributed by the original libpng author, Guy Schalnat, nor by
4323 Andreas Dilger, who had taken over from Guy during 1996 and 1997, and
4324 distributed versions 0.89 through 0.96, but rather by another member
4325 of the original PNG Group, Glenn Randers-Pehrson. Guy and Andreas are
4326 still alive and well, but they have moved on to other things.
4328 The old libpng functions png_read_init(), png_write_init(),
4329 png_info_init(), png_read_destroy(), and png_write_destroy() have been
4330 moved to PNG_INTERNAL in version 0.95 to discourage their use. These
4331 functions will be removed from libpng version 1.4.0.
4333 The preferred method of creating and initializing the libpng structures is
4334 via the png_create_read_struct(), png_create_write_struct(), and
4335 png_create_info_struct() because they isolate the size of the structures
4336 from the application, allow version error checking, and also allow the
4337 use of custom error handling routines during the initialization, which
4338 the old functions do not. The functions png_read_destroy() and
4339 png_write_destroy() do not actually free the memory that libpng
4340 allocated for these structs, but just reset the data structures, so they
4341 can be used instead of png_destroy_read_struct() and
4342 png_destroy_write_struct() if you feel there is too much system overhead
4343 allocating and freeing the png_struct for each image read.
4345 Setting the error callbacks via png_set_message_fn() before
4346 png_read_init() as was suggested in libpng-0.88 is no longer supported
4347 because this caused applications that do not use custom error functions
4348 to fail if the png_ptr was not initialized to zero. It is still possible
4349 to set the error callbacks AFTER png_read_init(), or to change them with
4350 png_set_error_fn(), which is essentially the same function, but with a new
4351 name to force compilation errors with applications that try to use the old
4354 Starting with version 1.0.7, you can find out which version of the library
4355 you are using at run-time:
4357 png_uint_32 libpng_vn = png_access_version_number();
4359 The number libpng_vn is constructed from the major version, minor
4360 version with leading zero, and release number with leading zero,
4361 (e.g., libpng_vn for version 1.0.7 is 10007).
4363 Note that this function does not take a png_ptr, so you can call it
4364 before you've created one.
4366 You can also check which version of png.h you used when compiling your
4369 png_uint_32 application_vn = PNG_LIBPNG_VER;
4371 IX. Changes to Libpng from version 1.0.x to 1.2.x
4373 Support for user memory management was enabled by default. To
4374 accomplish this, the functions png_create_read_struct_2(),
4375 png_create_write_struct_2(), png_set_mem_fn(), png_get_mem_ptr(),
4376 png_malloc_default(), and png_free_default() were added.
4378 Support for the iTXt chunk has been enabled by default as of
4381 Support for certain MNG features was enabled.
4383 Support for numbered error messages was added. However, we never got
4384 around to actually numbering the error messages. The function
4385 png_set_strip_error_numbers() was added (Note: the prototype for this
4386 function was inadvertently removed from png.h in PNG_NO_ASSEMBLER_CODE
4387 builds of libpng-1.2.15. It was restored in libpng-1.2.36).
4389 The png_malloc_warn() function was added at libpng-1.2.3. This issues
4390 a png_warning and returns NULL instead of aborting when it fails to
4391 acquire the requested memory allocation.
4393 Support for setting user limits on image width and height was enabled
4394 by default. The functions png_set_user_limits(), png_get_user_width_max(),
4395 and png_get_user_height_max() were added at libpng-1.2.6.
4397 The png_set_add_alpha() function was added at libpng-1.2.7.
4399 The function png_set_expand_gray_1_2_4_to_8() was added at libpng-1.2.9.
4400 Unlike png_set_gray_1_2_4_to_8(), the new function does not expand the
4401 tRNS chunk to alpha. The png_set_gray_1_2_4_to_8() function is
4404 A number of macro definitions in support of runtime selection of
4405 assembler code features (especially Intel MMX code support) were
4406 added at libpng-1.2.0:
4408 PNG_ASM_FLAG_MMX_SUPPORT_COMPILED
4409 PNG_ASM_FLAG_MMX_SUPPORT_IN_CPU
4410 PNG_ASM_FLAG_MMX_READ_COMBINE_ROW
4411 PNG_ASM_FLAG_MMX_READ_INTERLACE
4412 PNG_ASM_FLAG_MMX_READ_FILTER_SUB
4413 PNG_ASM_FLAG_MMX_READ_FILTER_UP
4414 PNG_ASM_FLAG_MMX_READ_FILTER_AVG
4415 PNG_ASM_FLAG_MMX_READ_FILTER_PAETH
4416 PNG_ASM_FLAGS_INITIALIZED
4422 We added the following functions in support of runtime
4423 selection of assembler code features:
4425 png_get_mmx_flagmask()
4426 png_set_mmx_thresholds()
4428 png_get_mmx_bitdepth_threshold()
4429 png_get_mmx_rowbytes_threshold()
4432 We replaced all of these functions with simple stubs in libpng-1.2.20,
4433 when the Intel assembler code was removed due to a licensing issue.
4435 These macros are deprecated:
4437 PNG_READ_TRANSFORMS_NOT_SUPPORTED
4438 PNG_PROGRESSIVE_READ_NOT_SUPPORTED
4439 PNG_NO_SEQUENTIAL_READ_SUPPORTED
4440 PNG_WRITE_TRANSFORMS_NOT_SUPPORTED
4441 PNG_READ_ANCILLARY_CHUNKS_NOT_SUPPORTED
4442 PNG_WRITE_ANCILLARY_CHUNKS_NOT_SUPPORTED
4444 They have been replaced, respectively, by:
4446 PNG_NO_READ_TRANSFORMS
4447 PNG_NO_PROGRESSIVE_READ
4448 PNG_NO_SEQUENTIAL_READ
4449 PNG_NO_WRITE_TRANSFORMS
4450 PNG_NO_READ_ANCILLARY_CHUNKS
4451 PNG_NO_WRITE_ANCILLARY_CHUNKS
4453 PNG_MAX_UINT was replaced with PNG_UINT_31_MAX. It has been
4454 deprecated since libpng-1.0.16 and libpng-1.2.6.
4457 png_check_sig(sig, num)
4459 !png_sig_cmp(sig, 0, num)
4460 It has been deprecated since libpng-0.90.
4463 png_set_gray_1_2_4_to_8()
4464 which also expands tRNS to alpha was replaced with
4465 png_set_expand_gray_1_2_4_to_8()
4466 which does not. It has been deprecated since libpng-1.0.18 and 1.2.9.
4468 X. Changes to Libpng from version 1.0.x/1.2.x to 1.4.x
4470 Private libpng prototypes and macro definitions were moved from
4471 png.h and pngconf.h into a new pngpriv.h header file.
4473 Functions png_set_benign_errors(), png_benign_error(), and
4474 png_chunk_benign_error() were added.
4476 Support for setting the maximum amount of memory that the application
4477 will allocate for reading chunks was added, as a security measure.
4478 The functions png_set_chunk_cache_max() and png_get_chunk_cache_max()
4479 were added to the library.
4481 We implemented support for I/O states by adding png_ptr member io_state
4482 and functions png_get_io_chunk_name() and png_get_io_state() in pngget.c
4484 We added PNG_TRANSFORM_GRAY_TO_RGB to the available high-level
4487 Checking for and reporting of errors in the IHDR chunk is more thorough.
4489 Support for global arrays was removed, to improve thread safety.
4491 Some obsolete/deprecated macros and functions have been removed.
4493 Typecasted NULL definitions such as
4494 #define png_voidp_NULL (png_voidp)NULL
4495 were eliminated. If you used these in your application, just use
4498 The png_struct and info_struct members "trans" and "trans_values" were
4499 changed to "trans_alpha" and "trans_color", respectively.
4501 The obsolete, unused pnggccrd.c and pngvcrd.c files and related makefiles
4504 The PNG_1_0_X and PNG_1_2_X macros were eliminated.
4506 The PNG_LEGACY_SUPPORTED macro was eliminated.
4508 Many WIN32_WCE #ifdefs were removed.
4510 The functions png_read_init(info_ptr), png_write_init(info_ptr),
4511 png_info_init(info_ptr), png_read_destroy(), and png_write_destroy()
4512 have been removed. They have been deprecated since libpng-0.95.
4514 The png_permit_empty_plte() was removed. It has been deprecated
4515 since libpng-1.0.9. Use png_permit_mng_features() instead.
4517 We removed the obsolete stub functions png_get_mmx_flagmask(),
4518 png_set_mmx_thresholds(), png_get_asm_flags(),
4519 png_get_mmx_bitdepth_threshold(), png_get_mmx_rowbytes_threshold(),
4520 png_set_asm_flags(), and png_mmx_supported()
4522 We removed the obsolete png_check_sig(), png_memcpy_check(), and
4523 png_memset_check() functions. Instead use !png_sig_cmp(), memcpy(),
4524 and memset(), respectively.
4526 The function png_set_gray_1_2_4_to_8() was removed. It has been
4527 deprecated since libpng-1.0.18 and 1.2.9, when it was replaced with
4528 png_set_expand_gray_1_2_4_to_8() because the former function also
4529 expanded any tRNS chunk to an alpha channel.
4531 Macros for png_get_uint_16, png_get_uint_32, and png_get_int_32
4532 were added and are used by default instead of the corresponding
4533 functions. Unfortunately,
4534 from libpng-1.4.0 until 1.4.4, the png_get_uint_16 macro (but not the
4535 function) incorrectly returned a value of type png_uint_32.
4537 We changed the prototype for png_malloc() from
4538 png_malloc(png_structp png_ptr, png_uint_32 size)
4540 png_malloc(png_structp png_ptr, png_alloc_size_t size)
4542 This also applies to the prototype for the user replacement malloc_fn().
4544 The png_calloc() function was added and is used in place of
4545 of "png_malloc(); memset();" except in the case in png_read_png()
4546 where the array consists of pointers; in this case a "for" loop is used
4547 after the png_malloc() to set the pointers to NULL, to give robust.
4548 behavior in case the application runs out of memory part-way through
4551 We changed the prototypes of png_get_compression_buffer_size() and
4552 png_set_compression_buffer_size() to work with png_size_t instead of
4555 Support for numbered error messages was removed by default, since we
4556 never got around to actually numbering the error messages. The function
4557 png_set_strip_error_numbers() was removed from the library by default.
4559 The png_zalloc() and png_zfree() functions are no longer exported.
4560 The png_zalloc() function no longer zeroes out the memory that it
4561 allocates. Applications that called png_zalloc(png_ptr, number, size)
4562 can call png_calloc(png_ptr, number*size) instead, and can call
4563 png_free() instead of png_zfree().
4565 Support for dithering was disabled by default in libpng-1.4.0, because
4566 it has not been well tested and doesn't actually "dither".
4568 removed, however, and could be enabled by building libpng with
4569 PNG_READ_DITHER_SUPPORTED defined. In libpng-1.4.2, this support
4570 was reenabled, but the function was renamed png_set_quantize() to
4571 reflect more accurately what it actually does. At the same time,
4572 the PNG_DITHER_[RED,GREEN_BLUE]_BITS macros were also renamed to
4573 PNG_QUANTIZE_[RED,GREEN,BLUE]_BITS, and PNG_READ_DITHER_SUPPORTED
4574 was renamed to PNG_READ_QUANTIZE_SUPPORTED.
4576 We removed the trailing '.' from the warning and error messages.
4578 XI. Changes to Libpng from version 1.4.x to 1.5.x
4580 From libpng-1.4.0 until 1.4.4, the png_get_uint_16 macro (but not the
4581 function) incorrectly returned a value of type png_uint_32.
4583 Checking for invalid palette index on read or write was added at libpng
4584 1.5.10. When an invalid index is found, libpng issues a benign error.
4585 This is enabled by default because this condition is an error according
4586 to the PNG specification, Clause 11.3.2, but the error can be ignored in
4589 png_set_check_for_invalid_index(png_ptr, allowed);
4592 0: disable benign error (accept the
4593 invalid data without warning).
4594 1: enable benign error (treat the
4595 invalid data as an error or a
4598 If the error is ignored, or if png_benign_error() treats it as a warning,
4599 any invalid pixels are decoded as opaque black by the decoder and written
4600 as-is by the encoder.
4602 Retrieving the maximum palette index found was added at libpng-1.5.15.
4603 This statement must appear after png_read_png() or png_read_image() while
4604 reading, and after png_write_png() or png_write_image() while writing.
4606 int max_palette = png_get_palette_max(png_ptr, info_ptr);
4608 This will return the maximum palette index found in the image, or "-1" if
4609 the palette was not checked, or "0" if no palette was found. Note that this
4610 does not account for any palette index used by ancillary chunks such as the
4611 bKGD chunk; you must check those separately to determine the maximum
4612 palette index actually used.
4614 A. Changes that affect users of libpng
4616 There are no substantial API changes between the non-deprecated parts of
4617 the 1.4.5 API and the 1.5.0 API; however, the ability to directly access
4618 members of the main libpng control structures, png_struct and png_info,
4619 deprecated in earlier versions of libpng, has been completely removed from
4622 We no longer include zlib.h in png.h. The include statement has been moved
4623 to pngstruct.h, where it is not accessible by applications. Applications that
4624 need access to information in zlib.h will need to add the '#include "zlib.h"'
4625 directive. It does not matter whether this is placed prior to or after
4626 the '"#include png.h"' directive.
4628 The png_sprintf(), png_strcpy(), and png_strncpy() macros are no longer used
4631 We moved the png_strlen(), png_memcpy(), png_memset(), and png_memcmp()
4632 macros into a private header file (pngpriv.h) that is not accessible to
4635 In png_get_iCCP, the type of "profile" was changed from png_charpp
4636 to png_bytepp, and in png_set_iCCP, from png_charp to png_const_bytep.
4638 There are changes of form in png.h, including new and changed macros to
4639 declare parts of the API. Some API functions with arguments that are
4640 pointers to data not modified within the function have been corrected to
4641 declare these arguments with PNG_CONST.
4643 Much of the internal use of C macros to control the library build has also
4644 changed and some of this is visible in the exported header files, in
4645 particular the use of macros to control data and API elements visible
4646 during application compilation may require significant revision to
4647 application code. (It is extremely rare for an application to do this.)
4649 Any program that compiled against libpng 1.4 and did not use deprecated
4650 features or access internal library structures should compile and work
4651 against libpng 1.5, except for the change in the prototype for
4652 png_get_iCCP() and png_set_iCCP() API functions mentioned above.
4654 libpng 1.5.0 adds PNG_ PASS macros to help in the reading and writing of
4655 interlaced images. The macros return the number of rows and columns in
4656 each pass and information that can be used to de-interlace and (if
4657 absolutely necessary) interlace an image.
4659 libpng 1.5.0 adds an API png_longjmp(png_ptr, value). This API calls
4660 the application-provided png_longjmp_ptr on the internal, but application
4661 initialized, longjmp buffer. It is provided as a convenience to avoid
4662 the need to use the png_jmpbuf macro, which had the unnecessary side
4663 effect of resetting the internal png_longjmp_ptr value.
4665 libpng 1.5.0 includes a complete fixed point API. By default this is
4666 present along with the corresponding floating point API. In general the
4667 fixed point API is faster and smaller than the floating point one because
4668 the PNG file format used fixed point, not floating point. This applies
4669 even if the library uses floating point in internal calculations. A new
4670 macro, PNG_FLOATING_ARITHMETIC_SUPPORTED, reveals whether the library
4671 uses floating point arithmetic (the default) or fixed point arithmetic
4672 internally for performance critical calculations such as gamma correction.
4673 In some cases, the gamma calculations may produce slightly different
4674 results. This has changed the results in png_rgb_to_gray and in alpha
4675 composition (png_set_background for example). This applies even if the
4676 original image was already linear (gamma == 1.0) and, therefore, it is
4677 not necessary to linearize the image. This is because libpng has *not*
4678 been changed to optimize that case correctly, yet.
4680 Fixed point support for the sCAL chunk comes with an important caveat;
4681 the sCAL specification uses a decimal encoding of floating point values
4682 and the accuracy of PNG fixed point values is insufficient for
4683 representation of these values. Consequently a "string" API
4684 (png_get_sCAL_s and png_set_sCAL_s) is the only reliable way of reading
4685 arbitrary sCAL chunks in the absence of either the floating point API or
4686 internal floating point calculations.
4688 Applications no longer need to include the optional distribution header
4689 file pngusr.h or define the corresponding macros during application
4690 build in order to see the correct variant of the libpng API. From 1.5.0
4691 application code can check for the corresponding _SUPPORTED macro:
4693 #ifdef PNG_INCH_CONVERSIONS_SUPPORTED
4694 /* code that uses the inch conversion APIs. */
4697 This macro will only be defined if the inch conversion functions have been
4698 compiled into libpng. The full set of macros, and whether or not support
4699 has been compiled in, are available in the header file pnglibconf.h.
4700 This header file is specific to the libpng build. Notice that prior to
4701 1.5.0 the _SUPPORTED macros would always have the default definition unless
4702 reset by pngusr.h or by explicit settings on the compiler command line.
4703 These settings may produce compiler warnings or errors in 1.5.0 because
4704 of macro redefinition.
4706 From libpng-1.4.0 until 1.4.4, the png_get_uint_16 macro (but not the
4707 function) incorrectly returned a value of type png_uint_32. libpng 1.5.0
4708 is consistent with the implementation in 1.4.5 and 1.2.x (where the macro
4711 Applications can now choose whether to use these macros or to call the
4712 corresponding function by defining PNG_USE_READ_MACROS or
4713 PNG_NO_USE_READ_MACROS before including png.h. Notice that this is
4714 only supported from 1.5.0 -defining PNG_NO_USE_READ_MACROS prior to 1.5.0
4715 will lead to a link failure.
4717 Prior to libpng-1.5.4, the zlib compressor used the same set of parameters
4718 when compressing the IDAT data and textual data such as zTXt and iCCP.
4719 In libpng-1.5.4 we reinitialized the zlib stream for each type of data.
4720 We added five png_set_text_*() functions for setting the parameters to
4721 use with textual data.
4723 Prior to libpng-1.5.4, the PNG_READ_16_TO_8_ACCURATE_SCALE_SUPPORTED
4724 option was off by default, and slightly inaccurate scaling occurred.
4725 This option can no longer be turned off, and the choice of accurate
4726 or inaccurate 16-to-8 scaling is by using the new png_set_scale_16_to_8()
4727 API for accurate scaling or the old png_set_strip_16_to_8() API for simple
4730 Prior to libpng-1.5.4, the png_set_user_limits() function could only be
4731 used to reduce the width and height limits from the value of
4732 PNG_USER_WIDTH_MAX and PNG_USER_HEIGHT_MAX, although this document said
4733 that it could be used to override them. Now this function will reduce or
4734 increase the limits.
4736 Starting in libpng-1.5.10, the user limits can be set en masse with the
4737 configuration option PNG_SAFE_LIMITS_SUPPORTED. If this option is enabled,
4738 a set of "safe" limits is applied in pngpriv.h. These can be overridden by
4739 application calls to png_set_user_limits(), png_set_user_chunk_cache_max(),
4740 and/or png_set_user_malloc_max() that increase or decrease the limits. Also,
4741 in libpng-1.5.10 the default width and height limits were increased
4742 from 1,000,000 to 0x7ffffff (i.e., made unlimited). Therefore, the
4745 png_user_width_max 0x7fffffff 1,000,000
4746 png_user_height_max 0x7fffffff 1,000,000
4747 png_user_chunk_cache_max 0 (unlimited) 128
4748 png_user_chunk_malloc_max 0 (unlimited) 8,000,000
4750 B. Changes to the build and configuration of libpng
4752 Details of internal changes to the library code can be found in the CHANGES
4753 file and in the GIT repository logs. These will be of no concern to the vast
4754 majority of library users or builders; however, the few who configure libpng
4755 to a non-default feature set may need to change how this is done.
4757 There should be no need for library builders to alter build scripts if
4758 these use the distributed build support - configure or the makefiles -
4759 however, users of the makefiles may care to update their build scripts
4760 to build pnglibconf.h where the corresponding makefile does not do so.
4762 Building libpng with a non-default configuration has changed completely.
4763 The old method using pngusr.h should still work correctly even though the
4764 way pngusr.h is used in the build has been changed; however, library
4765 builders will probably want to examine the changes to take advantage of
4766 new capabilities and to simplify their build system.
4768 B.1 Specific changes to library configuration capabilities
4770 The library now supports a complete fixed point implementation and can
4771 thus be used on systems that have no floating point support or very
4772 limited or slow support. Previously gamma correction, an essential part
4773 of complete PNG support, required reasonably fast floating point.
4775 As part of this the choice of internal implementation has been made
4776 independent of the choice of fixed versus floating point APIs and all the
4777 missing fixed point APIs have been implemented.
4779 The exact mechanism used to control attributes of API functions has
4780 changed. A single set of operating system independent macro definitions
4781 is used and operating system specific directives are defined in
4784 As part of this the mechanism used to choose procedure call standards on
4785 those systems that allow a choice has been changed. At present this only
4786 affects certain Microsoft (DOS, Windows) and IBM (OS/2) operating systems
4787 running on Intel processors. As before, PNGAPI is defined where required
4788 to control the exported API functions; however, two new macros, PNGCBAPI
4789 and PNGCAPI, are used instead for callback functions (PNGCBAPI) and
4790 (PNGCAPI) for functions that must match a C library prototype (currently
4791 only png_longjmp_ptr, which must match the C longjmp function.) The new
4792 approach is documented in pngconf.h
4794 Despite these changes, libpng 1.5.0 only supports the native C function
4795 calling standard on those platforms tested so far (__cdecl on Microsoft
4796 Windows). This is because the support requirements for alternative
4797 calling conventions seem to no longer exist. Developers who find it
4798 necessary to set PNG_API_RULE to 1 should advise the mailing list
4799 (png-mng-implement) of this and library builders who use Openwatcom and
4800 therefore set PNG_API_RULE to 2 should also contact the mailing list.
4802 A new test program, pngvalid, is provided in addition to pngtest.
4803 pngvalid validates the arithmetic accuracy of the gamma correction
4804 calculations and includes a number of validations of the file format.
4805 A subset of the full range of tests is run when "make check" is done
4806 (in the 'configure' build.) pngvalid also allows total allocated memory
4807 usage to be evaluated and performs additional memory overwrite validation.
4809 Many changes to individual feature macros have been made. The following
4810 are the changes most likely to be noticed by library builders who
4813 1) All feature macros now have consistent naming:
4815 #define PNG_NO_feature turns the feature off
4816 #define PNG_feature_SUPPORTED turns the feature on
4818 pnglibconf.h contains one line for each feature macro which is either:
4820 #define PNG_feature_SUPPORTED
4822 if the feature is supported or:
4824 /*#undef PNG_feature_SUPPORTED*/
4826 if it is not. Library code consistently checks for the 'SUPPORTED' macro.
4827 It does not, and libpng applications should not, check for the 'NO' macro
4828 which will not normally be defined even if the feature is not supported.
4829 The 'NO' macros are only used internally for setting or not setting the
4830 corresponding 'SUPPORTED' macros.
4832 Compatibility with the old names is provided as follows:
4834 PNG_INCH_CONVERSIONS turns on PNG_INCH_CONVERSIONS_SUPPORTED
4836 And the following definitions disable the corresponding feature:
4838 PNG_SETJMP_NOT_SUPPORTED disables SETJMP
4839 PNG_READ_TRANSFORMS_NOT_SUPPORTED disables READ_TRANSFORMS
4840 PNG_NO_READ_COMPOSITED_NODIV disables READ_COMPOSITE_NODIV
4841 PNG_WRITE_TRANSFORMS_NOT_SUPPORTED disables WRITE_TRANSFORMS
4842 PNG_READ_ANCILLARY_CHUNKS_NOT_SUPPORTED disables READ_ANCILLARY_CHUNKS
4843 PNG_WRITE_ANCILLARY_CHUNKS_NOT_SUPPORTED disables WRITE_ANCILLARY_CHUNKS
4845 Library builders should remove use of the above, inconsistent, names.
4847 2) Warning and error message formatting was previously conditional on
4848 the STDIO feature. The library has been changed to use the
4849 CONSOLE_IO feature instead. This means that if CONSOLE_IO is disabled
4850 the library no longer uses the printf(3) functions, even though the
4851 default read/write implementations use (FILE) style stdio.h functions.
4853 3) Three feature macros now control the fixed/floating point decisions:
4855 PNG_FLOATING_POINT_SUPPORTED enables the floating point APIs
4857 PNG_FIXED_POINT_SUPPORTED enables the fixed point APIs; however, in
4858 practice these are normally required internally anyway (because the PNG
4859 file format is fixed point), therefore in most cases PNG_NO_FIXED_POINT
4860 merely stops the function from being exported.
4862 PNG_FLOATING_ARITHMETIC_SUPPORTED chooses between the internal floating
4863 point implementation or the fixed point one. Typically the fixed point
4864 implementation is larger and slower than the floating point implementation
4865 on a system that supports floating point; however, it may be faster on a
4866 system which lacks floating point hardware and therefore uses a software
4869 4) Added PNG_{READ,WRITE}_INT_FUNCTIONS_SUPPORTED. This allows the
4870 functions to read and write ints to be disabled independently of
4871 PNG_USE_READ_MACROS, which allows libpng to be built with the functions
4872 even though the default is to use the macros - this allows applications
4873 to choose at app buildtime whether or not to use macros (previously
4874 impossible because the functions weren't in the default build.)
4876 B.2 Changes to the configuration mechanism
4878 Prior to libpng-1.5.0 library builders who needed to configure libpng
4879 had either to modify the exported pngconf.h header file to add system
4880 specific configuration or had to write feature selection macros into
4881 pngusr.h and cause this to be included into pngconf.h by defining
4882 PNG_USER_CONFIG. The latter mechanism had the disadvantage that an
4883 application built without PNG_USER_CONFIG defined would see the
4884 unmodified, default, libpng API and thus would probably fail to link.
4886 These mechanisms still work in the configure build and in any makefile
4887 build that builds pnglibconf.h, although the feature selection macros
4888 have changed somewhat as described above. In 1.5.0, however, pngusr.h is
4889 processed only once, when the exported header file pnglibconf.h is built.
4890 pngconf.h no longer includes pngusr.h, therefore pngusr.h is ignored after the
4891 build of pnglibconf.h and it is never included in an application build.
4893 The rarely used alternative of adding a list of feature macros to the
4894 CFLAGS setting in the build also still works; however, the macros will be
4895 copied to pnglibconf.h and this may produce macro redefinition warnings
4896 when the individual C files are compiled.
4898 All configuration now only works if pnglibconf.h is built from
4899 scripts/pnglibconf.dfa. This requires the program awk. Brian Kernighan
4900 (the original author of awk) maintains C source code of that awk and this
4901 and all known later implementations (often called by subtly different
4902 names - nawk and gawk for example) are adequate to build pnglibconf.h.
4903 The Sun Microsystems (now Oracle) program 'awk' is an earlier version
4904 and does not work; this may also apply to other systems that have a
4905 functioning awk called 'nawk'.
4907 Configuration options are now documented in scripts/pnglibconf.dfa. This
4908 file also includes dependency information that ensures a configuration is
4909 consistent; that is, if a feature is switched off dependent features are
4910 also removed. As a recommended alternative to using feature macros in
4911 pngusr.h a system builder may also define equivalent options in pngusr.dfa
4912 (or, indeed, any file) and add that to the configuration by setting
4913 DFA_XTRA to the file name. The makefiles in contrib/pngminim illustrate
4914 how to do this, and a case where pngusr.h is still required.
4916 XII. Changes to Libpng from version 1.5.x to 1.6.x
4918 A "simplified API" has been added (see documentation in png.h and a simple
4919 example in contrib/examples/pngtopng.c). The new publicly visible API
4920 includes the following:
4929 png_image_begin_read_from_file()
4930 png_image_begin_read_from_stdio()
4931 png_image_begin_read_from_memory()
4932 png_image_finish_read()
4935 png_image_write_to_file()
4936 png_image_write_to_stdio()
4938 Starting with libpng-1.6.0, you can configure libpng to prefix all exported
4939 symbols, using the PNG_PREFIX macro.
4941 We no longer include string.h in png.h. The include statement has been moved
4942 to pngpriv.h, where it is not accessible by applications. Applications that
4943 need access to information in string.h must add an '#include "string.h"'
4944 directive. It does not matter whether this is placed prior to or after
4945 the '"#include png.h"' directive.
4947 The following API are now DEPRECATED:
4949 png_convert_to_rfc1123() which has been replaced
4950 with png_convert_to_rfc1123_buffer()
4952 png_malloc_default()
4956 The following has been removed:
4957 png_get_io_chunk_name(), which has been replaced
4958 with png_get_io_chunk_type(). The new
4959 function returns a 32-bit integer instead of
4961 The png_sizeof(), png_strlen(), png_memcpy(), png_memcmp(), and
4962 png_memset() macros are no longer used in the libpng sources and
4963 have been removed. These had already been made invisible to applications
4964 (i.e., defined in the private pngpriv.h header file) since libpng-1.5.0.
4966 The signatures of many exported functions were changed, such that
4967 png_structp became png_structrp or png_const_structrp
4968 png_infop became png_inforp or png_const_inforp
4969 where "rp" indicates a "restricted pointer".
4971 Error detection in some chunks has improved; in particular the iCCP chunk
4972 reader now does pretty complete validation of the basic format. Some bad
4973 profiles that were previously accepted are now rejected, in particular the
4974 very old broken Microsoft/HP sRGB profile.
4976 The library now issues a warning if both background processing and RGB to
4977 gray are used when gamma correction happens. As with previous versions of
4978 the library the results are numerically very incorrect in this case.
4980 There are some minor arithmetic changes in some transforms such as
4981 png_set_background(), that might be detected by certain regression tests.
4983 Unknown chunk handling has been improved internally, without any API change.
4984 This adds more correct option control of the unknown handling, corrects
4985 a pre-existing bug where the per-chunk 'keep' setting is ignored, and makes
4986 it possible to skip IDAT chunks in the sequential reader.
4988 The machine-generated configure files are no longer included in branches
4989 libpng16 and later of the GIT repository. They continue to be included
4990 in the tarball releases, however.
4992 XIII. Detecting libpng
4994 The png_get_io_ptr() function has been present since libpng-0.88, has never
4995 changed, and is unaffected by conditional compilation macros. It is the
4996 best choice for use in configure scripts for detecting the presence of any
4997 libpng version since 0.88. In an autoconf "configure.in" you could use
4999 AC_CHECK_LIB(png, png_get_io_ptr, ...
5001 XV. Source code repository
5003 Since about February 2009, version 1.2.34, libpng has been under "git" source
5004 control. The git repository was built from old libpng-x.y.z.tar.gz files
5005 going back to version 0.70. You can access the git repository (read only)
5008 git://libpng.git.sourceforge.net/gitroot/libpng
5010 or you can browse it via "gitweb" at
5012 http://libpng.git.sourceforge.net/git/gitweb.cgi?p=libpng
5014 Patches can be sent to glennrp at users.sourceforge.net or to
5015 png-mng-implement at lists.sourceforge.net or you can upload them to
5016 the libpng bug tracker at
5018 http://libpng.sourceforge.net
5020 We also accept patches built from the tar or zip distributions, and
5021 simple verbal discriptions of bug fixes, reported either to the
5022 SourceForge bug tracker, to the png-mng-implement at lists.sf.net
5023 mailing list, or directly to glennrp.
5027 Our coding style is similar to the "Allman" style, with curly
5028 braces on separate lines:
5035 else if (another condition)
5040 The braces can be omitted from simple one-line actions:
5045 We use 3-space indentation, except for continued statements which
5046 are usually indented the same as the first line of the statement
5047 plus four more spaces.
5049 For macro definitions we use 2-space indentation, always leaving the "#"
5050 in the first column.
5052 #ifndef PNG_NO_FEATURE
5053 # ifndef PNG_FEATURE_SUPPORTED
5054 # define PNG_FEATURE_SUPPORTED
5058 Comments appear with the leading "/*" at the same indentation as
5059 the statement that follows the comment:
5061 /* Single-line comment */
5064 /* This is a multiple-line
5069 Very short comments can be placed after the end of the statement
5070 to which they pertain:
5072 statement; /* comment */
5074 We don't use C++ style ("//") comments. We have, however,
5075 used them in the past in some now-abandoned MMX assembler
5078 Functions and their curly braces are not indented, and
5079 exported functions are marked with PNGAPI:
5081 /* This is a public function that is visible to
5082 * application programmers. It does thus-and-so.
5085 png_exported_function(png_ptr, png_info, foo)
5090 The prototypes for all exported functions appear in png.h,
5091 above the comment that says
5093 /* Maintainer: Put new public prototypes here ... */
5095 We mark all non-exported functions with "/* PRIVATE */"":
5098 png_non_exported_function(png_ptr, png_info, foo)
5103 The prototypes for non-exported functions (except for those in
5106 above the comment that says
5108 /* Maintainer: Put new private prototypes here ^ */
5110 We put a space after the "sizeof" operator and we omit the
5111 optional parentheses around its argument when the argument
5112 is an expression, not a type name, and we always enclose the
5113 sizeof operator, with its argument, in parentheses:
5115 (sizeof (png_uint_32))
5118 Prior to libpng-1.6.0 we used a "png_sizeof()" macro, formatted as
5119 though it were a function.
5121 To avoid polluting the global namespace, the names of all exported
5122 functions and variables begin with "png_", and all publicly visible C
5123 preprocessor macros begin with "PNG". We request that applications that
5124 use libpng *not* begin any of their own symbols with either of these strings.
5126 We put a space after each comma and after each semicolon
5127 in "for" statements, and we put spaces before and after each
5128 C binary operator and after "for" or "while", and before
5129 "?". We don't put a space between a typecast and the expression
5130 being cast, nor do we put one between a function name and the
5131 left parenthesis that follows it:
5133 for (i = 2; i > 0; --i)
5134 y[i] = a(x) + (int)b;
5136 We prefer #ifdef and #ifndef to #if defined() and #if !defined()
5137 when there is only one macro being tested.
5139 We prefer to express integers that are used as bit masks in hex format,
5140 with an even number of lower-case hex digits (e.g., 0x00, 0xff, 0x0100).
5142 We prefer to use underscores in variable names rather than camelCase, except
5143 for a few type names that we inherit from zlib.h.
5145 We do not use the TAB character for indentation in the C sources.
5147 Lines do not exceed 80 characters.
5149 Other rules can be inferred by inspecting the libpng source.
5151 XVI. Y2K Compliance in libpng
5155 Since the PNG Development group is an ad-hoc body, we can't make
5156 an official declaration.
5158 This is your unofficial assurance that libpng from version 0.71 and
5159 upward through 1.6.0 are Y2K compliant. It is my belief that earlier
5160 versions were also Y2K compliant.
5162 Libpng only has two year fields. One is a 2-byte unsigned integer
5163 that will hold years up to 65535. The other, which is deprecated,
5164 holds the date in text format, and will hold years up to 9999.
5167 "png_uint_16 year" in png_time_struct.
5170 "char time_buffer[29]" in png_struct. This is no longer used
5171 in libpng-1.6.x and will be removed from libpng-1.7.0.
5173 There are seven time-related functions:
5175 png_convert_to_rfc_1123() in png.c
5176 (formerly png_convert_to_rfc_1152() in error)
5177 png_convert_from_struct_tm() in pngwrite.c, called
5179 png_convert_from_time_t() in pngwrite.c
5180 png_get_tIME() in pngget.c
5181 png_handle_tIME() in pngrutil.c, called in pngread.c
5182 png_set_tIME() in pngset.c
5183 png_write_tIME() in pngwutil.c, called in pngwrite.c
5185 All appear to handle dates properly in a Y2K environment. The
5186 png_convert_from_time_t() function calls gmtime() to convert from system
5187 clock time, which returns (year - 1900), which we properly convert to
5188 the full 4-digit year. There is a possibility that applications using
5189 libpng are not passing 4-digit years into the png_convert_to_rfc_1123()
5190 function, or that they are incorrectly passing only a 2-digit year
5191 instead of "year - 1900" into the png_convert_from_struct_tm() function,
5192 but this is not under our control. The libpng documentation has always
5193 stated that it works with 4-digit years, and the APIs have been
5196 The tIME chunk itself is also Y2K compliant. It uses a 2-byte unsigned
5197 integer to hold the year, and can hold years as large as 65535.
5199 zlib, upon which libpng depends, is also Y2K compliant. It contains
5200 no date-related code.
5203 Glenn Randers-Pehrson
5205 PNG Development Group