1 libpng.txt - A description on how to use and modify libpng
3 libpng version 1.2.40 - September 10, 2009
4 Updated and distributed by Glenn Randers-Pehrson
5 <glennrp at users.sourceforge.net>
6 Copyright (c) 1998-2009 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 1.2.40 - September 10, 2009
15 Updated and distributed by Glenn Randers-Pehrson
16 Copyright (c) 1998-2009 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
33 This file describes how to use and modify the PNG reference library
34 (known as libpng) for your own use. There are five sections to this
35 file: introduction, structures, reading, writing, and modification and
36 configuration notes for various special platforms. In addition to this
37 file, example.c is a good starting point for using the library, as
38 it is heavily commented and should include everything most people
39 will need. We assume that libpng is already installed; see the
40 INSTALL file for instructions on how to install libpng.
42 For examples of libpng usage, see the files "example.c", "pngtest.c",
43 and the files in the "contrib" directory, all of which are included in the
46 Libpng was written as a companion to the PNG specification, as a way
47 of reducing the amount of time and effort it takes to support the PNG
48 file format in application programs.
50 The PNG specification (second edition), November 2003, is available as
51 a W3C Recommendation and as an ISO Standard (ISO/IEC 15948:2003 (E)) at
52 <http://www.w3.org/TR/2003/REC-PNG-20031110/
53 The W3C and ISO documents have identical technical content.
55 The PNG-1.2 specification is available at
56 <http://www.libpng.org/pub/png/documents/>. It is technically equivalent
57 to the PNG specification (second edition) but has some additional material.
59 The PNG-1.0 specification is available
60 as RFC 2083 <http://www.libpng.org/pub/png/documents/> and as a
61 W3C Recommendation <http://www.w3.org/TR/REC.png.html>.
63 Some additional chunks are described in the special-purpose public chunks
64 documents at <http://www.libpng.org/pub/png/documents/>.
67 about PNG, and the latest version of libpng, can be found at the PNG home
68 page, <http://www.libpng.org/pub/png/>.
70 Most users will not have to modify the library significantly; advanced
71 users may want to modify it more. All attempts were made to make it as
72 complete as possible, while keeping the code easy to understand.
73 Currently, this library only supports C. Support for other languages
76 Libpng has been designed to handle multiple sessions at one time,
77 to be easily modifiable, to be portable to the vast majority of
78 machines (ANSI, K&R, 16-, 32-, and 64-bit) available, and to be easy
79 to use. The ultimate goal of libpng is to promote the acceptance of
80 the PNG file format in whatever way possible. While there is still
81 work to be done (see the TODO file), libpng should cover the
82 majority of the needs of its users.
84 Libpng uses zlib for its compression and decompression of PNG files.
85 Further information about zlib, and the latest version of zlib, can
86 be found at the zlib home page, <http://www.info-zip.org/pub/infozip/zlib/>.
87 The zlib compression utility is a general purpose utility that is
88 useful for more than PNG files, and can be used without libpng.
89 See the documentation delivered with zlib for more details.
90 You can usually find the source files for the zlib utility wherever you
91 find the libpng source files.
93 Libpng is thread safe, provided the threads are using different
94 instances of the structures. Each thread should have its own
95 png_struct and png_info instances, and thus its own image.
96 Libpng does not protect itself against two threads using the
97 same instance of a structure.
101 There are two main structures that are important to libpng, png_struct
102 and png_info. The first, png_struct, is an internal structure that
103 will not, for the most part, be used by a user except as the first
104 variable passed to every libpng function call.
106 The png_info structure is designed to provide information about the
107 PNG file. At one time, the fields of png_info were intended to be
108 directly accessible to the user. However, this tended to cause problems
109 with applications using dynamically loaded libraries, and as a result
110 a set of interface functions for png_info (the png_get_*() and png_set_*()
111 functions) was developed. The fields of png_info are still available for
112 older applications, but it is suggested that applications use the new
113 interfaces if at all possible.
115 Applications that do make direct access to the members of png_struct (except
116 for png_ptr->jmpbuf) must be recompiled whenever the library is updated,
117 and applications that make direct access to the members of png_info must
118 be recompiled if they were compiled or loaded with libpng version 1.0.6,
119 in which the members were in a different order. In version 1.0.7, the
120 members of the png_info structure reverted to the old order, as they were
121 in versions 0.97c through 1.0.5. Starting with version 2.0.0, both
122 structures are going to be hidden, and the contents of the structures will
123 only be accessible through the png_get/png_set functions.
125 The png.h header file is an invaluable reference for programming with libpng.
126 And while I'm on the topic, make sure you include the libpng header file:
132 We'll now walk you through the possible functions to call when reading
133 in a PNG file sequentially, briefly explaining the syntax and purpose
134 of each one. See example.c and png.h for more detail. While
135 progressive reading is covered in the next section, you will still
136 need some of the functions discussed in this section to read a PNG
141 You will want to do the I/O initialization(*) before you get into libpng,
142 so if it doesn't work, you don't have much to undo. Of course, you
143 will also want to insure that you are, in fact, dealing with a PNG
144 file. Libpng provides a simple check to see if a file is a PNG file.
145 To use it, pass in the first 1 to 8 bytes of the file to the function
146 png_sig_cmp(), and it will return 0 (false) if the bytes match the
147 corresponding bytes of the PNG signature, or nonzero (true) otherwise.
148 Of course, the more bytes you pass in, the greater the accuracy of the
151 If you are intending to keep the file pointer open for use in libpng,
152 you must ensure you don't read more than 8 bytes from the beginning
153 of the file, and you also have to make a call to png_set_sig_bytes_read()
154 with the number of bytes you read from the beginning. Libpng will
155 then only check the bytes (if any) that your program didn't read.
157 (*): If you are not using the standard I/O functions, you will need
158 to replace them with custom functions. See the discussion under
162 FILE *fp = fopen(file_name, "rb");
167 fread(header, 1, number, fp);
168 is_png = !png_sig_cmp(header, 0, number);
175 Next, png_struct and png_info need to be allocated and initialized. In
176 order to ensure that the size of these structures is correct even with a
177 dynamically linked libpng, there are functions to initialize and
178 allocate the structures. We also pass the library version, optional
179 pointers to error handling functions, and a pointer to a data struct for
180 use by the error functions, if necessary (the pointer and functions can
181 be NULL if the default error handlers are to be used). See the section
182 on Changes to Libpng below regarding the old initialization functions.
183 The structure allocation functions quietly return NULL if they fail to
184 create the structure, so your application should check for that.
186 png_structp png_ptr = png_create_read_struct
187 (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
188 user_error_fn, user_warning_fn);
192 png_infop info_ptr = png_create_info_struct(png_ptr);
195 png_destroy_read_struct(&png_ptr,
196 (png_infopp)NULL, (png_infopp)NULL);
200 png_infop end_info = png_create_info_struct(png_ptr);
203 png_destroy_read_struct(&png_ptr, &info_ptr,
208 If you want to use your own memory allocation routines,
209 define PNG_USER_MEM_SUPPORTED and use
210 png_create_read_struct_2() instead of png_create_read_struct():
212 png_structp png_ptr = png_create_read_struct_2
213 (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
214 user_error_fn, user_warning_fn, (png_voidp)
215 user_mem_ptr, user_malloc_fn, user_free_fn);
217 The error handling routines passed to png_create_read_struct()
218 and the memory alloc/free routines passed to png_create_struct_2()
219 are only necessary if you are not using the libpng supplied error
220 handling and memory alloc/free functions.
222 When libpng encounters an error, it expects to longjmp back
223 to your routine. Therefore, you will need to call setjmp and pass
224 your png_jmpbuf(png_ptr). If you read the file from different
225 routines, you will need to update the jmpbuf field every time you enter
226 a new routine that will call a png_*() function.
228 See your documentation of setjmp/longjmp for your compiler for more
229 information on setjmp/longjmp. See the discussion on libpng error
230 handling in the Customizing Libpng section below for more information
231 on the libpng error handling. If an error occurs, and libpng longjmp's
232 back to your setjmp, you will want to call png_destroy_read_struct() to
235 if (setjmp(png_jmpbuf(png_ptr)))
237 png_destroy_read_struct(&png_ptr, &info_ptr,
243 If you would rather avoid the complexity of setjmp/longjmp issues,
244 you can compile libpng with PNG_SETJMP_NOT_SUPPORTED, in which case
245 errors will result in a call to PNG_ABORT() which defaults to abort().
247 Now you need to set up the input code. The default for libpng is to
248 use the C function fread(). If you use this, you will need to pass a
249 valid FILE * in the function png_init_io(). Be sure that the file is
250 opened in binary mode. If you wish to handle reading data in another
251 way, you need not call the png_init_io() function, but you must then
252 implement the libpng I/O methods discussed in the Customizing Libpng
255 png_init_io(png_ptr, fp);
257 If you had previously opened the file and read any of the signature from
258 the beginning in order to see if this was a PNG file, you need to let
259 libpng know that there are some bytes missing from the start of the file.
261 png_set_sig_bytes(png_ptr, number);
263 Setting up callback code
265 You can set up a callback function to handle any unknown chunks in the
266 input stream. You must supply the function
268 read_chunk_callback(png_ptr ptr,
269 png_unknown_chunkp chunk);
271 /* The unknown chunk structure contains your
272 chunk data, along with similar data for any other
279 /* Note that libpng has already taken care of
282 /* put your code here. Search for your chunk in the
283 unknown chunk structure, process it, and return one
286 return (-n); /* chunk had an error */
287 return (0); /* did not recognize */
288 return (n); /* success */
291 (You can give your function another name that you like instead of
292 "read_chunk_callback")
294 To inform libpng about your function, use
296 png_set_read_user_chunk_fn(png_ptr, user_chunk_ptr,
297 read_chunk_callback);
299 This names not only the callback function, but also a user pointer that
300 you can retrieve with
302 png_get_user_chunk_ptr(png_ptr);
304 If you call the png_set_read_user_chunk_fn() function, then all unknown
305 chunks will be saved when read, in case your callback function will need
306 one or more of them. This behavior can be changed with the
307 png_set_keep_unknown_chunks() function, described below.
309 At this point, you can set up a callback function that will be
310 called after each row has been read, which you can use to control
311 a progress meter or the like. It's demonstrated in pngtest.c.
312 You must supply a function
314 void read_row_callback(png_ptr ptr, png_uint_32 row,
317 /* put your code here */
320 (You can give it another name that you like instead of "read_row_callback")
322 To inform libpng about your function, use
324 png_set_read_status_fn(png_ptr, read_row_callback);
326 Unknown-chunk handling
328 Now you get to set the way the library processes unknown chunks in the
329 input PNG stream. Both known and unknown chunks will be read. Normal
330 behavior is that known chunks will be parsed into information in
331 various info_ptr members while unknown chunks will be discarded. This
332 behavior can be wasteful if your application will never use some known
333 chunk types. To change this, you can call:
335 png_set_keep_unknown_chunks(png_ptr, keep,
336 chunk_list, num_chunks);
337 keep - 0: default unknown chunk handling
338 1: ignore; do not keep
339 2: keep only if safe-to-copy
340 3: keep even if unsafe-to-copy
341 You can use these definitions:
342 PNG_HANDLE_CHUNK_AS_DEFAULT 0
343 PNG_HANDLE_CHUNK_NEVER 1
344 PNG_HANDLE_CHUNK_IF_SAFE 2
345 PNG_HANDLE_CHUNK_ALWAYS 3
346 chunk_list - list of chunks affected (a byte string,
347 five bytes per chunk, NULL or '\0' if
349 num_chunks - number of chunks affected; if 0, all
350 unknown chunks are affected. If nonzero,
351 only the chunks in the list are affected
353 Unknown chunks declared in this way will be saved as raw data onto a
354 list of png_unknown_chunk structures. If a chunk that is normally
355 known to libpng is named in the list, it will be handled as unknown,
356 according to the "keep" directive. If a chunk is named in successive
357 instances of png_set_keep_unknown_chunks(), the final instance will
358 take precedence. The IHDR and IEND chunks should not be named in
359 chunk_list; if they are, libpng will process them normally anyway.
361 Here is an example of the usage of png_set_keep_unknown_chunks(),
362 where the private "vpAg" chunk will later be processed by a user chunk
365 png_byte vpAg[5]={118, 112, 65, 103, (png_byte) '\0'};
367 #if defined(PNG_UNKNOWN_CHUNKS_SUPPORTED)
368 png_byte unused_chunks[]=
370 104, 73, 83, 84, (png_byte) '\0', /* hIST */
371 105, 84, 88, 116, (png_byte) '\0', /* iTXt */
372 112, 67, 65, 76, (png_byte) '\0', /* pCAL */
373 115, 67, 65, 76, (png_byte) '\0', /* sCAL */
374 115, 80, 76, 84, (png_byte) '\0', /* sPLT */
375 116, 73, 77, 69, (png_byte) '\0', /* tIME */
381 #if defined(PNG_UNKNOWN_CHUNKS_SUPPORTED)
382 /* ignore all unknown chunks: */
383 png_set_keep_unknown_chunks(read_ptr, 1, NULL, 0);
384 /* except for vpAg: */
385 png_set_keep_unknown_chunks(read_ptr, 2, vpAg, 1);
386 /* also ignore unused known chunks: */
387 png_set_keep_unknown_chunks(read_ptr, 1, unused_chunks,
388 (int)sizeof(unused_chunks)/5);
393 The PNG specification allows the width and height of an image to be as
394 large as 2^31-1 (0x7fffffff), or about 2.147 billion rows and columns.
395 Since very few applications really need to process such large images,
396 we have imposed an arbitrary 1-million limit on rows and columns.
397 Larger images will be rejected immediately with a png_error() call. If
398 you wish to override this limit, you can use
400 png_set_user_limits(png_ptr, width_max, height_max);
402 to set your own limits, or use width_max = height_max = 0x7fffffffL
403 to allow all valid dimensions (libpng may reject some very large images
404 anyway because of potential buffer overflow conditions).
406 You should put this statement after you create the PNG structure and
407 before calling png_read_info(), png_read_png(), or png_process_data().
408 If you need to retrieve the limits that are being applied, use
410 width_max = png_get_user_width_max(png_ptr);
411 height_max = png_get_user_height_max(png_ptr);
413 The high-level read interface
415 At this point there are two ways to proceed; through the high-level
416 read interface, or through a sequence of low-level read operations.
417 You can use the high-level interface if (a) you are willing to read
418 the entire image into memory, and (b) the input transformations
419 you want to do are limited to the following set:
421 PNG_TRANSFORM_IDENTITY No transformation
422 PNG_TRANSFORM_STRIP_16 Strip 16-bit samples to
424 PNG_TRANSFORM_STRIP_ALPHA Discard the alpha channel
425 PNG_TRANSFORM_PACKING Expand 1, 2 and 4-bit
427 PNG_TRANSFORM_PACKSWAP Change order of packed
429 PNG_TRANSFORM_EXPAND Perform set_expand()
430 PNG_TRANSFORM_INVERT_MONO Invert monochrome images
431 PNG_TRANSFORM_SHIFT Normalize pixels to the
433 PNG_TRANSFORM_BGR Flip RGB to BGR, RGBA
435 PNG_TRANSFORM_SWAP_ALPHA Flip RGBA to ARGB or GA
437 PNG_TRANSFORM_INVERT_ALPHA Change alpha from opacity
439 PNG_TRANSFORM_SWAP_ENDIAN Byte-swap 16-bit samples
441 (This excludes setting a background color, doing gamma transformation,
442 dithering, and setting filler.) If this is the case, simply do this:
444 png_read_png(png_ptr, info_ptr, png_transforms, NULL)
446 where png_transforms is an integer containing the bitwise OR of
447 some set of transformation flags. This call is equivalent to png_read_info(),
448 followed the set of transformations indicated by the transform mask,
449 then png_read_image(), and finally png_read_end().
451 (The final parameter of this call is not yet used. Someday it might point
452 to transformation parameters required by some future input transform.)
454 You must use png_transforms and not call any png_set_transform() functions
455 when you use png_read_png().
457 After you have called png_read_png(), you can retrieve the image data
460 row_pointers = png_get_rows(png_ptr, info_ptr);
462 where row_pointers is an array of pointers to the pixel data for each row:
464 png_bytep row_pointers[height];
466 If you know your image size and pixel size ahead of time, you can allocate
467 row_pointers prior to calling png_read_png() with
469 if (height > PNG_UINT_32_MAX/png_sizeof(png_byte))
471 "Image is too tall to process in memory");
472 if (width > PNG_UINT_32_MAX/pixel_size)
474 "Image is too wide to process in memory");
475 row_pointers = png_malloc(png_ptr,
476 height*png_sizeof(png_bytep));
477 for (int i=0; i<height, i++)
478 row_pointers[i]=NULL; /* security precaution */
479 for (int i=0; i<height, i++)
480 row_pointers[i]=png_malloc(png_ptr,
482 png_set_rows(png_ptr, info_ptr, &row_pointers);
484 Alternatively you could allocate your image in one big block and define
485 row_pointers[i] to point into the proper places in your block.
487 If you use png_set_rows(), the application is responsible for freeing
488 row_pointers (and row_pointers[i], if they were separately allocated).
490 If you don't allocate row_pointers ahead of time, png_read_png() will
491 do it, and it'll be free'ed when you call png_destroy_*().
493 The low-level read interface
495 If you are going the low-level route, you are now ready to read all
496 the file information up to the actual image data. You do this with a
497 call to png_read_info().
499 png_read_info(png_ptr, info_ptr);
501 This will process all chunks up to but not including the image data.
503 Querying the info structure
505 Functions are used to get the information from the info_ptr once it
506 has been read. Note that these fields may not be completely filled
507 in until png_read_end() has read the chunk data following the image.
509 png_get_IHDR(png_ptr, info_ptr, &width, &height,
510 &bit_depth, &color_type, &interlace_type,
511 &compression_type, &filter_method);
513 width - holds the width of the image
514 in pixels (up to 2^31).
515 height - holds the height of the image
516 in pixels (up to 2^31).
517 bit_depth - holds the bit depth of one of the
518 image channels. (valid values are
519 1, 2, 4, 8, 16 and depend also on
520 the color_type. See also
521 significant bits (sBIT) below).
522 color_type - describes which color/alpha channels
525 (bit depths 1, 2, 4, 8, 16)
526 PNG_COLOR_TYPE_GRAY_ALPHA
528 PNG_COLOR_TYPE_PALETTE
529 (bit depths 1, 2, 4, 8)
532 PNG_COLOR_TYPE_RGB_ALPHA
535 PNG_COLOR_MASK_PALETTE
539 filter_method - (must be PNG_FILTER_TYPE_BASE
540 for PNG 1.0, and can also be
541 PNG_INTRAPIXEL_DIFFERENCING if
542 the PNG datastream is embedded in
543 a MNG-1.0 datastream)
544 compression_type - (must be PNG_COMPRESSION_TYPE_BASE
546 interlace_type - (PNG_INTERLACE_NONE or
548 Any or all of interlace_type, compression_type, of
549 filter_method can be NULL if you are
550 not interested in their values.
552 channels = png_get_channels(png_ptr, info_ptr);
553 channels - number of channels of info for the
554 color type (valid values are 1 (GRAY,
555 PALETTE), 2 (GRAY_ALPHA), 3 (RGB),
556 4 (RGB_ALPHA or RGB + filler byte))
557 rowbytes = png_get_rowbytes(png_ptr, info_ptr);
558 rowbytes - number of bytes needed to hold a row
560 signature = png_get_signature(png_ptr, info_ptr);
561 signature - holds the signature read from the
562 file (if any). The data is kept in
563 the same offset it would be if the
564 whole signature were read (i.e. if an
565 application had already read in 4
566 bytes of signature before starting
567 libpng, the remaining 4 bytes would
568 be in signature[4] through signature[7]
569 (see png_set_sig_bytes())).
572 width = png_get_image_width(png_ptr,
574 height = png_get_image_height(png_ptr,
576 bit_depth = png_get_bit_depth(png_ptr,
578 color_type = png_get_color_type(png_ptr,
580 filter_method = png_get_filter_type(png_ptr,
582 compression_type = png_get_compression_type(png_ptr,
584 interlace_type = png_get_interlace_type(png_ptr,
588 These are also important, but their validity depends on whether the chunk
589 has been read. The png_get_valid(png_ptr, info_ptr, PNG_INFO_<chunk>) and
590 png_get_<chunk>(png_ptr, info_ptr, ...) functions return non-zero if the
591 data has been read, or zero if it is missing. The parameters to the
592 png_get_<chunk> are set directly if they are simple data types, or a pointer
593 into the info_ptr is returned for any complex types.
595 png_get_PLTE(png_ptr, info_ptr, &palette,
597 palette - the palette for the file
599 num_palette - number of entries in the palette
601 png_get_gAMA(png_ptr, info_ptr, &gamma);
602 gamma - the gamma the file is written
605 png_get_sRGB(png_ptr, info_ptr, &srgb_intent);
606 srgb_intent - the rendering intent (PNG_INFO_sRGB)
607 The presence of the sRGB chunk
608 means that the pixel data is in the
609 sRGB color space. This chunk also
610 implies specific values of gAMA and
613 png_get_iCCP(png_ptr, info_ptr, &name,
614 &compression_type, &profile, &proflen);
615 name - The profile name.
616 compression - The compression type; always
617 PNG_COMPRESSION_TYPE_BASE for PNG 1.0.
618 You may give NULL to this argument to
620 profile - International Color Consortium color
621 profile data. May contain NULs.
622 proflen - length of profile data in bytes.
624 png_get_sBIT(png_ptr, info_ptr, &sig_bit);
625 sig_bit - the number of significant bits for
626 (PNG_INFO_sBIT) each of the gray,
627 red, green, and blue channels,
628 whichever are appropriate for the
629 given color type (png_color_16)
631 png_get_tRNS(png_ptr, info_ptr, &trans, &num_trans,
633 trans - array of transparent entries for
634 palette (PNG_INFO_tRNS)
635 trans_values - graylevel or color sample values of
636 the single transparent color for
637 non-paletted images (PNG_INFO_tRNS)
638 num_trans - number of transparent entries
641 png_get_hIST(png_ptr, info_ptr, &hist);
643 hist - histogram of palette (array of
646 png_get_tIME(png_ptr, info_ptr, &mod_time);
647 mod_time - time image was last modified
650 png_get_bKGD(png_ptr, info_ptr, &background);
651 background - background color (PNG_VALID_bKGD)
652 valid 16-bit red, green and blue
653 values, regardless of color_type
655 num_comments = png_get_text(png_ptr, info_ptr,
656 &text_ptr, &num_text);
657 num_comments - number of comments
658 text_ptr - array of png_text holding image
660 text_ptr[i].compression - type of compression used
661 on "text" PNG_TEXT_COMPRESSION_NONE
662 PNG_TEXT_COMPRESSION_zTXt
663 PNG_ITXT_COMPRESSION_NONE
664 PNG_ITXT_COMPRESSION_zTXt
665 text_ptr[i].key - keyword for comment. Must contain
667 text_ptr[i].text - text comments for current
668 keyword. Can be empty.
669 text_ptr[i].text_length - length of text string,
670 after decompression, 0 for iTXt
671 text_ptr[i].itxt_length - length of itxt string,
672 after decompression, 0 for tEXt/zTXt
673 text_ptr[i].lang - language of comment (empty
675 text_ptr[i].lang_key - keyword in UTF-8
676 (empty string for unknown).
677 num_text - number of comments (same as
678 num_comments; you can put NULL here
679 to avoid the duplication)
680 Note while png_set_text() will accept text, language,
681 and translated keywords that can be NULL pointers, the
682 structure returned by png_get_text will always contain
683 regular zero-terminated C strings. They might be
684 empty strings but they will never be NULL pointers.
686 num_spalettes = png_get_sPLT(png_ptr, info_ptr,
688 palette_ptr - array of palette structures holding
689 contents of one or more sPLT chunks
691 num_spalettes - number of sPLT chunks read.
693 png_get_oFFs(png_ptr, info_ptr, &offset_x, &offset_y,
695 offset_x - positive offset from the left edge
697 offset_y - positive offset from the top edge
699 unit_type - PNG_OFFSET_PIXEL, PNG_OFFSET_MICROMETER
701 png_get_pHYs(png_ptr, info_ptr, &res_x, &res_y,
703 res_x - pixels/unit physical resolution in
705 res_y - pixels/unit physical resolution in
707 unit_type - PNG_RESOLUTION_UNKNOWN,
710 png_get_sCAL(png_ptr, info_ptr, &unit, &width,
712 unit - physical scale units (an integer)
713 width - width of a pixel in physical scale units
714 height - height of a pixel in physical scale units
715 (width and height are doubles)
717 png_get_sCAL_s(png_ptr, info_ptr, &unit, &width,
719 unit - physical scale units (an integer)
720 width - width of a pixel in physical scale units
721 height - height of a pixel in physical scale units
722 (width and height are strings like "2.54")
724 num_unknown_chunks = png_get_unknown_chunks(png_ptr,
726 unknowns - array of png_unknown_chunk
727 structures holding unknown chunks
728 unknowns[i].name - name of unknown chunk
729 unknowns[i].data - data of unknown chunk
730 unknowns[i].size - size of unknown chunk's data
731 unknowns[i].location - position of chunk in file
733 The value of "i" corresponds to the order in which the
734 chunks were read from the PNG file or inserted with the
735 png_set_unknown_chunks() function.
737 The data from the pHYs chunk can be retrieved in several convenient
740 res_x = png_get_x_pixels_per_meter(png_ptr,
742 res_y = png_get_y_pixels_per_meter(png_ptr,
744 res_x_and_y = png_get_pixels_per_meter(png_ptr,
746 res_x = png_get_x_pixels_per_inch(png_ptr,
748 res_y = png_get_y_pixels_per_inch(png_ptr,
750 res_x_and_y = png_get_pixels_per_inch(png_ptr,
752 aspect_ratio = png_get_pixel_aspect_ratio(png_ptr,
755 (Each of these returns 0 [signifying "unknown"] if
756 the data is not present or if res_x is 0;
757 res_x_and_y is 0 if res_x != res_y)
759 The data from the oFFs chunk can be retrieved in several convenient
762 x_offset = png_get_x_offset_microns(png_ptr, info_ptr);
763 y_offset = png_get_y_offset_microns(png_ptr, info_ptr);
764 x_offset = png_get_x_offset_inches(png_ptr, info_ptr);
765 y_offset = png_get_y_offset_inches(png_ptr, info_ptr);
767 (Each of these returns 0 [signifying "unknown" if both
768 x and y are 0] if the data is not present or if the
769 chunk is present but the unit is the pixel)
771 For more information, see the png_info definition in png.h and the
772 PNG specification for chunk contents. Be careful with trusting
773 rowbytes, as some of the transformations could increase the space
774 needed to hold a row (expand, filler, gray_to_rgb, etc.).
775 See png_read_update_info(), below.
777 A quick word about text_ptr and num_text. PNG stores comments in
778 keyword/text pairs, one pair per chunk, with no limit on the number
779 of text chunks, and a 2^31 byte limit on their size. While there are
780 suggested keywords, there is no requirement to restrict the use to these
781 strings. It is strongly suggested that keywords and text be sensible
782 to humans (that's the point), so don't use abbreviations. Non-printing
783 symbols are not allowed. See the PNG specification for more details.
784 There is also no requirement to have text after the keyword.
786 Keywords should be limited to 79 Latin-1 characters without leading or
787 trailing spaces, but non-consecutive spaces are allowed within the
788 keyword. It is possible to have the same keyword any number of times.
789 The text_ptr is an array of png_text structures, each holding a
790 pointer to a language string, a pointer to a keyword and a pointer to
791 a text string. The text string, language code, and translated
792 keyword may be empty or NULL pointers. The keyword/text
793 pairs are put into the array in the order that they are received.
794 However, some or all of the text chunks may be after the image, so, to
795 make sure you have read all the text chunks, don't mess with these
796 until after you read the stuff after the image. This will be
797 mentioned again below in the discussion that goes with png_read_end().
799 Input transformations
801 After you've read the header information, you can set up the library
802 to handle any special transformations of the image data. The various
803 ways to transform the data will be described in the order that they
804 should occur. This is important, as some of these change the color
805 type and/or bit depth of the data, and some others only work on
806 certain color types and bit depths. Even though each transformation
807 checks to see if it has data that it can do something with, you should
808 make sure to only enable a transformation if it will be valid for the
809 data. For example, don't swap red and blue on grayscale data.
811 The colors used for the background and transparency values should be
812 supplied in the same format/depth as the current image data. They
813 are stored in the same format/depth as the image data in a bKGD or tRNS
814 chunk, so this is what libpng expects for this data. The colors are
815 transformed to keep in sync with the image data when an application
816 calls the png_read_update_info() routine (see below).
818 Data will be decoded into the supplied row buffers packed into bytes
819 unless the library has been told to transform it into another format.
820 For example, 4 bit/pixel paletted or grayscale data will be returned
821 2 pixels/byte with the leftmost pixel in the high-order bits of the
822 byte, unless png_set_packing() is called. 8-bit RGB data will be stored
823 in RGB RGB RGB format unless png_set_filler() or png_set_add_alpha()
824 is called to insert filler bytes, either before or after each RGB triplet.
825 16-bit RGB data will be returned RRGGBB RRGGBB, with the most significant
826 byte of the color value first, unless png_set_strip_16() is called to
827 transform it to regular RGB RGB triplets, or png_set_filler() or
828 png_set_add alpha() is called to insert filler bytes, either before or
829 after each RRGGBB triplet. Similarly, 8-bit or 16-bit grayscale data can
831 png_set_filler(), png_set_add_alpha(), or png_set_strip_16().
833 The following code transforms grayscale images of less than 8 to 8 bits,
834 changes paletted images to RGB, and adds a full alpha channel if there is
835 transparency information in a tRNS chunk. This is most useful on
836 grayscale images with bit depths of 2 or 4 or if there is a multiple-image
837 viewing application that wishes to treat all images in the same way.
839 if (color_type == PNG_COLOR_TYPE_PALETTE)
840 png_set_palette_to_rgb(png_ptr);
842 if (color_type == PNG_COLOR_TYPE_GRAY &&
843 bit_depth < 8) png_set_expand_gray_1_2_4_to_8(png_ptr);
845 if (png_get_valid(png_ptr, info_ptr,
846 PNG_INFO_tRNS)) png_set_tRNS_to_alpha(png_ptr);
848 These three functions are actually aliases for png_set_expand(), added
849 in libpng version 1.0.4, with the function names expanded to improve code
850 readability. In some future version they may actually do different
853 As of libpng version 1.2.9, png_set_expand_gray_1_2_4_to_8() was
854 added. It expands the sample depth without changing tRNS to alpha.
856 PNG can have files with 16 bits per channel. If you only can handle
857 8 bits per channel, this will strip the pixels down to 8 bit.
860 png_set_strip_16(png_ptr);
862 If, for some reason, you don't need the alpha channel on an image,
863 and you want to remove it rather than combining it with the background
864 (but the image author certainly had in mind that you *would* combine
865 it with the background, so that's what you should probably do):
867 if (color_type & PNG_COLOR_MASK_ALPHA)
868 png_set_strip_alpha(png_ptr);
870 In PNG files, the alpha channel in an image
871 is the level of opacity. If you need the alpha channel in an image to
872 be the level of transparency instead of opacity, you can invert the
873 alpha channel (or the tRNS chunk data) after it's read, so that 0 is
874 fully opaque and 255 (in 8-bit or paletted images) or 65535 (in 16-bit
875 images) is fully transparent, with
877 png_set_invert_alpha(png_ptr);
879 PNG files pack pixels of bit depths 1, 2, and 4 into bytes as small as
880 they can, resulting in, for example, 8 pixels per byte for 1 bit
881 files. This code expands to 1 pixel per byte without changing the
882 values of the pixels:
885 png_set_packing(png_ptr);
887 PNG files have possible bit depths of 1, 2, 4, 8, and 16. All pixels
888 stored in a PNG image have been "scaled" or "shifted" up to the next
889 higher possible bit depth (e.g. from 5 bits/sample in the range [0,31] to
890 8 bits/sample in the range [0, 255]). However, it is also possible to
891 convert the PNG pixel data back to the original bit depth of the image.
892 This call reduces the pixels back down to the original bit depth:
894 png_color_8p sig_bit;
896 if (png_get_sBIT(png_ptr, info_ptr, &sig_bit))
897 png_set_shift(png_ptr, sig_bit);
899 PNG files store 3-color pixels in red, green, blue order. This code
900 changes the storage of the pixels to blue, green, red:
902 if (color_type == PNG_COLOR_TYPE_RGB ||
903 color_type == PNG_COLOR_TYPE_RGB_ALPHA)
904 png_set_bgr(png_ptr);
906 PNG files store RGB pixels packed into 3 or 6 bytes. This code expands them
907 into 4 or 8 bytes for windowing systems that need them in this format:
909 if (color_type == PNG_COLOR_TYPE_RGB)
910 png_set_filler(png_ptr, filler, PNG_FILLER_BEFORE);
912 where "filler" is the 8 or 16-bit number to fill with, and the location is
913 either PNG_FILLER_BEFORE or PNG_FILLER_AFTER, depending upon whether
914 you want the filler before the RGB or after. This transformation
915 does not affect images that already have full alpha channels. To add an
916 opaque alpha channel, use filler=0xff or 0xffff and PNG_FILLER_AFTER which
917 will generate RGBA pixels.
919 Note that png_set_filler() does not change the color type. If you want
920 to do that, you can add a true alpha channel with
922 if (color_type == PNG_COLOR_TYPE_RGB ||
923 color_type == PNG_COLOR_TYPE_GRAY)
924 png_set_add_alpha(png_ptr, filler, PNG_FILLER_AFTER);
926 where "filler" contains the alpha value to assign to each pixel.
927 This function was added in libpng-1.2.7.
929 If you are reading an image with an alpha channel, and you need the
930 data as ARGB instead of the normal PNG format RGBA:
932 if (color_type == PNG_COLOR_TYPE_RGB_ALPHA)
933 png_set_swap_alpha(png_ptr);
935 For some uses, you may want a grayscale image to be represented as
936 RGB. This code will do that conversion:
938 if (color_type == PNG_COLOR_TYPE_GRAY ||
939 color_type == PNG_COLOR_TYPE_GRAY_ALPHA)
940 png_set_gray_to_rgb(png_ptr);
942 Conversely, you can convert an RGB or RGBA image to grayscale or grayscale
945 if (color_type == PNG_COLOR_TYPE_RGB ||
946 color_type == PNG_COLOR_TYPE_RGB_ALPHA)
947 png_set_rgb_to_gray_fixed(png_ptr, error_action,
948 int red_weight, int green_weight);
950 error_action = 1: silently do the conversion
951 error_action = 2: issue a warning if the original
952 image has any pixel where
953 red != green or red != blue
954 error_action = 3: issue an error and abort the
955 conversion if the original
956 image has any pixel where
957 red != green or red != blue
959 red_weight: weight of red component times 100000
960 green_weight: weight of green component times 100000
961 If either weight is negative, default
962 weights (21268, 71514) are used.
964 If you have set error_action = 1 or 2, you can
965 later check whether the image really was gray, after processing
966 the image rows, with the png_get_rgb_to_gray_status(png_ptr) function.
967 It will return a png_byte that is zero if the image was gray or
968 1 if there were any non-gray pixels. bKGD and sBIT data
969 will be silently converted to grayscale, using the green channel
970 data, regardless of the error_action setting.
972 With red_weight+green_weight<=100000,
973 the normalized graylevel is computed:
975 int rw = red_weight * 65536;
976 int gw = green_weight * 65536;
977 int bw = 65536 - (rw + gw);
978 gray = (rw*red + gw*green + bw*blue)/65536;
980 The default values approximate those recommended in the Charles
981 Poynton's Color FAQ, <http://www.inforamp.net/~poynton/>
982 Copyright (c) 1998-01-04 Charles Poynton <poynton at inforamp.net>
984 Y = 0.212671 * R + 0.715160 * G + 0.072169 * B
986 Libpng approximates this with
988 Y = 0.21268 * R + 0.7151 * G + 0.07217 * B
990 which can be expressed with integers as
992 Y = (6969 * R + 23434 * G + 2365 * B)/32768
994 The calculation is done in a linear colorspace, if the image gamma
997 If you have a grayscale and you are using png_set_expand_depth(),
998 png_set_expand(), or png_set_gray_to_rgb to change to truecolor or to
999 a higher bit-depth, you must either supply the background color as a gray
1000 value at the original file bit-depth (need_expand = 1) or else supply the
1001 background color as an RGB triplet at the final, expanded bit depth
1002 (need_expand = 0). Similarly, if you are reading a paletted image, you
1003 must either supply the background color as a palette index (need_expand = 1)
1004 or as an RGB triplet that may or may not be in the palette (need_expand = 0).
1006 png_color_16 my_background;
1007 png_color_16p image_background;
1009 if (png_get_bKGD(png_ptr, info_ptr, &image_background))
1010 png_set_background(png_ptr, image_background,
1011 PNG_BACKGROUND_GAMMA_FILE, 1, 1.0);
1013 png_set_background(png_ptr, &my_background,
1014 PNG_BACKGROUND_GAMMA_SCREEN, 0, 1.0);
1016 The png_set_background() function tells libpng to composite images
1017 with alpha or simple transparency against the supplied background
1018 color. If the PNG file contains a bKGD chunk (PNG_INFO_bKGD valid),
1019 you may use this color, or supply another color more suitable for
1020 the current display (e.g., the background color from a web page). You
1021 need to tell libpng whether the color is in the gamma space of the
1022 display (PNG_BACKGROUND_GAMMA_SCREEN for colors you supply), the file
1023 (PNG_BACKGROUND_GAMMA_FILE for colors from the bKGD chunk), or one
1024 that is neither of these gammas (PNG_BACKGROUND_GAMMA_UNIQUE - I don't
1025 know why anyone would use this, but it's here).
1027 To properly display PNG images on any kind of system, the application needs
1028 to know what the display gamma is. Ideally, the user will know this, and
1029 the application will allow them to set it. One method of allowing the user
1030 to set the display gamma separately for each system is to check for a
1031 SCREEN_GAMMA or DISPLAY_GAMMA environment variable, which will hopefully be
1034 Note that display_gamma is the overall gamma correction required to produce
1035 pleasing results, which depends on the lighting conditions in the surrounding
1036 environment. In a dim or brightly lit room, no compensation other than
1037 the physical gamma exponent of the monitor is needed, while in a dark room
1038 a slightly smaller exponent is better.
1040 double gamma, screen_gamma;
1042 if (/* We have a user-defined screen
1045 screen_gamma = user_defined_screen_gamma;
1047 /* One way that applications can share the same
1048 screen gamma value */
1049 else if ((gamma_str = getenv("SCREEN_GAMMA"))
1052 screen_gamma = (double)atof(gamma_str);
1054 /* If we don't have another value */
1057 screen_gamma = 2.2; /* A good guess for a
1058 PC monitor in a bright office or a dim room */
1059 screen_gamma = 2.0; /* A good guess for a
1060 PC monitor in a dark room */
1061 screen_gamma = 1.7 or 1.0; /* A good
1062 guess for Mac systems */
1065 The png_set_gamma() function handles gamma transformations of the data.
1066 Pass both the file gamma and the current screen_gamma. If the file does
1067 not have a gamma value, you can pass one anyway if you have an idea what
1068 it is (usually 0.45455 is a good guess for GIF images on PCs). Note
1069 that file gammas are inverted from screen gammas. See the discussions
1070 on gamma in the PNG specification for an excellent description of what
1071 gamma is, and why all applications should support it. It is strongly
1072 recommended that PNG viewers support gamma correction.
1074 if (png_get_gAMA(png_ptr, info_ptr, &gamma))
1075 png_set_gamma(png_ptr, screen_gamma, gamma);
1077 png_set_gamma(png_ptr, screen_gamma, 0.45455);
1079 If you need to reduce an RGB file to a paletted file, or if a paletted
1080 file has more entries then will fit on your screen, png_set_dither()
1081 will do that. Note that this is a simple match dither that merely
1082 finds the closest color available. This should work fairly well with
1083 optimized palettes, and fairly badly with linear color cubes. If you
1084 pass a palette that is larger then maximum_colors, the file will
1085 reduce the number of colors in the palette so it will fit into
1086 maximum_colors. If there is a histogram, it will use it to make
1087 more intelligent choices when reducing the palette. If there is no
1088 histogram, it may not do as good a job.
1090 if (color_type & PNG_COLOR_MASK_COLOR)
1092 if (png_get_valid(png_ptr, info_ptr,
1095 png_uint_16p histogram = NULL;
1097 png_get_hIST(png_ptr, info_ptr,
1099 png_set_dither(png_ptr, palette, num_palette,
1100 max_screen_colors, histogram, 1);
1104 png_color std_color_cube[MAX_SCREEN_COLORS] =
1107 png_set_dither(png_ptr, std_color_cube,
1108 MAX_SCREEN_COLORS, MAX_SCREEN_COLORS,
1113 PNG files describe monochrome as black being zero and white being one.
1114 The following code will reverse this (make black be one and white be
1117 if (bit_depth == 1 && color_type == PNG_COLOR_TYPE_GRAY)
1118 png_set_invert_mono(png_ptr);
1120 This function can also be used to invert grayscale and gray-alpha images:
1122 if (color_type == PNG_COLOR_TYPE_GRAY ||
1123 color_type == PNG_COLOR_TYPE_GRAY_ALPHA)
1124 png_set_invert_mono(png_ptr);
1126 PNG files store 16 bit pixels in network byte order (big-endian,
1127 ie. most significant bits first). This code changes the storage to the
1128 other way (little-endian, i.e. least significant bits first, the
1129 way PCs store them):
1131 if (bit_depth == 16)
1132 png_set_swap(png_ptr);
1134 If you are using packed-pixel images (1, 2, or 4 bits/pixel), and you
1135 need to change the order the pixels are packed into bytes, you can use:
1138 png_set_packswap(png_ptr);
1140 Finally, you can write your own transformation function if none of
1141 the existing ones meets your needs. This is done by setting a callback
1144 png_set_read_user_transform_fn(png_ptr,
1147 You must supply the function
1149 void read_transform_fn(png_ptr ptr, row_info_ptr
1150 row_info, png_bytep data)
1152 See pngtest.c for a working example. Your function will be called
1153 after all of the other transformations have been processed.
1155 You can also set up a pointer to a user structure for use by your
1156 callback function, and you can inform libpng that your transform
1157 function will change the number of channels or bit depth with the
1160 png_set_user_transform_info(png_ptr, user_ptr,
1161 user_depth, user_channels);
1163 The user's application, not libpng, is responsible for allocating and
1164 freeing any memory required for the user structure.
1166 You can retrieve the pointer via the function
1167 png_get_user_transform_ptr(). For example:
1169 voidp read_user_transform_ptr =
1170 png_get_user_transform_ptr(png_ptr);
1172 The last thing to handle is interlacing; this is covered in detail below,
1173 but you must call the function here if you want libpng to handle expansion
1174 of the interlaced image.
1176 number_of_passes = png_set_interlace_handling(png_ptr);
1178 After setting the transformations, libpng can update your png_info
1179 structure to reflect any transformations you've requested with this
1180 call. This is most useful to update the info structure's rowbytes
1181 field so you can use it to allocate your image memory. This function
1182 will also update your palette with the correct screen_gamma and
1183 background if these have been given with the calls above.
1185 png_read_update_info(png_ptr, info_ptr);
1187 After you call png_read_update_info(), you can allocate any
1188 memory you need to hold the image. The row data is simply
1189 raw byte data for all forms of images. As the actual allocation
1190 varies among applications, no example will be given. If you
1191 are allocating one large chunk, you will need to build an
1192 array of pointers to each row, as it will be needed for some
1193 of the functions below.
1197 After you've allocated memory, you can read the image data.
1198 The simplest way to do this is in one function call. If you are
1199 allocating enough memory to hold the whole image, you can just
1200 call png_read_image() and libpng will read in all the image data
1201 and put it in the memory area supplied. You will need to pass in
1202 an array of pointers to each row.
1204 This function automatically handles interlacing, so you don't need
1205 to call png_set_interlace_handling() or call this function multiple
1206 times, or any of that other stuff necessary with png_read_rows().
1208 png_read_image(png_ptr, row_pointers);
1210 where row_pointers is:
1212 png_bytep row_pointers[height];
1214 You can point to void or char or whatever you use for pixels.
1216 If you don't want to read in the whole image at once, you can
1217 use png_read_rows() instead. If there is no interlacing (check
1218 interlace_type == PNG_INTERLACE_NONE), this is simple:
1220 png_read_rows(png_ptr, row_pointers, NULL,
1223 where row_pointers is the same as in the png_read_image() call.
1225 If you are doing this just one row at a time, you can do this with
1226 a single row_pointer instead of an array of row_pointers:
1228 png_bytep row_pointer = row;
1229 png_read_row(png_ptr, row_pointer, NULL);
1231 If the file is interlaced (interlace_type != 0 in the IHDR chunk), things
1232 get somewhat harder. The only current (PNG Specification version 1.2)
1233 interlacing type for PNG is (interlace_type == PNG_INTERLACE_ADAM7)
1234 is a somewhat complicated 2D interlace scheme, known as Adam7, that
1235 breaks down an image into seven smaller images of varying size, based
1238 libpng can fill out those images or it can give them to you "as is".
1239 If you want them filled out, there are two ways to do that. The one
1240 mentioned in the PNG specification is to expand each pixel to cover
1241 those pixels that have not been read yet (the "rectangle" method).
1242 This results in a blocky image for the first pass, which gradually
1243 smooths out as more pixels are read. The other method is the "sparkle"
1244 method, where pixels are drawn only in their final locations, with the
1245 rest of the image remaining whatever colors they were initialized to
1246 before the start of the read. The first method usually looks better,
1247 but tends to be slower, as there are more pixels to put in the rows.
1249 If you don't want libpng to handle the interlacing details, just call
1250 png_read_rows() seven times to read in all seven images. Each of the
1251 images is a valid image by itself, or they can all be combined on an
1252 8x8 grid to form a single image (although if you intend to combine them
1253 you would be far better off using the libpng interlace handling).
1255 The first pass will return an image 1/8 as wide as the entire image
1256 (every 8th column starting in column 0) and 1/8 as high as the original
1257 (every 8th row starting in row 0), the second will be 1/8 as wide
1258 (starting in column 4) and 1/8 as high (also starting in row 0). The
1259 third pass will be 1/4 as wide (every 4th pixel starting in column 0) and
1260 1/8 as high (every 8th row starting in row 4), and the fourth pass will
1261 be 1/4 as wide and 1/4 as high (every 4th column starting in column 2,
1262 and every 4th row starting in row 0). The fifth pass will return an
1263 image 1/2 as wide, and 1/4 as high (starting at column 0 and row 2),
1264 while the sixth pass will be 1/2 as wide and 1/2 as high as the original
1265 (starting in column 1 and row 0). The seventh and final pass will be as
1266 wide as the original, and 1/2 as high, containing all of the odd
1267 numbered scanlines. Phew!
1269 If you want libpng to expand the images, call this before calling
1270 png_start_read_image() or png_read_update_info():
1272 if (interlace_type == PNG_INTERLACE_ADAM7)
1274 = png_set_interlace_handling(png_ptr);
1276 This will return the number of passes needed. Currently, this
1277 is seven, but may change if another interlace type is added.
1278 This function can be called even if the file is not interlaced,
1279 where it will return one pass.
1281 If you are not going to display the image after each pass, but are
1282 going to wait until the entire image is read in, use the sparkle
1283 effect. This effect is faster and the end result of either method
1284 is exactly the same. If you are planning on displaying the image
1285 after each pass, the "rectangle" effect is generally considered the
1288 If you only want the "sparkle" effect, just call png_read_rows() as
1289 normal, with the third parameter NULL. Make sure you make pass over
1290 the image number_of_passes times, and you don't change the data in the
1291 rows between calls. You can change the locations of the data, just
1292 not the data. Each pass only writes the pixels appropriate for that
1293 pass, and assumes the data from previous passes is still valid.
1295 png_read_rows(png_ptr, row_pointers, NULL,
1298 If you only want the first effect (the rectangles), do the same as
1299 before except pass the row buffer in the third parameter, and leave
1300 the second parameter NULL.
1302 png_read_rows(png_ptr, NULL, row_pointers,
1305 Finishing a sequential read
1307 After you are finished reading the image through the
1308 low-level interface, you can finish reading the file. If you are
1309 interested in comments or time, which may be stored either before or
1310 after the image data, you should pass the separate png_info struct if
1311 you want to keep the comments from before and after the image
1312 separate. If you are not interested, you can pass NULL.
1314 png_read_end(png_ptr, end_info);
1316 When you are done, you can free all memory allocated by libpng like this:
1318 png_destroy_read_struct(&png_ptr, &info_ptr,
1321 It is also possible to individually free the info_ptr members that
1322 point to libpng-allocated storage with the following function:
1324 png_free_data(png_ptr, info_ptr, mask, seq)
1325 mask - identifies data to be freed, a mask
1326 containing the bitwise OR of one or
1328 PNG_FREE_PLTE, PNG_FREE_TRNS,
1329 PNG_FREE_HIST, PNG_FREE_ICCP,
1330 PNG_FREE_PCAL, PNG_FREE_ROWS,
1331 PNG_FREE_SCAL, PNG_FREE_SPLT,
1332 PNG_FREE_TEXT, PNG_FREE_UNKN,
1333 or simply PNG_FREE_ALL
1334 seq - sequence number of item to be freed
1337 This function may be safely called when the relevant storage has
1338 already been freed, or has not yet been allocated, or was allocated
1339 by the user and not by libpng, and will in those
1340 cases do nothing. The "seq" parameter is ignored if only one item
1341 of the selected data type, such as PLTE, is allowed. If "seq" is not
1342 -1, and multiple items are allowed for the data type identified in
1343 the mask, such as text or sPLT, only the n'th item in the structure
1344 is freed, where n is "seq".
1346 The default behavior is only to free data that was allocated internally
1347 by libpng. This can be changed, so that libpng will not free the data,
1348 or so that it will free data that was allocated by the user with png_malloc()
1349 or png_zalloc() and passed in via a png_set_*() function, with
1351 png_data_freer(png_ptr, info_ptr, freer, mask)
1352 mask - which data elements are affected
1353 same choices as in png_free_data()
1355 PNG_DESTROY_WILL_FREE_DATA
1356 PNG_SET_WILL_FREE_DATA
1357 PNG_USER_WILL_FREE_DATA
1359 This function only affects data that has already been allocated.
1360 You can call this function after reading the PNG data but before calling
1361 any png_set_*() functions, to control whether the user or the png_set_*()
1362 function is responsible for freeing any existing data that might be present,
1363 and again after the png_set_*() functions to control whether the user
1364 or png_destroy_*() is supposed to free the data. When the user assumes
1365 responsibility for libpng-allocated data, the application must use
1366 png_free() to free it, and when the user transfers responsibility to libpng
1367 for data that the user has allocated, the user must have used png_malloc()
1368 or png_zalloc() to allocate it.
1370 If you allocated your row_pointers in a single block, as suggested above in
1371 the description of the high level read interface, you must not transfer
1372 responsibility for freeing it to the png_set_rows or png_read_destroy function,
1373 because they would also try to free the individual row_pointers[i].
1375 If you allocated text_ptr.text, text_ptr.lang, and text_ptr.translated_keyword
1376 separately, do not transfer responsibility for freeing text_ptr to libpng,
1377 because when libpng fills a png_text structure it combines these members with
1378 the key member, and png_free_data() will free only text_ptr.key. Similarly,
1379 if you transfer responsibility for free'ing text_ptr from libpng to your
1380 application, your application must not separately free those members.
1382 The png_free_data() function will turn off the "valid" flag for anything
1383 it frees. If you need to turn the flag off for a chunk that was freed by your
1384 application instead of by libpng, you can use
1386 png_set_invalid(png_ptr, info_ptr, mask);
1387 mask - identifies the chunks to be made invalid,
1388 containing the bitwise OR of one or
1390 PNG_INFO_gAMA, PNG_INFO_sBIT,
1391 PNG_INFO_cHRM, PNG_INFO_PLTE,
1392 PNG_INFO_tRNS, PNG_INFO_bKGD,
1393 PNG_INFO_hIST, PNG_INFO_pHYs,
1394 PNG_INFO_oFFs, PNG_INFO_tIME,
1395 PNG_INFO_pCAL, PNG_INFO_sRGB,
1396 PNG_INFO_iCCP, PNG_INFO_sPLT,
1397 PNG_INFO_sCAL, PNG_INFO_IDAT
1399 For a more compact example of reading a PNG image, see the file example.c.
1401 Reading PNG files progressively
1403 The progressive reader is slightly different then the non-progressive
1404 reader. Instead of calling png_read_info(), png_read_rows(), and
1405 png_read_end(), you make one call to png_process_data(), which calls
1406 callbacks when it has the info, a row, or the end of the image. You
1407 set up these callbacks with png_set_progressive_read_fn(). You don't
1408 have to worry about the input/output functions of libpng, as you are
1409 giving the library the data directly in png_process_data(). I will
1410 assume that you have read the section on reading PNG files above,
1411 so I will only highlight the differences (although I will show
1414 png_structp png_ptr;
1417 /* An example code fragment of how you would
1418 initialize the progressive reader in your
1421 initialize_png_reader()
1423 png_ptr = png_create_read_struct
1424 (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
1425 user_error_fn, user_warning_fn);
1428 info_ptr = png_create_info_struct(png_ptr);
1431 png_destroy_read_struct(&png_ptr, (png_infopp)NULL,
1436 if (setjmp(png_jmpbuf(png_ptr)))
1438 png_destroy_read_struct(&png_ptr, &info_ptr,
1443 /* This one's new. You can provide functions
1444 to be called when the header info is valid,
1445 when each row is completed, and when the image
1446 is finished. If you aren't using all functions,
1447 you can specify NULL parameters. Even when all
1448 three functions are NULL, you need to call
1449 png_set_progressive_read_fn(). You can use
1450 any struct as the user_ptr (cast to a void pointer
1451 for the function call), and retrieve the pointer
1452 from inside the callbacks using the function
1454 png_get_progressive_ptr(png_ptr);
1456 which will return a void pointer, which you have
1457 to cast appropriately.
1459 png_set_progressive_read_fn(png_ptr, (void *)user_ptr,
1460 info_callback, row_callback, end_callback);
1465 /* A code fragment that you call as you receive blocks
1468 process_data(png_bytep buffer, png_uint_32 length)
1470 if (setjmp(png_jmpbuf(png_ptr)))
1472 png_destroy_read_struct(&png_ptr, &info_ptr,
1477 /* This one's new also. Simply give it a chunk
1478 of data from the file stream (in order, of
1479 course). On machines with segmented memory
1480 models machines, don't give it any more than
1481 64K. The library seems to run fine with sizes
1482 of 4K. Although you can give it much less if
1483 necessary (I assume you can give it chunks of
1484 1 byte, I haven't tried less then 256 bytes
1485 yet). When this function returns, you may
1486 want to display any rows that were generated
1487 in the row callback if you don't already do
1490 png_process_data(png_ptr, info_ptr, buffer, length);
1494 /* This function is called (as set by
1495 png_set_progressive_read_fn() above) when enough data
1496 has been supplied so all of the header has been
1500 info_callback(png_structp png_ptr, png_infop info)
1502 /* Do any setup here, including setting any of
1503 the transformations mentioned in the Reading
1504 PNG files section. For now, you _must_ call
1505 either png_start_read_image() or
1506 png_read_update_info() after all the
1507 transformations are set (even if you don't set
1508 any). You may start getting rows before
1509 png_process_data() returns, so this is your
1510 last chance to prepare for that.
1514 /* This function is called when each row of image
1517 row_callback(png_structp png_ptr, png_bytep new_row,
1518 png_uint_32 row_num, int pass)
1520 /* If the image is interlaced, and you turned
1521 on the interlace handler, this function will
1522 be called for every row in every pass. Some
1523 of these rows will not be changed from the
1524 previous pass. When the row is not changed,
1525 the new_row variable will be NULL. The rows
1526 and passes are called in order, so you don't
1527 really need the row_num and pass, but I'm
1528 supplying them because it may make your life
1531 For the non-NULL rows of interlaced images,
1532 you must call png_progressive_combine_row()
1533 passing in the row and the old row. You can
1534 call this function for NULL rows (it will just
1535 return) and for non-interlaced images (it just
1536 does the memcpy for you) if it will make the
1537 code easier. Thus, you can just do this for
1541 png_progressive_combine_row(png_ptr, old_row,
1544 /* where old_row is what was displayed for
1545 previously for the row. Note that the first
1546 pass (pass == 0, really) will completely cover
1547 the old row, so the rows do not have to be
1548 initialized. After the first pass (and only
1549 for interlaced images), you will have to pass
1550 the current row, and the function will combine
1551 the old row and the new row.
1556 end_callback(png_structp png_ptr, png_infop info)
1558 /* This function is called after the whole image
1559 has been read, including any chunks after the
1560 image (up to and including the IEND). You
1561 will usually have the same info chunk as you
1562 had in the header, although some data may have
1563 been added to the comments and time fields.
1565 Most people won't do much here, perhaps setting
1566 a flag that marks the image as finished.
1574 Much of this is very similar to reading. However, everything of
1575 importance is repeated here, so you won't have to constantly look
1576 back up in the reading section to understand writing.
1580 You will want to do the I/O initialization before you get into libpng,
1581 so if it doesn't work, you don't have anything to undo. If you are not
1582 using the standard I/O functions, you will need to replace them with
1583 custom writing functions. See the discussion under Customizing libpng.
1585 FILE *fp = fopen(file_name, "wb");
1591 Next, png_struct and png_info need to be allocated and initialized.
1592 As these can be both relatively large, you may not want to store these
1593 on the stack, unless you have stack space to spare. Of course, you
1594 will want to check if they return NULL. If you are also reading,
1595 you won't want to name your read structure and your write structure
1596 both "png_ptr"; you can call them anything you like, such as
1597 "read_ptr" and "write_ptr". Look at pngtest.c, for example.
1599 png_structp png_ptr = png_create_write_struct
1600 (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
1601 user_error_fn, user_warning_fn);
1605 png_infop info_ptr = png_create_info_struct(png_ptr);
1608 png_destroy_write_struct(&png_ptr,
1613 If you want to use your own memory allocation routines,
1614 define PNG_USER_MEM_SUPPORTED and use
1615 png_create_write_struct_2() instead of png_create_write_struct():
1617 png_structp png_ptr = png_create_write_struct_2
1618 (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
1619 user_error_fn, user_warning_fn, (png_voidp)
1620 user_mem_ptr, user_malloc_fn, user_free_fn);
1622 After you have these structures, you will need to set up the
1623 error handling. When libpng encounters an error, it expects to
1624 longjmp() back to your routine. Therefore, you will need to call
1625 setjmp() and pass the png_jmpbuf(png_ptr). If you
1626 write the file from different routines, you will need to update
1627 the png_jmpbuf(png_ptr) every time you enter a new routine that will
1628 call a png_*() function. See your documentation of setjmp/longjmp
1629 for your compiler for more information on setjmp/longjmp. See
1630 the discussion on libpng error handling in the Customizing Libpng
1631 section below for more information on the libpng error handling.
1633 if (setjmp(png_jmpbuf(png_ptr)))
1635 png_destroy_write_struct(&png_ptr, &info_ptr);
1642 If you would rather avoid the complexity of setjmp/longjmp issues,
1643 you can compile libpng with PNG_SETJMP_NOT_SUPPORTED, in which case
1644 errors will result in a call to PNG_ABORT() which defaults to abort().
1646 Now you need to set up the output code. The default for libpng is to
1647 use the C function fwrite(). If you use this, you will need to pass a
1648 valid FILE * in the function png_init_io(). Be sure that the file is
1649 opened in binary mode. Again, if you wish to handle writing data in
1650 another way, see the discussion on libpng I/O handling in the Customizing
1651 Libpng section below.
1653 png_init_io(png_ptr, fp);
1655 If you are embedding your PNG into a datastream such as MNG, and don't
1656 want libpng to write the 8-byte signature, or if you have already
1657 written the signature in your application, use
1659 png_set_sig_bytes(png_ptr, 8);
1661 to inform libpng that it should not write a signature.
1665 At this point, you can set up a callback function that will be
1666 called after each row has been written, which you can use to control
1667 a progress meter or the like. It's demonstrated in pngtest.c.
1668 You must supply a function
1670 void write_row_callback(png_ptr, png_uint_32 row,
1673 /* put your code here */
1676 (You can give it another name that you like instead of "write_row_callback")
1678 To inform libpng about your function, use
1680 png_set_write_status_fn(png_ptr, write_row_callback);
1682 You now have the option of modifying how the compression library will
1683 run. The following functions are mainly for testing, but may be useful
1684 in some cases, like if you need to write PNG files extremely fast and
1685 are willing to give up some compression, or if you want to get the
1686 maximum possible compression at the expense of slower writing. If you
1687 have no special needs in this area, let the library do what it wants by
1688 not calling this function at all, as it has been tuned to deliver a good
1689 speed/compression ratio. The second parameter to png_set_filter() is
1690 the filter method, for which the only valid values are 0 (as of the
1691 July 1999 PNG specification, version 1.2) or 64 (if you are writing
1692 a PNG datastream that is to be embedded in a MNG datastream). The third
1693 parameter is a flag that indicates which filter type(s) are to be tested
1694 for each scanline. See the PNG specification for details on the specific filter
1698 /* turn on or off filtering, and/or choose
1699 specific filters. You can use either a single
1700 PNG_FILTER_VALUE_NAME or the bitwise OR of one
1701 or more PNG_FILTER_NAME masks. */
1702 png_set_filter(png_ptr, 0,
1703 PNG_FILTER_NONE | PNG_FILTER_VALUE_NONE |
1704 PNG_FILTER_SUB | PNG_FILTER_VALUE_SUB |
1705 PNG_FILTER_UP | PNG_FILTER_VALUE_UP |
1706 PNG_FILTER_AVG | PNG_FILTER_VALUE_AVG |
1707 PNG_FILTER_PAETH | PNG_FILTER_VALUE_PAETH|
1711 wants to start and stop using particular filters during compression,
1712 it should start out with all of the filters (to ensure that the previous
1713 row of pixels will be stored in case it's needed later), and then add
1714 and remove them after the start of compression.
1716 If you are writing a PNG datastream that is to be embedded in a MNG
1717 datastream, the second parameter can be either 0 or 64.
1719 The png_set_compression_*() functions interface to the zlib compression
1720 library, and should mostly be ignored unless you really know what you are
1721 doing. The only generally useful call is png_set_compression_level()
1722 which changes how much time zlib spends on trying to compress the image
1723 data. See the Compression Library (zlib.h and algorithm.txt, distributed
1724 with zlib) for details on the compression levels.
1726 /* set the zlib compression level */
1727 png_set_compression_level(png_ptr,
1728 Z_BEST_COMPRESSION);
1730 /* set other zlib parameters */
1731 png_set_compression_mem_level(png_ptr, 8);
1732 png_set_compression_strategy(png_ptr,
1733 Z_DEFAULT_STRATEGY);
1734 png_set_compression_window_bits(png_ptr, 15);
1735 png_set_compression_method(png_ptr, 8);
1736 png_set_compression_buffer_size(png_ptr, 8192)
1738 extern PNG_EXPORT(void,png_set_zbuf_size)
1740 Setting the contents of info for output
1742 You now need to fill in the png_info structure with all the data you
1743 wish to write before the actual image. Note that the only thing you
1744 are allowed to write after the image is the text chunks and the time
1745 chunk (as of PNG Specification 1.2, anyway). See png_write_end() and
1746 the latest PNG specification for more information on that. If you
1747 wish to write them before the image, fill them in now, and flag that
1748 data as being valid. If you want to wait until after the data, don't
1749 fill them until png_write_end(). For all the fields in png_info and
1750 their data types, see png.h. For explanations of what the fields
1751 contain, see the PNG specification.
1753 Some of the more important parts of the png_info are:
1755 png_set_IHDR(png_ptr, info_ptr, width, height,
1756 bit_depth, color_type, interlace_type,
1757 compression_type, filter_method)
1758 width - holds the width of the image
1759 in pixels (up to 2^31).
1760 height - holds the height of the image
1761 in pixels (up to 2^31).
1762 bit_depth - holds the bit depth of one of the
1764 (valid values are 1, 2, 4, 8, 16
1765 and depend also on the
1766 color_type. See also significant
1768 color_type - describes which color/alpha
1769 channels are present.
1771 (bit depths 1, 2, 4, 8, 16)
1772 PNG_COLOR_TYPE_GRAY_ALPHA
1774 PNG_COLOR_TYPE_PALETTE
1775 (bit depths 1, 2, 4, 8)
1778 PNG_COLOR_TYPE_RGB_ALPHA
1781 PNG_COLOR_MASK_PALETTE
1782 PNG_COLOR_MASK_COLOR
1783 PNG_COLOR_MASK_ALPHA
1785 interlace_type - PNG_INTERLACE_NONE or
1787 compression_type - (must be
1788 PNG_COMPRESSION_TYPE_DEFAULT)
1789 filter_method - (must be PNG_FILTER_TYPE_DEFAULT
1790 or, if you are writing a PNG to
1791 be embedded in a MNG datastream,
1793 PNG_INTRAPIXEL_DIFFERENCING)
1795 If you call png_set_IHDR(), the call must appear before any of the
1796 other png_set_*() functions, because they might require access to some of
1797 the IHDR settings. The remaining png_set_*() functions can be called
1800 If you wish, you can reset the compression_type, interlace_type, or
1801 filter_method later by calling png_set_IHDR() again; if you do this, the
1802 width, height, bit_depth, and color_type must be the same in each call.
1804 png_set_PLTE(png_ptr, info_ptr, palette,
1806 palette - the palette for the file
1807 (array of png_color)
1808 num_palette - number of entries in the palette
1810 png_set_gAMA(png_ptr, info_ptr, gamma);
1811 gamma - the gamma the image was created
1814 png_set_sRGB(png_ptr, info_ptr, srgb_intent);
1815 srgb_intent - the rendering intent
1816 (PNG_INFO_sRGB) The presence of
1817 the sRGB chunk means that the pixel
1818 data is in the sRGB color space.
1819 This chunk also implies specific
1820 values of gAMA and cHRM. Rendering
1821 intent is the CSS-1 property that
1822 has been defined by the International
1824 (http://www.color.org).
1826 PNG_sRGB_INTENT_SATURATION,
1827 PNG_sRGB_INTENT_PERCEPTUAL,
1828 PNG_sRGB_INTENT_ABSOLUTE, or
1829 PNG_sRGB_INTENT_RELATIVE.
1832 png_set_sRGB_gAMA_and_cHRM(png_ptr, info_ptr,
1834 srgb_intent - the rendering intent
1835 (PNG_INFO_sRGB) The presence of the
1836 sRGB chunk means that the pixel
1837 data is in the sRGB color space.
1838 This function also causes gAMA and
1839 cHRM chunks with the specific values
1840 that are consistent with sRGB to be
1843 png_set_iCCP(png_ptr, info_ptr, name, compression_type,
1845 name - The profile name.
1846 compression - The compression type; always
1847 PNG_COMPRESSION_TYPE_BASE for PNG 1.0.
1848 You may give NULL to this argument to
1850 profile - International Color Consortium color
1851 profile data. May contain NULs.
1852 proflen - length of profile data in bytes.
1854 png_set_sBIT(png_ptr, info_ptr, sig_bit);
1855 sig_bit - the number of significant bits for
1856 (PNG_INFO_sBIT) each of the gray, red,
1857 green, and blue channels, whichever are
1858 appropriate for the given color type
1861 png_set_tRNS(png_ptr, info_ptr, trans, num_trans,
1863 trans - array of transparent entries for
1864 palette (PNG_INFO_tRNS)
1865 trans_values - graylevel or color sample values
1866 (in order red, green, blue) of the
1867 single transparent color for
1868 non-paletted images (PNG_INFO_tRNS)
1869 num_trans - number of transparent entries
1872 png_set_hIST(png_ptr, info_ptr, hist);
1874 hist - histogram of palette (array of
1877 png_set_tIME(png_ptr, info_ptr, mod_time);
1878 mod_time - time image was last modified
1881 png_set_bKGD(png_ptr, info_ptr, background);
1882 background - background color (PNG_VALID_bKGD)
1884 png_set_text(png_ptr, info_ptr, text_ptr, num_text);
1885 text_ptr - array of png_text holding image
1887 text_ptr[i].compression - type of compression used
1888 on "text" PNG_TEXT_COMPRESSION_NONE
1889 PNG_TEXT_COMPRESSION_zTXt
1890 PNG_ITXT_COMPRESSION_NONE
1891 PNG_ITXT_COMPRESSION_zTXt
1892 text_ptr[i].key - keyword for comment. Must contain
1894 text_ptr[i].text - text comments for current
1895 keyword. Can be NULL or empty.
1896 text_ptr[i].text_length - length of text string,
1897 after decompression, 0 for iTXt
1898 text_ptr[i].itxt_length - length of itxt string,
1899 after decompression, 0 for tEXt/zTXt
1900 text_ptr[i].lang - language of comment (NULL or
1902 text_ptr[i].translated_keyword - keyword in UTF-8 (NULL
1903 or empty for unknown).
1904 num_text - number of comments
1906 png_set_sPLT(png_ptr, info_ptr, &palette_ptr,
1908 palette_ptr - array of png_sPLT_struct structures
1909 to be added to the list of palettes
1910 in the info structure.
1911 num_spalettes - number of palette structures to be
1914 png_set_oFFs(png_ptr, info_ptr, offset_x, offset_y,
1916 offset_x - positive offset from the left
1918 offset_y - positive offset from the top
1920 unit_type - PNG_OFFSET_PIXEL, PNG_OFFSET_MICROMETER
1922 png_set_pHYs(png_ptr, info_ptr, res_x, res_y,
1924 res_x - pixels/unit physical resolution
1926 res_y - pixels/unit physical resolution
1928 unit_type - PNG_RESOLUTION_UNKNOWN,
1929 PNG_RESOLUTION_METER
1931 png_set_sCAL(png_ptr, info_ptr, unit, width, height)
1932 unit - physical scale units (an integer)
1933 width - width of a pixel in physical scale units
1934 height - height of a pixel in physical scale units
1935 (width and height are doubles)
1937 png_set_sCAL_s(png_ptr, info_ptr, unit, width, height)
1938 unit - physical scale units (an integer)
1939 width - width of a pixel in physical scale units
1940 height - height of a pixel in physical scale units
1941 (width and height are strings like "2.54")
1943 png_set_unknown_chunks(png_ptr, info_ptr, &unknowns,
1945 unknowns - array of png_unknown_chunk
1946 structures holding unknown chunks
1947 unknowns[i].name - name of unknown chunk
1948 unknowns[i].data - data of unknown chunk
1949 unknowns[i].size - size of unknown chunk's data
1950 unknowns[i].location - position to write chunk in file
1951 0: do not write chunk
1952 PNG_HAVE_IHDR: before PLTE
1953 PNG_HAVE_PLTE: before IDAT
1954 PNG_AFTER_IDAT: after IDAT
1956 The "location" member is set automatically according to
1957 what part of the output file has already been written.
1958 You can change its value after calling png_set_unknown_chunks()
1959 as demonstrated in pngtest.c. Within each of the "locations",
1960 the chunks are sequenced according to their position in the
1961 structure (that is, the value of "i", which is the order in which
1962 the chunk was either read from the input file or defined with
1963 png_set_unknown_chunks).
1965 A quick word about text and num_text. text is an array of png_text
1966 structures. num_text is the number of valid structures in the array.
1967 Each png_text structure holds a language code, a keyword, a text value,
1968 and a compression type.
1970 The compression types have the same valid numbers as the compression
1971 types of the image data. Currently, the only valid number is zero.
1972 However, you can store text either compressed or uncompressed, unlike
1973 images, which always have to be compressed. So if you don't want the
1974 text compressed, set the compression type to PNG_TEXT_COMPRESSION_NONE.
1975 Because tEXt and zTXt chunks don't have a language field, if you
1976 specify PNG_TEXT_COMPRESSION_NONE or PNG_TEXT_COMPRESSION_zTXt
1977 any language code or translated keyword will not be written out.
1979 Until text gets around 1000 bytes, it is not worth compressing it.
1980 After the text has been written out to the file, the compression type
1981 is set to PNG_TEXT_COMPRESSION_NONE_WR or PNG_TEXT_COMPRESSION_zTXt_WR,
1982 so that it isn't written out again at the end (in case you are calling
1983 png_write_end() with the same struct.
1985 The keywords that are given in the PNG Specification are:
1987 Title Short (one line) title or
1989 Author Name of image's creator
1990 Description Description of image (possibly long)
1991 Copyright Copyright notice
1992 Creation Time Time of original image creation
1993 (usually RFC 1123 format, see below)
1994 Software Software used to create the image
1995 Disclaimer Legal disclaimer
1996 Warning Warning of nature of content
1997 Source Device used to create the image
1998 Comment Miscellaneous comment; conversion
1999 from other image format
2001 The keyword-text pairs work like this. Keywords should be short
2002 simple descriptions of what the comment is about. Some typical
2003 keywords are found in the PNG specification, as is some recommendations
2004 on keywords. You can repeat keywords in a file. You can even write
2005 some text before the image and some after. For example, you may want
2006 to put a description of the image before the image, but leave the
2007 disclaimer until after, so viewers working over modem connections
2008 don't have to wait for the disclaimer to go over the modem before
2009 they start seeing the image. Finally, keywords should be full
2010 words, not abbreviations. Keywords and text are in the ISO 8859-1
2011 (Latin-1) character set (a superset of regular ASCII) and can not
2012 contain NUL characters, and should not contain control or other
2013 unprintable characters. To make the comments widely readable, stick
2014 with basic ASCII, and avoid machine specific character set extensions
2015 like the IBM-PC character set. The keyword must be present, but
2016 you can leave off the text string on non-compressed pairs.
2017 Compressed pairs must have a text string, as only the text string
2018 is compressed anyway, so the compression would be meaningless.
2020 PNG supports modification time via the png_time structure. Two
2021 conversion routines are provided, png_convert_from_time_t() for
2022 time_t and png_convert_from_struct_tm() for struct tm. The
2023 time_t routine uses gmtime(). You don't have to use either of
2024 these, but if you wish to fill in the png_time structure directly,
2025 you should provide the time in universal time (GMT) if possible
2026 instead of your local time. Note that the year number is the full
2027 year (e.g. 1998, rather than 98 - PNG is year 2000 compliant!), and
2028 that months start with 1.
2030 If you want to store the time of the original image creation, you should
2031 use a plain tEXt chunk with the "Creation Time" keyword. This is
2032 necessary because the "creation time" of a PNG image is somewhat vague,
2033 depending on whether you mean the PNG file, the time the image was
2034 created in a non-PNG format, a still photo from which the image was
2035 scanned, or possibly the subject matter itself. In order to facilitate
2036 machine-readable dates, it is recommended that the "Creation Time"
2037 tEXt chunk use RFC 1123 format dates (e.g. "22 May 1997 18:07:10 GMT"),
2038 although this isn't a requirement. Unlike the tIME chunk, the
2039 "Creation Time" tEXt chunk is not expected to be automatically changed
2040 by the software. To facilitate the use of RFC 1123 dates, a function
2041 png_convert_to_rfc1123(png_timep) is provided to convert from PNG
2042 time to an RFC 1123 format string.
2044 Writing unknown chunks
2046 You can use the png_set_unknown_chunks function to queue up chunks
2047 for writing. You give it a chunk name, raw data, and a size; that's
2048 all there is to it. The chunks will be written by the next following
2049 png_write_info_before_PLTE, png_write_info, or png_write_end function.
2050 Any chunks previously read into the info structure's unknown-chunk
2051 list will also be written out in a sequence that satisfies the PNG
2052 specification's ordering rules.
2054 The high-level write interface
2056 At this point there are two ways to proceed; through the high-level
2057 write interface, or through a sequence of low-level write operations.
2058 You can use the high-level interface if your image data is present
2059 in the info structure. All defined output
2060 transformations are permitted, enabled by the following masks.
2062 PNG_TRANSFORM_IDENTITY No transformation
2063 PNG_TRANSFORM_PACKING Pack 1, 2 and 4-bit samples
2064 PNG_TRANSFORM_PACKSWAP Change order of packed
2066 PNG_TRANSFORM_INVERT_MONO Invert monochrome images
2067 PNG_TRANSFORM_SHIFT Normalize pixels to the
2069 PNG_TRANSFORM_BGR Flip RGB to BGR, RGBA
2071 PNG_TRANSFORM_SWAP_ALPHA Flip RGBA to ARGB or GA
2073 PNG_TRANSFORM_INVERT_ALPHA Change alpha from opacity
2075 PNG_TRANSFORM_SWAP_ENDIAN Byte-swap 16-bit samples
2076 PNG_TRANSFORM_STRIP_FILLER Strip out filler
2078 PNG_TRANSFORM_STRIP_FILLER_BEFORE Strip out leading
2080 PNG_TRANSFORM_STRIP_FILLER_AFTER Strip out trailing
2083 If you have valid image data in the info structure (you can use
2084 png_set_rows() to put image data in the info structure), simply do this:
2086 png_write_png(png_ptr, info_ptr, png_transforms, NULL)
2088 where png_transforms is an integer containing the bitwise OR of some set of
2089 transformation flags. This call is equivalent to png_write_info(),
2090 followed the set of transformations indicated by the transform mask,
2091 then png_write_image(), and finally png_write_end().
2093 (The final parameter of this call is not yet used. Someday it might point
2094 to transformation parameters required by some future output transform.)
2096 You must use png_transforms and not call any png_set_transform() functions
2097 when you use png_write_png().
2099 The low-level write interface
2101 If you are going the low-level route instead, you are now ready to
2102 write all the file information up to the actual image data. You do
2103 this with a call to png_write_info().
2105 png_write_info(png_ptr, info_ptr);
2107 Note that there is one transformation you may need to do before
2108 png_write_info(). In PNG files, the alpha channel in an image is the
2109 level of opacity. If your data is supplied as a level of
2110 transparency, you can invert the alpha channel before you write it, so
2111 that 0 is fully transparent and 255 (in 8-bit or paletted images) or
2112 65535 (in 16-bit images) is fully opaque, with
2114 png_set_invert_alpha(png_ptr);
2116 This must appear before png_write_info() instead of later with the
2117 other transformations because in the case of paletted images the tRNS
2118 chunk data has to be inverted before the tRNS chunk is written. If
2119 your image is not a paletted image, the tRNS data (which in such cases
2120 represents a single color to be rendered as transparent) won't need to
2121 be changed, and you can safely do this transformation after your
2122 png_write_info() call.
2124 If you need to write a private chunk that you want to appear before
2125 the PLTE chunk when PLTE is present, you can write the PNG info in
2126 two steps, and insert code to write your own chunk between them:
2128 png_write_info_before_PLTE(png_ptr, info_ptr);
2129 png_set_unknown_chunks(png_ptr, info_ptr, ...);
2130 png_write_info(png_ptr, info_ptr);
2132 After you've written the file information, you can set up the library
2133 to handle any special transformations of the image data. The various
2134 ways to transform the data will be described in the order that they
2135 should occur. This is important, as some of these change the color
2136 type and/or bit depth of the data, and some others only work on
2137 certain color types and bit depths. Even though each transformation
2138 checks to see if it has data that it can do something with, you should
2139 make sure to only enable a transformation if it will be valid for the
2140 data. For example, don't swap red and blue on grayscale data.
2142 PNG files store RGB pixels packed into 3 or 6 bytes. This code tells
2143 the library to strip input data that has 4 or 8 bytes per pixel down
2144 to 3 or 6 bytes (or strip 2 or 4-byte grayscale+filler data to 1 or 2
2147 png_set_filler(png_ptr, 0, PNG_FILLER_BEFORE);
2149 where the 0 is unused, and the location is either PNG_FILLER_BEFORE or
2150 PNG_FILLER_AFTER, depending upon whether the filler byte in the pixel
2151 is stored XRGB or RGBX.
2153 PNG files pack pixels of bit depths 1, 2, and 4 into bytes as small as
2154 they can, resulting in, for example, 8 pixels per byte for 1 bit files.
2155 If the data is supplied at 1 pixel per byte, use this code, which will
2156 correctly pack the pixels into a single byte:
2158 png_set_packing(png_ptr);
2160 PNG files reduce possible bit depths to 1, 2, 4, 8, and 16. If your
2161 data is of another bit depth, you can write an sBIT chunk into the
2162 file so that decoders can recover the original data if desired.
2164 /* Set the true bit depth of the image data */
2165 if (color_type & PNG_COLOR_MASK_COLOR)
2167 sig_bit.red = true_bit_depth;
2168 sig_bit.green = true_bit_depth;
2169 sig_bit.blue = true_bit_depth;
2173 sig_bit.gray = true_bit_depth;
2175 if (color_type & PNG_COLOR_MASK_ALPHA)
2177 sig_bit.alpha = true_bit_depth;
2180 png_set_sBIT(png_ptr, info_ptr, &sig_bit);
2182 If the data is stored in the row buffer in a bit depth other than
2183 one supported by PNG (e.g. 3 bit data in the range 0-7 for a 4-bit PNG),
2184 this will scale the values to appear to be the correct bit depth as
2187 png_set_shift(png_ptr, &sig_bit);
2189 PNG files store 16 bit pixels in network byte order (big-endian,
2190 ie. most significant bits first). This code would be used if they are
2191 supplied the other way (little-endian, i.e. least significant bits
2192 first, the way PCs store them):
2195 png_set_swap(png_ptr);
2197 If you are using packed-pixel images (1, 2, or 4 bits/pixel), and you
2198 need to change the order the pixels are packed into bytes, you can use:
2201 png_set_packswap(png_ptr);
2203 PNG files store 3 color pixels in red, green, blue order. This code
2204 would be used if they are supplied as blue, green, red:
2206 png_set_bgr(png_ptr);
2208 PNG files describe monochrome as black being zero and white being
2209 one. This code would be used if the pixels are supplied with this reversed
2210 (black being one and white being zero):
2212 png_set_invert_mono(png_ptr);
2214 Finally, you can write your own transformation function if none of
2215 the existing ones meets your needs. This is done by setting a callback
2218 png_set_write_user_transform_fn(png_ptr,
2219 write_transform_fn);
2221 You must supply the function
2223 void write_transform_fn(png_ptr ptr, row_info_ptr
2224 row_info, png_bytep data)
2226 See pngtest.c for a working example. Your function will be called
2227 before any of the other transformations are processed.
2229 You can also set up a pointer to a user structure for use by your
2232 png_set_user_transform_info(png_ptr, user_ptr, 0, 0);
2234 The user_channels and user_depth parameters of this function are ignored
2235 when writing; you can set them to zero as shown.
2237 You can retrieve the pointer via the function png_get_user_transform_ptr().
2240 voidp write_user_transform_ptr =
2241 png_get_user_transform_ptr(png_ptr);
2243 It is possible to have libpng flush any pending output, either manually,
2244 or automatically after a certain number of lines have been written. To
2245 flush the output stream a single time call:
2247 png_write_flush(png_ptr);
2249 and to have libpng flush the output stream periodically after a certain
2250 number of scanlines have been written, call:
2252 png_set_flush(png_ptr, nrows);
2254 Note that the distance between rows is from the last time png_write_flush()
2255 was called, or the first row of the image if it has never been called.
2256 So if you write 50 lines, and then png_set_flush 25, it will flush the
2257 output on the next scanline, and every 25 lines thereafter, unless
2258 png_write_flush() is called before 25 more lines have been written.
2259 If nrows is too small (less than about 10 lines for a 640 pixel wide
2260 RGB image) the image compression may decrease noticeably (although this
2261 may be acceptable for real-time applications). Infrequent flushing will
2262 only degrade the compression performance by a few percent over images
2263 that do not use flushing.
2265 Writing the image data
2267 That's it for the transformations. Now you can write the image data.
2268 The simplest way to do this is in one function call. If you have the
2269 whole image in memory, you can just call png_write_image() and libpng
2270 will write the image. You will need to pass in an array of pointers to
2271 each row. This function automatically handles interlacing, so you don't
2272 need to call png_set_interlace_handling() or call this function multiple
2273 times, or any of that other stuff necessary with png_write_rows().
2275 png_write_image(png_ptr, row_pointers);
2277 where row_pointers is:
2279 png_byte *row_pointers[height];
2281 You can point to void or char or whatever you use for pixels.
2283 If you don't want to write the whole image at once, you can
2284 use png_write_rows() instead. If the file is not interlaced,
2287 png_write_rows(png_ptr, row_pointers,
2290 row_pointers is the same as in the png_write_image() call.
2292 If you are just writing one row at a time, you can do this with
2293 a single row_pointer instead of an array of row_pointers:
2295 png_bytep row_pointer = row;
2297 png_write_row(png_ptr, row_pointer);
2299 When the file is interlaced, things can get a good deal more
2300 complicated. The only currently (as of the PNG Specification
2301 version 1.2, dated July 1999) defined interlacing scheme for PNG files
2302 is the "Adam7" interlace scheme, that breaks down an
2303 image into seven smaller images of varying size. libpng will build
2304 these images for you, or you can do them yourself. If you want to
2305 build them yourself, see the PNG specification for details of which
2306 pixels to write when.
2308 If you don't want libpng to handle the interlacing details, just
2309 use png_set_interlace_handling() and call png_write_rows() the
2310 correct number of times to write all seven sub-images.
2312 If you want libpng to build the sub-images, call this before you start
2316 png_set_interlace_handling(png_ptr);
2318 This will return the number of passes needed. Currently, this
2319 is seven, but may change if another interlace type is added.
2321 Then write the complete image number_of_passes times.
2323 png_write_rows(png_ptr, row_pointers,
2326 As some of these rows are not used, and thus return immediately,
2327 you may want to read about interlacing in the PNG specification,
2328 and only update the rows that are actually used.
2330 Finishing a sequential write
2332 After you are finished writing the image, you should finish writing
2333 the file. If you are interested in writing comments or time, you should
2334 pass an appropriately filled png_info pointer. If you are not interested,
2337 png_write_end(png_ptr, info_ptr);
2339 When you are done, you can free all memory used by libpng like this:
2341 png_destroy_write_struct(&png_ptr, &info_ptr);
2343 It is also possible to individually free the info_ptr members that
2344 point to libpng-allocated storage with the following function:
2346 png_free_data(png_ptr, info_ptr, mask, seq)
2347 mask - identifies data to be freed, a mask
2348 containing the bitwise OR of one or
2350 PNG_FREE_PLTE, PNG_FREE_TRNS,
2351 PNG_FREE_HIST, PNG_FREE_ICCP,
2352 PNG_FREE_PCAL, PNG_FREE_ROWS,
2353 PNG_FREE_SCAL, PNG_FREE_SPLT,
2354 PNG_FREE_TEXT, PNG_FREE_UNKN,
2355 or simply PNG_FREE_ALL
2356 seq - sequence number of item to be freed
2359 This function may be safely called when the relevant storage has
2360 already been freed, or has not yet been allocated, or was allocated
2361 by the user and not by libpng, and will in those
2362 cases do nothing. The "seq" parameter is ignored if only one item
2363 of the selected data type, such as PLTE, is allowed. If "seq" is not
2364 -1, and multiple items are allowed for the data type identified in
2365 the mask, such as text or sPLT, only the n'th item in the structure
2366 is freed, where n is "seq".
2368 If you allocated data such as a palette that you passed
2369 in to libpng with png_set_*, you must not free it until just before the call to
2370 png_destroy_write_struct().
2372 The default behavior is only to free data that was allocated internally
2373 by libpng. This can be changed, so that libpng will not free the data,
2374 or so that it will free data that was allocated by the user with png_malloc()
2375 or png_zalloc() and passed in via a png_set_*() function, with
2377 png_data_freer(png_ptr, info_ptr, freer, mask)
2378 mask - which data elements are affected
2379 same choices as in png_free_data()
2381 PNG_DESTROY_WILL_FREE_DATA
2382 PNG_SET_WILL_FREE_DATA
2383 PNG_USER_WILL_FREE_DATA
2385 For example, to transfer responsibility for some data from a read structure
2386 to a write structure, you could use
2388 png_data_freer(read_ptr, read_info_ptr,
2389 PNG_USER_WILL_FREE_DATA,
2390 PNG_FREE_PLTE|PNG_FREE_tRNS|PNG_FREE_hIST)
2391 png_data_freer(write_ptr, write_info_ptr,
2392 PNG_DESTROY_WILL_FREE_DATA,
2393 PNG_FREE_PLTE|PNG_FREE_tRNS|PNG_FREE_hIST)
2395 thereby briefly reassigning responsibility for freeing to the user but
2396 immediately afterwards reassigning it once more to the write_destroy
2397 function. Having done this, it would then be safe to destroy the read
2398 structure and continue to use the PLTE, tRNS, and hIST data in the write
2401 This function only affects data that has already been allocated.
2402 You can call this function before calling after the png_set_*() functions
2403 to control whether the user or png_destroy_*() is supposed to free the data.
2404 When the user assumes responsibility for libpng-allocated data, the
2405 application must use
2406 png_free() to free it, and when the user transfers responsibility to libpng
2407 for data that the user has allocated, the user must have used png_malloc()
2408 or png_zalloc() to allocate it.
2410 If you allocated text_ptr.text, text_ptr.lang, and text_ptr.translated_keyword
2411 separately, do not transfer responsibility for freeing text_ptr to libpng,
2412 because when libpng fills a png_text structure it combines these members with
2413 the key member, and png_free_data() will free only text_ptr.key. Similarly,
2414 if you transfer responsibility for free'ing text_ptr from libpng to your
2415 application, your application must not separately free those members.
2416 For a more compact example of writing a PNG image, see the file example.c.
2418 V. Modifying/Customizing libpng:
2420 There are two issues here. The first is changing how libpng does
2421 standard things like memory allocation, input/output, and error handling.
2422 The second deals with more complicated things like adding new chunks,
2423 adding new transformations, and generally changing how libpng works.
2424 Both of those are compile-time issues; that is, they are generally
2425 determined at the time the code is written, and there is rarely a need
2426 to provide the user with a means of changing them.
2428 Memory allocation, input/output, and error handling
2430 All of the memory allocation, input/output, and error handling in libpng
2431 goes through callbacks that are user-settable. The default routines are
2432 in pngmem.c, pngrio.c, pngwio.c, and pngerror.c, respectively. To change
2433 these functions, call the appropriate png_set_*_fn() function.
2435 Memory allocation is done through the functions png_malloc()
2436 and png_free(). These currently just call the standard C functions. If
2437 your pointers can't access more then 64K at a time, you will want to set
2438 MAXSEG_64K in zlib.h. Since it is unlikely that the method of handling
2439 memory allocation on a platform will change between applications, these
2440 functions must be modified in the library at compile time. If you prefer
2441 to use a different method of allocating and freeing data, you can use
2442 png_create_read_struct_2() or png_create_write_struct_2() to register
2443 your own functions as described above.
2444 These functions also provide a void pointer that can be retrieved via
2446 mem_ptr=png_get_mem_ptr(png_ptr);
2448 Your replacement memory functions must have prototypes as follows:
2450 png_voidp malloc_fn(png_structp png_ptr,
2452 void free_fn(png_structp png_ptr, png_voidp ptr);
2454 Your malloc_fn() must return NULL in case of failure. The png_malloc()
2455 function will normally call png_error() if it receives a NULL from the
2456 system memory allocator or from your replacement malloc_fn().
2458 Your free_fn() will never be called with a NULL ptr, since libpng's
2459 png_free() checks for NULL before calling free_fn().
2461 Input/Output in libpng is done through png_read() and png_write(),
2462 which currently just call fread() and fwrite(). The FILE * is stored in
2463 png_struct and is initialized via png_init_io(). If you wish to change
2464 the method of I/O, the library supplies callbacks that you can set
2465 through the function png_set_read_fn() and png_set_write_fn() at run
2466 time, instead of calling the png_init_io() function. These functions
2467 also provide a void pointer that can be retrieved via the function
2468 png_get_io_ptr(). For example:
2470 png_set_read_fn(png_structp read_ptr,
2471 voidp read_io_ptr, png_rw_ptr read_data_fn)
2473 png_set_write_fn(png_structp write_ptr,
2474 voidp write_io_ptr, png_rw_ptr write_data_fn,
2475 png_flush_ptr output_flush_fn);
2477 voidp read_io_ptr = png_get_io_ptr(read_ptr);
2478 voidp write_io_ptr = png_get_io_ptr(write_ptr);
2480 The replacement I/O functions must have prototypes as follows:
2482 void user_read_data(png_structp png_ptr,
2483 png_bytep data, png_size_t length);
2484 void user_write_data(png_structp png_ptr,
2485 png_bytep data, png_size_t length);
2486 void user_flush_data(png_structp png_ptr);
2488 The user_read_data() function is responsible for detecting and
2489 handling end-of-data errors.
2491 Supplying NULL for the read, write, or flush functions sets them back
2492 to using the default C stream functions, which expect the io_ptr to
2493 point to a standard *FILE structure. It is probably a mistake
2494 to use NULL for one of write_data_fn and output_flush_fn but not both
2495 of them, unless you have built libpng with PNG_NO_WRITE_FLUSH defined.
2496 It is an error to read from a write stream, and vice versa.
2498 Error handling in libpng is done through png_error() and png_warning().
2499 Errors handled through png_error() are fatal, meaning that png_error()
2500 should never return to its caller. Currently, this is handled via
2501 setjmp() and longjmp() (unless you have compiled libpng with
2502 PNG_SETJMP_NOT_SUPPORTED, in which case it is handled via PNG_ABORT()),
2503 but you could change this to do things like exit() if you should wish.
2505 On non-fatal errors, png_warning() is called
2506 to print a warning message, and then control returns to the calling code.
2507 By default png_error() and png_warning() print a message on stderr via
2508 fprintf() unless the library is compiled with PNG_NO_CONSOLE_IO defined
2509 (because you don't want the messages) or PNG_NO_STDIO defined (because
2510 fprintf() isn't available). If you wish to change the behavior of the error
2511 functions, you will need to set up your own message callbacks. These
2512 functions are normally supplied at the time that the png_struct is created.
2513 It is also possible to redirect errors and warnings to your own replacement
2514 functions after png_create_*_struct() has been called by calling:
2516 png_set_error_fn(png_structp png_ptr,
2517 png_voidp error_ptr, png_error_ptr error_fn,
2518 png_error_ptr warning_fn);
2520 png_voidp error_ptr = png_get_error_ptr(png_ptr);
2522 If NULL is supplied for either error_fn or warning_fn, then the libpng
2523 default function will be used, calling fprintf() and/or longjmp() if a
2524 problem is encountered. The replacement error functions should have
2525 parameters as follows:
2527 void user_error_fn(png_structp png_ptr,
2528 png_const_charp error_msg);
2529 void user_warning_fn(png_structp png_ptr,
2530 png_const_charp warning_msg);
2532 The motivation behind using setjmp() and longjmp() is the C++ throw and
2533 catch exception handling methods. This makes the code much easier to write,
2534 as there is no need to check every return code of every function call.
2535 However, there are some uncertainties about the status of local variables
2536 after a longjmp, so the user may want to be careful about doing anything after
2537 setjmp returns non-zero besides returning itself. Consult your compiler
2538 documentation for more details. For an alternative approach, you may wish
2539 to use the "cexcept" facility (see http://cexcept.sourceforge.net).
2543 If you need to read or write custom chunks, you may need to get deeper
2544 into the libpng code. The library now has mechanisms for storing
2545 and writing chunks of unknown type; you can even declare callbacks
2546 for custom chunks. However, this may not be good enough if the
2547 library code itself needs to know about interactions between your
2548 chunk and existing `intrinsic' chunks.
2550 If you need to write a new intrinsic chunk, first read the PNG
2551 specification. Acquire a first level of
2552 understanding of how it works. Pay particular attention to the
2553 sections that describe chunk names, and look at how other chunks were
2554 designed, so you can do things similarly. Second, check out the
2555 sections of libpng that read and write chunks. Try to find a chunk
2556 that is similar to yours and use it as a template. More details can
2557 be found in the comments inside the code. It is best to handle unknown
2558 chunks in a generic method, via callback functions, instead of by
2559 modifying libpng functions.
2561 If you wish to write your own transformation for the data, look through
2562 the part of the code that does the transformations, and check out some of
2563 the simpler ones to get an idea of how they work. Try to find a similar
2564 transformation to the one you want to add and copy off of it. More details
2565 can be found in the comments inside the code itself.
2567 Configuring for 16 bit platforms
2569 You will want to look into zconf.h to tell zlib (and thus libpng) that
2570 it cannot allocate more then 64K at a time. Even if you can, the memory
2571 won't be accessible. So limit zlib and libpng to 64K by defining MAXSEG_64K.
2575 For DOS users who only have access to the lower 640K, you will
2576 have to limit zlib's memory usage via a png_set_compression_mem_level()
2577 call. See zlib.h or zconf.h in the zlib library for more information.
2579 Configuring for Medium Model
2581 Libpng's support for medium model has been tested on most of the popular
2582 compilers. Make sure MAXSEG_64K gets defined, USE_FAR_KEYWORD gets
2583 defined, and FAR gets defined to far in pngconf.h, and you should be
2584 all set. Everything in the library (except for zlib's structure) is
2585 expecting far data. You must use the typedefs with the p or pp on
2586 the end for pointers (or at least look at them and be careful). Make
2587 note that the rows of data are defined as png_bytepp, which is an
2588 unsigned char far * far *.
2590 Configuring for gui/windowing platforms:
2592 You will need to write new error and warning functions that use the GUI
2593 interface, as described previously, and set them to be the error and
2594 warning functions at the time that png_create_*_struct() is called,
2595 in order to have them available during the structure initialization.
2596 They can be changed later via png_set_error_fn(). On some compilers,
2597 you may also have to change the memory allocators (png_malloc, etc.).
2599 Configuring for compiler xxx:
2601 All includes for libpng are in pngconf.h. If you need to add, change
2602 or delete an include, this is the place to do it.
2603 The includes that are not needed outside libpng are protected by the
2604 PNG_INTERNAL definition, which is only defined for those routines inside
2605 libpng itself. The files in libpng proper only include png.h, which
2610 There are special functions to configure the compression. Perhaps the
2611 most useful one changes the compression level, which currently uses
2612 input compression values in the range 0 - 9. The library normally
2613 uses the default compression level (Z_DEFAULT_COMPRESSION = 6). Tests
2614 have shown that for a large majority of images, compression values in
2615 the range 3-6 compress nearly as well as higher levels, and do so much
2616 faster. For online applications it may be desirable to have maximum speed
2617 (Z_BEST_SPEED = 1). With versions of zlib after v0.99, you can also
2618 specify no compression (Z_NO_COMPRESSION = 0), but this would create
2619 files larger than just storing the raw bitmap. You can specify the
2620 compression level by calling:
2622 png_set_compression_level(png_ptr, level);
2624 Another useful one is to reduce the memory level used by the library.
2625 The memory level defaults to 8, but it can be lowered if you are
2626 short on memory (running DOS, for example, where you only have 640K).
2627 Note that the memory level does have an effect on compression; among
2628 other things, lower levels will result in sections of incompressible
2629 data being emitted in smaller stored blocks, with a correspondingly
2630 larger relative overhead of up to 15% in the worst case.
2632 png_set_compression_mem_level(png_ptr, level);
2634 The other functions are for configuring zlib. They are not recommended
2635 for normal use and may result in writing an invalid PNG file. See
2636 zlib.h for more information on what these mean.
2638 png_set_compression_strategy(png_ptr,
2640 png_set_compression_window_bits(png_ptr,
2642 png_set_compression_method(png_ptr, method);
2643 png_set_compression_buffer_size(png_ptr, size);
2645 Controlling row filtering
2647 If you want to control whether libpng uses filtering or not, which
2648 filters are used, and how it goes about picking row filters, you
2649 can call one of these functions. The selection and configuration
2650 of row filters can have a significant impact on the size and
2651 encoding speed and a somewhat lesser impact on the decoding speed
2652 of an image. Filtering is enabled by default for RGB and grayscale
2653 images (with and without alpha), but not for paletted images nor
2654 for any images with bit depths less than 8 bits/pixel.
2656 The 'method' parameter sets the main filtering method, which is
2657 currently only '0' in the PNG 1.2 specification. The 'filters'
2658 parameter sets which filter(s), if any, should be used for each
2659 scanline. Possible values are PNG_ALL_FILTERS and PNG_NO_FILTERS
2660 to turn filtering on and off, respectively.
2662 Individual filter types are PNG_FILTER_NONE, PNG_FILTER_SUB,
2663 PNG_FILTER_UP, PNG_FILTER_AVG, PNG_FILTER_PAETH, which can be bitwise
2664 ORed together with '|' to specify one or more filters to use.
2665 These filters are described in more detail in the PNG specification.
2666 If you intend to change the filter type during the course of writing
2667 the image, you should start with flags set for all of the filters
2668 you intend to use so that libpng can initialize its internal
2669 structures appropriately for all of the filter types. (Note that this
2670 means the first row must always be adaptively filtered, because libpng
2671 currently does not allocate the filter buffers until png_write_row()
2672 is called for the first time.)
2674 filters = PNG_FILTER_NONE | PNG_FILTER_SUB
2675 PNG_FILTER_UP | PNG_FILTER_AVG |
2676 PNG_FILTER_PAETH | PNG_ALL_FILTERS;
2678 png_set_filter(png_ptr, PNG_FILTER_TYPE_BASE,
2680 The second parameter can also be
2681 PNG_INTRAPIXEL_DIFFERENCING if you are
2682 writing a PNG to be embedded in a MNG
2683 datastream. This parameter must be the
2684 same as the value of filter_method used
2687 It is also possible to influence how libpng chooses from among the
2688 available filters. This is done in one or both of two ways - by
2689 telling it how important it is to keep the same filter for successive
2690 rows, and by telling it the relative computational costs of the filters.
2692 double weights[3] = {1.5, 1.3, 1.1},
2693 costs[PNG_FILTER_VALUE_LAST] =
2694 {1.0, 1.3, 1.3, 1.5, 1.7};
2696 png_set_filter_heuristics(png_ptr,
2697 PNG_FILTER_HEURISTIC_WEIGHTED, 3,
2700 The weights are multiplying factors that indicate to libpng that the
2701 row filter should be the same for successive rows unless another row filter
2702 is that many times better than the previous filter. In the above example,
2703 if the previous 3 filters were SUB, SUB, NONE, the SUB filter could have a
2704 "sum of absolute differences" 1.5 x 1.3 times higher than other filters
2705 and still be chosen, while the NONE filter could have a sum 1.1 times
2706 higher than other filters and still be chosen. Unspecified weights are
2707 taken to be 1.0, and the specified weights should probably be declining
2708 like those above in order to emphasize recent filters over older filters.
2710 The filter costs specify for each filter type a relative decoding cost
2711 to be considered when selecting row filters. This means that filters
2712 with higher costs are less likely to be chosen over filters with lower
2713 costs, unless their "sum of absolute differences" is that much smaller.
2714 The costs do not necessarily reflect the exact computational speeds of
2715 the various filters, since this would unduly influence the final image
2718 Note that the numbers above were invented purely for this example and
2719 are given only to help explain the function usage. Little testing has
2720 been done to find optimum values for either the costs or the weights.
2722 Removing unwanted object code
2724 There are a bunch of #define's in pngconf.h that control what parts of
2725 libpng are compiled. All the defines end in _SUPPORTED. If you are
2726 never going to use a capability, you can change the #define to #undef
2727 before recompiling libpng and save yourself code and data space, or
2728 you can turn off individual capabilities with defines that begin with
2731 You can also turn all of the transforms and ancillary chunk capabilities
2732 off en masse with compiler directives that define
2733 PNG_NO_READ[or WRITE]_TRANSFORMS, or PNG_NO_READ[or WRITE]_ANCILLARY_CHUNKS,
2735 along with directives to turn on any of the capabilities that you do
2736 want. The PNG_NO_READ[or WRITE]_TRANSFORMS directives disable
2737 the extra transformations but still leave the library fully capable of reading
2738 and writing PNG files with all known public chunks
2739 Use of the PNG_NO_READ[or WRITE]_ANCILLARY_CHUNKS directive
2740 produces a library that is incapable of reading or writing ancillary chunks.
2741 If you are not using the progressive reading capability, you can
2742 turn that off with PNG_NO_PROGRESSIVE_READ (don't confuse
2743 this with the INTERLACING capability, which you'll still have).
2745 All the reading and writing specific code are in separate files, so the
2746 linker should only grab the files it needs. However, if you want to
2747 make sure, or if you are building a stand alone library, all the
2748 reading files start with pngr and all the writing files start with
2749 pngw. The files that don't match either (like png.c, pngtrans.c, etc.)
2750 are used for both reading and writing, and always need to be included.
2751 The progressive reader is in pngpread.c
2753 If you are creating or distributing a dynamically linked library (a .so
2754 or DLL file), you should not remove or disable any parts of the library,
2755 as this will cause applications linked with different versions of the
2756 library to fail if they call functions not available in your library.
2757 The size of the library itself should not be an issue, because only
2758 those sections that are actually used will be loaded into memory.
2760 Requesting debug printout
2762 The macro definition PNG_DEBUG can be used to request debugging
2763 printout. Set it to an integer value in the range 0 to 3. Higher
2764 numbers result in increasing amounts of debugging information. The
2765 information is printed to the "stderr" file, unless another file
2766 name is specified in the PNG_DEBUG_FILE macro definition.
2768 When PNG_DEBUG > 0, the following functions (macros) become available:
2770 png_debug(level, message)
2771 png_debug1(level, message, p1)
2772 png_debug2(level, message, p1, p2)
2774 in which "level" is compared to PNG_DEBUG to decide whether to print
2775 the message, "message" is the formatted string to be printed,
2776 and p1 and p2 are parameters that are to be embedded in the string
2777 according to printf-style formatting directives. For example,
2779 png_debug1(2, "foo=%d\n", foo);
2784 fprintf(PNG_DEBUG_FILE, "foo=%d\n", foo);
2786 When PNG_DEBUG is defined but is zero, the macros aren't defined, but you
2787 can still use PNG_DEBUG to control your own debugging:
2793 When PNG_DEBUG = 1, the macros are defined, but only png_debug statements
2794 having level = 0 will be printed. There aren't any such statements in
2795 this version of libpng, but if you insert some they will be printed.
2799 The MNG specification (available at http://www.libpng.org/pub/mng) allows
2800 certain extensions to PNG for PNG images that are embedded in MNG datastreams.
2801 Libpng can support some of these extensions. To enable them, use the
2802 png_permit_mng_features() function:
2804 feature_set = png_permit_mng_features(png_ptr, mask)
2805 mask is a png_uint_32 containing the bitwise OR of the
2806 features you want to enable. These include
2807 PNG_FLAG_MNG_EMPTY_PLTE
2808 PNG_FLAG_MNG_FILTER_64
2809 PNG_ALL_MNG_FEATURES
2810 feature_set is a png_uint_32 that is the bitwise AND of
2811 your mask with the set of MNG features that is
2812 supported by the version of libpng that you are using.
2814 It is an error to use this function when reading or writing a standalone
2815 PNG file with the PNG 8-byte signature. The PNG datastream must be wrapped
2816 in a MNG datastream. As a minimum, it must have the MNG 8-byte signature
2817 and the MHDR and MEND chunks. Libpng does not provide support for these
2818 or any other MNG chunks; your application must provide its own support for
2819 them. You may wish to consider using libmng (available at
2820 http://www.libmng.com) instead.
2822 VII. Changes to Libpng from version 0.88
2824 It should be noted that versions of libpng later than 0.96 are not
2825 distributed by the original libpng author, Guy Schalnat, nor by
2826 Andreas Dilger, who had taken over from Guy during 1996 and 1997, and
2827 distributed versions 0.89 through 0.96, but rather by another member
2828 of the original PNG Group, Glenn Randers-Pehrson. Guy and Andreas are
2829 still alive and well, but they have moved on to other things.
2831 The old libpng functions png_read_init(), png_write_init(),
2832 png_info_init(), png_read_destroy(), and png_write_destroy() have been
2833 moved to PNG_INTERNAL in version 0.95 to discourage their use. These
2834 functions will be removed from libpng version 2.0.0.
2836 The preferred method of creating and initializing the libpng structures is
2837 via the png_create_read_struct(), png_create_write_struct(), and
2838 png_create_info_struct() because they isolate the size of the structures
2839 from the application, allow version error checking, and also allow the
2840 use of custom error handling routines during the initialization, which
2841 the old functions do not. The functions png_read_destroy() and
2842 png_write_destroy() do not actually free the memory that libpng
2843 allocated for these structs, but just reset the data structures, so they
2844 can be used instead of png_destroy_read_struct() and
2845 png_destroy_write_struct() if you feel there is too much system overhead
2846 allocating and freeing the png_struct for each image read.
2848 Setting the error callbacks via png_set_message_fn() before
2849 png_read_init() as was suggested in libpng-0.88 is no longer supported
2850 because this caused applications that do not use custom error functions
2851 to fail if the png_ptr was not initialized to zero. It is still possible
2852 to set the error callbacks AFTER png_read_init(), or to change them with
2853 png_set_error_fn(), which is essentially the same function, but with a new
2854 name to force compilation errors with applications that try to use the old
2857 Starting with version 1.0.7, you can find out which version of the library
2858 you are using at run-time:
2860 png_uint_32 libpng_vn = png_access_version_number();
2862 The number libpng_vn is constructed from the major version, minor
2863 version with leading zero, and release number with leading zero,
2864 (e.g., libpng_vn for version 1.0.7 is 10007).
2866 You can also check which version of png.h you used when compiling your
2869 png_uint_32 application_vn = PNG_LIBPNG_VER;
2871 VIII. Changes to Libpng from version 1.0.x to 1.2.x
2873 Support for user memory management was enabled by default. To
2874 accomplish this, the functions png_create_read_struct_2(),
2875 png_create_write_struct_2(), png_set_mem_fn(), png_get_mem_ptr(),
2876 png_malloc_default(), and png_free_default() were added.
2878 Support for certain MNG features was enabled.
2880 Support for numbered error messages was added. However, we never got
2881 around to actually numbering the error messages. The function
2882 png_set_strip_error_numbers() was added (Note: the prototype for this
2883 function was inadvertently removed from png.h in PNG_NO_ASSEMBLER_CODE
2884 builds of libpng-1.2.15. It was restored in libpng-1.2.36).
2886 The png_malloc_warn() function was added at libpng-1.2.3. This issues
2887 a png_warning and returns NULL instead of aborting when it fails to
2888 acquire the requested memory allocation.
2890 Support for setting user limits on image width and height was enabled
2891 by default. The functions png_set_user_limits(), png_get_user_width_max(),
2892 and png_get_user_height_max() were added at libpng-1.2.6.
2894 The png_set_add_alpha() function was added at libpng-1.2.7.
2896 The function png_set_expand_gray_1_2_4_to_8() was added at libpng-1.2.9.
2897 Unlike png_set_gray_1_2_4_to_8(), the new function does not expand the
2898 tRNS chunk to alpha. The png_set_gray_1_2_4_to_8() function is
2901 A number of macro definitions in support of runtime selection of
2902 assembler code features (especially Intel MMX code support) were
2903 added at libpng-1.2.0:
2905 PNG_ASM_FLAG_MMX_SUPPORT_COMPILED
2906 PNG_ASM_FLAG_MMX_SUPPORT_IN_CPU
2907 PNG_ASM_FLAG_MMX_READ_COMBINE_ROW
2908 PNG_ASM_FLAG_MMX_READ_INTERLACE
2909 PNG_ASM_FLAG_MMX_READ_FILTER_SUB
2910 PNG_ASM_FLAG_MMX_READ_FILTER_UP
2911 PNG_ASM_FLAG_MMX_READ_FILTER_AVG
2912 PNG_ASM_FLAG_MMX_READ_FILTER_PAETH
2913 PNG_ASM_FLAGS_INITIALIZED
2919 We added the following functions in support of runtime
2920 selection of assembler code features:
2922 png_get_mmx_flagmask()
2923 png_set_mmx_thresholds()
2925 png_get_mmx_bitdepth_threshold()
2926 png_get_mmx_rowbytes_threshold()
2929 We replaced all of these functions with simple stubs in libpng-1.2.20,
2930 when the Intel assembler code was removed due to a licensing issue.
2936 The png_get_io_ptr() function has been present since libpng-0.88, has never
2937 changed, and is unaffected by conditional compilation macros. It is the
2938 best choice for use in configure scripts for detecting the presence of any
2939 libpng version since 0.88. In an autoconf "configure.in" you could use
2941 AC_CHECK_LIB(png, png_get_io_ptr, ...
2943 XI. Source code repository
2945 Since about February 2009, version 1.2.34, libpng has been under "git" source
2946 control. The git repository was built from old libpng-x.y.z.tar.gz files
2947 going back to version 0.70. You can access the git repository (read only)
2950 git://libpng.git.sourceforge.net/gitroot/libpng
2952 or you can browse it via "gitweb" at
2954 http://libpng.git.sourceforge.net/git/gitweb.cgi?p=libpng
2956 Patches can be sent to glennrp at users.sourceforge.net or to
2957 png-mng-implement at lists.sourceforge.net or you can upload them to
2958 the libpng bug tracker at
2960 http://libpng.sourceforge.net
2964 Our coding style is similar to the "Allman" style, with curly
2965 braces on separate lines:
2972 else if (another condition)
2977 The braces can be omitted from simple one-line actions:
2982 We use 3-space indentation, except for continued statements which
2983 are usually indented the same as the first line of the statement
2984 plus four more spaces.
2986 Comments appear with the leading "/*" at the same indentation as
2987 the statement that follows the comment:
2989 /* Single-line comment */
2997 Very short comments can be placed at the end of the statement
2998 to which they pertain:
3000 statement; /* comment */
3002 We don't use C++ style ("//") comments. We have, however,
3003 used them in the past in some now-abandoned MMX assembler
3006 Functions and their curly braces are not indented, and
3007 exported functions are marked with PNGAPI:
3009 /* This is a public function that is visible to
3010 * application programers. It does thus-and-so.
3013 png_exported_function(png_ptr, png_info, foo)
3018 The prototypes for all exported functions appear in png.h,
3019 above the comment that says
3021 /* Maintainer: Put new public prototypes here ... */
3023 We mark all non-exported functions with "/* PRIVATE */"":
3026 png_non_exported_function(png_ptr, png_info, foo)
3031 The prototypes for non-exported functions (except for those in
3033 the PNG_INTERNAL section of png.h
3034 above the comment that says
3036 /* Maintainer: Put new private prototypes here ^ and in libpngpf.3 */
3038 The names of all exported functions and variables begin
3039 with "png_", and all publicly visible C preprocessor
3040 macros begin with "PNG_".
3042 We put a space after each comma and after each semicolon
3043 in "for" statments, and we put spaces before and after each
3044 C binary operator and after "for" or "while". We don't
3045 put a space between a typecast and the expression being
3046 cast, nor do we put one between a function name and the
3047 left parenthesis that follows it:
3049 for (i = 2; i > 0; --i)
3050 x[i] = a(x) + (int)b;
3052 We prefer #ifdef and #ifndef to #if defined() and if !defined()
3053 when there is only one macro being tested.
3055 Other rules can be inferred by inspecting the libpng
3058 XIII. Y2K Compliance in libpng
3062 Since the PNG Development group is an ad-hoc body, we can't make
3063 an official declaration.
3065 This is your unofficial assurance that libpng from version 0.71 and
3066 upward through 1.2.40 are Y2K compliant. It is my belief that earlier
3067 versions were also Y2K compliant.
3069 Libpng only has three year fields. One is a 2-byte unsigned integer that
3070 will hold years up to 65535. The other two hold the date in text
3071 format, and will hold years up to 9999.
3074 "png_uint_16 year" in png_time_struct.
3077 "png_charp time_buffer" in png_struct and
3078 "near_time_buffer", which is a local character string in png.c.
3080 There are seven time-related functions:
3082 png_convert_to_rfc_1123() in png.c
3083 (formerly png_convert_to_rfc_1152() in error)
3084 png_convert_from_struct_tm() in pngwrite.c, called
3086 png_convert_from_time_t() in pngwrite.c
3087 png_get_tIME() in pngget.c
3088 png_handle_tIME() in pngrutil.c, called in pngread.c
3089 png_set_tIME() in pngset.c
3090 png_write_tIME() in pngwutil.c, called in pngwrite.c
3092 All appear to handle dates properly in a Y2K environment. The
3093 png_convert_from_time_t() function calls gmtime() to convert from system
3094 clock time, which returns (year - 1900), which we properly convert to
3095 the full 4-digit year. There is a possibility that applications using
3096 libpng are not passing 4-digit years into the png_convert_to_rfc_1123()
3097 function, or that they are incorrectly passing only a 2-digit year
3098 instead of "year - 1900" into the png_convert_from_struct_tm() function,
3099 but this is not under our control. The libpng documentation has always
3100 stated that it works with 4-digit years, and the APIs have been
3103 The tIME chunk itself is also Y2K compliant. It uses a 2-byte unsigned
3104 integer to hold the year, and can hold years as large as 65535.
3106 zlib, upon which libpng depends, is also Y2K compliant. It contains
3107 no date-related code.
3110 Glenn Randers-Pehrson
3112 PNG Development Group