1
libpng.txt - A description on how to use and modify libpng
3
libpng version 1.2.18 - May 15, 2007
4
Updated and distributed by Glenn Randers-Pehrson
5
<glennrp at users.sourceforge.net>
6
Copyright (c) 1998-2007 Glenn Randers-Pehrson
7
For conditions of distribution and use, see copyright
12
libpng 1.0 beta 6 version 0.96 May 28, 1997
13
Updated and distributed by Andreas Dilger
14
Copyright (c) 1996, 1997 Andreas Dilger
16
libpng 1.0 beta 2 - version 0.88 January 26, 1996
17
For conditions of distribution and use, see copyright
18
notice in png.h. Copyright (c) 1995, 1996 Guy Eric
19
Schalnat, Group 42, Inc.
21
Updated/rewritten per request in the libpng FAQ
22
Copyright (c) 1995, 1996 Frank J. T. Wojcik
23
December 18, 1995 & January 20, 1996
27
This file describes how to use and modify the PNG reference library
28
(known as libpng) for your own use. There are five sections to this
29
file: introduction, structures, reading, writing, and modification and
30
configuration notes for various special platforms. In addition to this
31
file, example.c is a good starting point for using the library, as
32
it is heavily commented and should include everything most people
33
will need. We assume that libpng is already installed; see the
34
INSTALL file for instructions on how to install libpng.
36
For examples of libpng usage, see the files "example.c", "pngtest.c",
37
and the files in the "contrib" directory, all of which are included in the
40
Libpng was written as a companion to the PNG specification, as a way
41
of reducing the amount of time and effort it takes to support the PNG
42
file format in application programs.
44
The PNG specification (second edition), November 2003, is available as
45
a W3C Recommendation and as an ISO Standard (ISO/IEC 15948:2003 (E)) at
46
<http://www.w3.org/TR/2003/REC-PNG-20031110/
47
The W3C and ISO documents have identical technical content.
49
The PNG-1.2 specification is available at
50
<http://www.libpng.org/pub/png/documents/>
52
The PNG-1.0 specification is available
53
as RFC 2083 <http://www.libpng.org/pub/png/documents/> and as a
54
W3C Recommendation <http://www.w3.org/TR/REC.png.html>. Some
55
additional chunks are described in the special-purpose public chunks
56
documents at <http://www.libpng.org/pub/png/documents/>.
59
about PNG, and the latest version of libpng, can be found at the PNG home
60
page, <http://www.libpng.org/pub/png/>.
62
Most users will not have to modify the library significantly; advanced
63
users may want to modify it more. All attempts were made to make it as
64
complete as possible, while keeping the code easy to understand.
65
Currently, this library only supports C. Support for other languages
68
Libpng has been designed to handle multiple sessions at one time,
69
to be easily modifiable, to be portable to the vast majority of
70
machines (ANSI, K&R, 16-, 32-, and 64-bit) available, and to be easy
71
to use. The ultimate goal of libpng is to promote the acceptance of
72
the PNG file format in whatever way possible. While there is still
73
work to be done (see the TODO file), libpng should cover the
74
majority of the needs of its users.
76
Libpng uses zlib for its compression and decompression of PNG files.
77
Further information about zlib, and the latest version of zlib, can
78
be found at the zlib home page, <http://www.info-zip.org/pub/infozip/zlib/>.
79
The zlib compression utility is a general purpose utility that is
80
useful for more than PNG files, and can be used without libpng.
81
See the documentation delivered with zlib for more details.
82
You can usually find the source files for the zlib utility wherever you
83
find the libpng source files.
85
Libpng is thread safe, provided the threads are using different
86
instances of the structures. Each thread should have its own
87
png_struct and png_info instances, and thus its own image.
88
Libpng does not protect itself against two threads using the
89
same instance of a structure. Note: thread safety may be defeated
90
by use of some of the MMX assembler code in pnggccrd.c, which is only
91
compiled when the user defines PNG_THREAD_UNSAFE_OK.
95
There are two main structures that are important to libpng, png_struct
96
and png_info. The first, png_struct, is an internal structure that
97
will not, for the most part, be used by a user except as the first
98
variable passed to every libpng function call.
100
The png_info structure is designed to provide information about the
101
PNG file. At one time, the fields of png_info were intended to be
102
directly accessible to the user. However, this tended to cause problems
103
with applications using dynamically loaded libraries, and as a result
104
a set of interface functions for png_info (the png_get_*() and png_set_*()
105
functions) was developed. The fields of png_info are still available for
106
older applications, but it is suggested that applications use the new
107
interfaces if at all possible.
109
Applications that do make direct access to the members of png_struct (except
110
for png_ptr->jmpbuf) must be recompiled whenever the library is updated,
111
and applications that make direct access to the members of png_info must
112
be recompiled if they were compiled or loaded with libpng version 1.0.6,
113
in which the members were in a different order. In version 1.0.7, the
114
members of the png_info structure reverted to the old order, as they were
115
in versions 0.97c through 1.0.5. Starting with version 2.0.0, both
116
structures are going to be hidden, and the contents of the structures will
117
only be accessible through the png_get/png_set functions.
119
The png.h header file is an invaluable reference for programming with libpng.
120
And while I'm on the topic, make sure you include the libpng header file:
126
We'll now walk you through the possible functions to call when reading
127
in a PNG file sequentially, briefly explaining the syntax and purpose
128
of each one. See example.c and png.h for more detail. While
129
progressive reading is covered in the next section, you will still
130
need some of the functions discussed in this section to read a PNG
135
You will want to do the I/O initialization(*) before you get into libpng,
136
so if it doesn't work, you don't have much to undo. Of course, you
137
will also want to insure that you are, in fact, dealing with a PNG
138
file. Libpng provides a simple check to see if a file is a PNG file.
139
To use it, pass in the first 1 to 8 bytes of the file to the function
140
png_sig_cmp(), and it will return 0 if the bytes match the corresponding
141
bytes of the PNG signature, or nonzero otherwise. Of course, the more bytes
142
you pass in, the greater the accuracy of the prediction.
144
If you are intending to keep the file pointer open for use in libpng,
145
you must ensure you don't read more than 8 bytes from the beginning
146
of the file, and you also have to make a call to png_set_sig_bytes_read()
147
with the number of bytes you read from the beginning. Libpng will
148
then only check the bytes (if any) that your program didn't read.
150
(*): If you are not using the standard I/O functions, you will need
151
to replace them with custom functions. See the discussion under
155
FILE *fp = fopen(file_name, "rb");
160
fread(header, 1, number, fp);
161
is_png = !png_sig_cmp(header, 0, number);
168
Next, png_struct and png_info need to be allocated and initialized. In
169
order to ensure that the size of these structures is correct even with a
170
dynamically linked libpng, there are functions to initialize and
171
allocate the structures. We also pass the library version, optional
172
pointers to error handling functions, and a pointer to a data struct for
173
use by the error functions, if necessary (the pointer and functions can
174
be NULL if the default error handlers are to be used). See the section
175
on Changes to Libpng below regarding the old initialization functions.
176
The structure allocation functions quietly return NULL if they fail to
177
create the structure, so your application should check for that.
179
png_structp png_ptr = png_create_read_struct
180
(PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
181
user_error_fn, user_warning_fn);
185
png_infop info_ptr = png_create_info_struct(png_ptr);
188
png_destroy_read_struct(&png_ptr,
189
(png_infopp)NULL, (png_infopp)NULL);
193
png_infop end_info = png_create_info_struct(png_ptr);
196
png_destroy_read_struct(&png_ptr, &info_ptr,
201
If you want to use your own memory allocation routines,
202
define PNG_USER_MEM_SUPPORTED and use
203
png_create_read_struct_2() instead of png_create_read_struct():
205
png_structp png_ptr = png_create_read_struct_2
206
(PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
207
user_error_fn, user_warning_fn, (png_voidp)
208
user_mem_ptr, user_malloc_fn, user_free_fn);
210
The error handling routines passed to png_create_read_struct()
211
and the memory alloc/free routines passed to png_create_struct_2()
212
are only necessary if you are not using the libpng supplied error
213
handling and memory alloc/free functions.
215
When libpng encounters an error, it expects to longjmp back
216
to your routine. Therefore, you will need to call setjmp and pass
217
your png_jmpbuf(png_ptr). If you read the file from different
218
routines, you will need to update the jmpbuf field every time you enter
219
a new routine that will call a png_*() function.
221
See your documentation of setjmp/longjmp for your compiler for more
222
information on setjmp/longjmp. See the discussion on libpng error
223
handling in the Customizing Libpng section below for more information
224
on the libpng error handling. If an error occurs, and libpng longjmp's
225
back to your setjmp, you will want to call png_destroy_read_struct() to
228
if (setjmp(png_jmpbuf(png_ptr)))
230
png_destroy_read_struct(&png_ptr, &info_ptr,
236
If you would rather avoid the complexity of setjmp/longjmp issues,
237
you can compile libpng with PNG_SETJMP_NOT_SUPPORTED, in which case
238
errors will result in a call to PNG_ABORT() which defaults to abort().
240
Now you need to set up the input code. The default for libpng is to
241
use the C function fread(). If you use this, you will need to pass a
242
valid FILE * in the function png_init_io(). Be sure that the file is
243
opened in binary mode. If you wish to handle reading data in another
244
way, you need not call the png_init_io() function, but you must then
245
implement the libpng I/O methods discussed in the Customizing Libpng
248
png_init_io(png_ptr, fp);
250
If you had previously opened the file and read any of the signature from
251
the beginning in order to see if this was a PNG file, you need to let
252
libpng know that there are some bytes missing from the start of the file.
254
png_set_sig_bytes(png_ptr, number);
256
Setting up callback code
258
You can set up a callback function to handle any unknown chunks in the
259
input stream. You must supply the function
261
read_chunk_callback(png_ptr ptr,
262
png_unknown_chunkp chunk);
264
/* The unknown chunk structure contains your
269
/* Note that libpng has already taken care of
272
/* put your code here. Return one of the
275
return (-n); /* chunk had an error */
276
return (0); /* did not recognize */
277
return (n); /* success */
280
(You can give your function another name that you like instead of
281
"read_chunk_callback")
283
To inform libpng about your function, use
285
png_set_read_user_chunk_fn(png_ptr, user_chunk_ptr,
286
read_chunk_callback);
288
This names not only the callback function, but also a user pointer that
289
you can retrieve with
291
png_get_user_chunk_ptr(png_ptr);
293
At this point, you can set up a callback function that will be
294
called after each row has been read, which you can use to control
295
a progress meter or the like. It's demonstrated in pngtest.c.
296
You must supply a function
298
void read_row_callback(png_ptr ptr, png_uint_32 row,
301
/* put your code here */
304
(You can give it another name that you like instead of "read_row_callback")
306
To inform libpng about your function, use
308
png_set_read_status_fn(png_ptr, read_row_callback);
310
Width and height limits
312
The PNG specification allows the width and height of an image to be as
313
large as 2^31-1 (0x7fffffff), or about 2.147 billion rows and columns.
314
Since very few applications really need to process such large images,
315
we have imposed an arbitrary 1-million limit on rows and columns.
316
Larger images will be rejected immediately with a png_error() call. If
317
you wish to override this limit, you can use
319
png_set_user_limits(png_ptr, width_max, height_max);
321
to set your own limits, or use width_max = height_max = 0x7fffffffL
322
to allow all valid dimensions (libpng may reject some very large images
323
anyway because of potential buffer overflow conditions).
325
You should put this statement after you create the PNG structure and
326
before calling png_read_info(), png_read_png(), or png_process_data().
327
If you need to retrieve the limits that are being applied, use
329
width_max = png_get_user_width_max(png_ptr);
330
height_max = png_get_user_height_max(png_ptr);
332
Unknown-chunk handling
334
Now you get to set the way the library processes unknown chunks in the
335
input PNG stream. Both known and unknown chunks will be read. Normal
336
behavior is that known chunks will be parsed into information in
337
various info_ptr members; unknown chunks will be discarded. To change
340
png_set_keep_unknown_chunks(png_ptr, keep,
341
chunk_list, num_chunks);
342
keep - 0: do not handle as unknown
344
2: keep only if safe-to-copy
345
3: keep even if unsafe-to-copy
346
You can use these definitions:
347
PNG_HANDLE_CHUNK_AS_DEFAULT 0
348
PNG_HANDLE_CHUNK_NEVER 1
349
PNG_HANDLE_CHUNK_IF_SAFE 2
350
PNG_HANDLE_CHUNK_ALWAYS 3
351
chunk_list - list of chunks affected (a byte string,
352
five bytes per chunk, NULL or '\0' if
354
num_chunks - number of chunks affected; if 0, all
355
unknown chunks are affected. If nonzero,
356
only the chunks in the list are affected
358
Unknown chunks declared in this way will be saved as raw data onto a
359
list of png_unknown_chunk structures. If a chunk that is normally
360
known to libpng is named in the list, it will be handled as unknown,
361
according to the "keep" directive. If a chunk is named in successive
362
instances of png_set_keep_unknown_chunks(), the final instance will
363
take precedence. The IHDR and IEND chunks should not be named in
364
chunk_list; if they are, libpng will process them normally anyway.
366
The high-level read interface
368
At this point there are two ways to proceed; through the high-level
369
read interface, or through a sequence of low-level read operations.
370
You can use the high-level interface if (a) you are willing to read
371
the entire image into memory, and (b) the input transformations
372
you want to do are limited to the following set:
374
PNG_TRANSFORM_IDENTITY No transformation
375
PNG_TRANSFORM_STRIP_16 Strip 16-bit samples to
377
PNG_TRANSFORM_STRIP_ALPHA Discard the alpha channel
378
PNG_TRANSFORM_PACKING Expand 1, 2 and 4-bit
380
PNG_TRANSFORM_PACKSWAP Change order of packed
382
PNG_TRANSFORM_EXPAND Perform set_expand()
383
PNG_TRANSFORM_INVERT_MONO Invert monochrome images
384
PNG_TRANSFORM_SHIFT Normalize pixels to the
386
PNG_TRANSFORM_BGR Flip RGB to BGR, RGBA
388
PNG_TRANSFORM_SWAP_ALPHA Flip RGBA to ARGB or GA
390
PNG_TRANSFORM_INVERT_ALPHA Change alpha from opacity
392
PNG_TRANSFORM_SWAP_ENDIAN Byte-swap 16-bit samples
394
(This excludes setting a background color, doing gamma transformation,
395
dithering, and setting filler.) If this is the case, simply do this:
397
png_read_png(png_ptr, info_ptr, png_transforms, NULL)
399
where png_transforms is an integer containing the bitwise OR of
400
some set of transformation flags. This call is equivalent to png_read_info(),
401
followed the set of transformations indicated by the transform mask,
402
then png_read_image(), and finally png_read_end().
404
(The final parameter of this call is not yet used. Someday it might point
405
to transformation parameters required by some future input transform.)
407
You must use png_transforms and not call any png_set_transform() functions
408
when you use png_read_png().
410
After you have called png_read_png(), you can retrieve the image data
413
row_pointers = png_get_rows(png_ptr, info_ptr);
415
where row_pointers is an array of pointers to the pixel data for each row:
417
png_bytep row_pointers[height];
419
If you know your image size and pixel size ahead of time, you can allocate
420
row_pointers prior to calling png_read_png() with
422
if (height > PNG_UINT_32_MAX/png_sizeof(png_byte))
424
"Image is too tall to process in memory");
425
if (width > PNG_UINT_32_MAX/pixel_size)
427
"Image is too wide to process in memory");
428
row_pointers = png_malloc(png_ptr,
429
height*png_sizeof(png_bytep));
430
for (int i=0; i<height, i++)
431
row_pointers[i]=png_malloc(png_ptr,
433
png_set_rows(png_ptr, info_ptr, &row_pointers);
435
Alternatively you could allocate your image in one big block and define
436
row_pointers[i] to point into the proper places in your block.
438
If you use png_set_rows(), the application is responsible for freeing
439
row_pointers (and row_pointers[i], if they were separately allocated).
441
If you don't allocate row_pointers ahead of time, png_read_png() will
442
do it, and it'll be free'ed when you call png_destroy_*().
444
The low-level read interface
446
If you are going the low-level route, you are now ready to read all
447
the file information up to the actual image data. You do this with a
448
call to png_read_info().
450
png_read_info(png_ptr, info_ptr);
452
This will process all chunks up to but not including the image data.
454
Querying the info structure
456
Functions are used to get the information from the info_ptr once it
457
has been read. Note that these fields may not be completely filled
458
in until png_read_end() has read the chunk data following the image.
460
png_get_IHDR(png_ptr, info_ptr, &width, &height,
461
&bit_depth, &color_type, &interlace_type,
462
&compression_type, &filter_method);
464
width - holds the width of the image
465
in pixels (up to 2^31).
466
height - holds the height of the image
467
in pixels (up to 2^31).
468
bit_depth - holds the bit depth of one of the
469
image channels. (valid values are
470
1, 2, 4, 8, 16 and depend also on
471
the color_type. See also
472
significant bits (sBIT) below).
473
color_type - describes which color/alpha channels
476
(bit depths 1, 2, 4, 8, 16)
477
PNG_COLOR_TYPE_GRAY_ALPHA
479
PNG_COLOR_TYPE_PALETTE
480
(bit depths 1, 2, 4, 8)
483
PNG_COLOR_TYPE_RGB_ALPHA
486
PNG_COLOR_MASK_PALETTE
490
filter_method - (must be PNG_FILTER_TYPE_BASE
491
for PNG 1.0, and can also be
492
PNG_INTRAPIXEL_DIFFERENCING if
493
the PNG datastream is embedded in
494
a MNG-1.0 datastream)
495
compression_type - (must be PNG_COMPRESSION_TYPE_BASE
497
interlace_type - (PNG_INTERLACE_NONE or
499
Any or all of interlace_type, compression_type, of
500
filter_method can be NULL if you are
501
not interested in their values.
503
channels = png_get_channels(png_ptr, info_ptr);
504
channels - number of channels of info for the
505
color type (valid values are 1 (GRAY,
506
PALETTE), 2 (GRAY_ALPHA), 3 (RGB),
507
4 (RGB_ALPHA or RGB + filler byte))
508
rowbytes = png_get_rowbytes(png_ptr, info_ptr);
509
rowbytes - number of bytes needed to hold a row
511
signature = png_get_signature(png_ptr, info_ptr);
512
signature - holds the signature read from the
513
file (if any). The data is kept in
514
the same offset it would be if the
515
whole signature were read (i.e. if an
516
application had already read in 4
517
bytes of signature before starting
518
libpng, the remaining 4 bytes would
519
be in signature[4] through signature[7]
520
(see png_set_sig_bytes())).
523
width = png_get_image_width(png_ptr,
525
height = png_get_image_height(png_ptr,
527
bit_depth = png_get_bit_depth(png_ptr,
529
color_type = png_get_color_type(png_ptr,
531
filter_method = png_get_filter_type(png_ptr,
533
compression_type = png_get_compression_type(png_ptr,
535
interlace_type = png_get_interlace_type(png_ptr,
539
These are also important, but their validity depends on whether the chunk
540
has been read. The png_get_valid(png_ptr, info_ptr, PNG_INFO_<chunk>) and
541
png_get_<chunk>(png_ptr, info_ptr, ...) functions return non-zero if the
542
data has been read, or zero if it is missing. The parameters to the
543
png_get_<chunk> are set directly if they are simple data types, or a pointer
544
into the info_ptr is returned for any complex types.
546
png_get_PLTE(png_ptr, info_ptr, &palette,
548
palette - the palette for the file
550
num_palette - number of entries in the palette
552
png_get_gAMA(png_ptr, info_ptr, &gamma);
553
gamma - the gamma the file is written
556
png_get_sRGB(png_ptr, info_ptr, &srgb_intent);
557
srgb_intent - the rendering intent (PNG_INFO_sRGB)
558
The presence of the sRGB chunk
559
means that the pixel data is in the
560
sRGB color space. This chunk also
561
implies specific values of gAMA and
564
png_get_iCCP(png_ptr, info_ptr, &name,
565
&compression_type, &profile, &proflen);
566
name - The profile name.
567
compression - The compression type; always
568
PNG_COMPRESSION_TYPE_BASE for PNG 1.0.
569
You may give NULL to this argument to
571
profile - International Color Consortium color
572
profile data. May contain NULs.
573
proflen - length of profile data in bytes.
575
png_get_sBIT(png_ptr, info_ptr, &sig_bit);
576
sig_bit - the number of significant bits for
577
(PNG_INFO_sBIT) each of the gray,
578
red, green, and blue channels,
579
whichever are appropriate for the
580
given color type (png_color_16)
582
png_get_tRNS(png_ptr, info_ptr, &trans, &num_trans,
584
trans - array of transparent entries for
585
palette (PNG_INFO_tRNS)
586
trans_values - graylevel or color sample values of
587
the single transparent color for
588
non-paletted images (PNG_INFO_tRNS)
589
num_trans - number of transparent entries
592
png_get_hIST(png_ptr, info_ptr, &hist);
594
hist - histogram of palette (array of
597
png_get_tIME(png_ptr, info_ptr, &mod_time);
598
mod_time - time image was last modified
601
png_get_bKGD(png_ptr, info_ptr, &background);
602
background - background color (PNG_VALID_bKGD)
603
valid 16-bit red, green and blue
604
values, regardless of color_type
606
num_comments = png_get_text(png_ptr, info_ptr,
607
&text_ptr, &num_text);
608
num_comments - number of comments
609
text_ptr - array of png_text holding image
611
text_ptr[i].compression - type of compression used
612
on "text" PNG_TEXT_COMPRESSION_NONE
613
PNG_TEXT_COMPRESSION_zTXt
614
PNG_ITXT_COMPRESSION_NONE
615
PNG_ITXT_COMPRESSION_zTXt
616
text_ptr[i].key - keyword for comment. Must contain
618
text_ptr[i].text - text comments for current
619
keyword. Can be empty.
620
text_ptr[i].text_length - length of text string,
621
after decompression, 0 for iTXt
622
text_ptr[i].itxt_length - length of itxt string,
623
after decompression, 0 for tEXt/zTXt
624
text_ptr[i].lang - language of comment (empty
626
text_ptr[i].lang_key - keyword in UTF-8
627
(empty string for unknown).
628
num_text - number of comments (same as
629
num_comments; you can put NULL here
630
to avoid the duplication)
631
Note while png_set_text() will accept text, language,
632
and translated keywords that can be NULL pointers, the
633
structure returned by png_get_text will always contain
634
regular zero-terminated C strings. They might be
635
empty strings but they will never be NULL pointers.
637
num_spalettes = png_get_sPLT(png_ptr, info_ptr,
639
palette_ptr - array of palette structures holding
640
contents of one or more sPLT chunks
642
num_spalettes - number of sPLT chunks read.
644
png_get_oFFs(png_ptr, info_ptr, &offset_x, &offset_y,
646
offset_x - positive offset from the left edge
648
offset_y - positive offset from the top edge
650
unit_type - PNG_OFFSET_PIXEL, PNG_OFFSET_MICROMETER
652
png_get_pHYs(png_ptr, info_ptr, &res_x, &res_y,
654
res_x - pixels/unit physical resolution in
656
res_y - pixels/unit physical resolution in
658
unit_type - PNG_RESOLUTION_UNKNOWN,
661
png_get_sCAL(png_ptr, info_ptr, &unit, &width,
663
unit - physical scale units (an integer)
664
width - width of a pixel in physical scale units
665
height - height of a pixel in physical scale units
666
(width and height are doubles)
668
png_get_sCAL_s(png_ptr, info_ptr, &unit, &width,
670
unit - physical scale units (an integer)
671
width - width of a pixel in physical scale units
672
height - height of a pixel in physical scale units
673
(width and height are strings like "2.54")
675
num_unknown_chunks = png_get_unknown_chunks(png_ptr,
677
unknowns - array of png_unknown_chunk
678
structures holding unknown chunks
679
unknowns[i].name - name of unknown chunk
680
unknowns[i].data - data of unknown chunk
681
unknowns[i].size - size of unknown chunk's data
682
unknowns[i].location - position of chunk in file
684
The value of "i" corresponds to the order in which the
685
chunks were read from the PNG file or inserted with the
686
png_set_unknown_chunks() function.
688
The data from the pHYs chunk can be retrieved in several convenient
691
res_x = png_get_x_pixels_per_meter(png_ptr,
693
res_y = png_get_y_pixels_per_meter(png_ptr,
695
res_x_and_y = png_get_pixels_per_meter(png_ptr,
697
res_x = png_get_x_pixels_per_inch(png_ptr,
699
res_y = png_get_y_pixels_per_inch(png_ptr,
701
res_x_and_y = png_get_pixels_per_inch(png_ptr,
703
aspect_ratio = png_get_pixel_aspect_ratio(png_ptr,
706
(Each of these returns 0 [signifying "unknown"] if
707
the data is not present or if res_x is 0;
708
res_x_and_y is 0 if res_x != res_y)
710
The data from the oFFs chunk can be retrieved in several convenient
713
x_offset = png_get_x_offset_microns(png_ptr, info_ptr);
714
y_offset = png_get_y_offset_microns(png_ptr, info_ptr);
715
x_offset = png_get_x_offset_inches(png_ptr, info_ptr);
716
y_offset = png_get_y_offset_inches(png_ptr, info_ptr);
718
(Each of these returns 0 [signifying "unknown" if both
719
x and y are 0] if the data is not present or if the
720
chunk is present but the unit is the pixel)
722
For more information, see the png_info definition in png.h and the
723
PNG specification for chunk contents. Be careful with trusting
724
rowbytes, as some of the transformations could increase the space
725
needed to hold a row (expand, filler, gray_to_rgb, etc.).
726
See png_read_update_info(), below.
728
A quick word about text_ptr and num_text. PNG stores comments in
729
keyword/text pairs, one pair per chunk, with no limit on the number
730
of text chunks, and a 2^31 byte limit on their size. While there are
731
suggested keywords, there is no requirement to restrict the use to these
732
strings. It is strongly suggested that keywords and text be sensible
733
to humans (that's the point), so don't use abbreviations. Non-printing
734
symbols are not allowed. See the PNG specification for more details.
735
There is also no requirement to have text after the keyword.
737
Keywords should be limited to 79 Latin-1 characters without leading or
738
trailing spaces, but non-consecutive spaces are allowed within the
739
keyword. It is possible to have the same keyword any number of times.
740
The text_ptr is an array of png_text structures, each holding a
741
pointer to a language string, a pointer to a keyword and a pointer to
742
a text string. The text string, language code, and translated
743
keyword may be empty or NULL pointers. The keyword/text
744
pairs are put into the array in the order that they are received.
745
However, some or all of the text chunks may be after the image, so, to
746
make sure you have read all the text chunks, don't mess with these
747
until after you read the stuff after the image. This will be
748
mentioned again below in the discussion that goes with png_read_end().
750
Input transformations
752
After you've read the header information, you can set up the library
753
to handle any special transformations of the image data. The various
754
ways to transform the data will be described in the order that they
755
should occur. This is important, as some of these change the color
756
type and/or bit depth of the data, and some others only work on
757
certain color types and bit depths. Even though each transformation
758
checks to see if it has data that it can do something with, you should
759
make sure to only enable a transformation if it will be valid for the
760
data. For example, don't swap red and blue on grayscale data.
762
The colors used for the background and transparency values should be
763
supplied in the same format/depth as the current image data. They
764
are stored in the same format/depth as the image data in a bKGD or tRNS
765
chunk, so this is what libpng expects for this data. The colors are
766
transformed to keep in sync with the image data when an application
767
calls the png_read_update_info() routine (see below).
769
Data will be decoded into the supplied row buffers packed into bytes
770
unless the library has been told to transform it into another format.
771
For example, 4 bit/pixel paletted or grayscale data will be returned
772
2 pixels/byte with the leftmost pixel in the high-order bits of the
773
byte, unless png_set_packing() is called. 8-bit RGB data will be stored
774
in RGB RGB RGB format unless png_set_filler() or png_set_add_alpha()
775
is called to insert filler bytes, either before or after each RGB triplet.
776
16-bit RGB data will be returned RRGGBB RRGGBB, with the most significant
777
byte of the color value first, unless png_set_strip_16() is called to
778
transform it to regular RGB RGB triplets, or png_set_filler() or
779
png_set_add alpha() is called to insert filler bytes, either before or
780
after each RRGGBB triplet. Similarly, 8-bit or 16-bit grayscale data can
782
png_set_filler(), png_set_add_alpha(), or png_set_strip_16().
784
The following code transforms grayscale images of less than 8 to 8 bits,
785
changes paletted images to RGB, and adds a full alpha channel if there is
786
transparency information in a tRNS chunk. This is most useful on
787
grayscale images with bit depths of 2 or 4 or if there is a multiple-image
788
viewing application that wishes to treat all images in the same way.
790
if (color_type == PNG_COLOR_TYPE_PALETTE)
791
png_set_palette_to_rgb(png_ptr);
793
if (color_type == PNG_COLOR_TYPE_GRAY &&
794
bit_depth < 8) png_set_expand_gray_1_2_4_to_8(png_ptr);
796
if (png_get_valid(png_ptr, info_ptr,
797
PNG_INFO_tRNS)) png_set_tRNS_to_alpha(png_ptr);
799
These three functions are actually aliases for png_set_expand(), added
800
in libpng version 1.0.4, with the function names expanded to improve code
801
readability. In some future version they may actually do different
804
As of libpng version 1.2.9, png_set_expand_gray_1_2_4_to_8() was
805
added. It expands the sample depth without changing tRNS to alpha.
806
At the same time, png_set_gray_1_2_4_to_8() was deprecated, and it
807
will be removed from a future version.
809
PNG can have files with 16 bits per channel. If you only can handle
810
8 bits per channel, this will strip the pixels down to 8 bit.
813
png_set_strip_16(png_ptr);
815
If, for some reason, you don't need the alpha channel on an image,
816
and you want to remove it rather than combining it with the background
817
(but the image author certainly had in mind that you *would* combine
818
it with the background, so that's what you should probably do):
820
if (color_type & PNG_COLOR_MASK_ALPHA)
821
png_set_strip_alpha(png_ptr);
823
In PNG files, the alpha channel in an image
824
is the level of opacity. If you need the alpha channel in an image to
825
be the level of transparency instead of opacity, you can invert the
826
alpha channel (or the tRNS chunk data) after it's read, so that 0 is
827
fully opaque and 255 (in 8-bit or paletted images) or 65535 (in 16-bit
828
images) is fully transparent, with
830
png_set_invert_alpha(png_ptr);
832
PNG files pack pixels of bit depths 1, 2, and 4 into bytes as small as
833
they can, resulting in, for example, 8 pixels per byte for 1 bit
834
files. This code expands to 1 pixel per byte without changing the
835
values of the pixels:
838
png_set_packing(png_ptr);
840
PNG files have possible bit depths of 1, 2, 4, 8, and 16. All pixels
841
stored in a PNG image have been "scaled" or "shifted" up to the next
842
higher possible bit depth (e.g. from 5 bits/sample in the range [0,31] to
843
8 bits/sample in the range [0, 255]). However, it is also possible to
844
convert the PNG pixel data back to the original bit depth of the image.
845
This call reduces the pixels back down to the original bit depth:
847
png_color_8p sig_bit;
849
if (png_get_sBIT(png_ptr, info_ptr, &sig_bit))
850
png_set_shift(png_ptr, sig_bit);
852
PNG files store 3-color pixels in red, green, blue order. This code
853
changes the storage of the pixels to blue, green, red:
855
if (color_type == PNG_COLOR_TYPE_RGB ||
856
color_type == PNG_COLOR_TYPE_RGB_ALPHA)
857
png_set_bgr(png_ptr);
859
PNG files store RGB pixels packed into 3 or 6 bytes. This code expands them
860
into 4 or 8 bytes for windowing systems that need them in this format:
862
if (color_type == PNG_COLOR_TYPE_RGB)
863
png_set_filler(png_ptr, filler, PNG_FILLER_BEFORE);
865
where "filler" is the 8 or 16-bit number to fill with, and the location is
866
either PNG_FILLER_BEFORE or PNG_FILLER_AFTER, depending upon whether
867
you want the filler before the RGB or after. This transformation
868
does not affect images that already have full alpha channels. To add an
869
opaque alpha channel, use filler=0xff or 0xffff and PNG_FILLER_AFTER which
870
will generate RGBA pixels.
872
Note that png_set_filler() does not change the color type. If you want
873
to do that, you can add a true alpha channel with
875
if (color_type == PNG_COLOR_TYPE_RGB ||
876
color_type == PNG_COLOR_TYPE_GRAY)
877
png_set_add_alpha(png_ptr, filler, PNG_FILLER_AFTER);
879
where "filler" contains the alpha value to assign to each pixel.
880
This function was added in libpng-1.2.7.
882
If you are reading an image with an alpha channel, and you need the
883
data as ARGB instead of the normal PNG format RGBA:
885
if (color_type == PNG_COLOR_TYPE_RGB_ALPHA)
886
png_set_swap_alpha(png_ptr);
888
For some uses, you may want a grayscale image to be represented as
889
RGB. This code will do that conversion:
891
if (color_type == PNG_COLOR_TYPE_GRAY ||
892
color_type == PNG_COLOR_TYPE_GRAY_ALPHA)
893
png_set_gray_to_rgb(png_ptr);
895
Conversely, you can convert an RGB or RGBA image to grayscale or grayscale
898
if (color_type == PNG_COLOR_TYPE_RGB ||
899
color_type == PNG_COLOR_TYPE_RGB_ALPHA)
900
png_set_rgb_to_gray_fixed(png_ptr, error_action,
901
int red_weight, int green_weight);
903
error_action = 1: silently do the conversion
904
error_action = 2: issue a warning if the original
905
image has any pixel where
906
red != green or red != blue
907
error_action = 3: issue an error and abort the
908
conversion if the original
909
image has any pixel where
910
red != green or red != blue
912
red_weight: weight of red component times 100000
913
green_weight: weight of green component times 100000
914
If either weight is negative, default
915
weights (21268, 71514) are used.
917
If you have set error_action = 1 or 2, you can
918
later check whether the image really was gray, after processing
919
the image rows, with the png_get_rgb_to_gray_status(png_ptr) function.
920
It will return a png_byte that is zero if the image was gray or
921
1 if there were any non-gray pixels. bKGD and sBIT data
922
will be silently converted to grayscale, using the green channel
923
data, regardless of the error_action setting.
925
With red_weight+green_weight<=100000,
926
the normalized graylevel is computed:
928
int rw = red_weight * 65536;
929
int gw = green_weight * 65536;
930
int bw = 65536 - (rw + gw);
931
gray = (rw*red + gw*green + bw*blue)/65536;
933
The default values approximate those recommended in the Charles
934
Poynton's Color FAQ, <http://www.inforamp.net/~poynton/>
935
Copyright (c) 1998-01-04 Charles Poynton <poynton at inforamp.net>
937
Y = 0.212671 * R + 0.715160 * G + 0.072169 * B
939
Libpng approximates this with
941
Y = 0.21268 * R + 0.7151 * G + 0.07217 * B
943
which can be expressed with integers as
945
Y = (6969 * R + 23434 * G + 2365 * B)/32768
947
The calculation is done in a linear colorspace, if the image gamma
950
If you have a grayscale and you are using png_set_expand_depth(),
951
png_set_expand(), or png_set_gray_to_rgb to change to truecolor or to
952
a higher bit-depth, you must either supply the background color as a gray
953
value at the original file bit-depth (need_expand = 1) or else supply the
954
background color as an RGB triplet at the final, expanded bit depth
955
(need_expand = 0). Similarly, if you are reading a paletted image, you
956
must either supply the background color as a palette index (need_expand = 1)
957
or as an RGB triplet that may or may not be in the palette (need_expand = 0).
959
png_color_16 my_background;
960
png_color_16p image_background;
962
if (png_get_bKGD(png_ptr, info_ptr, &image_background))
963
png_set_background(png_ptr, image_background,
964
PNG_BACKGROUND_GAMMA_FILE, 1, 1.0);
966
png_set_background(png_ptr, &my_background,
967
PNG_BACKGROUND_GAMMA_SCREEN, 0, 1.0);
969
The png_set_background() function tells libpng to composite images
970
with alpha or simple transparency against the supplied background
971
color. If the PNG file contains a bKGD chunk (PNG_INFO_bKGD valid),
972
you may use this color, or supply another color more suitable for
973
the current display (e.g., the background color from a web page). You
974
need to tell libpng whether the color is in the gamma space of the
975
display (PNG_BACKGROUND_GAMMA_SCREEN for colors you supply), the file
976
(PNG_BACKGROUND_GAMMA_FILE for colors from the bKGD chunk), or one
977
that is neither of these gammas (PNG_BACKGROUND_GAMMA_UNIQUE - I don't
978
know why anyone would use this, but it's here).
980
To properly display PNG images on any kind of system, the application needs
981
to know what the display gamma is. Ideally, the user will know this, and
982
the application will allow them to set it. One method of allowing the user
983
to set the display gamma separately for each system is to check for a
984
SCREEN_GAMMA or DISPLAY_GAMMA environment variable, which will hopefully be
987
Note that display_gamma is the overall gamma correction required to produce
988
pleasing results, which depends on the lighting conditions in the surrounding
989
environment. In a dim or brightly lit room, no compensation other than
990
the physical gamma exponent of the monitor is needed, while in a dark room
991
a slightly smaller exponent is better.
993
double gamma, screen_gamma;
995
if (/* We have a user-defined screen
998
screen_gamma = user_defined_screen_gamma;
1000
/* One way that applications can share the same
1001
screen gamma value */
1002
else if ((gamma_str = getenv("SCREEN_GAMMA"))
1005
screen_gamma = (double)atof(gamma_str);
1007
/* If we don't have another value */
1010
screen_gamma = 2.2; /* A good guess for a
1011
PC monitor in a bright office or a dim room */
1012
screen_gamma = 2.0; /* A good guess for a
1013
PC monitor in a dark room */
1014
screen_gamma = 1.7 or 1.0; /* A good
1015
guess for Mac systems */
1018
The png_set_gamma() function handles gamma transformations of the data.
1019
Pass both the file gamma and the current screen_gamma. If the file does
1020
not have a gamma value, you can pass one anyway if you have an idea what
1021
it is (usually 0.45455 is a good guess for GIF images on PCs). Note
1022
that file gammas are inverted from screen gammas. See the discussions
1023
on gamma in the PNG specification for an excellent description of what
1024
gamma is, and why all applications should support it. It is strongly
1025
recommended that PNG viewers support gamma correction.
1027
if (png_get_gAMA(png_ptr, info_ptr, &gamma))
1028
png_set_gamma(png_ptr, screen_gamma, gamma);
1030
png_set_gamma(png_ptr, screen_gamma, 0.45455);
1032
If you need to reduce an RGB file to a paletted file, or if a paletted
1033
file has more entries then will fit on your screen, png_set_dither()
1034
will do that. Note that this is a simple match dither that merely
1035
finds the closest color available. This should work fairly well with
1036
optimized palettes, and fairly badly with linear color cubes. If you
1037
pass a palette that is larger then maximum_colors, the file will
1038
reduce the number of colors in the palette so it will fit into
1039
maximum_colors. If there is a histogram, it will use it to make
1040
more intelligent choices when reducing the palette. If there is no
1041
histogram, it may not do as good a job.
1043
if (color_type & PNG_COLOR_MASK_COLOR)
1045
if (png_get_valid(png_ptr, info_ptr,
1048
png_uint_16p histogram = NULL;
1050
png_get_hIST(png_ptr, info_ptr,
1052
png_set_dither(png_ptr, palette, num_palette,
1053
max_screen_colors, histogram, 1);
1057
png_color std_color_cube[MAX_SCREEN_COLORS] =
1060
png_set_dither(png_ptr, std_color_cube,
1061
MAX_SCREEN_COLORS, MAX_SCREEN_COLORS,
1066
PNG files describe monochrome as black being zero and white being one.
1067
The following code will reverse this (make black be one and white be
1070
if (bit_depth == 1 && color_type == PNG_COLOR_TYPE_GRAY)
1071
png_set_invert_mono(png_ptr);
1073
This function can also be used to invert grayscale and gray-alpha images:
1075
if (color_type == PNG_COLOR_TYPE_GRAY ||
1076
color_type == PNG_COLOR_TYPE_GRAY_ALPHA)
1077
png_set_invert_mono(png_ptr);
1079
PNG files store 16 bit pixels in network byte order (big-endian,
1080
ie. most significant bits first). This code changes the storage to the
1081
other way (little-endian, i.e. least significant bits first, the
1082
way PCs store them):
1084
if (bit_depth == 16)
1085
png_set_swap(png_ptr);
1087
If you are using packed-pixel images (1, 2, or 4 bits/pixel), and you
1088
need to change the order the pixels are packed into bytes, you can use:
1091
png_set_packswap(png_ptr);
1093
Finally, you can write your own transformation function if none of
1094
the existing ones meets your needs. This is done by setting a callback
1097
png_set_read_user_transform_fn(png_ptr,
1100
You must supply the function
1102
void read_transform_fn(png_ptr ptr, row_info_ptr
1103
row_info, png_bytep data)
1105
See pngtest.c for a working example. Your function will be called
1106
after all of the other transformations have been processed.
1108
You can also set up a pointer to a user structure for use by your
1109
callback function, and you can inform libpng that your transform
1110
function will change the number of channels or bit depth with the
1113
png_set_user_transform_info(png_ptr, user_ptr,
1114
user_depth, user_channels);
1116
The user's application, not libpng, is responsible for allocating and
1117
freeing any memory required for the user structure.
1119
You can retrieve the pointer via the function
1120
png_get_user_transform_ptr(). For example:
1122
voidp read_user_transform_ptr =
1123
png_get_user_transform_ptr(png_ptr);
1125
The last thing to handle is interlacing; this is covered in detail below,
1126
but you must call the function here if you want libpng to handle expansion
1127
of the interlaced image.
1129
number_of_passes = png_set_interlace_handling(png_ptr);
1131
After setting the transformations, libpng can update your png_info
1132
structure to reflect any transformations you've requested with this
1133
call. This is most useful to update the info structure's rowbytes
1134
field so you can use it to allocate your image memory. This function
1135
will also update your palette with the correct screen_gamma and
1136
background if these have been given with the calls above.
1138
png_read_update_info(png_ptr, info_ptr);
1140
After you call png_read_update_info(), you can allocate any
1141
memory you need to hold the image. The row data is simply
1142
raw byte data for all forms of images. As the actual allocation
1143
varies among applications, no example will be given. If you
1144
are allocating one large chunk, you will need to build an
1145
array of pointers to each row, as it will be needed for some
1146
of the functions below.
1150
After you've allocated memory, you can read the image data.
1151
The simplest way to do this is in one function call. If you are
1152
allocating enough memory to hold the whole image, you can just
1153
call png_read_image() and libpng will read in all the image data
1154
and put it in the memory area supplied. You will need to pass in
1155
an array of pointers to each row.
1157
This function automatically handles interlacing, so you don't need
1158
to call png_set_interlace_handling() or call this function multiple
1159
times, or any of that other stuff necessary with png_read_rows().
1161
png_read_image(png_ptr, row_pointers);
1163
where row_pointers is:
1165
png_bytep row_pointers[height];
1167
You can point to void or char or whatever you use for pixels.
1169
If you don't want to read in the whole image at once, you can
1170
use png_read_rows() instead. If there is no interlacing (check
1171
interlace_type == PNG_INTERLACE_NONE), this is simple:
1173
png_read_rows(png_ptr, row_pointers, NULL,
1176
where row_pointers is the same as in the png_read_image() call.
1178
If you are doing this just one row at a time, you can do this with
1179
a single row_pointer instead of an array of row_pointers:
1181
png_bytep row_pointer = row;
1182
png_read_row(png_ptr, row_pointer, NULL);
1184
If the file is interlaced (interlace_type != 0 in the IHDR chunk), things
1185
get somewhat harder. The only current (PNG Specification version 1.2)
1186
interlacing type for PNG is (interlace_type == PNG_INTERLACE_ADAM7)
1187
is a somewhat complicated 2D interlace scheme, known as Adam7, that
1188
breaks down an image into seven smaller images of varying size, based
1191
libpng can fill out those images or it can give them to you "as is".
1192
If you want them filled out, there are two ways to do that. The one
1193
mentioned in the PNG specification is to expand each pixel to cover
1194
those pixels that have not been read yet (the "rectangle" method).
1195
This results in a blocky image for the first pass, which gradually
1196
smooths out as more pixels are read. The other method is the "sparkle"
1197
method, where pixels are drawn only in their final locations, with the
1198
rest of the image remaining whatever colors they were initialized to
1199
before the start of the read. The first method usually looks better,
1200
but tends to be slower, as there are more pixels to put in the rows.
1202
If you don't want libpng to handle the interlacing details, just call
1203
png_read_rows() seven times to read in all seven images. Each of the
1204
images is a valid image by itself, or they can all be combined on an
1205
8x8 grid to form a single image (although if you intend to combine them
1206
you would be far better off using the libpng interlace handling).
1208
The first pass will return an image 1/8 as wide as the entire image
1209
(every 8th column starting in column 0) and 1/8 as high as the original
1210
(every 8th row starting in row 0), the second will be 1/8 as wide
1211
(starting in column 4) and 1/8 as high (also starting in row 0). The
1212
third pass will be 1/4 as wide (every 4th pixel starting in column 0) and
1213
1/8 as high (every 8th row starting in row 4), and the fourth pass will
1214
be 1/4 as wide and 1/4 as high (every 4th column starting in column 2,
1215
and every 4th row starting in row 0). The fifth pass will return an
1216
image 1/2 as wide, and 1/4 as high (starting at column 0 and row 2),
1217
while the sixth pass will be 1/2 as wide and 1/2 as high as the original
1218
(starting in column 1 and row 0). The seventh and final pass will be as
1219
wide as the original, and 1/2 as high, containing all of the odd
1220
numbered scanlines. Phew!
1222
If you want libpng to expand the images, call this before calling
1223
png_start_read_image() or png_read_update_info():
1225
if (interlace_type == PNG_INTERLACE_ADAM7)
1227
= png_set_interlace_handling(png_ptr);
1229
This will return the number of passes needed. Currently, this
1230
is seven, but may change if another interlace type is added.
1231
This function can be called even if the file is not interlaced,
1232
where it will return one pass.
1234
If you are not going to display the image after each pass, but are
1235
going to wait until the entire image is read in, use the sparkle
1236
effect. This effect is faster and the end result of either method
1237
is exactly the same. If you are planning on displaying the image
1238
after each pass, the "rectangle" effect is generally considered the
1241
If you only want the "sparkle" effect, just call png_read_rows() as
1242
normal, with the third parameter NULL. Make sure you make pass over
1243
the image number_of_passes times, and you don't change the data in the
1244
rows between calls. You can change the locations of the data, just
1245
not the data. Each pass only writes the pixels appropriate for that
1246
pass, and assumes the data from previous passes is still valid.
1248
png_read_rows(png_ptr, row_pointers, NULL,
1251
If you only want the first effect (the rectangles), do the same as
1252
before except pass the row buffer in the third parameter, and leave
1253
the second parameter NULL.
1255
png_read_rows(png_ptr, NULL, row_pointers,
1258
Finishing a sequential read
1260
After you are finished reading the image through the
1261
low-level interface, you can finish reading the file. If you are
1262
interested in comments or time, which may be stored either before or
1263
after the image data, you should pass the separate png_info struct if
1264
you want to keep the comments from before and after the image
1265
separate. If you are not interested, you can pass NULL.
1267
png_read_end(png_ptr, end_info);
1269
When you are done, you can free all memory allocated by libpng like this:
1271
png_destroy_read_struct(&png_ptr, &info_ptr,
1274
It is also possible to individually free the info_ptr members that
1275
point to libpng-allocated storage with the following function:
1277
png_free_data(png_ptr, info_ptr, mask, seq)
1278
mask - identifies data to be freed, a mask
1279
containing the bitwise OR of one or
1281
PNG_FREE_PLTE, PNG_FREE_TRNS,
1282
PNG_FREE_HIST, PNG_FREE_ICCP,
1283
PNG_FREE_PCAL, PNG_FREE_ROWS,
1284
PNG_FREE_SCAL, PNG_FREE_SPLT,
1285
PNG_FREE_TEXT, PNG_FREE_UNKN,
1286
or simply PNG_FREE_ALL
1287
seq - sequence number of item to be freed
1290
This function may be safely called when the relevant storage has
1291
already been freed, or has not yet been allocated, or was allocated
1292
by the user and not by libpng, and will in those
1293
cases do nothing. The "seq" parameter is ignored if only one item
1294
of the selected data type, such as PLTE, is allowed. If "seq" is not
1295
-1, and multiple items are allowed for the data type identified in
1296
the mask, such as text or sPLT, only the n'th item in the structure
1297
is freed, where n is "seq".
1299
The default behavior is only to free data that was allocated internally
1300
by libpng. This can be changed, so that libpng will not free the data,
1301
or so that it will free data that was allocated by the user with png_malloc()
1302
or png_zalloc() and passed in via a png_set_*() function, with
1304
png_data_freer(png_ptr, info_ptr, freer, mask)
1305
mask - which data elements are affected
1306
same choices as in png_free_data()
1308
PNG_DESTROY_WILL_FREE_DATA
1309
PNG_SET_WILL_FREE_DATA
1310
PNG_USER_WILL_FREE_DATA
1312
This function only affects data that has already been allocated.
1313
You can call this function after reading the PNG data but before calling
1314
any png_set_*() functions, to control whether the user or the png_set_*()
1315
function is responsible for freeing any existing data that might be present,
1316
and again after the png_set_*() functions to control whether the user
1317
or png_destroy_*() is supposed to free the data. When the user assumes
1318
responsibility for libpng-allocated data, the application must use
1319
png_free() to free it, and when the user transfers responsibility to libpng
1320
for data that the user has allocated, the user must have used png_malloc()
1321
or png_zalloc() to allocate it.
1323
If you allocated your row_pointers in a single block, as suggested above in
1324
the description of the high level read interface, you must not transfer
1325
responsibility for freeing it to the png_set_rows or png_read_destroy function,
1326
because they would also try to free the individual row_pointers[i].
1328
If you allocated text_ptr.text, text_ptr.lang, and text_ptr.translated_keyword
1329
separately, do not transfer responsibility for freeing text_ptr to libpng,
1330
because when libpng fills a png_text structure it combines these members with
1331
the key member, and png_free_data() will free only text_ptr.key. Similarly,
1332
if you transfer responsibility for free'ing text_ptr from libpng to your
1333
application, your application must not separately free those members.
1335
The png_free_data() function will turn off the "valid" flag for anything
1336
it frees. If you need to turn the flag off for a chunk that was freed by your
1337
application instead of by libpng, you can use
1339
png_set_invalid(png_ptr, info_ptr, mask);
1340
mask - identifies the chunks to be made invalid,
1341
containing the bitwise OR of one or
1343
PNG_INFO_gAMA, PNG_INFO_sBIT,
1344
PNG_INFO_cHRM, PNG_INFO_PLTE,
1345
PNG_INFO_tRNS, PNG_INFO_bKGD,
1346
PNG_INFO_hIST, PNG_INFO_pHYs,
1347
PNG_INFO_oFFs, PNG_INFO_tIME,
1348
PNG_INFO_pCAL, PNG_INFO_sRGB,
1349
PNG_INFO_iCCP, PNG_INFO_sPLT,
1350
PNG_INFO_sCAL, PNG_INFO_IDAT
1352
For a more compact example of reading a PNG image, see the file example.c.
1354
Reading PNG files progressively
1356
The progressive reader is slightly different then the non-progressive
1357
reader. Instead of calling png_read_info(), png_read_rows(), and
1358
png_read_end(), you make one call to png_process_data(), which calls
1359
callbacks when it has the info, a row, or the end of the image. You
1360
set up these callbacks with png_set_progressive_read_fn(). You don't
1361
have to worry about the input/output functions of libpng, as you are
1362
giving the library the data directly in png_process_data(). I will
1363
assume that you have read the section on reading PNG files above,
1364
so I will only highlight the differences (although I will show
1367
png_structp png_ptr;
1370
/* An example code fragment of how you would
1371
initialize the progressive reader in your
1374
initialize_png_reader()
1376
png_ptr = png_create_read_struct
1377
(PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
1378
user_error_fn, user_warning_fn);
1381
info_ptr = png_create_info_struct(png_ptr);
1384
png_destroy_read_struct(&png_ptr, (png_infopp)NULL,
1389
if (setjmp(png_jmpbuf(png_ptr)))
1391
png_destroy_read_struct(&png_ptr, &info_ptr,
1396
/* This one's new. You can provide functions
1397
to be called when the header info is valid,
1398
when each row is completed, and when the image
1399
is finished. If you aren't using all functions,
1400
you can specify NULL parameters. Even when all
1401
three functions are NULL, you need to call
1402
png_set_progressive_read_fn(). You can use
1403
any struct as the user_ptr (cast to a void pointer
1404
for the function call), and retrieve the pointer
1405
from inside the callbacks using the function
1407
png_get_progressive_ptr(png_ptr);
1409
which will return a void pointer, which you have
1410
to cast appropriately.
1412
png_set_progressive_read_fn(png_ptr, (void *)user_ptr,
1413
info_callback, row_callback, end_callback);
1418
/* A code fragment that you call as you receive blocks
1421
process_data(png_bytep buffer, png_uint_32 length)
1423
if (setjmp(png_jmpbuf(png_ptr)))
1425
png_destroy_read_struct(&png_ptr, &info_ptr,
1430
/* This one's new also. Simply give it a chunk
1431
of data from the file stream (in order, of
1432
course). On machines with segmented memory
1433
models machines, don't give it any more than
1434
64K. The library seems to run fine with sizes
1435
of 4K. Although you can give it much less if
1436
necessary (I assume you can give it chunks of
1437
1 byte, I haven't tried less then 256 bytes
1438
yet). When this function returns, you may
1439
want to display any rows that were generated
1440
in the row callback if you don't already do
1443
png_process_data(png_ptr, info_ptr, buffer, length);
1447
/* This function is called (as set by
1448
png_set_progressive_read_fn() above) when enough data
1449
has been supplied so all of the header has been
1453
info_callback(png_structp png_ptr, png_infop info)
1455
/* Do any setup here, including setting any of
1456
the transformations mentioned in the Reading
1457
PNG files section. For now, you _must_ call
1458
either png_start_read_image() or
1459
png_read_update_info() after all the
1460
transformations are set (even if you don't set
1461
any). You may start getting rows before
1462
png_process_data() returns, so this is your
1463
last chance to prepare for that.
1467
/* This function is called when each row of image
1470
row_callback(png_structp png_ptr, png_bytep new_row,
1471
png_uint_32 row_num, int pass)
1473
/* If the image is interlaced, and you turned
1474
on the interlace handler, this function will
1475
be called for every row in every pass. Some
1476
of these rows will not be changed from the
1477
previous pass. When the row is not changed,
1478
the new_row variable will be NULL. The rows
1479
and passes are called in order, so you don't
1480
really need the row_num and pass, but I'm
1481
supplying them because it may make your life
1484
For the non-NULL rows of interlaced images,
1485
you must call png_progressive_combine_row()
1486
passing in the row and the old row. You can
1487
call this function for NULL rows (it will just
1488
return) and for non-interlaced images (it just
1489
does the memcpy for you) if it will make the
1490
code easier. Thus, you can just do this for
1494
png_progressive_combine_row(png_ptr, old_row,
1497
/* where old_row is what was displayed for
1498
previously for the row. Note that the first
1499
pass (pass == 0, really) will completely cover
1500
the old row, so the rows do not have to be
1501
initialized. After the first pass (and only
1502
for interlaced images), you will have to pass
1503
the current row, and the function will combine
1504
the old row and the new row.
1509
end_callback(png_structp png_ptr, png_infop info)
1511
/* This function is called after the whole image
1512
has been read, including any chunks after the
1513
image (up to and including the IEND). You
1514
will usually have the same info chunk as you
1515
had in the header, although some data may have
1516
been added to the comments and time fields.
1518
Most people won't do much here, perhaps setting
1519
a flag that marks the image as finished.
1527
Much of this is very similar to reading. However, everything of
1528
importance is repeated here, so you won't have to constantly look
1529
back up in the reading section to understand writing.
1533
You will want to do the I/O initialization before you get into libpng,
1534
so if it doesn't work, you don't have anything to undo. If you are not
1535
using the standard I/O functions, you will need to replace them with
1536
custom writing functions. See the discussion under Customizing libpng.
1538
FILE *fp = fopen(file_name, "wb");
1544
Next, png_struct and png_info need to be allocated and initialized.
1545
As these can be both relatively large, you may not want to store these
1546
on the stack, unless you have stack space to spare. Of course, you
1547
will want to check if they return NULL. If you are also reading,
1548
you won't want to name your read structure and your write structure
1549
both "png_ptr"; you can call them anything you like, such as
1550
"read_ptr" and "write_ptr". Look at pngtest.c, for example.
1552
png_structp png_ptr = png_create_write_struct
1553
(PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
1554
user_error_fn, user_warning_fn);
1558
png_infop info_ptr = png_create_info_struct(png_ptr);
1561
png_destroy_write_struct(&png_ptr,
1566
If you want to use your own memory allocation routines,
1567
define PNG_USER_MEM_SUPPORTED and use
1568
png_create_write_struct_2() instead of png_create_write_struct():
1570
png_structp png_ptr = png_create_write_struct_2
1571
(PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
1572
user_error_fn, user_warning_fn, (png_voidp)
1573
user_mem_ptr, user_malloc_fn, user_free_fn);
1575
After you have these structures, you will need to set up the
1576
error handling. When libpng encounters an error, it expects to
1577
longjmp() back to your routine. Therefore, you will need to call
1578
setjmp() and pass the png_jmpbuf(png_ptr). If you
1579
write the file from different routines, you will need to update
1580
the png_jmpbuf(png_ptr) every time you enter a new routine that will
1581
call a png_*() function. See your documentation of setjmp/longjmp
1582
for your compiler for more information on setjmp/longjmp. See
1583
the discussion on libpng error handling in the Customizing Libpng
1584
section below for more information on the libpng error handling.
1586
if (setjmp(png_jmpbuf(png_ptr)))
1588
png_destroy_write_struct(&png_ptr, &info_ptr);
1595
If you would rather avoid the complexity of setjmp/longjmp issues,
1596
you can compile libpng with PNG_SETJMP_NOT_SUPPORTED, in which case
1597
errors will result in a call to PNG_ABORT() which defaults to abort().
1599
Now you need to set up the output code. The default for libpng is to
1600
use the C function fwrite(). If you use this, you will need to pass a
1601
valid FILE * in the function png_init_io(). Be sure that the file is
1602
opened in binary mode. Again, if you wish to handle writing data in
1603
another way, see the discussion on libpng I/O handling in the Customizing
1604
Libpng section below.
1606
png_init_io(png_ptr, fp);
1608
If you are embedding your PNG into a datastream such as MNG, and don't
1609
want libpng to write the 8-byte signature, or if you have already
1610
written the signature in your application, use
1612
png_set_sig_bytes(png_ptr, 8);
1614
to inform libpng that it should not write a signature.
1618
At this point, you can set up a callback function that will be
1619
called after each row has been written, which you can use to control
1620
a progress meter or the like. It's demonstrated in pngtest.c.
1621
You must supply a function
1623
void write_row_callback(png_ptr, png_uint_32 row,
1626
/* put your code here */
1629
(You can give it another name that you like instead of "write_row_callback")
1631
To inform libpng about your function, use
1633
png_set_write_status_fn(png_ptr, write_row_callback);
1635
You now have the option of modifying how the compression library will
1636
run. The following functions are mainly for testing, but may be useful
1637
in some cases, like if you need to write PNG files extremely fast and
1638
are willing to give up some compression, or if you want to get the
1639
maximum possible compression at the expense of slower writing. If you
1640
have no special needs in this area, let the library do what it wants by
1641
not calling this function at all, as it has been tuned to deliver a good
1642
speed/compression ratio. The second parameter to png_set_filter() is
1643
the filter method, for which the only valid values are 0 (as of the
1644
July 1999 PNG specification, version 1.2) or 64 (if you are writing
1645
a PNG datastream that is to be embedded in a MNG datastream). The third
1646
parameter is a flag that indicates which filter type(s) are to be tested
1647
for each scanline. See the PNG specification for details on the specific filter
1651
/* turn on or off filtering, and/or choose
1652
specific filters. You can use either a single
1653
PNG_FILTER_VALUE_NAME or the bitwise OR of one
1654
or more PNG_FILTER_NAME masks. */
1655
png_set_filter(png_ptr, 0,
1656
PNG_FILTER_NONE | PNG_FILTER_VALUE_NONE |
1657
PNG_FILTER_SUB | PNG_FILTER_VALUE_SUB |
1658
PNG_FILTER_UP | PNG_FILTER_VALUE_UP |
1659
PNG_FILTER_AVE | PNG_FILTER_VALUE_AVE |
1660
PNG_FILTER_PAETH | PNG_FILTER_VALUE_PAETH|
1664
wants to start and stop using particular filters during compression,
1665
it should start out with all of the filters (to ensure that the previous
1666
row of pixels will be stored in case it's needed later), and then add
1667
and remove them after the start of compression.
1669
If you are writing a PNG datastream that is to be embedded in a MNG
1670
datastream, the second parameter can be either 0 or 64.
1672
The png_set_compression_*() functions interface to the zlib compression
1673
library, and should mostly be ignored unless you really know what you are
1674
doing. The only generally useful call is png_set_compression_level()
1675
which changes how much time zlib spends on trying to compress the image
1676
data. See the Compression Library (zlib.h and algorithm.txt, distributed
1677
with zlib) for details on the compression levels.
1679
/* set the zlib compression level */
1680
png_set_compression_level(png_ptr,
1681
Z_BEST_COMPRESSION);
1683
/* set other zlib parameters */
1684
png_set_compression_mem_level(png_ptr, 8);
1685
png_set_compression_strategy(png_ptr,
1686
Z_DEFAULT_STRATEGY);
1687
png_set_compression_window_bits(png_ptr, 15);
1688
png_set_compression_method(png_ptr, 8);
1689
png_set_compression_buffer_size(png_ptr, 8192)
1691
extern PNG_EXPORT(void,png_set_zbuf_size)
1693
Setting the contents of info for output
1695
You now need to fill in the png_info structure with all the data you
1696
wish to write before the actual image. Note that the only thing you
1697
are allowed to write after the image is the text chunks and the time
1698
chunk (as of PNG Specification 1.2, anyway). See png_write_end() and
1699
the latest PNG specification for more information on that. If you
1700
wish to write them before the image, fill them in now, and flag that
1701
data as being valid. If you want to wait until after the data, don't
1702
fill them until png_write_end(). For all the fields in png_info and
1703
their data types, see png.h. For explanations of what the fields
1704
contain, see the PNG specification.
1706
Some of the more important parts of the png_info are:
1708
png_set_IHDR(png_ptr, info_ptr, width, height,
1709
bit_depth, color_type, interlace_type,
1710
compression_type, filter_method)
1711
width - holds the width of the image
1712
in pixels (up to 2^31).
1713
height - holds the height of the image
1714
in pixels (up to 2^31).
1715
bit_depth - holds the bit depth of one of the
1717
(valid values are 1, 2, 4, 8, 16
1718
and depend also on the
1719
color_type. See also significant
1721
color_type - describes which color/alpha
1722
channels are present.
1724
(bit depths 1, 2, 4, 8, 16)
1725
PNG_COLOR_TYPE_GRAY_ALPHA
1727
PNG_COLOR_TYPE_PALETTE
1728
(bit depths 1, 2, 4, 8)
1731
PNG_COLOR_TYPE_RGB_ALPHA
1734
PNG_COLOR_MASK_PALETTE
1735
PNG_COLOR_MASK_COLOR
1736
PNG_COLOR_MASK_ALPHA
1738
interlace_type - PNG_INTERLACE_NONE or
1740
compression_type - (must be
1741
PNG_COMPRESSION_TYPE_DEFAULT)
1742
filter_method - (must be PNG_FILTER_TYPE_DEFAULT
1743
or, if you are writing a PNG to
1744
be embedded in a MNG datastream,
1746
PNG_INTRAPIXEL_DIFFERENCING)
1748
png_set_PLTE(png_ptr, info_ptr, palette,
1750
palette - the palette for the file
1751
(array of png_color)
1752
num_palette - number of entries in the palette
1754
png_set_gAMA(png_ptr, info_ptr, gamma);
1755
gamma - the gamma the image was created
1758
png_set_sRGB(png_ptr, info_ptr, srgb_intent);
1759
srgb_intent - the rendering intent
1760
(PNG_INFO_sRGB) The presence of
1761
the sRGB chunk means that the pixel
1762
data is in the sRGB color space.
1763
This chunk also implies specific
1764
values of gAMA and cHRM. Rendering
1765
intent is the CSS-1 property that
1766
has been defined by the International
1768
(http://www.color.org).
1770
PNG_sRGB_INTENT_SATURATION,
1771
PNG_sRGB_INTENT_PERCEPTUAL,
1772
PNG_sRGB_INTENT_ABSOLUTE, or
1773
PNG_sRGB_INTENT_RELATIVE.
1776
png_set_sRGB_gAMA_and_cHRM(png_ptr, info_ptr,
1778
srgb_intent - the rendering intent
1779
(PNG_INFO_sRGB) The presence of the
1780
sRGB chunk means that the pixel
1781
data is in the sRGB color space.
1782
This function also causes gAMA and
1783
cHRM chunks with the specific values
1784
that are consistent with sRGB to be
1787
png_set_iCCP(png_ptr, info_ptr, name, compression_type,
1789
name - The profile name.
1790
compression - The compression type; always
1791
PNG_COMPRESSION_TYPE_BASE for PNG 1.0.
1792
You may give NULL to this argument to
1794
profile - International Color Consortium color
1795
profile data. May contain NULs.
1796
proflen - length of profile data in bytes.
1798
png_set_sBIT(png_ptr, info_ptr, sig_bit);
1799
sig_bit - the number of significant bits for
1800
(PNG_INFO_sBIT) each of the gray, red,
1801
green, and blue channels, whichever are
1802
appropriate for the given color type
1805
png_set_tRNS(png_ptr, info_ptr, trans, num_trans,
1807
trans - array of transparent entries for
1808
palette (PNG_INFO_tRNS)
1809
trans_values - graylevel or color sample values of
1810
the single transparent color for
1811
non-paletted images (PNG_INFO_tRNS)
1812
num_trans - number of transparent entries
1815
png_set_hIST(png_ptr, info_ptr, hist);
1817
hist - histogram of palette (array of
1820
png_set_tIME(png_ptr, info_ptr, mod_time);
1821
mod_time - time image was last modified
1824
png_set_bKGD(png_ptr, info_ptr, background);
1825
background - background color (PNG_VALID_bKGD)
1827
png_set_text(png_ptr, info_ptr, text_ptr, num_text);
1828
text_ptr - array of png_text holding image
1830
text_ptr[i].compression - type of compression used
1831
on "text" PNG_TEXT_COMPRESSION_NONE
1832
PNG_TEXT_COMPRESSION_zTXt
1833
PNG_ITXT_COMPRESSION_NONE
1834
PNG_ITXT_COMPRESSION_zTXt
1835
text_ptr[i].key - keyword for comment. Must contain
1837
text_ptr[i].text - text comments for current
1838
keyword. Can be NULL or empty.
1839
text_ptr[i].text_length - length of text string,
1840
after decompression, 0 for iTXt
1841
text_ptr[i].itxt_length - length of itxt string,
1842
after decompression, 0 for tEXt/zTXt
1843
text_ptr[i].lang - language of comment (NULL or
1845
text_ptr[i].translated_keyword - keyword in UTF-8 (NULL
1846
or empty for unknown).
1847
num_text - number of comments
1849
png_set_sPLT(png_ptr, info_ptr, &palette_ptr,
1851
palette_ptr - array of png_sPLT_struct structures
1852
to be added to the list of palettes
1853
in the info structure.
1854
num_spalettes - number of palette structures to be
1857
png_set_oFFs(png_ptr, info_ptr, offset_x, offset_y,
1859
offset_x - positive offset from the left
1861
offset_y - positive offset from the top
1863
unit_type - PNG_OFFSET_PIXEL, PNG_OFFSET_MICROMETER
1865
png_set_pHYs(png_ptr, info_ptr, res_x, res_y,
1867
res_x - pixels/unit physical resolution
1869
res_y - pixels/unit physical resolution
1871
unit_type - PNG_RESOLUTION_UNKNOWN,
1872
PNG_RESOLUTION_METER
1874
png_set_sCAL(png_ptr, info_ptr, unit, width, height)
1875
unit - physical scale units (an integer)
1876
width - width of a pixel in physical scale units
1877
height - height of a pixel in physical scale units
1878
(width and height are doubles)
1880
png_set_sCAL_s(png_ptr, info_ptr, unit, width, height)
1881
unit - physical scale units (an integer)
1882
width - width of a pixel in physical scale units
1883
height - height of a pixel in physical scale units
1884
(width and height are strings like "2.54")
1886
png_set_unknown_chunks(png_ptr, info_ptr, &unknowns,
1888
unknowns - array of png_unknown_chunk
1889
structures holding unknown chunks
1890
unknowns[i].name - name of unknown chunk
1891
unknowns[i].data - data of unknown chunk
1892
unknowns[i].size - size of unknown chunk's data
1893
unknowns[i].location - position to write chunk in file
1894
0: do not write chunk
1895
PNG_HAVE_IHDR: before PLTE
1896
PNG_HAVE_PLTE: before IDAT
1897
PNG_AFTER_IDAT: after IDAT
1899
The "location" member is set automatically according to
1900
what part of the output file has already been written.
1901
You can change its value after calling png_set_unknown_chunks()
1902
as demonstrated in pngtest.c. Within each of the "locations",
1903
the chunks are sequenced according to their position in the
1904
structure (that is, the value of "i", which is the order in which
1905
the chunk was either read from the input file or defined with
1906
png_set_unknown_chunks).
1908
A quick word about text and num_text. text is an array of png_text
1909
structures. num_text is the number of valid structures in the array.
1910
Each png_text structure holds a language code, a keyword, a text value,
1911
and a compression type.
1913
The compression types have the same valid numbers as the compression
1914
types of the image data. Currently, the only valid number is zero.
1915
However, you can store text either compressed or uncompressed, unlike
1916
images, which always have to be compressed. So if you don't want the
1917
text compressed, set the compression type to PNG_TEXT_COMPRESSION_NONE.
1918
Because tEXt and zTXt chunks don't have a language field, if you
1919
specify PNG_TEXT_COMPRESSION_NONE or PNG_TEXT_COMPRESSION_zTXt
1920
any language code or translated keyword will not be written out.
1922
Until text gets around 1000 bytes, it is not worth compressing it.
1923
After the text has been written out to the file, the compression type
1924
is set to PNG_TEXT_COMPRESSION_NONE_WR or PNG_TEXT_COMPRESSION_zTXt_WR,
1925
so that it isn't written out again at the end (in case you are calling
1926
png_write_end() with the same struct.
1928
The keywords that are given in the PNG Specification are:
1930
Title Short (one line) title or
1932
Author Name of image's creator
1933
Description Description of image (possibly long)
1934
Copyright Copyright notice
1935
Creation Time Time of original image creation
1936
(usually RFC 1123 format, see below)
1937
Software Software used to create the image
1938
Disclaimer Legal disclaimer
1939
Warning Warning of nature of content
1940
Source Device used to create the image
1941
Comment Miscellaneous comment; conversion
1942
from other image format
1944
The keyword-text pairs work like this. Keywords should be short
1945
simple descriptions of what the comment is about. Some typical
1946
keywords are found in the PNG specification, as is some recommendations
1947
on keywords. You can repeat keywords in a file. You can even write
1948
some text before the image and some after. For example, you may want
1949
to put a description of the image before the image, but leave the
1950
disclaimer until after, so viewers working over modem connections
1951
don't have to wait for the disclaimer to go over the modem before
1952
they start seeing the image. Finally, keywords should be full
1953
words, not abbreviations. Keywords and text are in the ISO 8859-1
1954
(Latin-1) character set (a superset of regular ASCII) and can not
1955
contain NUL characters, and should not contain control or other
1956
unprintable characters. To make the comments widely readable, stick
1957
with basic ASCII, and avoid machine specific character set extensions
1958
like the IBM-PC character set. The keyword must be present, but
1959
you can leave off the text string on non-compressed pairs.
1960
Compressed pairs must have a text string, as only the text string
1961
is compressed anyway, so the compression would be meaningless.
1963
PNG supports modification time via the png_time structure. Two
1964
conversion routines are provided, png_convert_from_time_t() for
1965
time_t and png_convert_from_struct_tm() for struct tm. The
1966
time_t routine uses gmtime(). You don't have to use either of
1967
these, but if you wish to fill in the png_time structure directly,
1968
you should provide the time in universal time (GMT) if possible
1969
instead of your local time. Note that the year number is the full
1970
year (e.g. 1998, rather than 98 - PNG is year 2000 compliant!), and
1971
that months start with 1.
1973
If you want to store the time of the original image creation, you should
1974
use a plain tEXt chunk with the "Creation Time" keyword. This is
1975
necessary because the "creation time" of a PNG image is somewhat vague,
1976
depending on whether you mean the PNG file, the time the image was
1977
created in a non-PNG format, a still photo from which the image was
1978
scanned, or possibly the subject matter itself. In order to facilitate
1979
machine-readable dates, it is recommended that the "Creation Time"
1980
tEXt chunk use RFC 1123 format dates (e.g. "22 May 1997 18:07:10 GMT"),
1981
although this isn't a requirement. Unlike the tIME chunk, the
1982
"Creation Time" tEXt chunk is not expected to be automatically changed
1983
by the software. To facilitate the use of RFC 1123 dates, a function
1984
png_convert_to_rfc1123(png_timep) is provided to convert from PNG
1985
time to an RFC 1123 format string.
1987
Writing unknown chunks
1989
You can use the png_set_unknown_chunks function to queue up chunks
1990
for writing. You give it a chunk name, raw data, and a size; that's
1991
all there is to it. The chunks will be written by the next following
1992
png_write_info_before_PLTE, png_write_info, or png_write_end function.
1993
Any chunks previously read into the info structure's unknown-chunk
1994
list will also be written out in a sequence that satisfies the PNG
1995
specification's ordering rules.
1997
The high-level write interface
1999
At this point there are two ways to proceed; through the high-level
2000
write interface, or through a sequence of low-level write operations.
2001
You can use the high-level interface if your image data is present
2002
in the info structure. All defined output
2003
transformations are permitted, enabled by the following masks.
2005
PNG_TRANSFORM_IDENTITY No transformation
2006
PNG_TRANSFORM_PACKING Pack 1, 2 and 4-bit samples
2007
PNG_TRANSFORM_PACKSWAP Change order of packed
2009
PNG_TRANSFORM_INVERT_MONO Invert monochrome images
2010
PNG_TRANSFORM_SHIFT Normalize pixels to the
2012
PNG_TRANSFORM_BGR Flip RGB to BGR, RGBA
2014
PNG_TRANSFORM_SWAP_ALPHA Flip RGBA to ARGB or GA
2016
PNG_TRANSFORM_INVERT_ALPHA Change alpha from opacity
2018
PNG_TRANSFORM_SWAP_ENDIAN Byte-swap 16-bit samples
2019
PNG_TRANSFORM_STRIP_FILLER Strip out filler bytes.
2021
If you have valid image data in the info structure (you can use
2022
png_set_rows() to put image data in the info structure), simply do this:
2024
png_write_png(png_ptr, info_ptr, png_transforms, NULL)
2026
where png_transforms is an integer containing the bitwise OR of some set of
2027
transformation flags. This call is equivalent to png_write_info(),
2028
followed the set of transformations indicated by the transform mask,
2029
then png_write_image(), and finally png_write_end().
2031
(The final parameter of this call is not yet used. Someday it might point
2032
to transformation parameters required by some future output transform.)
2034
You must use png_transforms and not call any png_set_transform() functions
2035
when you use png_write_png().
2037
The low-level write interface
2039
If you are going the low-level route instead, you are now ready to
2040
write all the file information up to the actual image data. You do
2041
this with a call to png_write_info().
2043
png_write_info(png_ptr, info_ptr);
2045
Note that there is one transformation you may need to do before
2046
png_write_info(). In PNG files, the alpha channel in an image is the
2047
level of opacity. If your data is supplied as a level of
2048
transparency, you can invert the alpha channel before you write it, so
2049
that 0 is fully transparent and 255 (in 8-bit or paletted images) or
2050
65535 (in 16-bit images) is fully opaque, with
2052
png_set_invert_alpha(png_ptr);
2054
This must appear before png_write_info() instead of later with the
2055
other transformations because in the case of paletted images the tRNS
2056
chunk data has to be inverted before the tRNS chunk is written. If
2057
your image is not a paletted image, the tRNS data (which in such cases
2058
represents a single color to be rendered as transparent) won't need to
2059
be changed, and you can safely do this transformation after your
2060
png_write_info() call.
2062
If you need to write a private chunk that you want to appear before
2063
the PLTE chunk when PLTE is present, you can write the PNG info in
2064
two steps, and insert code to write your own chunk between them:
2066
png_write_info_before_PLTE(png_ptr, info_ptr);
2067
png_set_unknown_chunks(png_ptr, info_ptr, ...);
2068
png_write_info(png_ptr, info_ptr);
2070
After you've written the file information, you can set up the library
2071
to handle any special transformations of the image data. The various
2072
ways to transform the data will be described in the order that they
2073
should occur. This is important, as some of these change the color
2074
type and/or bit depth of the data, and some others only work on
2075
certain color types and bit depths. Even though each transformation
2076
checks to see if it has data that it can do something with, you should
2077
make sure to only enable a transformation if it will be valid for the
2078
data. For example, don't swap red and blue on grayscale data.
2080
PNG files store RGB pixels packed into 3 or 6 bytes. This code tells
2081
the library to strip input data that has 4 or 8 bytes per pixel down
2082
to 3 or 6 bytes (or strip 2 or 4-byte grayscale+filler data to 1 or 2
2085
png_set_filler(png_ptr, 0, PNG_FILLER_BEFORE);
2087
where the 0 is unused, and the location is either PNG_FILLER_BEFORE or
2088
PNG_FILLER_AFTER, depending upon whether the filler byte in the pixel
2089
is stored XRGB or RGBX.
2091
PNG files pack pixels of bit depths 1, 2, and 4 into bytes as small as
2092
they can, resulting in, for example, 8 pixels per byte for 1 bit files.
2093
If the data is supplied at 1 pixel per byte, use this code, which will
2094
correctly pack the pixels into a single byte:
2096
png_set_packing(png_ptr);
2098
PNG files reduce possible bit depths to 1, 2, 4, 8, and 16. If your
2099
data is of another bit depth, you can write an sBIT chunk into the
2100
file so that decoders can recover the original data if desired.
2102
/* Set the true bit depth of the image data */
2103
if (color_type & PNG_COLOR_MASK_COLOR)
2105
sig_bit.red = true_bit_depth;
2106
sig_bit.green = true_bit_depth;
2107
sig_bit.blue = true_bit_depth;
2111
sig_bit.gray = true_bit_depth;
2113
if (color_type & PNG_COLOR_MASK_ALPHA)
2115
sig_bit.alpha = true_bit_depth;
2118
png_set_sBIT(png_ptr, info_ptr, &sig_bit);
2120
If the data is stored in the row buffer in a bit depth other than
2121
one supported by PNG (e.g. 3 bit data in the range 0-7 for a 4-bit PNG),
2122
this will scale the values to appear to be the correct bit depth as
2125
png_set_shift(png_ptr, &sig_bit);
2127
PNG files store 16 bit pixels in network byte order (big-endian,
2128
ie. most significant bits first). This code would be used if they are
2129
supplied the other way (little-endian, i.e. least significant bits
2130
first, the way PCs store them):
2133
png_set_swap(png_ptr);
2135
If you are using packed-pixel images (1, 2, or 4 bits/pixel), and you
2136
need to change the order the pixels are packed into bytes, you can use:
2139
png_set_packswap(png_ptr);
2141
PNG files store 3 color pixels in red, green, blue order. This code
2142
would be used if they are supplied as blue, green, red:
2144
png_set_bgr(png_ptr);
2146
PNG files describe monochrome as black being zero and white being
2147
one. This code would be used if the pixels are supplied with this reversed
2148
(black being one and white being zero):
2150
png_set_invert_mono(png_ptr);
2152
Finally, you can write your own transformation function if none of
2153
the existing ones meets your needs. This is done by setting a callback
2156
png_set_write_user_transform_fn(png_ptr,
2157
write_transform_fn);
2159
You must supply the function
2161
void write_transform_fn(png_ptr ptr, row_info_ptr
2162
row_info, png_bytep data)
2164
See pngtest.c for a working example. Your function will be called
2165
before any of the other transformations are processed.
2167
You can also set up a pointer to a user structure for use by your
2170
png_set_user_transform_info(png_ptr, user_ptr, 0, 0);
2172
The user_channels and user_depth parameters of this function are ignored
2173
when writing; you can set them to zero as shown.
2175
You can retrieve the pointer via the function png_get_user_transform_ptr().
2178
voidp write_user_transform_ptr =
2179
png_get_user_transform_ptr(png_ptr);
2181
It is possible to have libpng flush any pending output, either manually,
2182
or automatically after a certain number of lines have been written. To
2183
flush the output stream a single time call:
2185
png_write_flush(png_ptr);
2187
and to have libpng flush the output stream periodically after a certain
2188
number of scanlines have been written, call:
2190
png_set_flush(png_ptr, nrows);
2192
Note that the distance between rows is from the last time png_write_flush()
2193
was called, or the first row of the image if it has never been called.
2194
So if you write 50 lines, and then png_set_flush 25, it will flush the
2195
output on the next scanline, and every 25 lines thereafter, unless
2196
png_write_flush() is called before 25 more lines have been written.
2197
If nrows is too small (less than about 10 lines for a 640 pixel wide
2198
RGB image) the image compression may decrease noticeably (although this
2199
may be acceptable for real-time applications). Infrequent flushing will
2200
only degrade the compression performance by a few percent over images
2201
that do not use flushing.
2203
Writing the image data
2205
That's it for the transformations. Now you can write the image data.
2206
The simplest way to do this is in one function call. If you have the
2207
whole image in memory, you can just call png_write_image() and libpng
2208
will write the image. You will need to pass in an array of pointers to
2209
each row. This function automatically handles interlacing, so you don't
2210
need to call png_set_interlace_handling() or call this function multiple
2211
times, or any of that other stuff necessary with png_write_rows().
2213
png_write_image(png_ptr, row_pointers);
2215
where row_pointers is:
2217
png_byte *row_pointers[height];
2219
You can point to void or char or whatever you use for pixels.
2221
If you don't want to write the whole image at once, you can
2222
use png_write_rows() instead. If the file is not interlaced,
2225
png_write_rows(png_ptr, row_pointers,
2228
row_pointers is the same as in the png_write_image() call.
2230
If you are just writing one row at a time, you can do this with
2231
a single row_pointer instead of an array of row_pointers:
2233
png_bytep row_pointer = row;
2235
png_write_row(png_ptr, row_pointer);
2237
When the file is interlaced, things can get a good deal more
2238
complicated. The only currently (as of the PNG Specification
2239
version 1.2, dated July 1999) defined interlacing scheme for PNG files
2240
is the "Adam7" interlace scheme, that breaks down an
2241
image into seven smaller images of varying size. libpng will build
2242
these images for you, or you can do them yourself. If you want to
2243
build them yourself, see the PNG specification for details of which
2244
pixels to write when.
2246
If you don't want libpng to handle the interlacing details, just
2247
use png_set_interlace_handling() and call png_write_rows() the
2248
correct number of times to write all seven sub-images.
2250
If you want libpng to build the sub-images, call this before you start
2254
png_set_interlace_handling(png_ptr);
2256
This will return the number of passes needed. Currently, this
2257
is seven, but may change if another interlace type is added.
2259
Then write the complete image number_of_passes times.
2261
png_write_rows(png_ptr, row_pointers,
2264
As some of these rows are not used, and thus return immediately,
2265
you may want to read about interlacing in the PNG specification,
2266
and only update the rows that are actually used.
2268
Finishing a sequential write
2270
After you are finished writing the image, you should finish writing
2271
the file. If you are interested in writing comments or time, you should
2272
pass an appropriately filled png_info pointer. If you are not interested,
2275
png_write_end(png_ptr, info_ptr);
2277
When you are done, you can free all memory used by libpng like this:
2279
png_destroy_write_struct(&png_ptr, &info_ptr);
2281
It is also possible to individually free the info_ptr members that
2282
point to libpng-allocated storage with the following function:
2284
png_free_data(png_ptr, info_ptr, mask, seq)
2285
mask - identifies data to be freed, a mask
2286
containing the bitwise OR of one or
2288
PNG_FREE_PLTE, PNG_FREE_TRNS,
2289
PNG_FREE_HIST, PNG_FREE_ICCP,
2290
PNG_FREE_PCAL, PNG_FREE_ROWS,
2291
PNG_FREE_SCAL, PNG_FREE_SPLT,
2292
PNG_FREE_TEXT, PNG_FREE_UNKN,
2293
or simply PNG_FREE_ALL
2294
seq - sequence number of item to be freed
2297
This function may be safely called when the relevant storage has
2298
already been freed, or has not yet been allocated, or was allocated
2299
by the user and not by libpng, and will in those
2300
cases do nothing. The "seq" parameter is ignored if only one item
2301
of the selected data type, such as PLTE, is allowed. If "seq" is not
2302
-1, and multiple items are allowed for the data type identified in
2303
the mask, such as text or sPLT, only the n'th item in the structure
2304
is freed, where n is "seq".
2306
If you allocated data such as a palette that you passed
2307
in to libpng with png_set_*, you must not free it until just before the call to
2308
png_destroy_write_struct().
2310
The default behavior is only to free data that was allocated internally
2311
by libpng. This can be changed, so that libpng will not free the data,
2312
or so that it will free data that was allocated by the user with png_malloc()
2313
or png_zalloc() and passed in via a png_set_*() function, with
2315
png_data_freer(png_ptr, info_ptr, freer, mask)
2316
mask - which data elements are affected
2317
same choices as in png_free_data()
2319
PNG_DESTROY_WILL_FREE_DATA
2320
PNG_SET_WILL_FREE_DATA
2321
PNG_USER_WILL_FREE_DATA
2323
For example, to transfer responsibility for some data from a read structure
2324
to a write structure, you could use
2326
png_data_freer(read_ptr, read_info_ptr,
2327
PNG_USER_WILL_FREE_DATA,
2328
PNG_FREE_PLTE|PNG_FREE_tRNS|PNG_FREE_hIST)
2329
png_data_freer(write_ptr, write_info_ptr,
2330
PNG_DESTROY_WILL_FREE_DATA,
2331
PNG_FREE_PLTE|PNG_FREE_tRNS|PNG_FREE_hIST)
2333
thereby briefly reassigning responsibility for freeing to the user but
2334
immediately afterwards reassigning it once more to the write_destroy
2335
function. Having done this, it would then be safe to destroy the read
2336
structure and continue to use the PLTE, tRNS, and hIST data in the write
2339
This function only affects data that has already been allocated.
2340
You can call this function before calling after the png_set_*() functions
2341
to control whether the user or png_destroy_*() is supposed to free the data.
2342
When the user assumes responsibility for libpng-allocated data, the
2343
application must use
2344
png_free() to free it, and when the user transfers responsibility to libpng
2345
for data that the user has allocated, the user must have used png_malloc()
2346
or png_zalloc() to allocate it.
2348
If you allocated text_ptr.text, text_ptr.lang, and text_ptr.translated_keyword
2349
separately, do not transfer responsibility for freeing text_ptr to libpng,
2350
because when libpng fills a png_text structure it combines these members with
2351
the key member, and png_free_data() will free only text_ptr.key. Similarly,
2352
if you transfer responsibility for free'ing text_ptr from libpng to your
2353
application, your application must not separately free those members.
2354
For a more compact example of writing a PNG image, see the file example.c.
2356
V. Modifying/Customizing libpng:
2358
There are three issues here. The first is changing how libpng does
2359
standard things like memory allocation, input/output, and error handling.
2360
The second deals with more complicated things like adding new chunks,
2361
adding new transformations, and generally changing how libpng works.
2362
Both of those are compile-time issues; that is, they are generally
2363
determined at the time the code is written, and there is rarely a need
2364
to provide the user with a means of changing them. The third is a
2365
run-time issue: choosing between and/or tuning one or more alternate
2366
versions of computationally intensive routines; specifically, optimized
2367
assembly-language (and therefore compiler- and platform-dependent)
2370
Memory allocation, input/output, and error handling
2372
All of the memory allocation, input/output, and error handling in libpng
2373
goes through callbacks that are user-settable. The default routines are
2374
in pngmem.c, pngrio.c, pngwio.c, and pngerror.c, respectively. To change
2375
these functions, call the appropriate png_set_*_fn() function.
2377
Memory allocation is done through the functions png_malloc()
2378
and png_free(). These currently just call the standard C functions. If
2379
your pointers can't access more then 64K at a time, you will want to set
2380
MAXSEG_64K in zlib.h. Since it is unlikely that the method of handling
2381
memory allocation on a platform will change between applications, these
2382
functions must be modified in the library at compile time. If you prefer
2383
to use a different method of allocating and freeing data, you can use
2384
png_create_read_struct_2() or png_create_write_struct_2() to register
2385
your own functions as described above.
2386
These functions also provide a void pointer that can be retrieved via
2388
mem_ptr=png_get_mem_ptr(png_ptr);
2390
Your replacement memory functions must have prototypes as follows:
2392
png_voidp malloc_fn(png_structp png_ptr,
2394
void free_fn(png_structp png_ptr, png_voidp ptr);
2396
Your malloc_fn() must return NULL in case of failure. The png_malloc()
2397
function will normally call png_error() if it receives a NULL from the
2398
system memory allocator or from your replacement malloc_fn().
2400
Input/Output in libpng is done through png_read() and png_write(),
2401
which currently just call fread() and fwrite(). The FILE * is stored in
2402
png_struct and is initialized via png_init_io(). If you wish to change
2403
the method of I/O, the library supplies callbacks that you can set
2404
through the function png_set_read_fn() and png_set_write_fn() at run
2405
time, instead of calling the png_init_io() function. These functions
2406
also provide a void pointer that can be retrieved via the function
2407
png_get_io_ptr(). For example:
2409
png_set_read_fn(png_structp read_ptr,
2410
voidp read_io_ptr, png_rw_ptr read_data_fn)
2412
png_set_write_fn(png_structp write_ptr,
2413
voidp write_io_ptr, png_rw_ptr write_data_fn,
2414
png_flush_ptr output_flush_fn);
2416
voidp read_io_ptr = png_get_io_ptr(read_ptr);
2417
voidp write_io_ptr = png_get_io_ptr(write_ptr);
2419
The replacement I/O functions must have prototypes as follows:
2421
void user_read_data(png_structp png_ptr,
2422
png_bytep data, png_size_t length);
2423
void user_write_data(png_structp png_ptr,
2424
png_bytep data, png_size_t length);
2425
void user_flush_data(png_structp png_ptr);
2427
Supplying NULL for the read, write, or flush functions sets them back
2428
to using the default C stream functions. It is an error to read from
2429
a write stream, and vice versa.
2431
Error handling in libpng is done through png_error() and png_warning().
2432
Errors handled through png_error() are fatal, meaning that png_error()
2433
should never return to its caller. Currently, this is handled via
2434
setjmp() and longjmp() (unless you have compiled libpng with
2435
PNG_SETJMP_NOT_SUPPORTED, in which case it is handled via PNG_ABORT()),
2436
but you could change this to do things like exit() if you should wish.
2438
On non-fatal errors, png_warning() is called
2439
to print a warning message, and then control returns to the calling code.
2440
By default png_error() and png_warning() print a message on stderr via
2441
fprintf() unless the library is compiled with PNG_NO_CONSOLE_IO defined
2442
(because you don't want the messages) or PNG_NO_STDIO defined (because
2443
fprintf() isn't available). If you wish to change the behavior of the error
2444
functions, you will need to set up your own message callbacks. These
2445
functions are normally supplied at the time that the png_struct is created.
2446
It is also possible to redirect errors and warnings to your own replacement
2447
functions after png_create_*_struct() has been called by calling:
2449
png_set_error_fn(png_structp png_ptr,
2450
png_voidp error_ptr, png_error_ptr error_fn,
2451
png_error_ptr warning_fn);
2453
png_voidp error_ptr = png_get_error_ptr(png_ptr);
2455
If NULL is supplied for either error_fn or warning_fn, then the libpng
2456
default function will be used, calling fprintf() and/or longjmp() if a
2457
problem is encountered. The replacement error functions should have
2458
parameters as follows:
2460
void user_error_fn(png_structp png_ptr,
2461
png_const_charp error_msg);
2462
void user_warning_fn(png_structp png_ptr,
2463
png_const_charp warning_msg);
2465
The motivation behind using setjmp() and longjmp() is the C++ throw and
2466
catch exception handling methods. This makes the code much easier to write,
2467
as there is no need to check every return code of every function call.
2468
However, there are some uncertainties about the status of local variables
2469
after a longjmp, so the user may want to be careful about doing anything after
2470
setjmp returns non-zero besides returning itself. Consult your compiler
2471
documentation for more details. For an alternative approach, you may wish
2472
to use the "cexcept" facility (see http://cexcept.sourceforge.net).
2476
If you need to read or write custom chunks, you may need to get deeper
2477
into the libpng code. The library now has mechanisms for storing
2478
and writing chunks of unknown type; you can even declare callbacks
2479
for custom chunks. However, this may not be good enough if the
2480
library code itself needs to know about interactions between your
2481
chunk and existing `intrinsic' chunks.
2483
If you need to write a new intrinsic chunk, first read the PNG
2484
specification. Acquire a first level of
2485
understanding of how it works. Pay particular attention to the
2486
sections that describe chunk names, and look at how other chunks were
2487
designed, so you can do things similarly. Second, check out the
2488
sections of libpng that read and write chunks. Try to find a chunk
2489
that is similar to yours and use it as a template. More details can
2490
be found in the comments inside the code. It is best to handle unknown
2491
chunks in a generic method, via callback functions, instead of by
2492
modifying libpng functions.
2494
If you wish to write your own transformation for the data, look through
2495
the part of the code that does the transformations, and check out some of
2496
the simpler ones to get an idea of how they work. Try to find a similar
2497
transformation to the one you want to add and copy off of it. More details
2498
can be found in the comments inside the code itself.
2500
Configuring for 16 bit platforms
2502
You will want to look into zconf.h to tell zlib (and thus libpng) that
2503
it cannot allocate more then 64K at a time. Even if you can, the memory
2504
won't be accessible. So limit zlib and libpng to 64K by defining MAXSEG_64K.
2508
For DOS users who only have access to the lower 640K, you will
2509
have to limit zlib's memory usage via a png_set_compression_mem_level()
2510
call. See zlib.h or zconf.h in the zlib library for more information.
2512
Configuring for Medium Model
2514
Libpng's support for medium model has been tested on most of the popular
2515
compilers. Make sure MAXSEG_64K gets defined, USE_FAR_KEYWORD gets
2516
defined, and FAR gets defined to far in pngconf.h, and you should be
2517
all set. Everything in the library (except for zlib's structure) is
2518
expecting far data. You must use the typedefs with the p or pp on
2519
the end for pointers (or at least look at them and be careful). Make
2520
note that the rows of data are defined as png_bytepp, which is an
2521
unsigned char far * far *.
2523
Configuring for gui/windowing platforms:
2525
You will need to write new error and warning functions that use the GUI
2526
interface, as described previously, and set them to be the error and
2527
warning functions at the time that png_create_*_struct() is called,
2528
in order to have them available during the structure initialization.
2529
They can be changed later via png_set_error_fn(). On some compilers,
2530
you may also have to change the memory allocators (png_malloc, etc.).
2532
Configuring for compiler xxx:
2534
All includes for libpng are in pngconf.h. If you need to add/change/delete
2535
an include, this is the place to do it. The includes that are not
2536
needed outside libpng are protected by the PNG_INTERNAL definition,
2537
which is only defined for those routines inside libpng itself. The
2538
files in libpng proper only include png.h, which includes pngconf.h.
2542
There are special functions to configure the compression. Perhaps the
2543
most useful one changes the compression level, which currently uses
2544
input compression values in the range 0 - 9. The library normally
2545
uses the default compression level (Z_DEFAULT_COMPRESSION = 6). Tests
2546
have shown that for a large majority of images, compression values in
2547
the range 3-6 compress nearly as well as higher levels, and do so much
2548
faster. For online applications it may be desirable to have maximum speed
2549
(Z_BEST_SPEED = 1). With versions of zlib after v0.99, you can also
2550
specify no compression (Z_NO_COMPRESSION = 0), but this would create
2551
files larger than just storing the raw bitmap. You can specify the
2552
compression level by calling:
2554
png_set_compression_level(png_ptr, level);
2556
Another useful one is to reduce the memory level used by the library.
2557
The memory level defaults to 8, but it can be lowered if you are
2558
short on memory (running DOS, for example, where you only have 640K).
2559
Note that the memory level does have an effect on compression; among
2560
other things, lower levels will result in sections of incompressible
2561
data being emitted in smaller stored blocks, with a correspondingly
2562
larger relative overhead of up to 15% in the worst case.
2564
png_set_compression_mem_level(png_ptr, level);
2566
The other functions are for configuring zlib. They are not recommended
2567
for normal use and may result in writing an invalid PNG file. See
2568
zlib.h for more information on what these mean.
2570
png_set_compression_strategy(png_ptr,
2572
png_set_compression_window_bits(png_ptr,
2574
png_set_compression_method(png_ptr, method);
2575
png_set_compression_buffer_size(png_ptr, size);
2577
Controlling row filtering
2579
If you want to control whether libpng uses filtering or not, which
2580
filters are used, and how it goes about picking row filters, you
2581
can call one of these functions. The selection and configuration
2582
of row filters can have a significant impact on the size and
2583
encoding speed and a somewhat lesser impact on the decoding speed
2584
of an image. Filtering is enabled by default for RGB and grayscale
2585
images (with and without alpha), but not for paletted images nor
2586
for any images with bit depths less than 8 bits/pixel.
2588
The 'method' parameter sets the main filtering method, which is
2589
currently only '0' in the PNG 1.2 specification. The 'filters'
2590
parameter sets which filter(s), if any, should be used for each
2591
scanline. Possible values are PNG_ALL_FILTERS and PNG_NO_FILTERS
2592
to turn filtering on and off, respectively.
2594
Individual filter types are PNG_FILTER_NONE, PNG_FILTER_SUB,
2595
PNG_FILTER_UP, PNG_FILTER_AVG, PNG_FILTER_PAETH, which can be bitwise
2596
ORed together with '|' to specify one or more filters to use.
2597
These filters are described in more detail in the PNG specification.
2598
If you intend to change the filter type during the course of writing
2599
the image, you should start with flags set for all of the filters
2600
you intend to use so that libpng can initialize its internal
2601
structures appropriately for all of the filter types. (Note that this
2602
means the first row must always be adaptively filtered, because libpng
2603
currently does not allocate the filter buffers until png_write_row()
2604
is called for the first time.)
2606
filters = PNG_FILTER_NONE | PNG_FILTER_SUB
2607
PNG_FILTER_UP | PNG_FILTER_AVE |
2608
PNG_FILTER_PAETH | PNG_ALL_FILTERS;
2610
png_set_filter(png_ptr, PNG_FILTER_TYPE_BASE,
2612
The second parameter can also be
2613
PNG_INTRAPIXEL_DIFFERENCING if you are
2614
writing a PNG to be embedded in a MNG
2615
datastream. This parameter must be the
2616
same as the value of filter_method used
2619
It is also possible to influence how libpng chooses from among the
2620
available filters. This is done in one or both of two ways - by
2621
telling it how important it is to keep the same filter for successive
2622
rows, and by telling it the relative computational costs of the filters.
2624
double weights[3] = {1.5, 1.3, 1.1},
2625
costs[PNG_FILTER_VALUE_LAST] =
2626
{1.0, 1.3, 1.3, 1.5, 1.7};
2628
png_set_filter_heuristics(png_ptr,
2629
PNG_FILTER_HEURISTIC_WEIGHTED, 3,
2632
The weights are multiplying factors that indicate to libpng that the
2633
row filter should be the same for successive rows unless another row filter
2634
is that many times better than the previous filter. In the above example,
2635
if the previous 3 filters were SUB, SUB, NONE, the SUB filter could have a
2636
"sum of absolute differences" 1.5 x 1.3 times higher than other filters
2637
and still be chosen, while the NONE filter could have a sum 1.1 times
2638
higher than other filters and still be chosen. Unspecified weights are
2639
taken to be 1.0, and the specified weights should probably be declining
2640
like those above in order to emphasize recent filters over older filters.
2642
The filter costs specify for each filter type a relative decoding cost
2643
to be considered when selecting row filters. This means that filters
2644
with higher costs are less likely to be chosen over filters with lower
2645
costs, unless their "sum of absolute differences" is that much smaller.
2646
The costs do not necessarily reflect the exact computational speeds of
2647
the various filters, since this would unduly influence the final image
2650
Note that the numbers above were invented purely for this example and
2651
are given only to help explain the function usage. Little testing has
2652
been done to find optimum values for either the costs or the weights.
2654
Removing unwanted object code
2656
There are a bunch of #define's in pngconf.h that control what parts of
2657
libpng are compiled. All the defines end in _SUPPORTED. If you are
2658
never going to use a capability, you can change the #define to #undef
2659
before recompiling libpng and save yourself code and data space, or
2660
you can turn off individual capabilities with defines that begin with
2663
You can also turn all of the transforms and ancillary chunk capabilities
2664
off en masse with compiler directives that define
2665
PNG_NO_READ[or WRITE]_TRANSFORMS, or PNG_NO_READ[or WRITE]_ANCILLARY_CHUNKS,
2667
along with directives to turn on any of the capabilities that you do
2668
want. The PNG_NO_READ[or WRITE]_TRANSFORMS directives disable
2669
the extra transformations but still leave the library fully capable of reading
2670
and writing PNG files with all known public chunks
2671
Use of the PNG_NO_READ[or WRITE]_ANCILLARY_CHUNKS directive
2672
produces a library that is incapable of reading or writing ancillary chunks.
2673
If you are not using the progressive reading capability, you can
2674
turn that off with PNG_NO_PROGRESSIVE_READ (don't confuse
2675
this with the INTERLACING capability, which you'll still have).
2677
All the reading and writing specific code are in separate files, so the
2678
linker should only grab the files it needs. However, if you want to
2679
make sure, or if you are building a stand alone library, all the
2680
reading files start with pngr and all the writing files start with
2681
pngw. The files that don't match either (like png.c, pngtrans.c, etc.)
2682
are used for both reading and writing, and always need to be included.
2683
The progressive reader is in pngpread.c
2685
If you are creating or distributing a dynamically linked library (a .so
2686
or DLL file), you should not remove or disable any parts of the library,
2687
as this will cause applications linked with different versions of the
2688
library to fail if they call functions not available in your library.
2689
The size of the library itself should not be an issue, because only
2690
those sections that are actually used will be loaded into memory.
2692
Requesting debug printout
2694
The macro definition PNG_DEBUG can be used to request debugging
2695
printout. Set it to an integer value in the range 0 to 3. Higher
2696
numbers result in increasing amounts of debugging information. The
2697
information is printed to the "stderr" file, unless another file
2698
name is specified in the PNG_DEBUG_FILE macro definition.
2700
When PNG_DEBUG > 0, the following functions (macros) become available:
2702
png_debug(level, message)
2703
png_debug1(level, message, p1)
2704
png_debug2(level, message, p1, p2)
2706
in which "level" is compared to PNG_DEBUG to decide whether to print
2707
the message, "message" is the formatted string to be printed,
2708
and p1 and p2 are parameters that are to be embedded in the string
2709
according to printf-style formatting directives. For example,
2711
png_debug1(2, "foo=%d\n", foo);
2716
fprintf(PNG_DEBUG_FILE, "foo=%d\n", foo);
2718
When PNG_DEBUG is defined but is zero, the macros aren't defined, but you
2719
can still use PNG_DEBUG to control your own debugging:
2725
When PNG_DEBUG = 1, the macros are defined, but only png_debug statements
2726
having level = 0 will be printed. There aren't any such statements in
2727
this version of libpng, but if you insert some they will be printed.
2729
VI. Runtime optimization
2731
A new feature in libpng 1.2.0 is the ability to dynamically switch between
2732
standard and optimized versions of some routines. Currently these are
2733
limited to three computationally intensive tasks when reading PNG files:
2734
decoding row filters, expanding interlacing, and combining interlaced or
2735
transparent row data with previous row data. Currently the optimized
2736
versions are available only for x86 (Intel, AMD, etc.) platforms with
2737
MMX support, though this may change in future versions. (For example,
2738
the non-MMX assembler optimizations for zlib might become similarly
2739
runtime-selectable in future releases, in which case libpng could be
2740
extended to support them. Alternatively, the compile-time choice of
2741
floating-point versus integer routines for gamma correction might become
2742
runtime-selectable.)
2744
Because such optimizations tend to be very platform- and compiler-dependent,
2745
both in how they are written and in how they perform, the new runtime code
2746
in libpng has been written to allow programs to query, enable, and disable
2747
either specific optimizations or all such optimizations. For example, to
2748
enable all possible optimizations (bearing in mind that some "optimizations"
2749
may actually run more slowly in rare cases):
2751
#if defined(PNG_LIBPNG_VER) && (PNG_LIBPNG_VER >= 10200)
2752
png_uint_32 mask, flags;
2754
flags = png_get_asm_flags(png_ptr);
2755
mask = png_get_asm_flagmask(PNG_SELECT_READ | PNG_SELECT_WRITE);
2756
png_set_asm_flags(png_ptr, flags | mask);
2759
To enable only optimizations relevant to reading PNGs, use PNG_SELECT_READ
2760
by itself when calling png_get_asm_flagmask(); similarly for optimizing
2761
only writing. To disable all optimizations:
2763
#if defined(PNG_LIBPNG_VER) && (PNG_LIBPNG_VER >= 10200)
2764
flags = png_get_asm_flags(png_ptr);
2765
mask = png_get_asm_flagmask(PNG_SELECT_READ | PNG_SELECT_WRITE);
2766
png_set_asm_flags(png_ptr, flags & ~mask);
2769
To enable or disable only MMX-related features, use png_get_mmx_flagmask()
2770
in place of png_get_asm_flagmask(). The mmx version takes one additional
2773
#if defined(PNG_LIBPNG_VER) && (PNG_LIBPNG_VER >= 10200)
2774
int selection = PNG_SELECT_READ | PNG_SELECT_WRITE;
2777
mask = png_get_mmx_flagmask(selection, &compilerID);
2780
On return, compilerID will indicate which version of the MMX assembler
2781
optimizations was compiled. Currently two flavors exist: Microsoft
2782
Visual C++ (compilerID == 1) and GNU C (a.k.a. gcc/gas, compilerID == 2).
2783
On non-x86 platforms or on systems compiled without MMX optimizations, a
2784
value of -1 is used.
2786
Note that both png_get_asm_flagmask() and png_get_mmx_flagmask() return
2787
all valid, settable optimization bits for the version of the library that's
2788
currently in use. In the case of shared (dynamically linked) libraries,
2789
this may include optimizations that did not exist at the time the code was
2790
written and compiled. It is also possible, of course, to enable only known,
2791
specific optimizations; for example:
2793
#if defined(PNG_LIBPNG_VER) && (PNG_LIBPNG_VER >= 10200)
2794
flags = PNG_ASM_FLAG_MMX_READ_COMBINE_ROW \
2795
| PNG_ASM_FLAG_MMX_READ_INTERLACE \
2796
| PNG_ASM_FLAG_MMX_READ_FILTER_SUB \
2797
| PNG_ASM_FLAG_MMX_READ_FILTER_UP \
2798
| PNG_ASM_FLAG_MMX_READ_FILTER_AVG \
2799
| PNG_ASM_FLAG_MMX_READ_FILTER_PAETH ;
2800
png_set_asm_flags(png_ptr, flags);
2803
This method would enable only the MMX read-optimizations available at the
2804
time of libpng 1.2.0's release, regardless of whether a later version of
2805
the DLL were actually being used. (Also note that these functions did not
2806
exist in versions older than 1.2.0, so any attempt to run a dynamically
2807
linked app on such an older version would fail.)
2809
To determine whether the processor supports MMX instructions at all, use
2810
the png_mmx_support() function:
2812
#if defined(PNG_LIBPNG_VER) && (PNG_LIBPNG_VER >= 10200)
2813
mmxsupport = png_mmx_support();
2816
It returns -1 if MMX support is not compiled into libpng, 0 if MMX code
2817
is compiled but MMX is not supported by the processor, or 1 if MMX support
2818
is fully available. Note that png_mmx_support(), png_get_mmx_flagmask(),
2819
and png_get_asm_flagmask() all may be called without allocating and ini-
2820
tializing any PNG structures (for example, as part of a usage screen or
2823
The following code can be used to prevent an application from using the
2824
thread_unsafe features, even if libpng was built with PNG_THREAD_UNSAFE_OK
2827
#if defined(PNG_USE_PNGGCCRD) && defined(PNG_ASSEMBLER_CODE_SUPPORTED) \
2828
&& defined(PNG_THREAD_UNSAFE_OK)
2829
/* Disable thread-unsafe features of pnggccrd */
2830
if (png_access_version_number() >= 10200)
2832
png_uint_32 mmx_disable_mask = 0;
2833
png_uint_32 asm_flags;
2835
mmx_disable_mask |= ( PNG_ASM_FLAG_MMX_READ_COMBINE_ROW \
2836
| PNG_ASM_FLAG_MMX_READ_FILTER_SUB \
2837
| PNG_ASM_FLAG_MMX_READ_FILTER_AVG \
2838
| PNG_ASM_FLAG_MMX_READ_FILTER_PAETH );
2839
asm_flags = png_get_asm_flags(png_ptr);
2840
png_set_asm_flags(png_ptr, asm_flags & ~mmx_disable_mask);
2844
For more extensive examples of runtime querying, enabling and disabling
2845
of optimized features, see contrib/gregbook/readpng2.c in the libpng
2846
source-code distribution.
2850
The MNG specification (available at http://www.libpng.org/pub/mng) allows
2851
certain extensions to PNG for PNG images that are embedded in MNG datastreams.
2852
Libpng can support some of these extensions. To enable them, use the
2853
png_permit_mng_features() function:
2855
feature_set = png_permit_mng_features(png_ptr, mask)
2856
mask is a png_uint_32 containing the bitwise OR of the
2857
features you want to enable. These include
2858
PNG_FLAG_MNG_EMPTY_PLTE
2859
PNG_FLAG_MNG_FILTER_64
2860
PNG_ALL_MNG_FEATURES
2861
feature_set is a png_uint_32 that is the bitwise AND of
2862
your mask with the set of MNG features that is
2863
supported by the version of libpng that you are using.
2865
It is an error to use this function when reading or writing a standalone
2866
PNG file with the PNG 8-byte signature. The PNG datastream must be wrapped
2867
in a MNG datastream. As a minimum, it must have the MNG 8-byte signature
2868
and the MHDR and MEND chunks. Libpng does not provide support for these
2869
or any other MNG chunks; your application must provide its own support for
2870
them. You may wish to consider using libmng (available at
2871
http://www.libmng.com) instead.
2873
VIII. Changes to Libpng from version 0.88
2875
It should be noted that versions of libpng later than 0.96 are not
2876
distributed by the original libpng author, Guy Schalnat, nor by
2877
Andreas Dilger, who had taken over from Guy during 1996 and 1997, and
2878
distributed versions 0.89 through 0.96, but rather by another member
2879
of the original PNG Group, Glenn Randers-Pehrson. Guy and Andreas are
2880
still alive and well, but they have moved on to other things.
2882
The old libpng functions png_read_init(), png_write_init(),
2883
png_info_init(), png_read_destroy(), and png_write_destroy() have been
2884
moved to PNG_INTERNAL in version 0.95 to discourage their use. These
2885
functions will be removed from libpng version 2.0.0.
2887
The preferred method of creating and initializing the libpng structures is
2888
via the png_create_read_struct(), png_create_write_struct(), and
2889
png_create_info_struct() because they isolate the size of the structures
2890
from the application, allow version error checking, and also allow the
2891
use of custom error handling routines during the initialization, which
2892
the old functions do not. The functions png_read_destroy() and
2893
png_write_destroy() do not actually free the memory that libpng
2894
allocated for these structs, but just reset the data structures, so they
2895
can be used instead of png_destroy_read_struct() and
2896
png_destroy_write_struct() if you feel there is too much system overhead
2897
allocating and freeing the png_struct for each image read.
2899
Setting the error callbacks via png_set_message_fn() before
2900
png_read_init() as was suggested in libpng-0.88 is no longer supported
2901
because this caused applications that do not use custom error functions
2902
to fail if the png_ptr was not initialized to zero. It is still possible
2903
to set the error callbacks AFTER png_read_init(), or to change them with
2904
png_set_error_fn(), which is essentially the same function, but with a new
2905
name to force compilation errors with applications that try to use the old
2908
Starting with version 1.0.7, you can find out which version of the library
2909
you are using at run-time:
2911
png_uint_32 libpng_vn = png_access_version_number();
2913
The number libpng_vn is constructed from the major version, minor
2914
version with leading zero, and release number with leading zero,
2915
(e.g., libpng_vn for version 1.0.7 is 10007).
2917
You can also check which version of png.h you used when compiling your
2920
png_uint_32 application_vn = PNG_LIBPNG_VER;
2922
IX. Y2K Compliance in libpng
2926
Since the PNG Development group is an ad-hoc body, we can't make
2927
an official declaration.
2929
This is your unofficial assurance that libpng from version 0.71 and
2930
upward through 1.2.18 are Y2K compliant. It is my belief that earlier
2931
versions were also Y2K compliant.
2933
Libpng only has three year fields. One is a 2-byte unsigned integer that
2934
will hold years up to 65535. The other two hold the date in text
2935
format, and will hold years up to 9999.
2938
"png_uint_16 year" in png_time_struct.
2941
"png_charp time_buffer" in png_struct and
2942
"near_time_buffer", which is a local character string in png.c.
2944
There are seven time-related functions:
2946
png_convert_to_rfc_1123() in png.c
2947
(formerly png_convert_to_rfc_1152() in error)
2948
png_convert_from_struct_tm() in pngwrite.c, called
2950
png_convert_from_time_t() in pngwrite.c
2951
png_get_tIME() in pngget.c
2952
png_handle_tIME() in pngrutil.c, called in pngread.c
2953
png_set_tIME() in pngset.c
2954
png_write_tIME() in pngwutil.c, called in pngwrite.c
2956
All appear to handle dates properly in a Y2K environment. The
2957
png_convert_from_time_t() function calls gmtime() to convert from system
2958
clock time, which returns (year - 1900), which we properly convert to
2959
the full 4-digit year. There is a possibility that applications using
2960
libpng are not passing 4-digit years into the png_convert_to_rfc_1123()
2961
function, or that they are incorrectly passing only a 2-digit year
2962
instead of "year - 1900" into the png_convert_from_struct_tm() function,
2963
but this is not under our control. The libpng documentation has always
2964
stated that it works with 4-digit years, and the APIs have been
2967
The tIME chunk itself is also Y2K compliant. It uses a 2-byte unsigned
2968
integer to hold the year, and can hold years as large as 65535.
2970
zlib, upon which libpng depends, is also Y2K compliant. It contains
2971
no date-related code.
2974
Glenn Randers-Pehrson
2976
PNG Development Group