1
libpng.txt - A description on how to use and modify libpng
3
libpng version 1.2.22 - October 13, 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.
93
There are two main structures that are important to libpng, png_struct
94
and png_info. The first, png_struct, is an internal structure that
95
will not, for the most part, be used by a user except as the first
96
variable passed to every libpng function call.
98
The png_info structure is designed to provide information about the
99
PNG file. At one time, the fields of png_info were intended to be
100
directly accessible to the user. However, this tended to cause problems
101
with applications using dynamically loaded libraries, and as a result
102
a set of interface functions for png_info (the png_get_*() and png_set_*()
103
functions) was developed. The fields of png_info are still available for
104
older applications, but it is suggested that applications use the new
105
interfaces if at all possible.
107
Applications that do make direct access to the members of png_struct (except
108
for png_ptr->jmpbuf) must be recompiled whenever the library is updated,
109
and applications that make direct access to the members of png_info must
110
be recompiled if they were compiled or loaded with libpng version 1.0.6,
111
in which the members were in a different order. In version 1.0.7, the
112
members of the png_info structure reverted to the old order, as they were
113
in versions 0.97c through 1.0.5. Starting with version 2.0.0, both
114
structures are going to be hidden, and the contents of the structures will
115
only be accessible through the png_get/png_set functions.
117
The png.h header file is an invaluable reference for programming with libpng.
118
And while I'm on the topic, make sure you include the libpng header file:
124
We'll now walk you through the possible functions to call when reading
125
in a PNG file sequentially, briefly explaining the syntax and purpose
126
of each one. See example.c and png.h for more detail. While
127
progressive reading is covered in the next section, you will still
128
need some of the functions discussed in this section to read a PNG
133
You will want to do the I/O initialization(*) before you get into libpng,
134
so if it doesn't work, you don't have much to undo. Of course, you
135
will also want to insure that you are, in fact, dealing with a PNG
136
file. Libpng provides a simple check to see if a file is a PNG file.
137
To use it, pass in the first 1 to 8 bytes of the file to the function
138
png_sig_cmp(), and it will return 0 if the bytes match the corresponding
139
bytes of the PNG signature, or nonzero otherwise. Of course, the more bytes
140
you pass in, the greater the accuracy of the prediction.
142
If you are intending to keep the file pointer open for use in libpng,
143
you must ensure you don't read more than 8 bytes from the beginning
144
of the file, and you also have to make a call to png_set_sig_bytes_read()
145
with the number of bytes you read from the beginning. Libpng will
146
then only check the bytes (if any) that your program didn't read.
148
(*): If you are not using the standard I/O functions, you will need
149
to replace them with custom functions. See the discussion under
153
FILE *fp = fopen(file_name, "rb");
158
fread(header, 1, number, fp);
159
is_png = !png_sig_cmp(header, 0, number);
166
Next, png_struct and png_info need to be allocated and initialized. In
167
order to ensure that the size of these structures is correct even with a
168
dynamically linked libpng, there are functions to initialize and
169
allocate the structures. We also pass the library version, optional
170
pointers to error handling functions, and a pointer to a data struct for
171
use by the error functions, if necessary (the pointer and functions can
172
be NULL if the default error handlers are to be used). See the section
173
on Changes to Libpng below regarding the old initialization functions.
174
The structure allocation functions quietly return NULL if they fail to
175
create the structure, so your application should check for that.
177
png_structp png_ptr = png_create_read_struct
178
(PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
179
user_error_fn, user_warning_fn);
183
png_infop info_ptr = png_create_info_struct(png_ptr);
186
png_destroy_read_struct(&png_ptr,
187
(png_infopp)NULL, (png_infopp)NULL);
191
png_infop end_info = png_create_info_struct(png_ptr);
194
png_destroy_read_struct(&png_ptr, &info_ptr,
199
If you want to use your own memory allocation routines,
200
define PNG_USER_MEM_SUPPORTED and use
201
png_create_read_struct_2() instead of png_create_read_struct():
203
png_structp png_ptr = png_create_read_struct_2
204
(PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
205
user_error_fn, user_warning_fn, (png_voidp)
206
user_mem_ptr, user_malloc_fn, user_free_fn);
208
The error handling routines passed to png_create_read_struct()
209
and the memory alloc/free routines passed to png_create_struct_2()
210
are only necessary if you are not using the libpng supplied error
211
handling and memory alloc/free functions.
213
When libpng encounters an error, it expects to longjmp back
214
to your routine. Therefore, you will need to call setjmp and pass
215
your png_jmpbuf(png_ptr). If you read the file from different
216
routines, you will need to update the jmpbuf field every time you enter
217
a new routine that will call a png_*() function.
219
See your documentation of setjmp/longjmp for your compiler for more
220
information on setjmp/longjmp. See the discussion on libpng error
221
handling in the Customizing Libpng section below for more information
222
on the libpng error handling. If an error occurs, and libpng longjmp's
223
back to your setjmp, you will want to call png_destroy_read_struct() to
226
if (setjmp(png_jmpbuf(png_ptr)))
228
png_destroy_read_struct(&png_ptr, &info_ptr,
234
If you would rather avoid the complexity of setjmp/longjmp issues,
235
you can compile libpng with PNG_SETJMP_NOT_SUPPORTED, in which case
236
errors will result in a call to PNG_ABORT() which defaults to abort().
238
Now you need to set up the input code. The default for libpng is to
239
use the C function fread(). If you use this, you will need to pass a
240
valid FILE * in the function png_init_io(). Be sure that the file is
241
opened in binary mode. If you wish to handle reading data in another
242
way, you need not call the png_init_io() function, but you must then
243
implement the libpng I/O methods discussed in the Customizing Libpng
246
png_init_io(png_ptr, fp);
248
If you had previously opened the file and read any of the signature from
249
the beginning in order to see if this was a PNG file, you need to let
250
libpng know that there are some bytes missing from the start of the file.
252
png_set_sig_bytes(png_ptr, number);
254
Setting up callback code
256
You can set up a callback function to handle any unknown chunks in the
257
input stream. You must supply the function
259
read_chunk_callback(png_ptr ptr,
260
png_unknown_chunkp chunk);
262
/* The unknown chunk structure contains your
267
/* Note that libpng has already taken care of
270
/* put your code here. Return one of the
273
return (-n); /* chunk had an error */
274
return (0); /* did not recognize */
275
return (n); /* success */
278
(You can give your function another name that you like instead of
279
"read_chunk_callback")
281
To inform libpng about your function, use
283
png_set_read_user_chunk_fn(png_ptr, user_chunk_ptr,
284
read_chunk_callback);
286
This names not only the callback function, but also a user pointer that
287
you can retrieve with
289
png_get_user_chunk_ptr(png_ptr);
291
At this point, you can set up a callback function that will be
292
called after each row has been read, which you can use to control
293
a progress meter or the like. It's demonstrated in pngtest.c.
294
You must supply a function
296
void read_row_callback(png_ptr ptr, png_uint_32 row,
299
/* put your code here */
302
(You can give it another name that you like instead of "read_row_callback")
304
To inform libpng about your function, use
306
png_set_read_status_fn(png_ptr, read_row_callback);
308
Width and height limits
310
The PNG specification allows the width and height of an image to be as
311
large as 2^31-1 (0x7fffffff), or about 2.147 billion rows and columns.
312
Since very few applications really need to process such large images,
313
we have imposed an arbitrary 1-million limit on rows and columns.
314
Larger images will be rejected immediately with a png_error() call. If
315
you wish to override this limit, you can use
317
png_set_user_limits(png_ptr, width_max, height_max);
319
to set your own limits, or use width_max = height_max = 0x7fffffffL
320
to allow all valid dimensions (libpng may reject some very large images
321
anyway because of potential buffer overflow conditions).
323
You should put this statement after you create the PNG structure and
324
before calling png_read_info(), png_read_png(), or png_process_data().
325
If you need to retrieve the limits that are being applied, use
327
width_max = png_get_user_width_max(png_ptr);
328
height_max = png_get_user_height_max(png_ptr);
330
Unknown-chunk handling
332
Now you get to set the way the library processes unknown chunks in the
333
input PNG stream. Both known and unknown chunks will be read. Normal
334
behavior is that known chunks will be parsed into information in
335
various info_ptr members; unknown chunks will be discarded. To change
338
png_set_keep_unknown_chunks(png_ptr, keep,
339
chunk_list, num_chunks);
340
keep - 0: do not handle as unknown
342
2: keep only if safe-to-copy
343
3: keep even if unsafe-to-copy
344
You can use these definitions:
345
PNG_HANDLE_CHUNK_AS_DEFAULT 0
346
PNG_HANDLE_CHUNK_NEVER 1
347
PNG_HANDLE_CHUNK_IF_SAFE 2
348
PNG_HANDLE_CHUNK_ALWAYS 3
349
chunk_list - list of chunks affected (a byte string,
350
five bytes per chunk, NULL or '\0' if
352
num_chunks - number of chunks affected; if 0, all
353
unknown chunks are affected. If nonzero,
354
only the chunks in the list are affected
356
Unknown chunks declared in this way will be saved as raw data onto a
357
list of png_unknown_chunk structures. If a chunk that is normally
358
known to libpng is named in the list, it will be handled as unknown,
359
according to the "keep" directive. If a chunk is named in successive
360
instances of png_set_keep_unknown_chunks(), the final instance will
361
take precedence. The IHDR and IEND chunks should not be named in
362
chunk_list; if they are, libpng will process them normally anyway.
364
The high-level read interface
366
At this point there are two ways to proceed; through the high-level
367
read interface, or through a sequence of low-level read operations.
368
You can use the high-level interface if (a) you are willing to read
369
the entire image into memory, and (b) the input transformations
370
you want to do are limited to the following set:
372
PNG_TRANSFORM_IDENTITY No transformation
373
PNG_TRANSFORM_STRIP_16 Strip 16-bit samples to
375
PNG_TRANSFORM_STRIP_ALPHA Discard the alpha channel
376
PNG_TRANSFORM_PACKING Expand 1, 2 and 4-bit
378
PNG_TRANSFORM_PACKSWAP Change order of packed
380
PNG_TRANSFORM_EXPAND Perform set_expand()
381
PNG_TRANSFORM_INVERT_MONO Invert monochrome images
382
PNG_TRANSFORM_SHIFT Normalize pixels to the
384
PNG_TRANSFORM_BGR Flip RGB to BGR, RGBA
386
PNG_TRANSFORM_SWAP_ALPHA Flip RGBA to ARGB or GA
388
PNG_TRANSFORM_INVERT_ALPHA Change alpha from opacity
390
PNG_TRANSFORM_SWAP_ENDIAN Byte-swap 16-bit samples
392
(This excludes setting a background color, doing gamma transformation,
393
dithering, and setting filler.) If this is the case, simply do this:
395
png_read_png(png_ptr, info_ptr, png_transforms, NULL)
397
where png_transforms is an integer containing the bitwise OR of
398
some set of transformation flags. This call is equivalent to png_read_info(),
399
followed the set of transformations indicated by the transform mask,
400
then png_read_image(), and finally png_read_end().
402
(The final parameter of this call is not yet used. Someday it might point
403
to transformation parameters required by some future input transform.)
405
You must use png_transforms and not call any png_set_transform() functions
406
when you use png_read_png().
408
After you have called png_read_png(), you can retrieve the image data
411
row_pointers = png_get_rows(png_ptr, info_ptr);
413
where row_pointers is an array of pointers to the pixel data for each row:
415
png_bytep row_pointers[height];
417
If you know your image size and pixel size ahead of time, you can allocate
418
row_pointers prior to calling png_read_png() with
420
if (height > PNG_UINT_32_MAX/png_sizeof(png_byte))
422
"Image is too tall to process in memory");
423
if (width > PNG_UINT_32_MAX/pixel_size)
425
"Image is too wide to process in memory");
426
row_pointers = png_malloc(png_ptr,
427
height*png_sizeof(png_bytep));
428
for (int i=0; i<height, i++)
429
row_pointers[i]=png_malloc(png_ptr,
431
png_set_rows(png_ptr, info_ptr, &row_pointers);
433
Alternatively you could allocate your image in one big block and define
434
row_pointers[i] to point into the proper places in your block.
436
If you use png_set_rows(), the application is responsible for freeing
437
row_pointers (and row_pointers[i], if they were separately allocated).
439
If you don't allocate row_pointers ahead of time, png_read_png() will
440
do it, and it'll be free'ed when you call png_destroy_*().
442
The low-level read interface
444
If you are going the low-level route, you are now ready to read all
445
the file information up to the actual image data. You do this with a
446
call to png_read_info().
448
png_read_info(png_ptr, info_ptr);
450
This will process all chunks up to but not including the image data.
452
Querying the info structure
454
Functions are used to get the information from the info_ptr once it
455
has been read. Note that these fields may not be completely filled
456
in until png_read_end() has read the chunk data following the image.
458
png_get_IHDR(png_ptr, info_ptr, &width, &height,
459
&bit_depth, &color_type, &interlace_type,
460
&compression_type, &filter_method);
462
width - holds the width of the image
463
in pixels (up to 2^31).
464
height - holds the height of the image
465
in pixels (up to 2^31).
466
bit_depth - holds the bit depth of one of the
467
image channels. (valid values are
468
1, 2, 4, 8, 16 and depend also on
469
the color_type. See also
470
significant bits (sBIT) below).
471
color_type - describes which color/alpha channels
474
(bit depths 1, 2, 4, 8, 16)
475
PNG_COLOR_TYPE_GRAY_ALPHA
477
PNG_COLOR_TYPE_PALETTE
478
(bit depths 1, 2, 4, 8)
481
PNG_COLOR_TYPE_RGB_ALPHA
484
PNG_COLOR_MASK_PALETTE
488
filter_method - (must be PNG_FILTER_TYPE_BASE
489
for PNG 1.0, and can also be
490
PNG_INTRAPIXEL_DIFFERENCING if
491
the PNG datastream is embedded in
492
a MNG-1.0 datastream)
493
compression_type - (must be PNG_COMPRESSION_TYPE_BASE
495
interlace_type - (PNG_INTERLACE_NONE or
497
Any or all of interlace_type, compression_type, of
498
filter_method can be NULL if you are
499
not interested in their values.
501
channels = png_get_channels(png_ptr, info_ptr);
502
channels - number of channels of info for the
503
color type (valid values are 1 (GRAY,
504
PALETTE), 2 (GRAY_ALPHA), 3 (RGB),
505
4 (RGB_ALPHA or RGB + filler byte))
506
rowbytes = png_get_rowbytes(png_ptr, info_ptr);
507
rowbytes - number of bytes needed to hold a row
509
signature = png_get_signature(png_ptr, info_ptr);
510
signature - holds the signature read from the
511
file (if any). The data is kept in
512
the same offset it would be if the
513
whole signature were read (i.e. if an
514
application had already read in 4
515
bytes of signature before starting
516
libpng, the remaining 4 bytes would
517
be in signature[4] through signature[7]
518
(see png_set_sig_bytes())).
521
width = png_get_image_width(png_ptr,
523
height = png_get_image_height(png_ptr,
525
bit_depth = png_get_bit_depth(png_ptr,
527
color_type = png_get_color_type(png_ptr,
529
filter_method = png_get_filter_type(png_ptr,
531
compression_type = png_get_compression_type(png_ptr,
533
interlace_type = png_get_interlace_type(png_ptr,
537
These are also important, but their validity depends on whether the chunk
538
has been read. The png_get_valid(png_ptr, info_ptr, PNG_INFO_<chunk>) and
539
png_get_<chunk>(png_ptr, info_ptr, ...) functions return non-zero if the
540
data has been read, or zero if it is missing. The parameters to the
541
png_get_<chunk> are set directly if they are simple data types, or a pointer
542
into the info_ptr is returned for any complex types.
544
png_get_PLTE(png_ptr, info_ptr, &palette,
546
palette - the palette for the file
548
num_palette - number of entries in the palette
550
png_get_gAMA(png_ptr, info_ptr, &gamma);
551
gamma - the gamma the file is written
554
png_get_sRGB(png_ptr, info_ptr, &srgb_intent);
555
srgb_intent - the rendering intent (PNG_INFO_sRGB)
556
The presence of the sRGB chunk
557
means that the pixel data is in the
558
sRGB color space. This chunk also
559
implies specific values of gAMA and
562
png_get_iCCP(png_ptr, info_ptr, &name,
563
&compression_type, &profile, &proflen);
564
name - The profile name.
565
compression - The compression type; always
566
PNG_COMPRESSION_TYPE_BASE for PNG 1.0.
567
You may give NULL to this argument to
569
profile - International Color Consortium color
570
profile data. May contain NULs.
571
proflen - length of profile data in bytes.
573
png_get_sBIT(png_ptr, info_ptr, &sig_bit);
574
sig_bit - the number of significant bits for
575
(PNG_INFO_sBIT) each of the gray,
576
red, green, and blue channels,
577
whichever are appropriate for the
578
given color type (png_color_16)
580
png_get_tRNS(png_ptr, info_ptr, &trans, &num_trans,
582
trans - array of transparent entries for
583
palette (PNG_INFO_tRNS)
584
trans_values - graylevel or color sample values of
585
the single transparent color for
586
non-paletted images (PNG_INFO_tRNS)
587
num_trans - number of transparent entries
590
png_get_hIST(png_ptr, info_ptr, &hist);
592
hist - histogram of palette (array of
595
png_get_tIME(png_ptr, info_ptr, &mod_time);
596
mod_time - time image was last modified
599
png_get_bKGD(png_ptr, info_ptr, &background);
600
background - background color (PNG_VALID_bKGD)
601
valid 16-bit red, green and blue
602
values, regardless of color_type
604
num_comments = png_get_text(png_ptr, info_ptr,
605
&text_ptr, &num_text);
606
num_comments - number of comments
607
text_ptr - array of png_text holding image
609
text_ptr[i].compression - type of compression used
610
on "text" PNG_TEXT_COMPRESSION_NONE
611
PNG_TEXT_COMPRESSION_zTXt
612
PNG_ITXT_COMPRESSION_NONE
613
PNG_ITXT_COMPRESSION_zTXt
614
text_ptr[i].key - keyword for comment. Must contain
616
text_ptr[i].text - text comments for current
617
keyword. Can be empty.
618
text_ptr[i].text_length - length of text string,
619
after decompression, 0 for iTXt
620
text_ptr[i].itxt_length - length of itxt string,
621
after decompression, 0 for tEXt/zTXt
622
text_ptr[i].lang - language of comment (empty
624
text_ptr[i].lang_key - keyword in UTF-8
625
(empty string for unknown).
626
num_text - number of comments (same as
627
num_comments; you can put NULL here
628
to avoid the duplication)
629
Note while png_set_text() will accept text, language,
630
and translated keywords that can be NULL pointers, the
631
structure returned by png_get_text will always contain
632
regular zero-terminated C strings. They might be
633
empty strings but they will never be NULL pointers.
635
num_spalettes = png_get_sPLT(png_ptr, info_ptr,
637
palette_ptr - array of palette structures holding
638
contents of one or more sPLT chunks
640
num_spalettes - number of sPLT chunks read.
642
png_get_oFFs(png_ptr, info_ptr, &offset_x, &offset_y,
644
offset_x - positive offset from the left edge
646
offset_y - positive offset from the top edge
648
unit_type - PNG_OFFSET_PIXEL, PNG_OFFSET_MICROMETER
650
png_get_pHYs(png_ptr, info_ptr, &res_x, &res_y,
652
res_x - pixels/unit physical resolution in
654
res_y - pixels/unit physical resolution in
656
unit_type - PNG_RESOLUTION_UNKNOWN,
659
png_get_sCAL(png_ptr, info_ptr, &unit, &width,
661
unit - physical scale units (an integer)
662
width - width of a pixel in physical scale units
663
height - height of a pixel in physical scale units
664
(width and height are doubles)
666
png_get_sCAL_s(png_ptr, info_ptr, &unit, &width,
668
unit - physical scale units (an integer)
669
width - width of a pixel in physical scale units
670
height - height of a pixel in physical scale units
671
(width and height are strings like "2.54")
673
num_unknown_chunks = png_get_unknown_chunks(png_ptr,
675
unknowns - array of png_unknown_chunk
676
structures holding unknown chunks
677
unknowns[i].name - name of unknown chunk
678
unknowns[i].data - data of unknown chunk
679
unknowns[i].size - size of unknown chunk's data
680
unknowns[i].location - position of chunk in file
682
The value of "i" corresponds to the order in which the
683
chunks were read from the PNG file or inserted with the
684
png_set_unknown_chunks() function.
686
The data from the pHYs chunk can be retrieved in several convenient
689
res_x = png_get_x_pixels_per_meter(png_ptr,
691
res_y = png_get_y_pixels_per_meter(png_ptr,
693
res_x_and_y = png_get_pixels_per_meter(png_ptr,
695
res_x = png_get_x_pixels_per_inch(png_ptr,
697
res_y = png_get_y_pixels_per_inch(png_ptr,
699
res_x_and_y = png_get_pixels_per_inch(png_ptr,
701
aspect_ratio = png_get_pixel_aspect_ratio(png_ptr,
704
(Each of these returns 0 [signifying "unknown"] if
705
the data is not present or if res_x is 0;
706
res_x_and_y is 0 if res_x != res_y)
708
The data from the oFFs chunk can be retrieved in several convenient
711
x_offset = png_get_x_offset_microns(png_ptr, info_ptr);
712
y_offset = png_get_y_offset_microns(png_ptr, info_ptr);
713
x_offset = png_get_x_offset_inches(png_ptr, info_ptr);
714
y_offset = png_get_y_offset_inches(png_ptr, info_ptr);
716
(Each of these returns 0 [signifying "unknown" if both
717
x and y are 0] if the data is not present or if the
718
chunk is present but the unit is the pixel)
720
For more information, see the png_info definition in png.h and the
721
PNG specification for chunk contents. Be careful with trusting
722
rowbytes, as some of the transformations could increase the space
723
needed to hold a row (expand, filler, gray_to_rgb, etc.).
724
See png_read_update_info(), below.
726
A quick word about text_ptr and num_text. PNG stores comments in
727
keyword/text pairs, one pair per chunk, with no limit on the number
728
of text chunks, and a 2^31 byte limit on their size. While there are
729
suggested keywords, there is no requirement to restrict the use to these
730
strings. It is strongly suggested that keywords and text be sensible
731
to humans (that's the point), so don't use abbreviations. Non-printing
732
symbols are not allowed. See the PNG specification for more details.
733
There is also no requirement to have text after the keyword.
735
Keywords should be limited to 79 Latin-1 characters without leading or
736
trailing spaces, but non-consecutive spaces are allowed within the
737
keyword. It is possible to have the same keyword any number of times.
738
The text_ptr is an array of png_text structures, each holding a
739
pointer to a language string, a pointer to a keyword and a pointer to
740
a text string. The text string, language code, and translated
741
keyword may be empty or NULL pointers. The keyword/text
742
pairs are put into the array in the order that they are received.
743
However, some or all of the text chunks may be after the image, so, to
744
make sure you have read all the text chunks, don't mess with these
745
until after you read the stuff after the image. This will be
746
mentioned again below in the discussion that goes with png_read_end().
748
Input transformations
750
After you've read the header information, you can set up the library
751
to handle any special transformations of the image data. The various
752
ways to transform the data will be described in the order that they
753
should occur. This is important, as some of these change the color
754
type and/or bit depth of the data, and some others only work on
755
certain color types and bit depths. Even though each transformation
756
checks to see if it has data that it can do something with, you should
757
make sure to only enable a transformation if it will be valid for the
758
data. For example, don't swap red and blue on grayscale data.
760
The colors used for the background and transparency values should be
761
supplied in the same format/depth as the current image data. They
762
are stored in the same format/depth as the image data in a bKGD or tRNS
763
chunk, so this is what libpng expects for this data. The colors are
764
transformed to keep in sync with the image data when an application
765
calls the png_read_update_info() routine (see below).
767
Data will be decoded into the supplied row buffers packed into bytes
768
unless the library has been told to transform it into another format.
769
For example, 4 bit/pixel paletted or grayscale data will be returned
770
2 pixels/byte with the leftmost pixel in the high-order bits of the
771
byte, unless png_set_packing() is called. 8-bit RGB data will be stored
772
in RGB RGB RGB format unless png_set_filler() or png_set_add_alpha()
773
is called to insert filler bytes, either before or after each RGB triplet.
774
16-bit RGB data will be returned RRGGBB RRGGBB, with the most significant
775
byte of the color value first, unless png_set_strip_16() is called to
776
transform it to regular RGB RGB triplets, or png_set_filler() or
777
png_set_add alpha() is called to insert filler bytes, either before or
778
after each RRGGBB triplet. Similarly, 8-bit or 16-bit grayscale data can
780
png_set_filler(), png_set_add_alpha(), or png_set_strip_16().
782
The following code transforms grayscale images of less than 8 to 8 bits,
783
changes paletted images to RGB, and adds a full alpha channel if there is
784
transparency information in a tRNS chunk. This is most useful on
785
grayscale images with bit depths of 2 or 4 or if there is a multiple-image
786
viewing application that wishes to treat all images in the same way.
788
if (color_type == PNG_COLOR_TYPE_PALETTE)
789
png_set_palette_to_rgb(png_ptr);
791
if (color_type == PNG_COLOR_TYPE_GRAY &&
792
bit_depth < 8) png_set_expand_gray_1_2_4_to_8(png_ptr);
794
if (png_get_valid(png_ptr, info_ptr,
795
PNG_INFO_tRNS)) png_set_tRNS_to_alpha(png_ptr);
797
These three functions are actually aliases for png_set_expand(), added
798
in libpng version 1.0.4, with the function names expanded to improve code
799
readability. In some future version they may actually do different
802
As of libpng version 1.2.9, png_set_expand_gray_1_2_4_to_8() was
803
added. It expands the sample depth without changing tRNS to alpha.
804
At the same time, png_set_gray_1_2_4_to_8() was deprecated, and it
805
will be removed from a future version.
807
PNG can have files with 16 bits per channel. If you only can handle
808
8 bits per channel, this will strip the pixels down to 8 bit.
811
png_set_strip_16(png_ptr);
813
If, for some reason, you don't need the alpha channel on an image,
814
and you want to remove it rather than combining it with the background
815
(but the image author certainly had in mind that you *would* combine
816
it with the background, so that's what you should probably do):
818
if (color_type & PNG_COLOR_MASK_ALPHA)
819
png_set_strip_alpha(png_ptr);
821
In PNG files, the alpha channel in an image
822
is the level of opacity. If you need the alpha channel in an image to
823
be the level of transparency instead of opacity, you can invert the
824
alpha channel (or the tRNS chunk data) after it's read, so that 0 is
825
fully opaque and 255 (in 8-bit or paletted images) or 65535 (in 16-bit
826
images) is fully transparent, with
828
png_set_invert_alpha(png_ptr);
830
PNG files pack pixels of bit depths 1, 2, and 4 into bytes as small as
831
they can, resulting in, for example, 8 pixels per byte for 1 bit
832
files. This code expands to 1 pixel per byte without changing the
833
values of the pixels:
836
png_set_packing(png_ptr);
838
PNG files have possible bit depths of 1, 2, 4, 8, and 16. All pixels
839
stored in a PNG image have been "scaled" or "shifted" up to the next
840
higher possible bit depth (e.g. from 5 bits/sample in the range [0,31] to
841
8 bits/sample in the range [0, 255]). However, it is also possible to
842
convert the PNG pixel data back to the original bit depth of the image.
843
This call reduces the pixels back down to the original bit depth:
845
png_color_8p sig_bit;
847
if (png_get_sBIT(png_ptr, info_ptr, &sig_bit))
848
png_set_shift(png_ptr, sig_bit);
850
PNG files store 3-color pixels in red, green, blue order. This code
851
changes the storage of the pixels to blue, green, red:
853
if (color_type == PNG_COLOR_TYPE_RGB ||
854
color_type == PNG_COLOR_TYPE_RGB_ALPHA)
855
png_set_bgr(png_ptr);
857
PNG files store RGB pixels packed into 3 or 6 bytes. This code expands them
858
into 4 or 8 bytes for windowing systems that need them in this format:
860
if (color_type == PNG_COLOR_TYPE_RGB)
861
png_set_filler(png_ptr, filler, PNG_FILLER_BEFORE);
863
where "filler" is the 8 or 16-bit number to fill with, and the location is
864
either PNG_FILLER_BEFORE or PNG_FILLER_AFTER, depending upon whether
865
you want the filler before the RGB or after. This transformation
866
does not affect images that already have full alpha channels. To add an
867
opaque alpha channel, use filler=0xff or 0xffff and PNG_FILLER_AFTER which
868
will generate RGBA pixels.
870
Note that png_set_filler() does not change the color type. If you want
871
to do that, you can add a true alpha channel with
873
if (color_type == PNG_COLOR_TYPE_RGB ||
874
color_type == PNG_COLOR_TYPE_GRAY)
875
png_set_add_alpha(png_ptr, filler, PNG_FILLER_AFTER);
877
where "filler" contains the alpha value to assign to each pixel.
878
This function was added in libpng-1.2.7.
880
If you are reading an image with an alpha channel, and you need the
881
data as ARGB instead of the normal PNG format RGBA:
883
if (color_type == PNG_COLOR_TYPE_RGB_ALPHA)
884
png_set_swap_alpha(png_ptr);
886
For some uses, you may want a grayscale image to be represented as
887
RGB. This code will do that conversion:
889
if (color_type == PNG_COLOR_TYPE_GRAY ||
890
color_type == PNG_COLOR_TYPE_GRAY_ALPHA)
891
png_set_gray_to_rgb(png_ptr);
893
Conversely, you can convert an RGB or RGBA image to grayscale or grayscale
896
if (color_type == PNG_COLOR_TYPE_RGB ||
897
color_type == PNG_COLOR_TYPE_RGB_ALPHA)
898
png_set_rgb_to_gray_fixed(png_ptr, error_action,
899
int red_weight, int green_weight);
901
error_action = 1: silently do the conversion
902
error_action = 2: issue a warning if the original
903
image has any pixel where
904
red != green or red != blue
905
error_action = 3: issue an error and abort the
906
conversion if the original
907
image has any pixel where
908
red != green or red != blue
910
red_weight: weight of red component times 100000
911
green_weight: weight of green component times 100000
912
If either weight is negative, default
913
weights (21268, 71514) are used.
915
If you have set error_action = 1 or 2, you can
916
later check whether the image really was gray, after processing
917
the image rows, with the png_get_rgb_to_gray_status(png_ptr) function.
918
It will return a png_byte that is zero if the image was gray or
919
1 if there were any non-gray pixels. bKGD and sBIT data
920
will be silently converted to grayscale, using the green channel
921
data, regardless of the error_action setting.
923
With red_weight+green_weight<=100000,
924
the normalized graylevel is computed:
926
int rw = red_weight * 65536;
927
int gw = green_weight * 65536;
928
int bw = 65536 - (rw + gw);
929
gray = (rw*red + gw*green + bw*blue)/65536;
931
The default values approximate those recommended in the Charles
932
Poynton's Color FAQ, <http://www.inforamp.net/~poynton/>
933
Copyright (c) 1998-01-04 Charles Poynton <poynton at inforamp.net>
935
Y = 0.212671 * R + 0.715160 * G + 0.072169 * B
937
Libpng approximates this with
939
Y = 0.21268 * R + 0.7151 * G + 0.07217 * B
941
which can be expressed with integers as
943
Y = (6969 * R + 23434 * G + 2365 * B)/32768
945
The calculation is done in a linear colorspace, if the image gamma
948
If you have a grayscale and you are using png_set_expand_depth(),
949
png_set_expand(), or png_set_gray_to_rgb to change to truecolor or to
950
a higher bit-depth, you must either supply the background color as a gray
951
value at the original file bit-depth (need_expand = 1) or else supply the
952
background color as an RGB triplet at the final, expanded bit depth
953
(need_expand = 0). Similarly, if you are reading a paletted image, you
954
must either supply the background color as a palette index (need_expand = 1)
955
or as an RGB triplet that may or may not be in the palette (need_expand = 0).
957
png_color_16 my_background;
958
png_color_16p image_background;
960
if (png_get_bKGD(png_ptr, info_ptr, &image_background))
961
png_set_background(png_ptr, image_background,
962
PNG_BACKGROUND_GAMMA_FILE, 1, 1.0);
964
png_set_background(png_ptr, &my_background,
965
PNG_BACKGROUND_GAMMA_SCREEN, 0, 1.0);
967
The png_set_background() function tells libpng to composite images
968
with alpha or simple transparency against the supplied background
969
color. If the PNG file contains a bKGD chunk (PNG_INFO_bKGD valid),
970
you may use this color, or supply another color more suitable for
971
the current display (e.g., the background color from a web page). You
972
need to tell libpng whether the color is in the gamma space of the
973
display (PNG_BACKGROUND_GAMMA_SCREEN for colors you supply), the file
974
(PNG_BACKGROUND_GAMMA_FILE for colors from the bKGD chunk), or one
975
that is neither of these gammas (PNG_BACKGROUND_GAMMA_UNIQUE - I don't
976
know why anyone would use this, but it's here).
978
To properly display PNG images on any kind of system, the application needs
979
to know what the display gamma is. Ideally, the user will know this, and
980
the application will allow them to set it. One method of allowing the user
981
to set the display gamma separately for each system is to check for a
982
SCREEN_GAMMA or DISPLAY_GAMMA environment variable, which will hopefully be
985
Note that display_gamma is the overall gamma correction required to produce
986
pleasing results, which depends on the lighting conditions in the surrounding
987
environment. In a dim or brightly lit room, no compensation other than
988
the physical gamma exponent of the monitor is needed, while in a dark room
989
a slightly smaller exponent is better.
991
double gamma, screen_gamma;
993
if (/* We have a user-defined screen
996
screen_gamma = user_defined_screen_gamma;
998
/* One way that applications can share the same
999
screen gamma value */
1000
else if ((gamma_str = getenv("SCREEN_GAMMA"))
1003
screen_gamma = (double)atof(gamma_str);
1005
/* If we don't have another value */
1008
screen_gamma = 2.2; /* A good guess for a
1009
PC monitor in a bright office or a dim room */
1010
screen_gamma = 2.0; /* A good guess for a
1011
PC monitor in a dark room */
1012
screen_gamma = 1.7 or 1.0; /* A good
1013
guess for Mac systems */
1016
The png_set_gamma() function handles gamma transformations of the data.
1017
Pass both the file gamma and the current screen_gamma. If the file does
1018
not have a gamma value, you can pass one anyway if you have an idea what
1019
it is (usually 0.45455 is a good guess for GIF images on PCs). Note
1020
that file gammas are inverted from screen gammas. See the discussions
1021
on gamma in the PNG specification for an excellent description of what
1022
gamma is, and why all applications should support it. It is strongly
1023
recommended that PNG viewers support gamma correction.
1025
if (png_get_gAMA(png_ptr, info_ptr, &gamma))
1026
png_set_gamma(png_ptr, screen_gamma, gamma);
1028
png_set_gamma(png_ptr, screen_gamma, 0.45455);
1030
If you need to reduce an RGB file to a paletted file, or if a paletted
1031
file has more entries then will fit on your screen, png_set_dither()
1032
will do that. Note that this is a simple match dither that merely
1033
finds the closest color available. This should work fairly well with
1034
optimized palettes, and fairly badly with linear color cubes. If you
1035
pass a palette that is larger then maximum_colors, the file will
1036
reduce the number of colors in the palette so it will fit into
1037
maximum_colors. If there is a histogram, it will use it to make
1038
more intelligent choices when reducing the palette. If there is no
1039
histogram, it may not do as good a job.
1041
if (color_type & PNG_COLOR_MASK_COLOR)
1043
if (png_get_valid(png_ptr, info_ptr,
1046
png_uint_16p histogram = NULL;
1048
png_get_hIST(png_ptr, info_ptr,
1050
png_set_dither(png_ptr, palette, num_palette,
1051
max_screen_colors, histogram, 1);
1055
png_color std_color_cube[MAX_SCREEN_COLORS] =
1058
png_set_dither(png_ptr, std_color_cube,
1059
MAX_SCREEN_COLORS, MAX_SCREEN_COLORS,
1064
PNG files describe monochrome as black being zero and white being one.
1065
The following code will reverse this (make black be one and white be
1068
if (bit_depth == 1 && color_type == PNG_COLOR_TYPE_GRAY)
1069
png_set_invert_mono(png_ptr);
1071
This function can also be used to invert grayscale and gray-alpha images:
1073
if (color_type == PNG_COLOR_TYPE_GRAY ||
1074
color_type == PNG_COLOR_TYPE_GRAY_ALPHA)
1075
png_set_invert_mono(png_ptr);
1077
PNG files store 16 bit pixels in network byte order (big-endian,
1078
ie. most significant bits first). This code changes the storage to the
1079
other way (little-endian, i.e. least significant bits first, the
1080
way PCs store them):
1082
if (bit_depth == 16)
1083
png_set_swap(png_ptr);
1085
If you are using packed-pixel images (1, 2, or 4 bits/pixel), and you
1086
need to change the order the pixels are packed into bytes, you can use:
1089
png_set_packswap(png_ptr);
1091
Finally, you can write your own transformation function if none of
1092
the existing ones meets your needs. This is done by setting a callback
1095
png_set_read_user_transform_fn(png_ptr,
1098
You must supply the function
1100
void read_transform_fn(png_ptr ptr, row_info_ptr
1101
row_info, png_bytep data)
1103
See pngtest.c for a working example. Your function will be called
1104
after all of the other transformations have been processed.
1106
You can also set up a pointer to a user structure for use by your
1107
callback function, and you can inform libpng that your transform
1108
function will change the number of channels or bit depth with the
1111
png_set_user_transform_info(png_ptr, user_ptr,
1112
user_depth, user_channels);
1114
The user's application, not libpng, is responsible for allocating and
1115
freeing any memory required for the user structure.
1117
You can retrieve the pointer via the function
1118
png_get_user_transform_ptr(). For example:
1120
voidp read_user_transform_ptr =
1121
png_get_user_transform_ptr(png_ptr);
1123
The last thing to handle is interlacing; this is covered in detail below,
1124
but you must call the function here if you want libpng to handle expansion
1125
of the interlaced image.
1127
number_of_passes = png_set_interlace_handling(png_ptr);
1129
After setting the transformations, libpng can update your png_info
1130
structure to reflect any transformations you've requested with this
1131
call. This is most useful to update the info structure's rowbytes
1132
field so you can use it to allocate your image memory. This function
1133
will also update your palette with the correct screen_gamma and
1134
background if these have been given with the calls above.
1136
png_read_update_info(png_ptr, info_ptr);
1138
After you call png_read_update_info(), you can allocate any
1139
memory you need to hold the image. The row data is simply
1140
raw byte data for all forms of images. As the actual allocation
1141
varies among applications, no example will be given. If you
1142
are allocating one large chunk, you will need to build an
1143
array of pointers to each row, as it will be needed for some
1144
of the functions below.
1148
After you've allocated memory, you can read the image data.
1149
The simplest way to do this is in one function call. If you are
1150
allocating enough memory to hold the whole image, you can just
1151
call png_read_image() and libpng will read in all the image data
1152
and put it in the memory area supplied. You will need to pass in
1153
an array of pointers to each row.
1155
This function automatically handles interlacing, so you don't need
1156
to call png_set_interlace_handling() or call this function multiple
1157
times, or any of that other stuff necessary with png_read_rows().
1159
png_read_image(png_ptr, row_pointers);
1161
where row_pointers is:
1163
png_bytep row_pointers[height];
1165
You can point to void or char or whatever you use for pixels.
1167
If you don't want to read in the whole image at once, you can
1168
use png_read_rows() instead. If there is no interlacing (check
1169
interlace_type == PNG_INTERLACE_NONE), this is simple:
1171
png_read_rows(png_ptr, row_pointers, NULL,
1174
where row_pointers is the same as in the png_read_image() call.
1176
If you are doing this just one row at a time, you can do this with
1177
a single row_pointer instead of an array of row_pointers:
1179
png_bytep row_pointer = row;
1180
png_read_row(png_ptr, row_pointer, NULL);
1182
If the file is interlaced (interlace_type != 0 in the IHDR chunk), things
1183
get somewhat harder. The only current (PNG Specification version 1.2)
1184
interlacing type for PNG is (interlace_type == PNG_INTERLACE_ADAM7)
1185
is a somewhat complicated 2D interlace scheme, known as Adam7, that
1186
breaks down an image into seven smaller images of varying size, based
1189
libpng can fill out those images or it can give them to you "as is".
1190
If you want them filled out, there are two ways to do that. The one
1191
mentioned in the PNG specification is to expand each pixel to cover
1192
those pixels that have not been read yet (the "rectangle" method).
1193
This results in a blocky image for the first pass, which gradually
1194
smooths out as more pixels are read. The other method is the "sparkle"
1195
method, where pixels are drawn only in their final locations, with the
1196
rest of the image remaining whatever colors they were initialized to
1197
before the start of the read. The first method usually looks better,
1198
but tends to be slower, as there are more pixels to put in the rows.
1200
If you don't want libpng to handle the interlacing details, just call
1201
png_read_rows() seven times to read in all seven images. Each of the
1202
images is a valid image by itself, or they can all be combined on an
1203
8x8 grid to form a single image (although if you intend to combine them
1204
you would be far better off using the libpng interlace handling).
1206
The first pass will return an image 1/8 as wide as the entire image
1207
(every 8th column starting in column 0) and 1/8 as high as the original
1208
(every 8th row starting in row 0), the second will be 1/8 as wide
1209
(starting in column 4) and 1/8 as high (also starting in row 0). The
1210
third pass will be 1/4 as wide (every 4th pixel starting in column 0) and
1211
1/8 as high (every 8th row starting in row 4), and the fourth pass will
1212
be 1/4 as wide and 1/4 as high (every 4th column starting in column 2,
1213
and every 4th row starting in row 0). The fifth pass will return an
1214
image 1/2 as wide, and 1/4 as high (starting at column 0 and row 2),
1215
while the sixth pass will be 1/2 as wide and 1/2 as high as the original
1216
(starting in column 1 and row 0). The seventh and final pass will be as
1217
wide as the original, and 1/2 as high, containing all of the odd
1218
numbered scanlines. Phew!
1220
If you want libpng to expand the images, call this before calling
1221
png_start_read_image() or png_read_update_info():
1223
if (interlace_type == PNG_INTERLACE_ADAM7)
1225
= png_set_interlace_handling(png_ptr);
1227
This will return the number of passes needed. Currently, this
1228
is seven, but may change if another interlace type is added.
1229
This function can be called even if the file is not interlaced,
1230
where it will return one pass.
1232
If you are not going to display the image after each pass, but are
1233
going to wait until the entire image is read in, use the sparkle
1234
effect. This effect is faster and the end result of either method
1235
is exactly the same. If you are planning on displaying the image
1236
after each pass, the "rectangle" effect is generally considered the
1239
If you only want the "sparkle" effect, just call png_read_rows() as
1240
normal, with the third parameter NULL. Make sure you make pass over
1241
the image number_of_passes times, and you don't change the data in the
1242
rows between calls. You can change the locations of the data, just
1243
not the data. Each pass only writes the pixels appropriate for that
1244
pass, and assumes the data from previous passes is still valid.
1246
png_read_rows(png_ptr, row_pointers, NULL,
1249
If you only want the first effect (the rectangles), do the same as
1250
before except pass the row buffer in the third parameter, and leave
1251
the second parameter NULL.
1253
png_read_rows(png_ptr, NULL, row_pointers,
1256
Finishing a sequential read
1258
After you are finished reading the image through the
1259
low-level interface, you can finish reading the file. If you are
1260
interested in comments or time, which may be stored either before or
1261
after the image data, you should pass the separate png_info struct if
1262
you want to keep the comments from before and after the image
1263
separate. If you are not interested, you can pass NULL.
1265
png_read_end(png_ptr, end_info);
1267
When you are done, you can free all memory allocated by libpng like this:
1269
png_destroy_read_struct(&png_ptr, &info_ptr,
1272
It is also possible to individually free the info_ptr members that
1273
point to libpng-allocated storage with the following function:
1275
png_free_data(png_ptr, info_ptr, mask, seq)
1276
mask - identifies data to be freed, a mask
1277
containing the bitwise OR of one or
1279
PNG_FREE_PLTE, PNG_FREE_TRNS,
1280
PNG_FREE_HIST, PNG_FREE_ICCP,
1281
PNG_FREE_PCAL, PNG_FREE_ROWS,
1282
PNG_FREE_SCAL, PNG_FREE_SPLT,
1283
PNG_FREE_TEXT, PNG_FREE_UNKN,
1284
or simply PNG_FREE_ALL
1285
seq - sequence number of item to be freed
1288
This function may be safely called when the relevant storage has
1289
already been freed, or has not yet been allocated, or was allocated
1290
by the user and not by libpng, and will in those
1291
cases do nothing. The "seq" parameter is ignored if only one item
1292
of the selected data type, such as PLTE, is allowed. If "seq" is not
1293
-1, and multiple items are allowed for the data type identified in
1294
the mask, such as text or sPLT, only the n'th item in the structure
1295
is freed, where n is "seq".
1297
The default behavior is only to free data that was allocated internally
1298
by libpng. This can be changed, so that libpng will not free the data,
1299
or so that it will free data that was allocated by the user with png_malloc()
1300
or png_zalloc() and passed in via a png_set_*() function, with
1302
png_data_freer(png_ptr, info_ptr, freer, mask)
1303
mask - which data elements are affected
1304
same choices as in png_free_data()
1306
PNG_DESTROY_WILL_FREE_DATA
1307
PNG_SET_WILL_FREE_DATA
1308
PNG_USER_WILL_FREE_DATA
1310
This function only affects data that has already been allocated.
1311
You can call this function after reading the PNG data but before calling
1312
any png_set_*() functions, to control whether the user or the png_set_*()
1313
function is responsible for freeing any existing data that might be present,
1314
and again after the png_set_*() functions to control whether the user
1315
or png_destroy_*() is supposed to free the data. When the user assumes
1316
responsibility for libpng-allocated data, the application must use
1317
png_free() to free it, and when the user transfers responsibility to libpng
1318
for data that the user has allocated, the user must have used png_malloc()
1319
or png_zalloc() to allocate it.
1321
If you allocated your row_pointers in a single block, as suggested above in
1322
the description of the high level read interface, you must not transfer
1323
responsibility for freeing it to the png_set_rows or png_read_destroy function,
1324
because they would also try to free the individual row_pointers[i].
1326
If you allocated text_ptr.text, text_ptr.lang, and text_ptr.translated_keyword
1327
separately, do not transfer responsibility for freeing text_ptr to libpng,
1328
because when libpng fills a png_text structure it combines these members with
1329
the key member, and png_free_data() will free only text_ptr.key. Similarly,
1330
if you transfer responsibility for free'ing text_ptr from libpng to your
1331
application, your application must not separately free those members.
1333
The png_free_data() function will turn off the "valid" flag for anything
1334
it frees. If you need to turn the flag off for a chunk that was freed by your
1335
application instead of by libpng, you can use
1337
png_set_invalid(png_ptr, info_ptr, mask);
1338
mask - identifies the chunks to be made invalid,
1339
containing the bitwise OR of one or
1341
PNG_INFO_gAMA, PNG_INFO_sBIT,
1342
PNG_INFO_cHRM, PNG_INFO_PLTE,
1343
PNG_INFO_tRNS, PNG_INFO_bKGD,
1344
PNG_INFO_hIST, PNG_INFO_pHYs,
1345
PNG_INFO_oFFs, PNG_INFO_tIME,
1346
PNG_INFO_pCAL, PNG_INFO_sRGB,
1347
PNG_INFO_iCCP, PNG_INFO_sPLT,
1348
PNG_INFO_sCAL, PNG_INFO_IDAT
1350
For a more compact example of reading a PNG image, see the file example.c.
1352
Reading PNG files progressively
1354
The progressive reader is slightly different then the non-progressive
1355
reader. Instead of calling png_read_info(), png_read_rows(), and
1356
png_read_end(), you make one call to png_process_data(), which calls
1357
callbacks when it has the info, a row, or the end of the image. You
1358
set up these callbacks with png_set_progressive_read_fn(). You don't
1359
have to worry about the input/output functions of libpng, as you are
1360
giving the library the data directly in png_process_data(). I will
1361
assume that you have read the section on reading PNG files above,
1362
so I will only highlight the differences (although I will show
1365
png_structp png_ptr;
1368
/* An example code fragment of how you would
1369
initialize the progressive reader in your
1372
initialize_png_reader()
1374
png_ptr = png_create_read_struct
1375
(PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
1376
user_error_fn, user_warning_fn);
1379
info_ptr = png_create_info_struct(png_ptr);
1382
png_destroy_read_struct(&png_ptr, (png_infopp)NULL,
1387
if (setjmp(png_jmpbuf(png_ptr)))
1389
png_destroy_read_struct(&png_ptr, &info_ptr,
1394
/* This one's new. You can provide functions
1395
to be called when the header info is valid,
1396
when each row is completed, and when the image
1397
is finished. If you aren't using all functions,
1398
you can specify NULL parameters. Even when all
1399
three functions are NULL, you need to call
1400
png_set_progressive_read_fn(). You can use
1401
any struct as the user_ptr (cast to a void pointer
1402
for the function call), and retrieve the pointer
1403
from inside the callbacks using the function
1405
png_get_progressive_ptr(png_ptr);
1407
which will return a void pointer, which you have
1408
to cast appropriately.
1410
png_set_progressive_read_fn(png_ptr, (void *)user_ptr,
1411
info_callback, row_callback, end_callback);
1416
/* A code fragment that you call as you receive blocks
1419
process_data(png_bytep buffer, png_uint_32 length)
1421
if (setjmp(png_jmpbuf(png_ptr)))
1423
png_destroy_read_struct(&png_ptr, &info_ptr,
1428
/* This one's new also. Simply give it a chunk
1429
of data from the file stream (in order, of
1430
course). On machines with segmented memory
1431
models machines, don't give it any more than
1432
64K. The library seems to run fine with sizes
1433
of 4K. Although you can give it much less if
1434
necessary (I assume you can give it chunks of
1435
1 byte, I haven't tried less then 256 bytes
1436
yet). When this function returns, you may
1437
want to display any rows that were generated
1438
in the row callback if you don't already do
1441
png_process_data(png_ptr, info_ptr, buffer, length);
1445
/* This function is called (as set by
1446
png_set_progressive_read_fn() above) when enough data
1447
has been supplied so all of the header has been
1451
info_callback(png_structp png_ptr, png_infop info)
1453
/* Do any setup here, including setting any of
1454
the transformations mentioned in the Reading
1455
PNG files section. For now, you _must_ call
1456
either png_start_read_image() or
1457
png_read_update_info() after all the
1458
transformations are set (even if you don't set
1459
any). You may start getting rows before
1460
png_process_data() returns, so this is your
1461
last chance to prepare for that.
1465
/* This function is called when each row of image
1468
row_callback(png_structp png_ptr, png_bytep new_row,
1469
png_uint_32 row_num, int pass)
1471
/* If the image is interlaced, and you turned
1472
on the interlace handler, this function will
1473
be called for every row in every pass. Some
1474
of these rows will not be changed from the
1475
previous pass. When the row is not changed,
1476
the new_row variable will be NULL. The rows
1477
and passes are called in order, so you don't
1478
really need the row_num and pass, but I'm
1479
supplying them because it may make your life
1482
For the non-NULL rows of interlaced images,
1483
you must call png_progressive_combine_row()
1484
passing in the row and the old row. You can
1485
call this function for NULL rows (it will just
1486
return) and for non-interlaced images (it just
1487
does the memcpy for you) if it will make the
1488
code easier. Thus, you can just do this for
1492
png_progressive_combine_row(png_ptr, old_row,
1495
/* where old_row is what was displayed for
1496
previously for the row. Note that the first
1497
pass (pass == 0, really) will completely cover
1498
the old row, so the rows do not have to be
1499
initialized. After the first pass (and only
1500
for interlaced images), you will have to pass
1501
the current row, and the function will combine
1502
the old row and the new row.
1507
end_callback(png_structp png_ptr, png_infop info)
1509
/* This function is called after the whole image
1510
has been read, including any chunks after the
1511
image (up to and including the IEND). You
1512
will usually have the same info chunk as you
1513
had in the header, although some data may have
1514
been added to the comments and time fields.
1516
Most people won't do much here, perhaps setting
1517
a flag that marks the image as finished.
1525
Much of this is very similar to reading. However, everything of
1526
importance is repeated here, so you won't have to constantly look
1527
back up in the reading section to understand writing.
1531
You will want to do the I/O initialization before you get into libpng,
1532
so if it doesn't work, you don't have anything to undo. If you are not
1533
using the standard I/O functions, you will need to replace them with
1534
custom writing functions. See the discussion under Customizing libpng.
1536
FILE *fp = fopen(file_name, "wb");
1542
Next, png_struct and png_info need to be allocated and initialized.
1543
As these can be both relatively large, you may not want to store these
1544
on the stack, unless you have stack space to spare. Of course, you
1545
will want to check if they return NULL. If you are also reading,
1546
you won't want to name your read structure and your write structure
1547
both "png_ptr"; you can call them anything you like, such as
1548
"read_ptr" and "write_ptr". Look at pngtest.c, for example.
1550
png_structp png_ptr = png_create_write_struct
1551
(PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
1552
user_error_fn, user_warning_fn);
1556
png_infop info_ptr = png_create_info_struct(png_ptr);
1559
png_destroy_write_struct(&png_ptr,
1564
If you want to use your own memory allocation routines,
1565
define PNG_USER_MEM_SUPPORTED and use
1566
png_create_write_struct_2() instead of png_create_write_struct():
1568
png_structp png_ptr = png_create_write_struct_2
1569
(PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
1570
user_error_fn, user_warning_fn, (png_voidp)
1571
user_mem_ptr, user_malloc_fn, user_free_fn);
1573
After you have these structures, you will need to set up the
1574
error handling. When libpng encounters an error, it expects to
1575
longjmp() back to your routine. Therefore, you will need to call
1576
setjmp() and pass the png_jmpbuf(png_ptr). If you
1577
write the file from different routines, you will need to update
1578
the png_jmpbuf(png_ptr) every time you enter a new routine that will
1579
call a png_*() function. See your documentation of setjmp/longjmp
1580
for your compiler for more information on setjmp/longjmp. See
1581
the discussion on libpng error handling in the Customizing Libpng
1582
section below for more information on the libpng error handling.
1584
if (setjmp(png_jmpbuf(png_ptr)))
1586
png_destroy_write_struct(&png_ptr, &info_ptr);
1593
If you would rather avoid the complexity of setjmp/longjmp issues,
1594
you can compile libpng with PNG_SETJMP_NOT_SUPPORTED, in which case
1595
errors will result in a call to PNG_ABORT() which defaults to abort().
1597
Now you need to set up the output code. The default for libpng is to
1598
use the C function fwrite(). If you use this, you will need to pass a
1599
valid FILE * in the function png_init_io(). Be sure that the file is
1600
opened in binary mode. Again, if you wish to handle writing data in
1601
another way, see the discussion on libpng I/O handling in the Customizing
1602
Libpng section below.
1604
png_init_io(png_ptr, fp);
1606
If you are embedding your PNG into a datastream such as MNG, and don't
1607
want libpng to write the 8-byte signature, or if you have already
1608
written the signature in your application, use
1610
png_set_sig_bytes(png_ptr, 8);
1612
to inform libpng that it should not write a signature.
1616
At this point, you can set up a callback function that will be
1617
called after each row has been written, which you can use to control
1618
a progress meter or the like. It's demonstrated in pngtest.c.
1619
You must supply a function
1621
void write_row_callback(png_ptr, png_uint_32 row,
1624
/* put your code here */
1627
(You can give it another name that you like instead of "write_row_callback")
1629
To inform libpng about your function, use
1631
png_set_write_status_fn(png_ptr, write_row_callback);
1633
You now have the option of modifying how the compression library will
1634
run. The following functions are mainly for testing, but may be useful
1635
in some cases, like if you need to write PNG files extremely fast and
1636
are willing to give up some compression, or if you want to get the
1637
maximum possible compression at the expense of slower writing. If you
1638
have no special needs in this area, let the library do what it wants by
1639
not calling this function at all, as it has been tuned to deliver a good
1640
speed/compression ratio. The second parameter to png_set_filter() is
1641
the filter method, for which the only valid values are 0 (as of the
1642
July 1999 PNG specification, version 1.2) or 64 (if you are writing
1643
a PNG datastream that is to be embedded in a MNG datastream). The third
1644
parameter is a flag that indicates which filter type(s) are to be tested
1645
for each scanline. See the PNG specification for details on the specific filter
1649
/* turn on or off filtering, and/or choose
1650
specific filters. You can use either a single
1651
PNG_FILTER_VALUE_NAME or the bitwise OR of one
1652
or more PNG_FILTER_NAME masks. */
1653
png_set_filter(png_ptr, 0,
1654
PNG_FILTER_NONE | PNG_FILTER_VALUE_NONE |
1655
PNG_FILTER_SUB | PNG_FILTER_VALUE_SUB |
1656
PNG_FILTER_UP | PNG_FILTER_VALUE_UP |
1657
PNG_FILTER_AVE | PNG_FILTER_VALUE_AVE |
1658
PNG_FILTER_PAETH | PNG_FILTER_VALUE_PAETH|
1662
wants to start and stop using particular filters during compression,
1663
it should start out with all of the filters (to ensure that the previous
1664
row of pixels will be stored in case it's needed later), and then add
1665
and remove them after the start of compression.
1667
If you are writing a PNG datastream that is to be embedded in a MNG
1668
datastream, the second parameter can be either 0 or 64.
1670
The png_set_compression_*() functions interface to the zlib compression
1671
library, and should mostly be ignored unless you really know what you are
1672
doing. The only generally useful call is png_set_compression_level()
1673
which changes how much time zlib spends on trying to compress the image
1674
data. See the Compression Library (zlib.h and algorithm.txt, distributed
1675
with zlib) for details on the compression levels.
1677
/* set the zlib compression level */
1678
png_set_compression_level(png_ptr,
1679
Z_BEST_COMPRESSION);
1681
/* set other zlib parameters */
1682
png_set_compression_mem_level(png_ptr, 8);
1683
png_set_compression_strategy(png_ptr,
1684
Z_DEFAULT_STRATEGY);
1685
png_set_compression_window_bits(png_ptr, 15);
1686
png_set_compression_method(png_ptr, 8);
1687
png_set_compression_buffer_size(png_ptr, 8192)
1689
extern PNG_EXPORT(void,png_set_zbuf_size)
1691
Setting the contents of info for output
1693
You now need to fill in the png_info structure with all the data you
1694
wish to write before the actual image. Note that the only thing you
1695
are allowed to write after the image is the text chunks and the time
1696
chunk (as of PNG Specification 1.2, anyway). See png_write_end() and
1697
the latest PNG specification for more information on that. If you
1698
wish to write them before the image, fill them in now, and flag that
1699
data as being valid. If you want to wait until after the data, don't
1700
fill them until png_write_end(). For all the fields in png_info and
1701
their data types, see png.h. For explanations of what the fields
1702
contain, see the PNG specification.
1704
Some of the more important parts of the png_info are:
1706
png_set_IHDR(png_ptr, info_ptr, width, height,
1707
bit_depth, color_type, interlace_type,
1708
compression_type, filter_method)
1709
width - holds the width of the image
1710
in pixels (up to 2^31).
1711
height - holds the height of the image
1712
in pixels (up to 2^31).
1713
bit_depth - holds the bit depth of one of the
1715
(valid values are 1, 2, 4, 8, 16
1716
and depend also on the
1717
color_type. See also significant
1719
color_type - describes which color/alpha
1720
channels are present.
1722
(bit depths 1, 2, 4, 8, 16)
1723
PNG_COLOR_TYPE_GRAY_ALPHA
1725
PNG_COLOR_TYPE_PALETTE
1726
(bit depths 1, 2, 4, 8)
1729
PNG_COLOR_TYPE_RGB_ALPHA
1732
PNG_COLOR_MASK_PALETTE
1733
PNG_COLOR_MASK_COLOR
1734
PNG_COLOR_MASK_ALPHA
1736
interlace_type - PNG_INTERLACE_NONE or
1738
compression_type - (must be
1739
PNG_COMPRESSION_TYPE_DEFAULT)
1740
filter_method - (must be PNG_FILTER_TYPE_DEFAULT
1741
or, if you are writing a PNG to
1742
be embedded in a MNG datastream,
1744
PNG_INTRAPIXEL_DIFFERENCING)
1746
png_set_PLTE(png_ptr, info_ptr, palette,
1748
palette - the palette for the file
1749
(array of png_color)
1750
num_palette - number of entries in the palette
1752
png_set_gAMA(png_ptr, info_ptr, gamma);
1753
gamma - the gamma the image was created
1756
png_set_sRGB(png_ptr, info_ptr, srgb_intent);
1757
srgb_intent - the rendering intent
1758
(PNG_INFO_sRGB) The presence of
1759
the sRGB chunk means that the pixel
1760
data is in the sRGB color space.
1761
This chunk also implies specific
1762
values of gAMA and cHRM. Rendering
1763
intent is the CSS-1 property that
1764
has been defined by the International
1766
(http://www.color.org).
1768
PNG_sRGB_INTENT_SATURATION,
1769
PNG_sRGB_INTENT_PERCEPTUAL,
1770
PNG_sRGB_INTENT_ABSOLUTE, or
1771
PNG_sRGB_INTENT_RELATIVE.
1774
png_set_sRGB_gAMA_and_cHRM(png_ptr, info_ptr,
1776
srgb_intent - the rendering intent
1777
(PNG_INFO_sRGB) The presence of the
1778
sRGB chunk means that the pixel
1779
data is in the sRGB color space.
1780
This function also causes gAMA and
1781
cHRM chunks with the specific values
1782
that are consistent with sRGB to be
1785
png_set_iCCP(png_ptr, info_ptr, name, compression_type,
1787
name - The profile name.
1788
compression - The compression type; always
1789
PNG_COMPRESSION_TYPE_BASE for PNG 1.0.
1790
You may give NULL to this argument to
1792
profile - International Color Consortium color
1793
profile data. May contain NULs.
1794
proflen - length of profile data in bytes.
1796
png_set_sBIT(png_ptr, info_ptr, sig_bit);
1797
sig_bit - the number of significant bits for
1798
(PNG_INFO_sBIT) each of the gray, red,
1799
green, and blue channels, whichever are
1800
appropriate for the given color type
1803
png_set_tRNS(png_ptr, info_ptr, trans, num_trans,
1805
trans - array of transparent entries for
1806
palette (PNG_INFO_tRNS)
1807
trans_values - graylevel or color sample values of
1808
the single transparent color for
1809
non-paletted images (PNG_INFO_tRNS)
1810
num_trans - number of transparent entries
1813
png_set_hIST(png_ptr, info_ptr, hist);
1815
hist - histogram of palette (array of
1818
png_set_tIME(png_ptr, info_ptr, mod_time);
1819
mod_time - time image was last modified
1822
png_set_bKGD(png_ptr, info_ptr, background);
1823
background - background color (PNG_VALID_bKGD)
1825
png_set_text(png_ptr, info_ptr, text_ptr, num_text);
1826
text_ptr - array of png_text holding image
1828
text_ptr[i].compression - type of compression used
1829
on "text" PNG_TEXT_COMPRESSION_NONE
1830
PNG_TEXT_COMPRESSION_zTXt
1831
PNG_ITXT_COMPRESSION_NONE
1832
PNG_ITXT_COMPRESSION_zTXt
1833
text_ptr[i].key - keyword for comment. Must contain
1835
text_ptr[i].text - text comments for current
1836
keyword. Can be NULL or empty.
1837
text_ptr[i].text_length - length of text string,
1838
after decompression, 0 for iTXt
1839
text_ptr[i].itxt_length - length of itxt string,
1840
after decompression, 0 for tEXt/zTXt
1841
text_ptr[i].lang - language of comment (NULL or
1843
text_ptr[i].translated_keyword - keyword in UTF-8 (NULL
1844
or empty for unknown).
1845
num_text - number of comments
1847
png_set_sPLT(png_ptr, info_ptr, &palette_ptr,
1849
palette_ptr - array of png_sPLT_struct structures
1850
to be added to the list of palettes
1851
in the info structure.
1852
num_spalettes - number of palette structures to be
1855
png_set_oFFs(png_ptr, info_ptr, offset_x, offset_y,
1857
offset_x - positive offset from the left
1859
offset_y - positive offset from the top
1861
unit_type - PNG_OFFSET_PIXEL, PNG_OFFSET_MICROMETER
1863
png_set_pHYs(png_ptr, info_ptr, res_x, res_y,
1865
res_x - pixels/unit physical resolution
1867
res_y - pixels/unit physical resolution
1869
unit_type - PNG_RESOLUTION_UNKNOWN,
1870
PNG_RESOLUTION_METER
1872
png_set_sCAL(png_ptr, info_ptr, unit, width, height)
1873
unit - physical scale units (an integer)
1874
width - width of a pixel in physical scale units
1875
height - height of a pixel in physical scale units
1876
(width and height are doubles)
1878
png_set_sCAL_s(png_ptr, info_ptr, unit, width, height)
1879
unit - physical scale units (an integer)
1880
width - width of a pixel in physical scale units
1881
height - height of a pixel in physical scale units
1882
(width and height are strings like "2.54")
1884
png_set_unknown_chunks(png_ptr, info_ptr, &unknowns,
1886
unknowns - array of png_unknown_chunk
1887
structures holding unknown chunks
1888
unknowns[i].name - name of unknown chunk
1889
unknowns[i].data - data of unknown chunk
1890
unknowns[i].size - size of unknown chunk's data
1891
unknowns[i].location - position to write chunk in file
1892
0: do not write chunk
1893
PNG_HAVE_IHDR: before PLTE
1894
PNG_HAVE_PLTE: before IDAT
1895
PNG_AFTER_IDAT: after IDAT
1897
The "location" member is set automatically according to
1898
what part of the output file has already been written.
1899
You can change its value after calling png_set_unknown_chunks()
1900
as demonstrated in pngtest.c. Within each of the "locations",
1901
the chunks are sequenced according to their position in the
1902
structure (that is, the value of "i", which is the order in which
1903
the chunk was either read from the input file or defined with
1904
png_set_unknown_chunks).
1906
A quick word about text and num_text. text is an array of png_text
1907
structures. num_text is the number of valid structures in the array.
1908
Each png_text structure holds a language code, a keyword, a text value,
1909
and a compression type.
1911
The compression types have the same valid numbers as the compression
1912
types of the image data. Currently, the only valid number is zero.
1913
However, you can store text either compressed or uncompressed, unlike
1914
images, which always have to be compressed. So if you don't want the
1915
text compressed, set the compression type to PNG_TEXT_COMPRESSION_NONE.
1916
Because tEXt and zTXt chunks don't have a language field, if you
1917
specify PNG_TEXT_COMPRESSION_NONE or PNG_TEXT_COMPRESSION_zTXt
1918
any language code or translated keyword will not be written out.
1920
Until text gets around 1000 bytes, it is not worth compressing it.
1921
After the text has been written out to the file, the compression type
1922
is set to PNG_TEXT_COMPRESSION_NONE_WR or PNG_TEXT_COMPRESSION_zTXt_WR,
1923
so that it isn't written out again at the end (in case you are calling
1924
png_write_end() with the same struct.
1926
The keywords that are given in the PNG Specification are:
1928
Title Short (one line) title or
1930
Author Name of image's creator
1931
Description Description of image (possibly long)
1932
Copyright Copyright notice
1933
Creation Time Time of original image creation
1934
(usually RFC 1123 format, see below)
1935
Software Software used to create the image
1936
Disclaimer Legal disclaimer
1937
Warning Warning of nature of content
1938
Source Device used to create the image
1939
Comment Miscellaneous comment; conversion
1940
from other image format
1942
The keyword-text pairs work like this. Keywords should be short
1943
simple descriptions of what the comment is about. Some typical
1944
keywords are found in the PNG specification, as is some recommendations
1945
on keywords. You can repeat keywords in a file. You can even write
1946
some text before the image and some after. For example, you may want
1947
to put a description of the image before the image, but leave the
1948
disclaimer until after, so viewers working over modem connections
1949
don't have to wait for the disclaimer to go over the modem before
1950
they start seeing the image. Finally, keywords should be full
1951
words, not abbreviations. Keywords and text are in the ISO 8859-1
1952
(Latin-1) character set (a superset of regular ASCII) and can not
1953
contain NUL characters, and should not contain control or other
1954
unprintable characters. To make the comments widely readable, stick
1955
with basic ASCII, and avoid machine specific character set extensions
1956
like the IBM-PC character set. The keyword must be present, but
1957
you can leave off the text string on non-compressed pairs.
1958
Compressed pairs must have a text string, as only the text string
1959
is compressed anyway, so the compression would be meaningless.
1961
PNG supports modification time via the png_time structure. Two
1962
conversion routines are provided, png_convert_from_time_t() for
1963
time_t and png_convert_from_struct_tm() for struct tm. The
1964
time_t routine uses gmtime(). You don't have to use either of
1965
these, but if you wish to fill in the png_time structure directly,
1966
you should provide the time in universal time (GMT) if possible
1967
instead of your local time. Note that the year number is the full
1968
year (e.g. 1998, rather than 98 - PNG is year 2000 compliant!), and
1969
that months start with 1.
1971
If you want to store the time of the original image creation, you should
1972
use a plain tEXt chunk with the "Creation Time" keyword. This is
1973
necessary because the "creation time" of a PNG image is somewhat vague,
1974
depending on whether you mean the PNG file, the time the image was
1975
created in a non-PNG format, a still photo from which the image was
1976
scanned, or possibly the subject matter itself. In order to facilitate
1977
machine-readable dates, it is recommended that the "Creation Time"
1978
tEXt chunk use RFC 1123 format dates (e.g. "22 May 1997 18:07:10 GMT"),
1979
although this isn't a requirement. Unlike the tIME chunk, the
1980
"Creation Time" tEXt chunk is not expected to be automatically changed
1981
by the software. To facilitate the use of RFC 1123 dates, a function
1982
png_convert_to_rfc1123(png_timep) is provided to convert from PNG
1983
time to an RFC 1123 format string.
1985
Writing unknown chunks
1987
You can use the png_set_unknown_chunks function to queue up chunks
1988
for writing. You give it a chunk name, raw data, and a size; that's
1989
all there is to it. The chunks will be written by the next following
1990
png_write_info_before_PLTE, png_write_info, or png_write_end function.
1991
Any chunks previously read into the info structure's unknown-chunk
1992
list will also be written out in a sequence that satisfies the PNG
1993
specification's ordering rules.
1995
The high-level write interface
1997
At this point there are two ways to proceed; through the high-level
1998
write interface, or through a sequence of low-level write operations.
1999
You can use the high-level interface if your image data is present
2000
in the info structure. All defined output
2001
transformations are permitted, enabled by the following masks.
2003
PNG_TRANSFORM_IDENTITY No transformation
2004
PNG_TRANSFORM_PACKING Pack 1, 2 and 4-bit samples
2005
PNG_TRANSFORM_PACKSWAP Change order of packed
2007
PNG_TRANSFORM_INVERT_MONO Invert monochrome images
2008
PNG_TRANSFORM_SHIFT Normalize pixels to the
2010
PNG_TRANSFORM_BGR Flip RGB to BGR, RGBA
2012
PNG_TRANSFORM_SWAP_ALPHA Flip RGBA to ARGB or GA
2014
PNG_TRANSFORM_INVERT_ALPHA Change alpha from opacity
2016
PNG_TRANSFORM_SWAP_ENDIAN Byte-swap 16-bit samples
2017
PNG_TRANSFORM_STRIP_FILLER Strip out filler bytes.
2019
If you have valid image data in the info structure (you can use
2020
png_set_rows() to put image data in the info structure), simply do this:
2022
png_write_png(png_ptr, info_ptr, png_transforms, NULL)
2024
where png_transforms is an integer containing the bitwise OR of some set of
2025
transformation flags. This call is equivalent to png_write_info(),
2026
followed the set of transformations indicated by the transform mask,
2027
then png_write_image(), and finally png_write_end().
2029
(The final parameter of this call is not yet used. Someday it might point
2030
to transformation parameters required by some future output transform.)
2032
You must use png_transforms and not call any png_set_transform() functions
2033
when you use png_write_png().
2035
The low-level write interface
2037
If you are going the low-level route instead, you are now ready to
2038
write all the file information up to the actual image data. You do
2039
this with a call to png_write_info().
2041
png_write_info(png_ptr, info_ptr);
2043
Note that there is one transformation you may need to do before
2044
png_write_info(). In PNG files, the alpha channel in an image is the
2045
level of opacity. If your data is supplied as a level of
2046
transparency, you can invert the alpha channel before you write it, so
2047
that 0 is fully transparent and 255 (in 8-bit or paletted images) or
2048
65535 (in 16-bit images) is fully opaque, with
2050
png_set_invert_alpha(png_ptr);
2052
This must appear before png_write_info() instead of later with the
2053
other transformations because in the case of paletted images the tRNS
2054
chunk data has to be inverted before the tRNS chunk is written. If
2055
your image is not a paletted image, the tRNS data (which in such cases
2056
represents a single color to be rendered as transparent) won't need to
2057
be changed, and you can safely do this transformation after your
2058
png_write_info() call.
2060
If you need to write a private chunk that you want to appear before
2061
the PLTE chunk when PLTE is present, you can write the PNG info in
2062
two steps, and insert code to write your own chunk between them:
2064
png_write_info_before_PLTE(png_ptr, info_ptr);
2065
png_set_unknown_chunks(png_ptr, info_ptr, ...);
2066
png_write_info(png_ptr, info_ptr);
2068
After you've written the file information, you can set up the library
2069
to handle any special transformations of the image data. The various
2070
ways to transform the data will be described in the order that they
2071
should occur. This is important, as some of these change the color
2072
type and/or bit depth of the data, and some others only work on
2073
certain color types and bit depths. Even though each transformation
2074
checks to see if it has data that it can do something with, you should
2075
make sure to only enable a transformation if it will be valid for the
2076
data. For example, don't swap red and blue on grayscale data.
2078
PNG files store RGB pixels packed into 3 or 6 bytes. This code tells
2079
the library to strip input data that has 4 or 8 bytes per pixel down
2080
to 3 or 6 bytes (or strip 2 or 4-byte grayscale+filler data to 1 or 2
2083
png_set_filler(png_ptr, 0, PNG_FILLER_BEFORE);
2085
where the 0 is unused, and the location is either PNG_FILLER_BEFORE or
2086
PNG_FILLER_AFTER, depending upon whether the filler byte in the pixel
2087
is stored XRGB or RGBX.
2089
PNG files pack pixels of bit depths 1, 2, and 4 into bytes as small as
2090
they can, resulting in, for example, 8 pixels per byte for 1 bit files.
2091
If the data is supplied at 1 pixel per byte, use this code, which will
2092
correctly pack the pixels into a single byte:
2094
png_set_packing(png_ptr);
2096
PNG files reduce possible bit depths to 1, 2, 4, 8, and 16. If your
2097
data is of another bit depth, you can write an sBIT chunk into the
2098
file so that decoders can recover the original data if desired.
2100
/* Set the true bit depth of the image data */
2101
if (color_type & PNG_COLOR_MASK_COLOR)
2103
sig_bit.red = true_bit_depth;
2104
sig_bit.green = true_bit_depth;
2105
sig_bit.blue = true_bit_depth;
2109
sig_bit.gray = true_bit_depth;
2111
if (color_type & PNG_COLOR_MASK_ALPHA)
2113
sig_bit.alpha = true_bit_depth;
2116
png_set_sBIT(png_ptr, info_ptr, &sig_bit);
2118
If the data is stored in the row buffer in a bit depth other than
2119
one supported by PNG (e.g. 3 bit data in the range 0-7 for a 4-bit PNG),
2120
this will scale the values to appear to be the correct bit depth as
2123
png_set_shift(png_ptr, &sig_bit);
2125
PNG files store 16 bit pixels in network byte order (big-endian,
2126
ie. most significant bits first). This code would be used if they are
2127
supplied the other way (little-endian, i.e. least significant bits
2128
first, the way PCs store them):
2131
png_set_swap(png_ptr);
2133
If you are using packed-pixel images (1, 2, or 4 bits/pixel), and you
2134
need to change the order the pixels are packed into bytes, you can use:
2137
png_set_packswap(png_ptr);
2139
PNG files store 3 color pixels in red, green, blue order. This code
2140
would be used if they are supplied as blue, green, red:
2142
png_set_bgr(png_ptr);
2144
PNG files describe monochrome as black being zero and white being
2145
one. This code would be used if the pixels are supplied with this reversed
2146
(black being one and white being zero):
2148
png_set_invert_mono(png_ptr);
2150
Finally, you can write your own transformation function if none of
2151
the existing ones meets your needs. This is done by setting a callback
2154
png_set_write_user_transform_fn(png_ptr,
2155
write_transform_fn);
2157
You must supply the function
2159
void write_transform_fn(png_ptr ptr, row_info_ptr
2160
row_info, png_bytep data)
2162
See pngtest.c for a working example. Your function will be called
2163
before any of the other transformations are processed.
2165
You can also set up a pointer to a user structure for use by your
2168
png_set_user_transform_info(png_ptr, user_ptr, 0, 0);
2170
The user_channels and user_depth parameters of this function are ignored
2171
when writing; you can set them to zero as shown.
2173
You can retrieve the pointer via the function png_get_user_transform_ptr().
2176
voidp write_user_transform_ptr =
2177
png_get_user_transform_ptr(png_ptr);
2179
It is possible to have libpng flush any pending output, either manually,
2180
or automatically after a certain number of lines have been written. To
2181
flush the output stream a single time call:
2183
png_write_flush(png_ptr);
2185
and to have libpng flush the output stream periodically after a certain
2186
number of scanlines have been written, call:
2188
png_set_flush(png_ptr, nrows);
2190
Note that the distance between rows is from the last time png_write_flush()
2191
was called, or the first row of the image if it has never been called.
2192
So if you write 50 lines, and then png_set_flush 25, it will flush the
2193
output on the next scanline, and every 25 lines thereafter, unless
2194
png_write_flush() is called before 25 more lines have been written.
2195
If nrows is too small (less than about 10 lines for a 640 pixel wide
2196
RGB image) the image compression may decrease noticeably (although this
2197
may be acceptable for real-time applications). Infrequent flushing will
2198
only degrade the compression performance by a few percent over images
2199
that do not use flushing.
2201
Writing the image data
2203
That's it for the transformations. Now you can write the image data.
2204
The simplest way to do this is in one function call. If you have the
2205
whole image in memory, you can just call png_write_image() and libpng
2206
will write the image. You will need to pass in an array of pointers to
2207
each row. This function automatically handles interlacing, so you don't
2208
need to call png_set_interlace_handling() or call this function multiple
2209
times, or any of that other stuff necessary with png_write_rows().
2211
png_write_image(png_ptr, row_pointers);
2213
where row_pointers is:
2215
png_byte *row_pointers[height];
2217
You can point to void or char or whatever you use for pixels.
2219
If you don't want to write the whole image at once, you can
2220
use png_write_rows() instead. If the file is not interlaced,
2223
png_write_rows(png_ptr, row_pointers,
2226
row_pointers is the same as in the png_write_image() call.
2228
If you are just writing one row at a time, you can do this with
2229
a single row_pointer instead of an array of row_pointers:
2231
png_bytep row_pointer = row;
2233
png_write_row(png_ptr, row_pointer);
2235
When the file is interlaced, things can get a good deal more
2236
complicated. The only currently (as of the PNG Specification
2237
version 1.2, dated July 1999) defined interlacing scheme for PNG files
2238
is the "Adam7" interlace scheme, that breaks down an
2239
image into seven smaller images of varying size. libpng will build
2240
these images for you, or you can do them yourself. If you want to
2241
build them yourself, see the PNG specification for details of which
2242
pixels to write when.
2244
If you don't want libpng to handle the interlacing details, just
2245
use png_set_interlace_handling() and call png_write_rows() the
2246
correct number of times to write all seven sub-images.
2248
If you want libpng to build the sub-images, call this before you start
2252
png_set_interlace_handling(png_ptr);
2254
This will return the number of passes needed. Currently, this
2255
is seven, but may change if another interlace type is added.
2257
Then write the complete image number_of_passes times.
2259
png_write_rows(png_ptr, row_pointers,
2262
As some of these rows are not used, and thus return immediately,
2263
you may want to read about interlacing in the PNG specification,
2264
and only update the rows that are actually used.
2266
Finishing a sequential write
2268
After you are finished writing the image, you should finish writing
2269
the file. If you are interested in writing comments or time, you should
2270
pass an appropriately filled png_info pointer. If you are not interested,
2273
png_write_end(png_ptr, info_ptr);
2275
When you are done, you can free all memory used by libpng like this:
2277
png_destroy_write_struct(&png_ptr, &info_ptr);
2279
It is also possible to individually free the info_ptr members that
2280
point to libpng-allocated storage with the following function:
2282
png_free_data(png_ptr, info_ptr, mask, seq)
2283
mask - identifies data to be freed, a mask
2284
containing the bitwise OR of one or
2286
PNG_FREE_PLTE, PNG_FREE_TRNS,
2287
PNG_FREE_HIST, PNG_FREE_ICCP,
2288
PNG_FREE_PCAL, PNG_FREE_ROWS,
2289
PNG_FREE_SCAL, PNG_FREE_SPLT,
2290
PNG_FREE_TEXT, PNG_FREE_UNKN,
2291
or simply PNG_FREE_ALL
2292
seq - sequence number of item to be freed
2295
This function may be safely called when the relevant storage has
2296
already been freed, or has not yet been allocated, or was allocated
2297
by the user and not by libpng, and will in those
2298
cases do nothing. The "seq" parameter is ignored if only one item
2299
of the selected data type, such as PLTE, is allowed. If "seq" is not
2300
-1, and multiple items are allowed for the data type identified in
2301
the mask, such as text or sPLT, only the n'th item in the structure
2302
is freed, where n is "seq".
2304
If you allocated data such as a palette that you passed
2305
in to libpng with png_set_*, you must not free it until just before the call to
2306
png_destroy_write_struct().
2308
The default behavior is only to free data that was allocated internally
2309
by libpng. This can be changed, so that libpng will not free the data,
2310
or so that it will free data that was allocated by the user with png_malloc()
2311
or png_zalloc() and passed in via a png_set_*() function, with
2313
png_data_freer(png_ptr, info_ptr, freer, mask)
2314
mask - which data elements are affected
2315
same choices as in png_free_data()
2317
PNG_DESTROY_WILL_FREE_DATA
2318
PNG_SET_WILL_FREE_DATA
2319
PNG_USER_WILL_FREE_DATA
2321
For example, to transfer responsibility for some data from a read structure
2322
to a write structure, you could use
2324
png_data_freer(read_ptr, read_info_ptr,
2325
PNG_USER_WILL_FREE_DATA,
2326
PNG_FREE_PLTE|PNG_FREE_tRNS|PNG_FREE_hIST)
2327
png_data_freer(write_ptr, write_info_ptr,
2328
PNG_DESTROY_WILL_FREE_DATA,
2329
PNG_FREE_PLTE|PNG_FREE_tRNS|PNG_FREE_hIST)
2331
thereby briefly reassigning responsibility for freeing to the user but
2332
immediately afterwards reassigning it once more to the write_destroy
2333
function. Having done this, it would then be safe to destroy the read
2334
structure and continue to use the PLTE, tRNS, and hIST data in the write
2337
This function only affects data that has already been allocated.
2338
You can call this function before calling after the png_set_*() functions
2339
to control whether the user or png_destroy_*() is supposed to free the data.
2340
When the user assumes responsibility for libpng-allocated data, the
2341
application must use
2342
png_free() to free it, and when the user transfers responsibility to libpng
2343
for data that the user has allocated, the user must have used png_malloc()
2344
or png_zalloc() to allocate it.
2346
If you allocated text_ptr.text, text_ptr.lang, and text_ptr.translated_keyword
2347
separately, do not transfer responsibility for freeing text_ptr to libpng,
2348
because when libpng fills a png_text structure it combines these members with
2349
the key member, and png_free_data() will free only text_ptr.key. Similarly,
2350
if you transfer responsibility for free'ing text_ptr from libpng to your
2351
application, your application must not separately free those members.
2352
For a more compact example of writing a PNG image, see the file example.c.
2354
V. Modifying/Customizing libpng:
2356
There are two issues here. The first is changing how libpng does
2357
standard things like memory allocation, input/output, and error handling.
2358
The second deals with more complicated things like adding new chunks,
2359
adding new transformations, and generally changing how libpng works.
2360
Both of those are compile-time issues; that is, they are generally
2361
determined at the time the code is written, and there is rarely a need
2362
to provide the user with a means of changing them.
2364
Memory allocation, input/output, and error handling
2366
All of the memory allocation, input/output, and error handling in libpng
2367
goes through callbacks that are user-settable. The default routines are
2368
in pngmem.c, pngrio.c, pngwio.c, and pngerror.c, respectively. To change
2369
these functions, call the appropriate png_set_*_fn() function.
2371
Memory allocation is done through the functions png_malloc()
2372
and png_free(). These currently just call the standard C functions. If
2373
your pointers can't access more then 64K at a time, you will want to set
2374
MAXSEG_64K in zlib.h. Since it is unlikely that the method of handling
2375
memory allocation on a platform will change between applications, these
2376
functions must be modified in the library at compile time. If you prefer
2377
to use a different method of allocating and freeing data, you can use
2378
png_create_read_struct_2() or png_create_write_struct_2() to register
2379
your own functions as described above.
2380
These functions also provide a void pointer that can be retrieved via
2382
mem_ptr=png_get_mem_ptr(png_ptr);
2384
Your replacement memory functions must have prototypes as follows:
2386
png_voidp malloc_fn(png_structp png_ptr,
2388
void free_fn(png_structp png_ptr, png_voidp ptr);
2390
Your malloc_fn() must return NULL in case of failure. The png_malloc()
2391
function will normally call png_error() if it receives a NULL from the
2392
system memory allocator or from your replacement malloc_fn().
2394
Input/Output in libpng is done through png_read() and png_write(),
2395
which currently just call fread() and fwrite(). The FILE * is stored in
2396
png_struct and is initialized via png_init_io(). If you wish to change
2397
the method of I/O, the library supplies callbacks that you can set
2398
through the function png_set_read_fn() and png_set_write_fn() at run
2399
time, instead of calling the png_init_io() function. These functions
2400
also provide a void pointer that can be retrieved via the function
2401
png_get_io_ptr(). For example:
2403
png_set_read_fn(png_structp read_ptr,
2404
voidp read_io_ptr, png_rw_ptr read_data_fn)
2406
png_set_write_fn(png_structp write_ptr,
2407
voidp write_io_ptr, png_rw_ptr write_data_fn,
2408
png_flush_ptr output_flush_fn);
2410
voidp read_io_ptr = png_get_io_ptr(read_ptr);
2411
voidp write_io_ptr = png_get_io_ptr(write_ptr);
2413
The replacement I/O functions must have prototypes as follows:
2415
void user_read_data(png_structp png_ptr,
2416
png_bytep data, png_size_t length);
2417
void user_write_data(png_structp png_ptr,
2418
png_bytep data, png_size_t length);
2419
void user_flush_data(png_structp png_ptr);
2421
Supplying NULL for the read, write, or flush functions sets them back
2422
to using the default C stream functions. It is an error to read from
2423
a write stream, and vice versa.
2425
Error handling in libpng is done through png_error() and png_warning().
2426
Errors handled through png_error() are fatal, meaning that png_error()
2427
should never return to its caller. Currently, this is handled via
2428
setjmp() and longjmp() (unless you have compiled libpng with
2429
PNG_SETJMP_NOT_SUPPORTED, in which case it is handled via PNG_ABORT()),
2430
but you could change this to do things like exit() if you should wish.
2432
On non-fatal errors, png_warning() is called
2433
to print a warning message, and then control returns to the calling code.
2434
By default png_error() and png_warning() print a message on stderr via
2435
fprintf() unless the library is compiled with PNG_NO_CONSOLE_IO defined
2436
(because you don't want the messages) or PNG_NO_STDIO defined (because
2437
fprintf() isn't available). If you wish to change the behavior of the error
2438
functions, you will need to set up your own message callbacks. These
2439
functions are normally supplied at the time that the png_struct is created.
2440
It is also possible to redirect errors and warnings to your own replacement
2441
functions after png_create_*_struct() has been called by calling:
2443
png_set_error_fn(png_structp png_ptr,
2444
png_voidp error_ptr, png_error_ptr error_fn,
2445
png_error_ptr warning_fn);
2447
png_voidp error_ptr = png_get_error_ptr(png_ptr);
2449
If NULL is supplied for either error_fn or warning_fn, then the libpng
2450
default function will be used, calling fprintf() and/or longjmp() if a
2451
problem is encountered. The replacement error functions should have
2452
parameters as follows:
2454
void user_error_fn(png_structp png_ptr,
2455
png_const_charp error_msg);
2456
void user_warning_fn(png_structp png_ptr,
2457
png_const_charp warning_msg);
2459
The motivation behind using setjmp() and longjmp() is the C++ throw and
2460
catch exception handling methods. This makes the code much easier to write,
2461
as there is no need to check every return code of every function call.
2462
However, there are some uncertainties about the status of local variables
2463
after a longjmp, so the user may want to be careful about doing anything after
2464
setjmp returns non-zero besides returning itself. Consult your compiler
2465
documentation for more details. For an alternative approach, you may wish
2466
to use the "cexcept" facility (see http://cexcept.sourceforge.net).
2470
If you need to read or write custom chunks, you may need to get deeper
2471
into the libpng code. The library now has mechanisms for storing
2472
and writing chunks of unknown type; you can even declare callbacks
2473
for custom chunks. However, this may not be good enough if the
2474
library code itself needs to know about interactions between your
2475
chunk and existing `intrinsic' chunks.
2477
If you need to write a new intrinsic chunk, first read the PNG
2478
specification. Acquire a first level of
2479
understanding of how it works. Pay particular attention to the
2480
sections that describe chunk names, and look at how other chunks were
2481
designed, so you can do things similarly. Second, check out the
2482
sections of libpng that read and write chunks. Try to find a chunk
2483
that is similar to yours and use it as a template. More details can
2484
be found in the comments inside the code. It is best to handle unknown
2485
chunks in a generic method, via callback functions, instead of by
2486
modifying libpng functions.
2488
If you wish to write your own transformation for the data, look through
2489
the part of the code that does the transformations, and check out some of
2490
the simpler ones to get an idea of how they work. Try to find a similar
2491
transformation to the one you want to add and copy off of it. More details
2492
can be found in the comments inside the code itself.
2494
Configuring for 16 bit platforms
2496
You will want to look into zconf.h to tell zlib (and thus libpng) that
2497
it cannot allocate more then 64K at a time. Even if you can, the memory
2498
won't be accessible. So limit zlib and libpng to 64K by defining MAXSEG_64K.
2502
For DOS users who only have access to the lower 640K, you will
2503
have to limit zlib's memory usage via a png_set_compression_mem_level()
2504
call. See zlib.h or zconf.h in the zlib library for more information.
2506
Configuring for Medium Model
2508
Libpng's support for medium model has been tested on most of the popular
2509
compilers. Make sure MAXSEG_64K gets defined, USE_FAR_KEYWORD gets
2510
defined, and FAR gets defined to far in pngconf.h, and you should be
2511
all set. Everything in the library (except for zlib's structure) is
2512
expecting far data. You must use the typedefs with the p or pp on
2513
the end for pointers (or at least look at them and be careful). Make
2514
note that the rows of data are defined as png_bytepp, which is an
2515
unsigned char far * far *.
2517
Configuring for gui/windowing platforms:
2519
You will need to write new error and warning functions that use the GUI
2520
interface, as described previously, and set them to be the error and
2521
warning functions at the time that png_create_*_struct() is called,
2522
in order to have them available during the structure initialization.
2523
They can be changed later via png_set_error_fn(). On some compilers,
2524
you may also have to change the memory allocators (png_malloc, etc.).
2526
Configuring for compiler xxx:
2528
All includes for libpng are in pngconf.h. If you need to add/change/delete
2529
an include, this is the place to do it. The includes that are not
2530
needed outside libpng are protected by the PNG_INTERNAL definition,
2531
which is only defined for those routines inside libpng itself. The
2532
files in libpng proper only include png.h, which includes pngconf.h.
2536
There are special functions to configure the compression. Perhaps the
2537
most useful one changes the compression level, which currently uses
2538
input compression values in the range 0 - 9. The library normally
2539
uses the default compression level (Z_DEFAULT_COMPRESSION = 6). Tests
2540
have shown that for a large majority of images, compression values in
2541
the range 3-6 compress nearly as well as higher levels, and do so much
2542
faster. For online applications it may be desirable to have maximum speed
2543
(Z_BEST_SPEED = 1). With versions of zlib after v0.99, you can also
2544
specify no compression (Z_NO_COMPRESSION = 0), but this would create
2545
files larger than just storing the raw bitmap. You can specify the
2546
compression level by calling:
2548
png_set_compression_level(png_ptr, level);
2550
Another useful one is to reduce the memory level used by the library.
2551
The memory level defaults to 8, but it can be lowered if you are
2552
short on memory (running DOS, for example, where you only have 640K).
2553
Note that the memory level does have an effect on compression; among
2554
other things, lower levels will result in sections of incompressible
2555
data being emitted in smaller stored blocks, with a correspondingly
2556
larger relative overhead of up to 15% in the worst case.
2558
png_set_compression_mem_level(png_ptr, level);
2560
The other functions are for configuring zlib. They are not recommended
2561
for normal use and may result in writing an invalid PNG file. See
2562
zlib.h for more information on what these mean.
2564
png_set_compression_strategy(png_ptr,
2566
png_set_compression_window_bits(png_ptr,
2568
png_set_compression_method(png_ptr, method);
2569
png_set_compression_buffer_size(png_ptr, size);
2571
Controlling row filtering
2573
If you want to control whether libpng uses filtering or not, which
2574
filters are used, and how it goes about picking row filters, you
2575
can call one of these functions. The selection and configuration
2576
of row filters can have a significant impact on the size and
2577
encoding speed and a somewhat lesser impact on the decoding speed
2578
of an image. Filtering is enabled by default for RGB and grayscale
2579
images (with and without alpha), but not for paletted images nor
2580
for any images with bit depths less than 8 bits/pixel.
2582
The 'method' parameter sets the main filtering method, which is
2583
currently only '0' in the PNG 1.2 specification. The 'filters'
2584
parameter sets which filter(s), if any, should be used for each
2585
scanline. Possible values are PNG_ALL_FILTERS and PNG_NO_FILTERS
2586
to turn filtering on and off, respectively.
2588
Individual filter types are PNG_FILTER_NONE, PNG_FILTER_SUB,
2589
PNG_FILTER_UP, PNG_FILTER_AVG, PNG_FILTER_PAETH, which can be bitwise
2590
ORed together with '|' to specify one or more filters to use.
2591
These filters are described in more detail in the PNG specification.
2592
If you intend to change the filter type during the course of writing
2593
the image, you should start with flags set for all of the filters
2594
you intend to use so that libpng can initialize its internal
2595
structures appropriately for all of the filter types. (Note that this
2596
means the first row must always be adaptively filtered, because libpng
2597
currently does not allocate the filter buffers until png_write_row()
2598
is called for the first time.)
2600
filters = PNG_FILTER_NONE | PNG_FILTER_SUB
2601
PNG_FILTER_UP | PNG_FILTER_AVE |
2602
PNG_FILTER_PAETH | PNG_ALL_FILTERS;
2604
png_set_filter(png_ptr, PNG_FILTER_TYPE_BASE,
2606
The second parameter can also be
2607
PNG_INTRAPIXEL_DIFFERENCING if you are
2608
writing a PNG to be embedded in a MNG
2609
datastream. This parameter must be the
2610
same as the value of filter_method used
2613
It is also possible to influence how libpng chooses from among the
2614
available filters. This is done in one or both of two ways - by
2615
telling it how important it is to keep the same filter for successive
2616
rows, and by telling it the relative computational costs of the filters.
2618
double weights[3] = {1.5, 1.3, 1.1},
2619
costs[PNG_FILTER_VALUE_LAST] =
2620
{1.0, 1.3, 1.3, 1.5, 1.7};
2622
png_set_filter_heuristics(png_ptr,
2623
PNG_FILTER_HEURISTIC_WEIGHTED, 3,
2626
The weights are multiplying factors that indicate to libpng that the
2627
row filter should be the same for successive rows unless another row filter
2628
is that many times better than the previous filter. In the above example,
2629
if the previous 3 filters were SUB, SUB, NONE, the SUB filter could have a
2630
"sum of absolute differences" 1.5 x 1.3 times higher than other filters
2631
and still be chosen, while the NONE filter could have a sum 1.1 times
2632
higher than other filters and still be chosen. Unspecified weights are
2633
taken to be 1.0, and the specified weights should probably be declining
2634
like those above in order to emphasize recent filters over older filters.
2636
The filter costs specify for each filter type a relative decoding cost
2637
to be considered when selecting row filters. This means that filters
2638
with higher costs are less likely to be chosen over filters with lower
2639
costs, unless their "sum of absolute differences" is that much smaller.
2640
The costs do not necessarily reflect the exact computational speeds of
2641
the various filters, since this would unduly influence the final image
2644
Note that the numbers above were invented purely for this example and
2645
are given only to help explain the function usage. Little testing has
2646
been done to find optimum values for either the costs or the weights.
2648
Removing unwanted object code
2650
There are a bunch of #define's in pngconf.h that control what parts of
2651
libpng are compiled. All the defines end in _SUPPORTED. If you are
2652
never going to use a capability, you can change the #define to #undef
2653
before recompiling libpng and save yourself code and data space, or
2654
you can turn off individual capabilities with defines that begin with
2657
You can also turn all of the transforms and ancillary chunk capabilities
2658
off en masse with compiler directives that define
2659
PNG_NO_READ[or WRITE]_TRANSFORMS, or PNG_NO_READ[or WRITE]_ANCILLARY_CHUNKS,
2661
along with directives to turn on any of the capabilities that you do
2662
want. The PNG_NO_READ[or WRITE]_TRANSFORMS directives disable
2663
the extra transformations but still leave the library fully capable of reading
2664
and writing PNG files with all known public chunks
2665
Use of the PNG_NO_READ[or WRITE]_ANCILLARY_CHUNKS directive
2666
produces a library that is incapable of reading or writing ancillary chunks.
2667
If you are not using the progressive reading capability, you can
2668
turn that off with PNG_NO_PROGRESSIVE_READ (don't confuse
2669
this with the INTERLACING capability, which you'll still have).
2671
All the reading and writing specific code are in separate files, so the
2672
linker should only grab the files it needs. However, if you want to
2673
make sure, or if you are building a stand alone library, all the
2674
reading files start with pngr and all the writing files start with
2675
pngw. The files that don't match either (like png.c, pngtrans.c, etc.)
2676
are used for both reading and writing, and always need to be included.
2677
The progressive reader is in pngpread.c
2679
If you are creating or distributing a dynamically linked library (a .so
2680
or DLL file), you should not remove or disable any parts of the library,
2681
as this will cause applications linked with different versions of the
2682
library to fail if they call functions not available in your library.
2683
The size of the library itself should not be an issue, because only
2684
those sections that are actually used will be loaded into memory.
2686
Requesting debug printout
2688
The macro definition PNG_DEBUG can be used to request debugging
2689
printout. Set it to an integer value in the range 0 to 3. Higher
2690
numbers result in increasing amounts of debugging information. The
2691
information is printed to the "stderr" file, unless another file
2692
name is specified in the PNG_DEBUG_FILE macro definition.
2694
When PNG_DEBUG > 0, the following functions (macros) become available:
2696
png_debug(level, message)
2697
png_debug1(level, message, p1)
2698
png_debug2(level, message, p1, p2)
2700
in which "level" is compared to PNG_DEBUG to decide whether to print
2701
the message, "message" is the formatted string to be printed,
2702
and p1 and p2 are parameters that are to be embedded in the string
2703
according to printf-style formatting directives. For example,
2705
png_debug1(2, "foo=%d\n", foo);
2710
fprintf(PNG_DEBUG_FILE, "foo=%d\n", foo);
2712
When PNG_DEBUG is defined but is zero, the macros aren't defined, but you
2713
can still use PNG_DEBUG to control your own debugging:
2719
When PNG_DEBUG = 1, the macros are defined, but only png_debug statements
2720
having level = 0 will be printed. There aren't any such statements in
2721
this version of libpng, but if you insert some they will be printed.
2725
The MNG specification (available at http://www.libpng.org/pub/mng) allows
2726
certain extensions to PNG for PNG images that are embedded in MNG datastreams.
2727
Libpng can support some of these extensions. To enable them, use the
2728
png_permit_mng_features() function:
2730
feature_set = png_permit_mng_features(png_ptr, mask)
2731
mask is a png_uint_32 containing the bitwise OR of the
2732
features you want to enable. These include
2733
PNG_FLAG_MNG_EMPTY_PLTE
2734
PNG_FLAG_MNG_FILTER_64
2735
PNG_ALL_MNG_FEATURES
2736
feature_set is a png_uint_32 that is the bitwise AND of
2737
your mask with the set of MNG features that is
2738
supported by the version of libpng that you are using.
2740
It is an error to use this function when reading or writing a standalone
2741
PNG file with the PNG 8-byte signature. The PNG datastream must be wrapped
2742
in a MNG datastream. As a minimum, it must have the MNG 8-byte signature
2743
and the MHDR and MEND chunks. Libpng does not provide support for these
2744
or any other MNG chunks; your application must provide its own support for
2745
them. You may wish to consider using libmng (available at
2746
http://www.libmng.com) instead.
2748
VIII. Changes to Libpng from version 0.88
2750
It should be noted that versions of libpng later than 0.96 are not
2751
distributed by the original libpng author, Guy Schalnat, nor by
2752
Andreas Dilger, who had taken over from Guy during 1996 and 1997, and
2753
distributed versions 0.89 through 0.96, but rather by another member
2754
of the original PNG Group, Glenn Randers-Pehrson. Guy and Andreas are
2755
still alive and well, but they have moved on to other things.
2757
The old libpng functions png_read_init(), png_write_init(),
2758
png_info_init(), png_read_destroy(), and png_write_destroy() have been
2759
moved to PNG_INTERNAL in version 0.95 to discourage their use. These
2760
functions will be removed from libpng version 2.0.0.
2762
The preferred method of creating and initializing the libpng structures is
2763
via the png_create_read_struct(), png_create_write_struct(), and
2764
png_create_info_struct() because they isolate the size of the structures
2765
from the application, allow version error checking, and also allow the
2766
use of custom error handling routines during the initialization, which
2767
the old functions do not. The functions png_read_destroy() and
2768
png_write_destroy() do not actually free the memory that libpng
2769
allocated for these structs, but just reset the data structures, so they
2770
can be used instead of png_destroy_read_struct() and
2771
png_destroy_write_struct() if you feel there is too much system overhead
2772
allocating and freeing the png_struct for each image read.
2774
Setting the error callbacks via png_set_message_fn() before
2775
png_read_init() as was suggested in libpng-0.88 is no longer supported
2776
because this caused applications that do not use custom error functions
2777
to fail if the png_ptr was not initialized to zero. It is still possible
2778
to set the error callbacks AFTER png_read_init(), or to change them with
2779
png_set_error_fn(), which is essentially the same function, but with a new
2780
name to force compilation errors with applications that try to use the old
2783
Starting with version 1.0.7, you can find out which version of the library
2784
you are using at run-time:
2786
png_uint_32 libpng_vn = png_access_version_number();
2788
The number libpng_vn is constructed from the major version, minor
2789
version with leading zero, and release number with leading zero,
2790
(e.g., libpng_vn for version 1.0.7 is 10007).
2792
You can also check which version of png.h you used when compiling your
2795
png_uint_32 application_vn = PNG_LIBPNG_VER;
2797
IX. Y2K Compliance in libpng
2801
Since the PNG Development group is an ad-hoc body, we can't make
2802
an official declaration.
2804
This is your unofficial assurance that libpng from version 0.71 and
2805
upward through 1.2.22 are Y2K compliant. It is my belief that earlier
2806
versions were also Y2K compliant.
2808
Libpng only has three year fields. One is a 2-byte unsigned integer that
2809
will hold years up to 65535. The other two hold the date in text
2810
format, and will hold years up to 9999.
2813
"png_uint_16 year" in png_time_struct.
2816
"png_charp time_buffer" in png_struct and
2817
"near_time_buffer", which is a local character string in png.c.
2819
There are seven time-related functions:
2821
png_convert_to_rfc_1123() in png.c
2822
(formerly png_convert_to_rfc_1152() in error)
2823
png_convert_from_struct_tm() in pngwrite.c, called
2825
png_convert_from_time_t() in pngwrite.c
2826
png_get_tIME() in pngget.c
2827
png_handle_tIME() in pngrutil.c, called in pngread.c
2828
png_set_tIME() in pngset.c
2829
png_write_tIME() in pngwutil.c, called in pngwrite.c
2831
All appear to handle dates properly in a Y2K environment. The
2832
png_convert_from_time_t() function calls gmtime() to convert from system
2833
clock time, which returns (year - 1900), which we properly convert to
2834
the full 4-digit year. There is a possibility that applications using
2835
libpng are not passing 4-digit years into the png_convert_to_rfc_1123()
2836
function, or that they are incorrectly passing only a 2-digit year
2837
instead of "year - 1900" into the png_convert_from_struct_tm() function,
2838
but this is not under our control. The libpng documentation has always
2839
stated that it works with 4-digit years, and the APIs have been
2842
The tIME chunk itself is also Y2K compliant. It uses a 2-byte unsigned
2843
integer to hold the year, and can hold years as large as 65535.
2845
zlib, upon which libpng depends, is also Y2K compliant. It contains
2846
no date-related code.
2849
Glenn Randers-Pehrson
2851
PNG Development Group