Types¶

The png.h header file defines a number of integral types used by the APIs. Most of these are fairly obvious; for example types corresponding to integers of particular sizes and types for passing color values.

One exception is how non-integral numbers are handled. For application convenience most APIs that take such numbers have C (double) arguments; however, internally PNG, and libpng, use 32 bit signed integers and encode the value by multiplying by 100,000. As of libpng 1.5.0 a convenience macro PNG_FP_1 is defined in png.h along with a type (png_fixed_point) which is simply (png_int_32).

All APIs that take (double) arguments also have a matching API that takes the corresponding fixed point integer arguments. The fixed point API has the same name as the floating point one with "_fixed" appended. The actual range of values permitted in the APIs is frequently less than the full range of (png_fixed_point) (-21474 to +21474). When APIs require a non-negative argument the type is recorded as png_uint_32 above. Consult the header file and the text below for more information.

Special care must be take with sCAL chunk handling because the chunk itself uses non-integral values encoded as strings containing decimal floating point numbers. See the comments in the header file.

Configuration¶

The main header file function declarations are frequently protected by C preprocessing directives of the form:

#ifdef PNG_feature_SUPPORTED
declare-function
#endif
...
#ifdef PNG_feature_SUPPORTED
use-function
#endif

The library can be built without support for these APIs, although a standard build will have all implemented APIs. Application programs should check the feature macros before using an API for maximum portability. From libpng 1.5.0 the feature macros set during the build of libpng are recorded in the header file "pnglibconf.h" and this file is always included by png.h.

If you don't need to change the library configuration from the default, skip to the next section ("Reading").

Notice that some of the makefiles in the 'scripts' directory and (in 1.5.0) all of the build project files in the 'projects' directory simply copy scripts/pnglibconf.h.prebuilt to pnglibconf.h. This means that these build systems do not permit easy auto-configuration of the library - they only support the default configuration.

The easiest way to make minor changes to the libpng configuration when auto-configuration is supported is to add definitions to the command line using (typically) CPPFLAGS. For example:

CPPFLAGS=-DPNG_NO_FLOATING_ARITHMETIC

will change the internal libpng math implementation for gamma correction and other arithmetic calculations to fixed point, avoiding the need for fast floating point support. The result can be seen in the generated pnglibconf.h - make sure it contains the changed feature macro setting.

If you need to make more extensive configuration changes - more than one or two feature macro settings - you can either add -DPNG_USER_CONFIG to the build command line and put a list of feature macro settings in pngusr.h or you can set DFA_XTRA (a makefile variable) to a file containing the same information in the form of 'option' settings.

A. Changing pnglibconf.h

A variety of methods exist to build libpng. Not all of these support reconfiguration of pnglibconf.h. To reconfigure pnglibconf.h it must either be rebuilt from scripts/pnglibconf.dfa using awk or it must be edited by hand.

Hand editing is achieved by copying scripts/pnglibconf.h.prebuilt to pnglibconf.h and changing the lines defining the supported features, paying very close attention to the 'option' information in scripts/pnglibconf.dfa that describes those features and their requirements. This is easy to get wrong.

B. Configuration using DFA_XTRA

Rebuilding from pnglibconf.dfa is easy if a functioning 'awk', or a later variant such as 'nawk' or 'gawk', is available. The configure build will automatically find an appropriate awk and build pnglibconf.h. The scripts/pnglibconf.mak file contains a set of make rules for doing the same thing if configure is not used, and many of the makefiles in the scripts directory use this approach.

When rebuilding simply write a new file containing changed options and set DFA_XTRA to the name of this file. This causes the build to append the new file to the end of scripts/pnglibconf.dfa. The pngusr.dfa file should contain lines of the following forms:

everything = off

This turns all optional features off. Include it at the start of pngusr.dfa to make it easier to build a minimal configuration. You will need to turn at least some features on afterward to enable either reading or writing code, or both.

option feature on option feature off

Enable or disable a single feature. This will automatically enable other features required by a feature that is turned on or disable other features that require a feature which is turned off. Conflicting settings will cause an error message to be emitted by awk.

setting feature default value

Changes the default value of setting 'feature' to 'value'. There are a small number of settings listed at the top of pnglibconf.h, they are documented in the source code. Most of these values have performance implications for the library but most of them have no visible effect on the API. Some can also be overridden from the API.

This method of building a customized pnglibconf.h is illustrated in contrib/pngminim/*. See the "$(PNGCONF):" target in the makefile and pngusr.dfa in these directories.

C. Configuration using PNG_USR_CONFIG

If -DPNG_USR_CONFIG is added to the CFLAGS when pnglibconf.h is built the file pngusr.h will automatically be included before the options in scripts/pnglibconf.dfa are processed. Your pngusr.h file should contain only macro definitions turning features on or off or setting settings.

Apart from the global setting "everything = off" all the options listed above can be set using macros in pngusr.h:

#define PNG_feature_SUPPORTED

is equivalent to:

option feature on

#define PNG_NO_feature

is equivalent to:

option feature off

#define PNG_feature value

is equivalent to:

setting feature default value

Notice that in both cases, pngusr.dfa and pngusr.h, the contents of the pngusr file you supply override the contents of scripts/pnglibconf.dfa

If confusing or incomprehensible behavior results it is possible to examine the intermediate file pnglibconf.dfn to find the full set of dependency information for each setting and option. Simply locate the feature in the file and read the C comments that precede it.

This method is also illustrated in the contrib/pngminim/* makefiles and pngusr.h.

Setup¶

You will want to do the I/O initialization(*) before you get into libpng, so if it doesn't work, you don't have much to undo. Of course, you will also want to insure that you are, in fact, dealing with a PNG file. Libpng provides a simple check to see if a file is a PNG file. To use it, pass in the first 1 to 8 bytes of the file to the function png_sig_cmp(), and it will return 0 (false) if the bytes match the corresponding bytes of the PNG signature, or nonzero (true) otherwise. Of course, the more bytes you pass in, the greater the accuracy of the prediction.

If you are intending to keep the file pointer open for use in libpng, you must ensure you don't read more than 8 bytes from the beginning of the file, and you also have to make a call to png_set_sig_bytes_read() with the number of bytes you read from the beginning. Libpng will then only check the bytes (if any) that your program didn't read.

(*): If you are not using the standard I/O functions, you will need to replace them with custom functions. See the discussion under Customizing libpng.

FILE *fp = fopen(file_name, "rb");
if (!fp)
{
return (ERROR);
}

fread(header, 1, number, fp);
is_png = !png_sig_cmp(header, 0, number);

if (!is_png)
{
return (NOT_PNG);
}

Next, png_struct and png_info need to be allocated and initialized. In order to ensure that the size of these structures is correct even with a dynamically linked libpng, there are functions to initialize and allocate the structures. We also pass the library version, optional pointers to error handling functions, and a pointer to a data struct for use by the error functions, if necessary (the pointer and functions can be NULL if the default error handlers are to be used). See the section on Changes to Libpng below regarding the old initialization functions. The structure allocation functions quietly return NULL if they fail to create the structure, so your application should check for that.

png_structp png_ptr = png_create_read_struct
(PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
user_error_fn, user_warning_fn);

if (!png_ptr)
return (ERROR);

png_infop info_ptr = png_create_info_struct(png_ptr);

if (!info_ptr)
{
png_destroy_read_struct(&png_ptr,
(png_infopp)NULL, (png_infopp)NULL);
return (ERROR);
}

If you want to use your own memory allocation routines, use a libpng that was built with PNG_USER_MEM_SUPPORTED defined, and use png_create_read_struct_2() instead of png_create_read_struct():

png_structp png_ptr = png_create_read_struct_2
(PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
user_error_fn, user_warning_fn, (png_voidp)
user_mem_ptr, user_malloc_fn, user_free_fn);

The error handling routines passed to png_create_read_struct() and the memory alloc/free routines passed to png_create_struct_2() are only necessary if you are not using the libpng supplied error handling and memory alloc/free functions.

When libpng encounters an error, it expects to longjmp back to your routine. Therefore, you will need to call setjmp and pass your png_jmpbuf(png_ptr). If you read the file from different routines, you will need to update the longjmp buffer every time you enter a new routine that will call a png_*() function.

See your documentation of setjmp/longjmp for your compiler for more information on setjmp/longjmp. See the discussion on libpng error handling in the Customizing Libpng section below for more information on the libpng error handling. If an error occurs, and libpng longjmp's back to your setjmp, you will want to call png_destroy_read_struct() to free any memory.

if (setjmp(png_jmpbuf(png_ptr)))
{
png_destroy_read_struct(&png_ptr, &info_ptr,
&end_info);
fclose(fp);
return (ERROR);
}

Pass (png_infopp)NULL instead of &end_info if you didn't create an end_info structure.

If you would rather avoid the complexity of setjmp/longjmp issues, you can compile libpng with PNG_NO_SETJMP, in which case errors will result in a call to PNG_ABORT() which defaults to abort().

You can #define PNG_ABORT() to a function that does something more useful than abort(), as long as your function does not return.

Now you need to set up the input code. The default for libpng is to use the C function fread(). If you use this, you will need to pass a valid FILE * in the function png_init_io(). Be sure that the file is opened in binary mode. If you wish to handle reading data in another way, you need not call the png_init_io() function, but you must then implement the libpng I/O methods discussed in the Customizing Libpng section below.

png_init_io(png_ptr, fp);

If you had previously opened the file and read any of the signature from the beginning in order to see if this was a PNG file, you need to let libpng know that there are some bytes missing from the start of the file.

png_set_sig_bytes(png_ptr, number);

You can change the zlib compression buffer size to be used while reading compressed data with

png_set_compression_buffer_size(png_ptr, buffer_size);

where the default size is 8192 bytes. Note that the buffer size is changed immediately and the buffer is reallocated immediately, instead of setting a flag to be acted upon later.

If you want CRC errors to be handled in a different manner than the default, use

png_set_crc_action(png_ptr, crit_action, ancil_action);

The values for png_set_crc_action() say how libpng is to handle CRC errors in ancillary and critical chunks, and whether to use the data contained therein. Note that it is impossible to "discard" data in a critical chunk.

Choices for (int) crit_action are
PNG_CRC_DEFAULT 0 error/quit
PNG_CRC_ERROR_QUIT 1 error/quit
PNG_CRC_WARN_USE 3 warn/use data
PNG_CRC_QUIET_USE 4 quiet/use data
PNG_CRC_NO_CHANGE 5 use the current value

Choices for (int) ancil_action are
PNG_CRC_DEFAULT 0 error/quit
PNG_CRC_ERROR_QUIT 1 error/quit
PNG_CRC_WARN_DISCARD 2 warn/discard data
PNG_CRC_WARN_USE 3 warn/use data
PNG_CRC_QUIET_USE 4 quiet/use data
PNG_CRC_NO_CHANGE 5 use the current value

Setting up callback code¶

You can set up a callback function to handle any unknown chunks in the input stream. You must supply the function

read_chunk_callback(png_structp png_ptr,
png_unknown_chunkp chunk);
{
/* The unknown chunk structure contains your
chunk data, along with similar data for any other
unknown chunks: */

png_byte name[5];
png_byte *data;
png_size_t size;

/* Note that libpng has already taken care of
the CRC handling */

/* put your code here. Search for your chunk in the
unknown chunk structure, process it, and return one
of the following: */

return (-n); /* chunk had an error */
return (0); /* did not recognize */
return (n); /* success */
}

(You can give your function another name that you like instead of "read_chunk_callback")

To inform libpng about your function, use

png_set_read_user_chunk_fn(png_ptr, user_chunk_ptr,
read_chunk_callback);

This names not only the callback function, but also a user pointer that you can retrieve with

png_get_user_chunk_ptr(png_ptr);

If you call the png_set_read_user_chunk_fn() function, then all unknown chunks will be saved when read, in case your callback function will need one or more of them. This behavior can be changed with the png_set_keep_unknown_chunks() function, described below.

At this point, you can set up a callback function that will be called after each row has been read, which you can use to control a progress meter or the like. It's demonstrated in pngtest.c. You must supply a function

void read_row_callback(png_structp png_ptr,
png_uint_32 row, int pass);
{
/* put your code here */
}

(You can give it another name that you like instead of "read_row_callback")

To inform libpng about your function, use

png_set_read_status_fn(png_ptr, read_row_callback);

When this function is called the row has already been completely processed and the 'row' and 'pass' refer to the next row to be handled. For the non-interlaced case the row that was just handled is simply one less than the passed in row number, and pass will always be 0. For the interlaced case the same applies unless the row value is 0, in which case the row just handled was the last one from one of the preceding passes. Because interlacing may skip a pass you cannot be sure that the preceding pass is just 'pass-1', if you really need to know what the last pass is record (row,pass) from the callback and use the last recorded value each time.

As with the user transform you can find the output row using the PNG_ROW_FROM_PASS_ROW macro.

Unknown-chunk handling¶

Now you get to set the way the library processes unknown chunks in the input PNG stream. Both known and unknown chunks will be read. Normal behavior is that known chunks will be parsed into information in various info_ptr members while unknown chunks will be discarded. This behavior can be wasteful if your application will never use some known chunk types. To change this, you can call:

png_set_keep_unknown_chunks(png_ptr, keep,
chunk_list, num_chunks);
keep - 0: default unknown chunk handling
1: ignore; do not keep
2: keep only if safe-to-copy
3: keep even if unsafe-to-copy

You can use these definitions:
PNG_HANDLE_CHUNK_AS_DEFAULT 0
PNG_HANDLE_CHUNK_NEVER 1
PNG_HANDLE_CHUNK_IF_SAFE 2
PNG_HANDLE_CHUNK_ALWAYS 3

chunk_list - list of chunks affected (a byte string,
five bytes per chunk, NULL or ' ' if
num_chunks is 0)

num_chunks - number of chunks affected; if 0, all
unknown chunks are affected. If nonzero,
only the chunks in the list are affected

Unknown chunks declared in this way will be saved as raw data onto a list of png_unknown_chunk structures. If a chunk that is normally known to libpng is named in the list, it will be handled as unknown, according to the "keep" directive. If a chunk is named in successive instances of png_set_keep_unknown_chunks(), the final instance will take precedence. The IHDR and IEND chunks should not be named in chunk_list; if they are, libpng will process them normally anyway. If you know that your application will never make use of some particular chunks, use PNG_HANDLE_CHUNK_NEVER (or 1) as demonstrated below.

Here is an example of the usage of png_set_keep_unknown_chunks(), where the private "vpAg" chunk will later be processed by a user chunk callback function:

png_byte vpAg[5]={118, 112, 65, 103, (png_byte) ' '};

#if defined(PNG_UNKNOWN_CHUNKS_SUPPORTED)
png_byte unused_chunks[]=
{
104, 73, 83, 84, (png_byte) ' ', /* hIST */
105, 84, 88, 116, (png_byte) ' ', /* iTXt */
112, 67, 65, 76, (png_byte) ' ', /* pCAL */
115, 67, 65, 76, (png_byte) ' ', /* sCAL */
115, 80, 76, 84, (png_byte) ' ', /* sPLT */
116, 73, 77, 69, (png_byte) ' ', /* tIME */
};
#endif

...

#if defined(PNG_UNKNOWN_CHUNKS_SUPPORTED)
/* ignore all unknown chunks: */
png_set_keep_unknown_chunks(read_ptr, 1, NULL, 0);

/* except for vpAg: */
png_set_keep_unknown_chunks(read_ptr, 2, vpAg, 1);

/* also ignore unused known chunks: */
png_set_keep_unknown_chunks(read_ptr, 1, unused_chunks,
(int)sizeof(unused_chunks)/5);
#endif

User limits¶

The PNG specification allows the width and height of an image to be as large as 2^31-1 (0x7fffffff), or about 2.147 billion rows and columns. Since very few applications really need to process such large images, we have imposed an arbitrary 1-million limit on rows and columns. Larger images will be rejected immediately with a png_error() call. If you wish to change this limit, you can use

png_set_user_limits(png_ptr, width_max, height_max);

to set your own limits, or use width_max = height_max = 0x7fffffffL to allow all valid dimensions (libpng may reject some very large images anyway because of potential buffer overflow conditions).

You should put this statement after you create the PNG structure and before calling png_read_info(), png_read_png(), or png_process_data().

When writing a PNG datastream, put this statement before calling png_write_info() or png_write_png().

If you need to retrieve the limits that are being applied, use

width_max = png_get_user_width_max(png_ptr);
height_max = png_get_user_height_max(png_ptr);

The PNG specification sets no limit on the number of ancillary chunks allowed in a PNG datastream. You can impose a limit on the total number of sPLT, tEXt, iTXt, zTXt, and unknown chunks that will be stored, with

png_set_chunk_cache_max(png_ptr, user_chunk_cache_max);

where 0x7fffffffL means unlimited. You can retrieve this limit with

chunk_cache_max = png_get_chunk_cache_max(png_ptr);

This limit also applies to the number of buffers that can be allocated by png_decompress_chunk() while decompressing iTXt, zTXt, and iCCP chunks.

You can also set a limit on the amount of memory that a compressed chunk other than IDAT can occupy, with

png_set_chunk_malloc_max(png_ptr, user_chunk_malloc_max);

and you can retrieve the limit with

chunk_malloc_max = png_get_chunk_malloc_max(png_ptr);

Any chunks that would cause either of these limits to be exceeded will be ignored.

Information about your system¶

If you intend to display the PNG or to incorporate it in other image data you need to tell libpng information about your display or drawing surface so that libpng can convert the values in the image to match the display.

From libpng-1.5.4 this information can be set before reading the PNG file header. In earlier versions png_set_gamma() existed but behaved incorrectly if called before the PNG file header had been read and png_set_alpha_mode() did not exist.

If you need to support versions prior to libpng-1.5.4 test the version number as illustrated below using "PNG_LIBPNG_VER >= 10504" and follow the procedures described in the appropriate manual page.

You give libpng the encoding expected by your system expressed as a 'gamma' value. You can also specify a default encoding for the PNG file in case the required information is missing from the file. By default libpng assumes that the PNG data matches your system, to keep this default call:

png_set_gamma(png_ptr, screen_gamma, 1/screen_gamma/*file gamma*/);

or you can use the fixed point equivalent:

png_set_gamma_fixed(png_ptr, PNG_FP_1*screen_gamma, PNG_FP_1/screen_gamma);

If you don't know the gamma for your system it is probably 2.2 - a good approximation to the IEC standard for display systems (sRGB). If images are too contrasty or washed out you got the value wrong - check your system documentation!

Many systems permit the system gamma to be changed via a lookup table in the display driver, a few systems, including older Macs, change the response by default. As of 1.5.4 three special values are available to handle common situations:

PNG_DEFAULT_sRGB: Indicates that the system conforms to the IEC 61966-2-1
standard. This matches almost all systems.
PNG_GAMMA_MAC_18: Indicates that the system is an older (pre Mac OS 10.6)
Apple Macintosh system with the default settings.
PNG_GAMMA_LINEAR: Just the fixed point value for 1.0 - indicates that the
system expects data with no gamma encoding.

You would use the linear (unencoded) value if you need to process the pixel values further because this avoids the need to decode and reencode each component value whenever arithmetic is performed. A lot of graphics software uses linear values for this reason, often with higher precision component values to preserve overall accuracy.

The second thing you may need to tell libpng about is how your system handles alpha channel information. Some, but not all, PNG files contain an alpha channel. To display these files correctly you need to compose the data onto a suitable background, as described in the PNG specification.

Libpng only supports composing onto a single color (using png_set_background; see below). Otherwise you must do the composition yourself and, in this case, you may need to call png_set_alpha_mode:

#if PNG_LIBPNG_VER >= 10504
png_set_alpha_mode(png_ptr, mode, screen_gamma);
#else
png_set_gamma(png_ptr, screen_gamma, 1.0/screen_gamma);
#endif

The screen_gamma value is the same as the argument to png_set_gamma; however, how it affects the output depends on the mode. png_set_alpha_mode() sets the file gamma default to 1/screen_gamma, so normally you don't need to call png_set_gamma. If you need different defaults call png_set_gamma() before png_set_alpha_mode() - if you call it after it will override the settings made by png_set_alpha_mode().

The mode is as follows:

PNG_ALPHA_PNG: The data is encoded according to the PNG specification. Red, green and blue, or gray, components are gamma encoded color values and are not premultiplied by the alpha value. The alpha value is a linear measure of the contribution of the pixel to the corresponding final output pixel.

You should normally use this format if you intend to perform color correction on the color values; most, maybe all, color correction software has no handling for the alpha channel and, anyway, the math to handle pre-multiplied component values is unnecessarily complex.

Before you do any arithmetic on the component values you need to remove the gamma encoding and multiply out the alpha channel. See the PNG specification for more detail. It is important to note that when an image with an alpha channel is scaled, linear encoded, pre-multiplied component values must be used!

The remaining modes assume you don't need to do any further color correction or that if you do, your color correction software knows all about alpha (it probably doesn't!)

PNG_ALPHA_STANDARD: The data libpng produces is encoded in the standard way assumed by most correctly written graphics software. The gamma encoding will be removed by libpng and the linear component values will be pre-multiplied by the alpha channel.

With this format the final image must be re-encoded to match the display gamma before the image is displayed. If your system doesn't do that, yet still seems to perform arithmetic on the pixels without decoding them, it is broken - check out the modes below.

With PNG_ALPHA_STANDARD libpng always produces linear component values, whatever screen_gamma you supply. The screen_gamma value is, however, used as a default for the file gamma if the PNG file has no gamma information.

If you call png_set_gamma() after png_set_alpha_mode() you will override the linear encoding. Instead the pre-multiplied pixel values will be gamma encoded but the alpha channel will still be linear. This may actually match the requirements of some broken software, but it is unlikely.

While linear 8-bit data is often used it has insufficient precision for any image with a reasonable dynamic range. To avoid problems, and if your software supports it, use png_set_expand_16() to force all components to 16 bits.

PNG_ALPHA_OPTIMIZED: This mode is the same as PNG_ALPHA_STANDARD except that completely opaque pixels are gamma encoded according to the screen_gamma value. Pixels with alpha less than 1.0 will still have linear components.

Use this format if you have control over your compositing software and so don't do other arithmetic (such as scaling) on the data you get from libpng. Your compositing software can simply copy opaque pixels to the output but still has linear values for the non-opaque pixels.

In normal compositing, where the alpha channel encodes partial pixel coverage (as opposed to broad area translucency), the inaccuracies of the 8-bit representation of non-opaque pixels are irrelevant.

You can also try this format if your software is broken; it might look better.

PNG_ALPHA_BROKEN: This is PNG_ALPHA_STANDARD; however, all component values, including the alpha channel are gamma encoded. This is an appropriate format to try if your software, or more likely hardware, is totally broken, i.e., if it performs linear arithmetic directly on gamma encoded values.

In most cases of broken software or hardware the bug in the final display manifests as a subtle halo around composited parts of the image. You may not even perceive this as a halo; the composited part of the image may simply appear separate from the background, as though it had been cut out of paper and pasted on afterward.

If you don't have to deal with bugs in software or hardware, or if you can fix them, there are three recommended ways of using png_set_alpha_mode():

png_set_alpha_mode(png_ptr, PNG_ALPHA_PNG,
screen_gamma);

You can do color correction on the result (libpng does not currently support color correction internally). When you handle the alpha channel you need to undo the gamma encoding and multiply out the alpha.

png_set_alpha_mode(png_ptr, PNG_ALPHA_STANDARD,
screen_gamma);
png_set_expand_16(png_ptr);

If you are using the high level interface, don't call png_set_expand_16(); instead pass PNG_TRANSFORM_EXPAND_16 to the interface.

With this mode you can't do color correction, but you can do arithmetic, including composition and scaling, on the data without further processing.

png_set_alpha_mode(png_ptr, PNG_ALPHA_OPTIMIZED,
screen_gamma);

You can avoid the expansion to 16-bit components with this mode, but you lose the ability to scale the image or perform other linear arithmetic. All you can do is compose the result onto a matching output. Since this mode is libpng-specific you also need to write your own composition software.

If you don't need, or can't handle, the alpha channel you can call png_set_background() to remove it by compositing against a fixed color. Don't call png_set_strip_alpha() to do this - it will leave spurious pixel values in transparent parts of this image.

png_set_background(png_ptr, &background_color,
PNG_BACKGROUND_GAMMA_SCREEN, 0, 1);

The background_color is an RGB or grayscale value according to the data format libpng will produce for you. Because you don't yet know the format of the PNG file, if you call png_set_background at this point you must arrange for the format produced by libpng to always have 8-bit or 16-bit components and then store the color as an 8-bit or 16-bit color as appropriate. The color contains separate gray and RGB component values, so you can let libpng produce gray or RGB output according to the input format, but low bit depth grayscale images must always be converted to at least 8-bit format. (Even though low bit depth grayscale images can't have an alpha channel they can have a transparent color!)

You set the transforms you need later, either as flags to the high level interface or libpng API calls for the low level interface. For reference the settings and API calls required are:

8-bit values:
PNG_TRANSFORM_SCALE_16 | PNG_EXPAND
png_set_expand(png_ptr); png_set_scale_16(png_ptr);

If you must get exactly the same inaccurate results
produced by default in versions prior to libpng-1.5.4,
use PNG_TRANSFORM_STRIP_16 and png_set_strip_16(png_ptr)
instead.

16-bit values:
PNG_TRANSFORM_EXPAND_16
png_set_expand_16(png_ptr);

In either case palette image data will be expanded to RGB. If you just want color data you can add PNG_TRANSFORM_GRAY_TO_RGB or png_set_gray_to_rgb(png_ptr) to the list.

Calling png_set_background before the PNG file header is read will not work prior to libpng-1.5.4. Because the failure may result in unexpected warnings or errors it is therefore much safer to call png_set_background after the head has been read. Unfortunately this means that prior to libpng-1.5.4 it cannot be used with the high level interface.

The high-level read interface¶

At this point there are two ways to proceed; through the high-level read interface, or through a sequence of low-level read operations. You can use the high-level interface if (a) you are willing to read the entire image into memory, and (b) the input transformations you want to do are limited to the following set:

PNG_TRANSFORM_IDENTITY No transformation
PNG_TRANSFORM_SCALE_16 Strip 16-bit samples to
8-bit accurately
PNG_TRANSFORM_STRIP_16 Chop 16-bit samples to
8-bit less accurately
PNG_TRANSFORM_STRIP_ALPHA Discard the alpha channel
PNG_TRANSFORM_PACKING Expand 1, 2 and 4-bit
samples to bytes
PNG_TRANSFORM_PACKSWAP Change order of packed
pixels to LSB first
PNG_TRANSFORM_EXPAND Perform set_expand()
PNG_TRANSFORM_INVERT_MONO Invert monochrome images
PNG_TRANSFORM_SHIFT Normalize pixels to the
sBIT depth
PNG_TRANSFORM_BGR Flip RGB to BGR, RGBA
to BGRA
PNG_TRANSFORM_SWAP_ALPHA Flip RGBA to ARGB or GA
to AG
PNG_TRANSFORM_INVERT_ALPHA Change alpha from opacity
to transparency
PNG_TRANSFORM_SWAP_ENDIAN Byte-swap 16-bit samples
PNG_TRANSFORM_GRAY_TO_RGB Expand grayscale samples
to RGB (or GA to RGBA)
PNG_TRANSFORM_EXPAND_16 Expand samples to 16 bits

(This excludes setting a background color, doing gamma transformation, quantizing, and setting filler.) If this is the case, simply do this:

png_read_png(png_ptr, info_ptr, png_transforms, NULL)

where png_transforms is an integer containing the bitwise OR of some set of transformation flags. This call is equivalent to png_read_info(), followed the set of transformations indicated by the transform mask, then png_read_image(), and finally png_read_end().

(The final parameter of this call is not yet used. Someday it might point to transformation parameters required by some future input transform.)

You must use png_transforms and not call any png_set_transform() functions when you use png_read_png().

After you have called png_read_png(), you can retrieve the image data with

row_pointers = png_get_rows(png_ptr, info_ptr);

where row_pointers is an array of pointers to the pixel data for each row:

png_bytep row_pointers[height];

If you know your image size and pixel size ahead of time, you can allocate row_pointers prior to calling png_read_png() with

if (height > PNG_UINT_32_MAX/png_sizeof(png_byte))
png_error (png_ptr,
"Image is too tall to process in memory");

if (width > PNG_UINT_32_MAX/pixel_size)
png_error (png_ptr,
"Image is too wide to process in memory");

row_pointers = png_malloc(png_ptr,
height*png_sizeof(png_bytep));

for (int i=0; i<height, i++)
row_pointers[i]=NULL; /* security precaution */

for (int i=0; i<height, i++)
row_pointers[i]=png_malloc(png_ptr,
width*pixel_size);

png_set_rows(png_ptr, info_ptr, &row_pointers);

Alternatively you could allocate your image in one big block and define row_pointers[i] to point into the proper places in your block.

If you use png_set_rows(), the application is responsible for freeing row_pointers (and row_pointers[i], if they were separately allocated).

If you don't allocate row_pointers ahead of time, png_read_png() will do it, and it'll be free'ed by libpng when you call png_destroy_*().

The low-level read interface¶

If you are going the low-level route, you are now ready to read all the file information up to the actual image data. You do this with a call to png_read_info().

png_read_info(png_ptr, info_ptr);

This will process all chunks up to but not including the image data.

This also copies some of the data from the PNG file into the decode structure for use in later transformations. Important information copied in is:

1) The PNG file gamma from the gAMA chunk. This overwrites the default value provided by an earlier call to png_set_gamma or png_set_alpha_mode.

2) Prior to libpng-1.5.4 the background color from a bKGd chunk. This damages the information provided by an earlier call to png_set_background resulting in unexpected behavior. Libpng-1.5.4 no longer does this.

3) The number of significant bits in each component value. Libpng uses this to optimize gamma handling by reducing the internal lookup table sizes.

4) The transparent color information from a tRNS chunk. This can be modified by a later call to png_set_tRNS.

Querying the info structure¶

Functions are used to get the information from the info_ptr once it has been read. Note that these fields may not be completely filled in until png_read_end() has read the chunk data following the image.

png_get_IHDR(png_ptr, info_ptr, &width, &height,
&bit_depth, &color_type, &interlace_type,
&compression_type, &filter_method);

width - holds the width of the image
in pixels (up to 2^31).

height - holds the height of the image
in pixels (up to 2^31).

bit_depth - holds the bit depth of one of the
image channels. (valid values are
1, 2, 4, 8, 16 and depend also on
the color_type. See also
significant bits (sBIT) below).

color_type - describes which color/alpha channels
are present.
PNG_COLOR_TYPE_GRAY
(bit depths 1, 2, 4, 8, 16)
PNG_COLOR_TYPE_GRAY_ALPHA
(bit depths 8, 16)
PNG_COLOR_TYPE_PALETTE
(bit depths 1, 2, 4, 8)
PNG_COLOR_TYPE_RGB
(bit_depths 8, 16)
PNG_COLOR_TYPE_RGB_ALPHA
(bit_depths 8, 16)

PNG_COLOR_MASK_PALETTE
PNG_COLOR_MASK_COLOR
PNG_COLOR_MASK_ALPHA

interlace_type - (PNG_INTERLACE_NONE or
PNG_INTERLACE_ADAM7)

compression_type - (must be PNG_COMPRESSION_TYPE_BASE
for PNG 1.0)

filter_method - (must be PNG_FILTER_TYPE_BASE
for PNG 1.0, and can also be
PNG_INTRAPIXEL_DIFFERENCING if
the PNG datastream is embedded in
a MNG-1.0 datastream)

Any or all of interlace_type, compression_type, or
filter_method can be NULL if you are
not interested in their values.

Note that png_get_IHDR() returns 32-bit data into
the application's width and height variables.
This is an unsafe situation if these are 16-bit
variables. In such situations, the
png_get_image_width() and png_get_image_height()
functions described below are safer.

width = png_get_image_width(png_ptr,
info_ptr);

height = png_get_image_height(png_ptr,
info_ptr);

bit_depth = png_get_bit_depth(png_ptr,
info_ptr);

color_type = png_get_color_type(png_ptr,
info_ptr);

interlace_type = png_get_interlace_type(png_ptr,
info_ptr);

compression_type = png_get_compression_type(png_ptr,
info_ptr);

filter_method = png_get_filter_type(png_ptr,
info_ptr);

channels = png_get_channels(png_ptr, info_ptr);

channels - number of channels of info for the
color type (valid values are 1 (GRAY,
PALETTE), 2 (GRAY_ALPHA), 3 (RGB),
4 (RGB_ALPHA or RGB + filler byte))

rowbytes = png_get_rowbytes(png_ptr, info_ptr);

rowbytes - number of bytes needed to hold a row

signature = png_get_signature(png_ptr, info_ptr);

signature - holds the signature read from the
file (if any). The data is kept in
the same offset it would be if the
whole signature were read (i.e. if an
application had already read in 4
bytes of signature before starting
libpng, the remaining 4 bytes would
be in signature[4] through signature[7]
(see png_set_sig_bytes())).

These are also important, but their validity depends on whether the chunk has been read. The png_get_valid(png_ptr, info_ptr, PNG_INFO_<chunk>) and png_get_<chunk>(png_ptr, info_ptr, ...) functions return non-zero if the data has been read, or zero if it is missing. The parameters to the png_get_<chunk> are set directly if they are simple data types, or a pointer into the info_ptr is returned for any complex types.

The colorspace data from gAMA, cHRM, sRGB, iCCP, and sBIT chunks is simply returned to give the application information about how the image was encoded. Libpng itself only does transformations using the file gamma when combining semitransparent pixels with the background color.

png_get_PLTE(png_ptr, info_ptr, &palette,
&num_palette);

palette - the palette for the file
(array of png_color)

num_palette - number of entries in the palette

png_get_gAMA(png_ptr, info_ptr, &file_gamma);
png_get_gAMA_fixed(png_ptr, info_ptr, &int_file_gamma);

file_gamma - the gamma at which the file was
written (PNG_INFO_gAMA)

int_file_gamma - 100,000 times the gamma at which the
file is written

png_get_cHRM(png_ptr, info_ptr, &white_x, &white_y, &red_x,
&red_y, &green_x, &green_y, &blue_x, &blue_y)
png_get_cHRM_XYZ(png_ptr, info_ptr, &red_X, &red_Y, &red_Z, &green_X,
&green_Y, &green_Z, &blue_X, &blue_Y, &blue_Z)
png_get_cHRM_fixed(png_ptr, info_ptr, &int_white_x, &int_white_y,
&int_red_x, &int_red_y, &int_green_x, &int_green_y,
&int_blue_x, &int_blue_y)
png_get_cHRM_XYZ_fixed(png_ptr, info_ptr, &int_red_X, &int_red_Y,
&int_red_Z, &int_green_X, &int_green_Y, &int_green_Z,
&int_blue_X, &int_blue_Y, &int_blue_Z)

{white,red,green,blue}_{x,y}
A color space encoding specified using the
chromaticities of the end points and the
white point. (PNG_INFO_cHRM)

{red,green,blue}_{X,Y,Z}
A color space encoding specified using the encoding end
points - the CIE tristimulus specification of the intended
color of the red, green and blue channels in the PNG RGB
data. The white point is simply the sum of the three end
points. (PNG_INFO_cHRM)

png_get_sRGB(png_ptr, info_ptr, &srgb_intent);

file_srgb_intent - the rendering intent (PNG_INFO_sRGB)
The presence of the sRGB chunk
means that the pixel data is in the
sRGB color space. This chunk also
implies specific values of gAMA and
cHRM.

png_get_iCCP(png_ptr, info_ptr, &name,
&compression_type, &profile, &proflen);

name - The profile name.

compression_type - The compression type; always
PNG_COMPRESSION_TYPE_BASE for PNG 1.0.
You may give NULL to this argument to
ignore it.

profile - International Color Consortium color
profile data. May contain NULs.

proflen - length of profile data in bytes.

png_get_sBIT(png_ptr, info_ptr, &sig_bit);

sig_bit - the number of significant bits for
(PNG_INFO_sBIT) each of the gray,
red, green, and blue channels,
whichever are appropriate for the
given color type (png_color_16)

png_get_tRNS(png_ptr, info_ptr, &trans_alpha,
&num_trans, &trans_color);

trans_alpha - array of alpha (transparency)
entries for palette (PNG_INFO_tRNS)

num_trans - number of transparent entries
(PNG_INFO_tRNS)

trans_color - graylevel or color sample values of
the single transparent color for
non-paletted images (PNG_INFO_tRNS)

png_get_hIST(png_ptr, info_ptr, &hist);
(PNG_INFO_hIST)

hist - histogram of palette (array of
png_uint_16)

png_get_tIME(png_ptr, info_ptr, &mod_time);

mod_time - time image was last modified
(PNG_VALID_tIME)

png_get_bKGD(png_ptr, info_ptr, &background);

background - background color (of type
png_color_16p) (PNG_VALID_bKGD)
valid 16-bit red, green and blue
values, regardless of color_type

num_comments = png_get_text(png_ptr, info_ptr,
&text_ptr, &num_text);

num_comments - number of comments

text_ptr - array of png_text holding image
comments

text_ptr[i].compression - type of compression used
on "text" PNG_TEXT_COMPRESSION_NONE
PNG_TEXT_COMPRESSION_zTXt
PNG_ITXT_COMPRESSION_NONE
PNG_ITXT_COMPRESSION_zTXt

text_ptr[i].key - keyword for comment. Must contain
1-79 characters.

text_ptr[i].text - text comments for current
keyword. Can be empty.

text_ptr[i].text_length - length of text string,
after decompression, 0 for iTXt

text_ptr[i].itxt_length - length of itxt string,
after decompression, 0 for tEXt/zTXt

text_ptr[i].lang - language of comment (empty
string for unknown).

text_ptr[i].lang_key - keyword in UTF-8
(empty string for unknown).

Note that the itxt_length, lang, and lang_key
members of the text_ptr structure only exist when the
library is built with iTXt chunk support. Prior to
libpng-1.4.0 the library was built by default without
iTXt support. Also note that when iTXt is supported,
they contain NULL pointers when the "compression"
field contains PNG_TEXT_COMPRESSION_NONE or
PNG_TEXT_COMPRESSION_zTXt.

num_text - number of comments (same as
num_comments; you can put NULL here
to avoid the duplication)

Note while png_set_text() will accept text, language,
and translated keywords that can be NULL pointers, the
structure returned by png_get_text will always contain
regular zero-terminated C strings. They might be
empty strings but they will never be NULL pointers.

num_spalettes = png_get_sPLT(png_ptr, info_ptr,
&palette_ptr);

num_spalettes - number of sPLT chunks read.

palette_ptr - array of palette structures holding
contents of one or more sPLT chunks
read.

png_get_oFFs(png_ptr, info_ptr, &offset_x, &offset_y,
&unit_type);

offset_x - positive offset from the left edge
of the screen (can be negative)

offset_y - positive offset from the top edge
of the screen (can be negative)

unit_type - PNG_OFFSET_PIXEL, PNG_OFFSET_MICROMETER

png_get_pHYs(png_ptr, info_ptr, &res_x, &res_y,
&unit_type);

res_x - pixels/unit physical resolution in
x direction

res_y - pixels/unit physical resolution in
x direction

unit_type - PNG_RESOLUTION_UNKNOWN,
PNG_RESOLUTION_METER

png_get_sCAL(png_ptr, info_ptr, &unit, &width,
&height)

unit - physical scale units (an integer)

width - width of a pixel in physical scale units

height - height of a pixel in physical scale units
(width and height are doubles)

png_get_sCAL_s(png_ptr, info_ptr, &unit, &width,
&height)

unit - physical scale units (an integer)

width - width of a pixel in physical scale units
(expressed as a string)

height - height of a pixel in physical scale units
(width and height are strings like "2.54")

num_unknown_chunks = png_get_unknown_chunks(png_ptr,
info_ptr, &unknowns)

unknowns - array of png_unknown_chunk
structures holding unknown chunks

unknowns[i].name - name of unknown chunk

unknowns[i].data - data of unknown chunk

unknowns[i].size - size of unknown chunk's data

unknowns[i].location - position of chunk in file

The value of "i" corresponds to the order in which the
chunks were read from the PNG file or inserted with the
png_set_unknown_chunks() function.

The value of "location" is a bitwise "or" of

PNG_HAVE_IHDR (0x01)
PNG_HAVE_PLTE (0x02)
PNG_AFTER_IDAT (0x08)

The data from the pHYs chunk can be retrieved in several convenient forms:

res_x = png_get_x_pixels_per_meter(png_ptr,
info_ptr)

res_y = png_get_y_pixels_per_meter(png_ptr,
info_ptr)

res_x_and_y = png_get_pixels_per_meter(png_ptr,
info_ptr)

res_x = png_get_x_pixels_per_inch(png_ptr,
info_ptr)

res_y = png_get_y_pixels_per_inch(png_ptr,
info_ptr)

res_x_and_y = png_get_pixels_per_inch(png_ptr,
info_ptr)

aspect_ratio = png_get_pixel_aspect_ratio(png_ptr,
info_ptr)

Each of these returns 0 [signifying "unknown"] if
the data is not present or if res_x is 0;
res_x_and_y is 0 if res_x != res_y

Note that because of the way the resolutions are
stored internally, the inch conversions won't
come out to exactly even number. For example,
72 dpi is stored as 0.28346 pixels/meter, and
when this is retrieved it is 71.9988 dpi, so
be sure to round the returned value appropriately
if you want to display a reasonable-looking result.

The data from the oFFs chunk can be retrieved in several convenient forms:

x_offset = png_get_x_offset_microns(png_ptr, info_ptr);

y_offset = png_get_y_offset_microns(png_ptr, info_ptr);

x_offset = png_get_x_offset_inches(png_ptr, info_ptr);

y_offset = png_get_y_offset_inches(png_ptr, info_ptr);

Each of these returns 0 [signifying "unknown" if both
x and y are 0] if the data is not present or if the
chunk is present but the unit is the pixel. The
remark about inexact inch conversions applies here
as well, because a value in inches can't always be
converted to microns and back without some loss
of precision.

For more information, see the PNG specification for chunk contents. Be careful with trusting rowbytes, as some of the transformations could increase the space needed to hold a row (expand, filler, gray_to_rgb, etc.). See png_read_update_info(), below.

A quick word about text_ptr and num_text. PNG stores comments in keyword/text pairs, one pair per chunk, with no limit on the number of text chunks, and a 2^31 byte limit on their size. While there are suggested keywords, there is no requirement to restrict the use to these strings. It is strongly suggested that keywords and text be sensible to humans (that's the point), so don't use abbreviations. Non-printing symbols are not allowed. See the PNG specification for more details. There is also no requirement to have text after the keyword.

Keywords should be limited to 79 Latin-1 characters without leading or trailing spaces, but non-consecutive spaces are allowed within the keyword. It is possible to have the same keyword any number of times. The text_ptr is an array of png_text structures, each holding a pointer to a language string, a pointer to a keyword and a pointer to a text string. The text string, language code, and translated keyword may be empty or NULL pointers. The keyword/text pairs are put into the array in the order that they are received. However, some or all of the text chunks may be after the image, so, to make sure you have read all the text chunks, don't mess with these until after you read the stuff after the image. This will be mentioned again below in the discussion that goes with png_read_end().

Input transformations¶

After you've read the header information, you can set up the library to handle any special transformations of the image data. The various ways to transform the data will be described in the order that they should occur. This is important, as some of these change the color type and/or bit depth of the data, and some others only work on certain color types and bit depths.

Transformations you request are ignored if they don't have any meaning for a particular input data format. However some transformations can have an effect as a result of a previous transformation. If you specify a contradictory set of transformations, for example both adding and removing the alpha channel, you cannot predict the final result.

The color used for the transparency values should be supplied in the same format/depth as the current image data. It is stored in the same format/depth as the image data in a tRNS chunk, so this is what libpng expects for this data.

The color used for the background value depends on the need_expand argument as described below.

Data will be decoded into the supplied row buffers packed into bytes unless the library has been told to transform it into another format. For example, 4 bit/pixel paletted or grayscale data will be returned 2 pixels/byte with the leftmost pixel in the high-order bits of the byte, unless png_set_packing() is called. 8-bit RGB data will be stored in RGB RGB RGB format unless png_set_filler() or png_set_add_alpha() is called to insert filler bytes, either before or after each RGB triplet. 16-bit RGB data will be returned RRGGBB RRGGBB, with the most significant byte of the color value first, unless png_set_scale_16() is called to transform it to regular RGB RGB triplets, or png_set_filler() or png_set_add alpha() is called to insert filler bytes, either before or after each RRGGBB triplet. Similarly, 8-bit or 16-bit grayscale data can be modified with png_set_filler(), png_set_add_alpha(), png_set_strip_16(), or png_set_scale_16().

The following code transforms grayscale images of less than 8 to 8 bits, changes paletted images to RGB, and adds a full alpha channel if there is transparency information in a tRNS chunk. This is most useful on grayscale images with bit depths of 2 or 4 or if there is a multiple-image viewing application that wishes to treat all images in the same way.

if (color_type == PNG_COLOR_TYPE_PALETTE)
png_set_palette_to_rgb(png_ptr);

if (png_get_valid(png_ptr, info_ptr,
PNG_INFO_tRNS)) png_set_tRNS_to_alpha(png_ptr);

if (color_type == PNG_COLOR_TYPE_GRAY &&
bit_depth < 8) png_set_expand_gray_1_2_4_to_8(png_ptr);

The first two functions are actually aliases for png_set_expand(), added in libpng version 1.0.4, with the function names expanded to improve code readability. In some future version they may actually do different things.

As of libpng version 1.2.9, png_set_expand_gray_1_2_4_to_8() was added. It expands the sample depth without changing tRNS to alpha.

As of libpng version 1.5.2, png_set_expand_16() was added. It behaves as png_set_expand(); however, the resultant channels have 16 bits rather than 8. Use this when the output color or gray channels are made linear to avoid fairly severe accuracy loss.

if (bit_depth < 16)
png_set_expand_16(png_ptr);

PNG can have files with 16 bits per channel. If you only can handle 8 bits per channel, this will strip the pixels down to 8-bit.

if (bit_depth == 16) #if PNG_LIBPNG_VER >= 10504
png_set_scale_16(png_ptr); #else
png_set_strip_16(png_ptr); #endif

(The more accurate "png_set_scale_16()" API became available in libpng version 1.5.4).

If you need to process the alpha channel on the image separately from the image data (for example if you convert it to a bitmap mask) it is possible to have libpng strip the channel leaving just RGB or gray data:

if (color_type & PNG_COLOR_MASK_ALPHA)
png_set_strip_alpha(png_ptr);

If you strip the alpha channel you need to find some other way of dealing with the information. If, instead, you want to convert the image to an opaque version with no alpha channel use png_set_background; see below.

As of libpng version 1.5.2, almost all useful expansions are supported, the major ommissions are conversion of grayscale to indexed images (which can be done trivially in the application) and conversion of indexed to grayscale (which can be done by a trivial manipulation of the palette.)

In the following table, the 01 means grayscale with depth<8, 31 means indexed with depth<8, other numerals represent the color type, "T" means the tRNS chunk is present, A means an alpha channel is present, and O means tRNS or alpha is present but all pixels in the image are opaque.

FROM 01 31 0 0T 0O 2 2T 2O 3 3T 3O 4A 4O 6A 6O
TO
01 - [G] - - - - - - - - - - - - -
31 [Q] Q [Q] [Q] [Q] Q Q Q Q Q Q [Q] [Q] Q Q
0 1 G + . . G G G G G G B B GB GB
0T lt Gt t + . Gt G G Gt G G Bt Bt GBt GBt
0O lt Gt t . + Gt Gt G Gt Gt G Bt Bt GBt GBt
2 C P C C C + . . C - - CB CB B B
2T Ct - Ct C C t + t - - - CBt CBt Bt Bt
2O Ct - Ct C C t t + - - - CBt CBt Bt Bt
3 [Q] p [Q] [Q] [Q] Q Q Q + . . [Q] [Q] Q Q
3T [Qt] p [Qt][Q] [Q] Qt Qt Qt t + t [Qt][Qt] Qt Qt
3O [Qt] p [Qt][Q] [Q] Qt Qt Qt t t + [Qt][Qt] Qt Qt
4A lA G A T T GA GT GT GA GT GT + BA G GBA
4O lA GBA A T T GA GT GT GA GT GT BA + GBA G
6A CA PA CA C C A T tT PA P P C CBA + BA
6O CA PBA CA C C A tT T PA P P CBA C BA +

Within the matrix,
"+" identifies entries where 'from' and 'to' are the same.
"-" means the transformation is not supported.
"." means nothing is necessary (a tRNS chunk can just be ignored).
"t" means the transformation is obtained by png_set_tRNS.
"A" means the transformation is obtained by png_set_add_alpha().
"X" means the transformation is obtained by png_set_expand().
"1" means the transformation is obtained by
png_set_expand_gray_1_2_4_to_8() (and by png_set_expand()
if there is no transparency in the original or the final
format).
"C" means the transformation is obtained by png_set_gray_to_rgb().
"G" means the transformation is obtained by png_set_rgb_to_gray().
"P" means the transformation is obtained by
png_set_expand_palette_to_rgb().
"p" means the transformation is obtained by png_set_packing().
"Q" means the transformation is obtained by png_set_quantize().
"T" means the transformation is obtained by
png_set_tRNS_to_alpha().
"B" means the transformation is obtained by
png_set_background(), or png_strip_alpha().

When an entry has multiple transforms listed all are required to cause the right overall transformation. When two transforms are separated by a comma either will do the job. When transforms are enclosed in [] the transform should do the job but this is currently unimplemented - a different format will result if the suggested transformations are used.

In PNG files, the alpha channel in an image is the level of opacity. If you need the alpha channel in an image to be the level of transparency instead of opacity, you can invert the alpha channel (or the tRNS chunk data) after it's read, so that 0 is fully opaque and 255 (in 8-bit or paletted images) or 65535 (in 16-bit images) is fully transparent, with

png_set_invert_alpha(png_ptr);

PNG files pack pixels of bit depths 1, 2, and 4 into bytes as small as they can, resulting in, for example, 8 pixels per byte for 1 bit files. This code expands to 1 pixel per byte without changing the values of the pixels:

if (bit_depth < 8)
png_set_packing(png_ptr);

PNG files have possible bit depths of 1, 2, 4, 8, and 16. All pixels stored in a PNG image have been "scaled" or "shifted" up to the next higher possible bit depth (e.g. from 5 bits/sample in the range [0,31] to 8 bits/sample in the range [0, 255]). However, it is also possible to convert the PNG pixel data back to the original bit depth of the image. This call reduces the pixels back down to the original bit depth:

png_color_8p sig_bit;

if (png_get_sBIT(png_ptr, info_ptr, &sig_bit))
png_set_shift(png_ptr, sig_bit);

PNG files store 3-color pixels in red, green, blue order. This code changes the storage of the pixels to blue, green, red:

if (color_type == PNG_COLOR_TYPE_RGB ||
color_type == PNG_COLOR_TYPE_RGB_ALPHA)
png_set_bgr(png_ptr);

PNG files store RGB pixels packed into 3 or 6 bytes. This code expands them into 4 or 8 bytes for windowing systems that need them in this format:

if (color_type == PNG_COLOR_TYPE_RGB)
png_set_filler(png_ptr, filler, PNG_FILLER_BEFORE);

where "filler" is the 8 or 16-bit number to fill with, and the location is either PNG_FILLER_BEFORE or PNG_FILLER_AFTER, depending upon whether you want the filler before the RGB or after. This transformation does not affect images that already have full alpha channels. To add an opaque alpha channel, use filler=0xff or 0xffff and PNG_FILLER_AFTER which will generate RGBA pixels.

Note that png_set_filler() does not change the color type. If you want to do that, you can add a true alpha channel with

if (color_type == PNG_COLOR_TYPE_RGB ||
color_type == PNG_COLOR_TYPE_GRAY)
png_set_add_alpha(png_ptr, filler, PNG_FILLER_AFTER);

where "filler" contains the alpha value to assign to each pixel. This function was added in libpng-1.2.7.

If you are reading an image with an alpha channel, and you need the data as ARGB instead of the normal PNG format RGBA:

if (color_type == PNG_COLOR_TYPE_RGB_ALPHA)
png_set_swap_alpha(png_ptr);

For some uses, you may want a grayscale image to be represented as RGB. This code will do that conversion:

if (color_type == PNG_COLOR_TYPE_GRAY ||
color_type == PNG_COLOR_TYPE_GRAY_ALPHA)
png_set_gray_to_rgb(png_ptr);

Conversely, you can convert an RGB or RGBA image to grayscale or grayscale with alpha.

if (color_type == PNG_COLOR_TYPE_RGB ||
color_type == PNG_COLOR_TYPE_RGB_ALPHA)
png_set_rgb_to_gray(png_ptr, error_action,
double red_weight, double green_weight);

error_action = 1: silently do the conversion

error_action = 2: issue a warning if the original
image has any pixel where
red != green or red != blue

error_action = 3: issue an error and abort the
conversion if the original
image has any pixel where
red != green or red != blue

red_weight: weight of red component

green_weight: weight of green component
If either weight is negative, default
weights are used.

In the corresponding fixed point API the red_weight and green_weight values are simply scaled by 100,000:

png_set_rgb_to_gray(png_ptr, error_action,
png_fixed_point red_weight,
png_fixed_point green_weight);

If you have set error_action = 1 or 2, you can later check whether the image really was gray, after processing the image rows, with the png_get_rgb_to_gray_status(png_ptr) function. It will return a png_byte that is zero if the image was gray or 1 if there were any non-gray pixels. Background and sBIT data will be silently converted to grayscale, using the green channel data for sBIT, regardless of the error_action setting.

The default values come from the PNG file cHRM chunk if present; otherwise, the defaults correspond to the ITU-R recommendation 709, and also the sRGB color space, as recommended in the Charles Poynton's Colour FAQ, <http://www.poynton.com/>, in section 9:

<http://www.poynton.com/notes/colour_and_gamma/ColorFAQ.html#RTFToC9>

Y = 0.2126 * R + 0.7152 * G + 0.0722 * B

Previous versions of this document, 1998 through 2002, recommended a slightly different formula:

Y = 0.212671 * R + 0.715160 * G + 0.072169 * B

Libpng uses an integer approximation:

Y = (6968 * R + 23434 * G + 2366 * B)/32768

The calculation is done in a linear colorspace, if the image gamma can be determined.

The png_set_background() function has been described already; it tells libpng to composite images with alpha or simple transparency against the supplied background color. For compatibility with versions of libpng earlier than libpng-1.5.4 it is recommended that you call the function after reading the file header, even if you don't want to use the color in a bKGD chunk, if one exists.

If the PNG file contains a bKGD chunk (PNG_INFO_bKGD valid), you may use this color, or supply another color more suitable for the current display (e.g., the background color from a web page). You need to tell libpng how the color is represented, both the format of the component values in the color (the number of bits) and the gamma encoding of the color. The function takes two arguments, background_gamma_mode and need_expand to convey this information; however, only two combinations are likely to be useful:

png_color_16 my_background;
png_color_16p image_background;

if (png_get_bKGD(png_ptr, info_ptr, &image_background))
png_set_background(png_ptr, image_background,
PNG_BACKGROUND_GAMMA_FILE, 1/*needs to be expanded*/, 1);
else
png_set_background(png_ptr, &my_background,
PNG_BACKGROUND_GAMMA_SCREEN, 0/*do not expand*/, 1);

The second call was described above - my_background is in the format of the final, display, output produced by libpng. Because you now know the format of the PNG it is possible to avoid the need to choose either 8-bit or 16-bit output and to retain palette images (the palette colors will be modified appropriately and the tRNS chunk removed.) However, if you are doing this, take great care not to ask for transformations without checking first that they apply!

In the first call the background color has the original bit depth and color type of the PNG file. So, for palette images the color is supplied as a palette index and for low bit greyscale images the color is a reduced bit value in image_background->gray.

If you didn't call png_set_gamma() before reading the file header, for example if you need your code to remain compatible with older versions of libpng prior to libpng-1.5.4, this is the place to call it.

Do not call it if you called png_set_alpha_mode(); doing so will damage the settings put in place by png_set_alpha_mode(). (If png_set_alpha_mode() is supported then you can certainly do png_set_gamma() before reading the PNG header.)

This API unconditionally sets the screen and file gamma values, so it will override the value in the PNG file unless it is called before the PNG file reading starts. For this reason you must always call it with the PNG file value when you call it in this position:

if (png_get_gAMA(png_ptr, info_ptr, &file_gamma))
png_set_gamma(png_ptr, screen_gamma, file_gamma);

else
png_set_gamma(png_ptr, screen_gamma, 0.45455);

If you need to reduce an RGB file to a paletted file, or if a paletted file has more entries then will fit on your screen, png_set_quantize() will do that. Note that this is a simple match quantization that merely finds the closest color available. This should work fairly well with optimized palettes, but fairly badly with linear color cubes. If you pass a palette that is larger than maximum_colors, the file will reduce the number of colors in the palette so it will fit into maximum_colors. If there is a histogram, libpng will use it to make more intelligent choices when reducing the palette. If there is no histogram, it may not do as good a job.

if (color_type & PNG_COLOR_MASK_COLOR)
{
if (png_get_valid(png_ptr, info_ptr,
PNG_INFO_PLTE))
{
png_uint_16p histogram = NULL;

png_get_hIST(png_ptr, info_ptr,
&histogram);
png_set_quantize(png_ptr, palette, num_palette,
max_screen_colors, histogram, 1);
}

else
{
png_color std_color_cube[MAX_SCREEN_COLORS] =
{ ... colors ... };

png_set_quantize(png_ptr, std_color_cube,
MAX_SCREEN_COLORS, MAX_SCREEN_COLORS,
NULL,0);
}
}

PNG files describe monochrome as black being zero and white being one. The following code will reverse this (make black be one and white be zero):

if (bit_depth == 1 && color_type == PNG_COLOR_TYPE_GRAY)
png_set_invert_mono(png_ptr);

This function can also be used to invert grayscale and gray-alpha images:

if (color_type == PNG_COLOR_TYPE_GRAY ||
color_type == PNG_COLOR_TYPE_GRAY_ALPHA)
png_set_invert_mono(png_ptr);

PNG files store 16-bit pixels in network byte order (big-endian, ie. most significant bits first). This code changes the storage to the other way (little-endian, i.e. least significant bits first, the way PCs store them):

if (bit_depth == 16)
png_set_swap(png_ptr);

If you are using packed-pixel images (1, 2, or 4 bits/pixel), and you need to change the order the pixels are packed into bytes, you can use:

if (bit_depth < 8)
png_set_packswap(png_ptr);

Finally, you can write your own transformation function if none of the existing ones meets your needs. This is done by setting a callback with

png_set_read_user_transform_fn(png_ptr,
read_transform_fn);

You must supply the function

void read_transform_fn(png_structp png_ptr, png_row_infop
row_info, png_bytep data)

See pngtest.c for a working example. Your function will be called after all of the other transformations have been processed. Take care with interlaced images if you do the interlace yourself - the width of the row is the width in 'row_info', not the overall image width.

If supported, libpng provides two information routines that you can use to find where you are in processing the image:

png_get_current_pass_number(png_structp png_ptr);
png_get_current_row_number(png_structp png_ptr);

Don't try using these outside a transform callback - firstly they are only supported if user transforms are supported, secondly they may well return unexpected results unless the row is actually being processed at the moment they are called.

With interlaced images the value returned is the row in the input sub-image image. Use PNG_ROW_FROM_PASS_ROW(row, pass) and PNG_COL_FROM_PASS_COL(col, pass) to find the output pixel (x,y) given an interlaced sub-image pixel (row,col,pass).

The discussion of interlace handling above contains more information on how to use these values.

You can also set up a pointer to a user structure for use by your callback function, and you can inform libpng that your transform function will change the number of channels or bit depth with the function

png_set_user_transform_info(png_ptr, user_ptr,
user_depth, user_channels);

The user's application, not libpng, is responsible for allocating and freeing any memory required for the user structure.

You can retrieve the pointer via the function png_get_user_transform_ptr(). For example:

voidp read_user_transform_ptr =
png_get_user_transform_ptr(png_ptr);

The last thing to handle is interlacing; this is covered in detail below, but you must call the function here if you want libpng to handle expansion of the interlaced image.

number_of_passes = png_set_interlace_handling(png_ptr);

After setting the transformations, libpng can update your png_info structure to reflect any transformations you've requested with this call.

png_read_update_info(png_ptr, info_ptr);

This is most useful to update the info structure's rowbytes field so you can use it to allocate your image memory. This function will also update your palette with the correct screen_gamma and background if these have been given with the calls above. You may only call png_read_update_info() once with a particular info_ptr.

After you call png_read_update_info(), you can allocate any memory you need to hold the image. The row data is simply raw byte data for all forms of images. As the actual allocation varies among applications, no example will be given. If you are allocating one large chunk, you will need to build an array of pointers to each row, as it will be needed for some of the functions below.

Remember: Before you call png_read_update_info(), the png_get_*() functions return the values corresponding to the original PNG image. After you call png_read_update_info the values refer to the image that libpng will output. Consequently you must call all the png_set_ functions before you call png_read_update_info(). This is particularly important for png_set_interlace_handling() - if you are going to call png_read_update_info() you must call png_set_interlace_handling() before it unless you want to receive interlaced output.

Reading image data¶

After you've allocated memory, you can read the image data. The simplest way to do this is in one function call. If you are allocating enough memory to hold the whole image, you can just call png_read_image() and libpng will read in all the image data and put it in the memory area supplied. You will need to pass in an array of pointers to each row.

This function automatically handles interlacing, so you don't need to call png_set_interlace_handling() (unless you call png_read_update_info()) or call this function multiple times, or any of that other stuff necessary with png_read_rows().

png_read_image(png_ptr, row_pointers);

where row_pointers is:

png_bytep row_pointers[height];

You can point to void or char or whatever you use for pixels.

If you don't want to read in the whole image at once, you can use png_read_rows() instead. If there is no interlacing (check interlace_type == PNG_INTERLACE_NONE), this is simple:

png_read_rows(png_ptr, row_pointers, NULL,
number_of_rows);

where row_pointers is the same as in the png_read_image() call.

If you are doing this just one row at a time, you can do this with a single row_pointer instead of an array of row_pointers:

png_bytep row_pointer = row;
png_read_row(png_ptr, row_pointer, NULL);

If the file is interlaced (interlace_type != 0 in the IHDR chunk), things get somewhat harder. The only current (PNG Specification version 1.2) interlacing type for PNG is (interlace_type == PNG_INTERLACE_ADAM7); a somewhat complicated 2D interlace scheme, known as Adam7, that breaks down an image into seven smaller images of varying size, based on an 8x8 grid. This number is defined (from libpng 1.5) as PNG_INTERLACE_ADAM7_PASSES in png.h

libpng can fill out those images or it can give them to you "as is". It is almost always better to have libpng handle the interlacing for you. If you want the images filled out, there are two ways to do that. The one mentioned in the PNG specification is to expand each pixel to cover those pixels that have not been read yet (the "rectangle" method). This results in a blocky image for the first pass, which gradually smooths out as more pixels are read. The other method is the "sparkle" method, where pixels are drawn only in their final locations, with the rest of the image remaining whatever colors they were initialized to before the start of the read. The first method usually looks better, but tends to be slower, as there are more pixels to put in the rows.

If, as is likely, you want libpng to expand the images, call this before calling png_start_read_image() or png_read_update_info():

if (interlace_type == PNG_INTERLACE_ADAM7)
number_of_passes
= png_set_interlace_handling(png_ptr);

This will return the number of passes needed. Currently, this is seven, but may change if another interlace type is added. This function can be called even if the file is not interlaced, where it will return one pass. You then need to read the whole image 'number_of_passes' times. Each time will distribute the pixels from the current pass to the correct place in the output image, so you need to supply the same rows to png_read_rows in each pass.

If you are not going to display the image after each pass, but are going to wait until the entire image is read in, use the sparkle effect. This effect is faster and the end result of either method is exactly the same. If you are planning on displaying the image after each pass, the "rectangle" effect is generally considered the better looking one.

If you only want the "sparkle" effect, just call png_read_rows() as normal, with the third parameter NULL. Make sure you make pass over the image number_of_passes times, and you don't change the data in the rows between calls. You can change the locations of the data, just not the data. Each pass only writes the pixels appropriate for that pass, and assumes the data from previous passes is still valid.

png_read_rows(png_ptr, row_pointers, NULL,
number_of_rows);

If you only want the first effect (the rectangles), do the same as before except pass the row buffer in the third parameter, and leave the second parameter NULL.

png_read_rows(png_ptr, NULL, row_pointers,
number_of_rows);

If you don't want libpng to handle the interlacing details, just call png_read_rows() PNG_INTERLACE_ADAM7_PASSES times to read in all the images. Each of the images is a valid image by itself; however, you will almost certainly need to distribute the pixels from each sub-image to the correct place. This is where everything gets very tricky.

If you want to retrieve the separate images you must pass the correct number of rows to each successive call of png_read_rows(). The calculation gets pretty complicated for small images, where some sub-images may not even exist because either their width or height ends up zero. libpng provides two macros to help you in 1.5 and later versions:

png_uint_32 width = PNG_PASS_COLS(image_width, pass_number);
png_uint_32 height = PNG_PASS_ROWS(image_height, pass_number);

Respectively these tell you the width and height of the sub-image corresponding to the numbered pass. 'pass' is in in the range 0 to 6 - this can be confusing because the specification refers to the same passes as 1 to 7! Be careful, you must check both the width and height before calling png_read_rows() and not call it for that pass if either is zero.

You can, of course, read each sub-image row by row. If you want to produce optimal code to make a pixel-by-pixel transformation of an interlaced image this is the best approach; read each row of each pass, transform it, and write it out to a new interlaced image.

If you want to de-interlace the image yourself libpng provides further macros to help that tell you where to place the pixels in the output image. Because the interlacing scheme is rectangular - sub-image pixels are always arranged on a rectangular grid - all you need to know for each pass is the starting column and row in the output image of the first pixel plus the spacing between each pixel. As of libpng 1.5 there are four macros to retrieve this information:

png_uint_32 x = PNG_PASS_START_COL(pass);
png_uint_32 y = PNG_PASS_START_ROW(pass);
png_uint_32 xStep = 1U << PNG_PASS_COL_SHIFT(pass);
png_uint_32 yStep = 1U << PNG_PASS_ROW_SHIFT(pass);

These allow you to write the obvious loop:

png_uint_32 input_y = 0;
png_uint_32 output_y = PNG_PASS_START_ROW(pass);

while (output_y < output_image_height)
{
png_uint_32 input_x = 0;
png_uint_32 output_x = PNG_PASS_START_COL(pass);

while (output_x < output_image_width)
{
image[output_y][output_x] =
subimage[pass][input_y][input_x++];

output_x += xStep;
}

++input_y;
output_y += yStep;
}

Notice that the steps between successive output rows and columns are returned as shifts. This is possible because the pixels in the subimages are always a power of 2 apart - 1, 2, 4 or 8 pixels - in the original image. In practice you may need to directly calculate the output coordinate given an input coordinate. libpng provides two further macros for this purpose:

png_uint_32 output_x = PNG_COL_FROM_PASS_COL(input_x, pass);
png_uint_32 output_y = PNG_ROW_FROM_PASS_ROW(input_y, pass);

Finally a pair of macros are provided to tell you if a particular image row or column appears in a given pass:

int col_in_pass = PNG_COL_IN_INTERLACE_PASS(output_x, pass);
int row_in_pass = PNG_ROW_IN_INTERLACE_PASS(output_y, pass);

Bear in mind that you will probably also need to check the width and height of the pass in addition to the above to be sure the pass even exists!

With any luck you are convinced by now that you don't want to do your own interlace handling. In reality normally the only good reason for doing this is if you are processing PNG files on a pixel-by-pixel basis and don't want to load the whole file into memory when it is interlaced.

libpng includes a test program, pngvalid, that illustrates reading and writing of interlaced images. If you can't get interlacing to work in your code and don't want to leave it to libpng (the recommended approach), see how pngvalid.c does it.

Finishing a sequential read¶

After you are finished reading the image through the low-level interface, you can finish reading the file. If you are interested in comments or time, which may be stored either before or after the image data, you should pass the separate png_info struct if you want to keep the comments from before and after the image separate.

png_infop end_info = png_create_info_struct(png_ptr);

if (!end_info)
{
png_destroy_read_struct(&png_ptr, &info_ptr,
(png_infopp)NULL);
return (ERROR);
}

png_read_end(png_ptr, end_info);

If you are not interested, you should still call png_read_end() but you can pass NULL, avoiding the need to create an end_info structure.

png_read_end(png_ptr, (png_infop)NULL);

If you don't call png_read_end(), then your file pointer will be left pointing to the first chunk after the last IDAT, which is probably not what you want if you expect to read something beyond the end of the PNG datastream.

When you are done, you can free all memory allocated by libpng like this:

png_destroy_read_struct(&png_ptr, &info_ptr,
&end_info);

or, if you didn't create an end_info structure,

png_destroy_read_struct(&png_ptr, &info_ptr,
(png_infopp)NULL);

It is also possible to individually free the info_ptr members that point to libpng-allocated storage with the following function:

png_free_data(png_ptr, info_ptr, mask, seq)

mask - identifies data to be freed, a mask
containing the bitwise OR of one or
more of
PNG_FREE_PLTE, PNG_FREE_TRNS,
PNG_FREE_HIST, PNG_FREE_ICCP,
PNG_FREE_PCAL, PNG_FREE_ROWS,
PNG_FREE_SCAL, PNG_FREE_SPLT,
PNG_FREE_TEXT, PNG_FREE_UNKN,
or simply PNG_FREE_ALL

seq - sequence number of item to be freed
(-1 for all items)

This function may be safely called when the relevant storage has already been freed, or has not yet been allocated, or was allocated by the user and not by libpng, and will in those cases do nothing. The "seq" parameter is ignored if only one item of the selected data type, such as PLTE, is allowed. If "seq" is not -1, and multiple items are allowed for the data type identified in the mask, such as text or sPLT, only the n'th item in the structure is freed, where n is "seq".

The default behavior is only to free data that was allocated internally by libpng. This can be changed, so that libpng will not free the data, or so that it will free data that was allocated by the user with png_malloc() or png_calloc() and passed in via a png_set_*() function, with

png_data_freer(png_ptr, info_ptr, freer, mask)

freer - one of
PNG_DESTROY_WILL_FREE_DATA
PNG_SET_WILL_FREE_DATA
PNG_USER_WILL_FREE_DATA

mask - which data elements are affected
same choices as in png_free_data()

This function only affects data that has already been allocated. You can call this function after reading the PNG data but before calling any png_set_*() functions, to control whether the user or the png_set_*() function is responsible for freeing any existing data that might be present, and again after the png_set_*() functions to control whether the user or png_destroy_*() is supposed to free the data. When the user assumes responsibility for libpng-allocated data, the application must use png_free() to free it, and when the user transfers responsibility to libpng for data that the user has allocated, the user must have used png_malloc() or png_calloc() to allocate it.

If you allocated your row_pointers in a single block, as suggested above in the description of the high level read interface, you must not transfer responsibility for freeing it to the png_set_rows or png_read_destroy function, because they would also try to free the individual row_pointers[i].

If you allocated text_ptr.text, text_ptr.lang, and text_ptr.translated_keyword separately, do not transfer responsibility for freeing text_ptr to libpng, because when libpng fills a png_text structure it combines these members with the key member, and png_free_data() will free only text_ptr.key. Similarly, if you transfer responsibility for free'ing text_ptr from libpng to your application, your application must not separately free those members.

The png_free_data() function will turn off the "valid" flag for anything it frees. If you need to turn the flag off for a chunk that was freed by your application instead of by libpng, you can use

png_set_invalid(png_ptr, info_ptr, mask);

mask - identifies the chunks to be made invalid,
containing the bitwise OR of one or
more of
PNG_INFO_gAMA, PNG_INFO_sBIT,
PNG_INFO_cHRM, PNG_INFO_PLTE,
PNG_INFO_tRNS, PNG_INFO_bKGD,
PNG_INFO_hIST, PNG_INFO_pHYs,
PNG_INFO_oFFs, PNG_INFO_tIME,
PNG_INFO_pCAL, PNG_INFO_sRGB,
PNG_INFO_iCCP, PNG_INFO_sPLT,
PNG_INFO_sCAL, PNG_INFO_IDAT

For a more compact example of reading a PNG image, see the file example.c.

Reading PNG files progressively¶

The progressive reader is slightly different then the non-progressive reader. Instead of calling png_read_info(), png_read_rows(), and png_read_end(), you make one call to png_process_data(), which calls callbacks when it has the info, a row, or the end of the image. You set up these callbacks with png_set_progressive_read_fn(). You don't have to worry about the input/output functions of libpng, as you are giving the library the data directly in png_process_data(). I will assume that you have read the section on reading PNG files above, so I will only highlight the differences (although I will show all of the code).

png_structp png_ptr; png_infop info_ptr;

/* An example code fragment of how you would
initialize the progressive reader in your
application. */
int
initialize_png_reader()
{
png_ptr = png_create_read_struct
(PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
user_error_fn, user_warning_fn);

if (!png_ptr)
return (ERROR);

info_ptr = png_create_info_struct(png_ptr);

if (!info_ptr)
{
png_destroy_read_struct(&png_ptr,
(png_infopp)NULL, (png_infopp)NULL);
return (ERROR);
}

if (setjmp(png_jmpbuf(png_ptr)))
{
png_destroy_read_struct(&png_ptr, &info_ptr,
(png_infopp)NULL);
return (ERROR);
}

/* This one's new. You can provide functions
to be called when the header info is valid,
when each row is completed, and when the image
is finished. If you aren't using all functions,
you can specify NULL parameters. Even when all
three functions are NULL, you need to call
png_set_progressive_read_fn(). You can use
any struct as the user_ptr (cast to a void pointer
for the function call), and retrieve the pointer
from inside the callbacks using the function

png_get_progressive_ptr(png_ptr);

which will return a void pointer, which you have
to cast appropriately.
*/
png_set_progressive_read_fn(png_ptr, (void *)user_ptr,
info_callback, row_callback, end_callback);

return 0;
}

/* A code fragment that you call as you receive blocks
of data */
int
process_data(png_bytep buffer, png_uint_32 length)
{
if (setjmp(png_jmpbuf(png_ptr)))
{
png_destroy_read_struct(&png_ptr, &info_ptr,
(png_infopp)NULL);
return (ERROR);
}

/* This one's new also. Simply give it a chunk
of data from the file stream (in order, of
course). On machines with segmented memory
models machines, don't give it any more than
64K. The library seems to run fine with sizes
of 4K. Although you can give it much less if
necessary (I assume you can give it chunks of
1 byte, I haven't tried less then 256 bytes
yet). When this function returns, you may
want to display any rows that were generated
in the row callback if you don't already do
so there.
*/
png_process_data(png_ptr, info_ptr, buffer, length);

/* At this point you can call png_process_data_skip if
you want to handle data the library will skip yourself;
it simply returns the number of bytes to skip (and stops
libpng skipping that number of bytes on the next
png_process_data call).
return 0;
}

/* This function is called (as set by
png_set_progressive_read_fn() above) when enough data
has been supplied so all of the header has been
read.
*/
void
info_callback(png_structp png_ptr, png_infop info)
{
/* Do any setup here, including setting any of
the transformations mentioned in the Reading
PNG files section. For now, you _must_ call
either png_start_read_image() or
png_read_update_info() after all the
transformations are set (even if you don't set
any). You may start getting rows before
png_process_data() returns, so this is your
last chance to prepare for that.

This is where you turn on interlace handling,
assuming you don't want to do it yourself.

If you need to you can stop the processing of
your original input data at this point by calling
png_process_data_pause. This returns the number
of unprocessed bytes from the last png_process_data
call - it is up to you to ensure that the next call
sees these bytes again. If you don't want to bother
with this you can get libpng to cache the unread
bytes by setting the 'save' parameter (see png.h) but
then libpng will have to copy the data internally.
*/
}

/* This function is called when each row of image
data is complete */
void
row_callback(png_structp png_ptr, png_bytep new_row,
png_uint_32 row_num, int pass)
{
/* If the image is interlaced, and you turned
on the interlace handler, this function will
be called for every row in every pass. Some
of these rows will not be changed from the
previous pass. When the row is not changed,
the new_row variable will be NULL. The rows
and passes are called in order, so you don't
really need the row_num and pass, but I'm
supplying them because it may make your life
easier.

If you did not turn on interlace handling then
the callback is called for each row of each
sub-image when the image is interlaced. In this
case 'row_num' is the row in the sub-image, not
the row in the output image as it is in all other
cases.

For the non-NULL rows of interlaced images when
you have switched on libpng interlace handling,
you must call png_progressive_combine_row()
passing in the row and the old row. You can
call this function for NULL rows (it will just
return) and for non-interlaced images (it just
does the memcpy for you) if it will make the
code easier. Thus, you can just do this for
all cases if you switch on interlace handling;
*/

png_progressive_combine_row(png_ptr, old_row,
new_row);

/* where old_row is what was displayed for
previously for the row. Note that the first
pass (pass == 0, really) will completely cover
the old row, so the rows do not have to be
initialized. After the first pass (and only
for interlaced images), you will have to pass
the current row, and the function will combine
the old row and the new row.

You can also call png_process_data_pause in this
callback - see above.
*/
}

void
end_callback(png_structp png_ptr, png_infop info)
{
/* This function is called after the whole image
has been read, including any chunks after the
image (up to and including the IEND). You
will usually have the same info chunk as you
had in the header, although some data may have
been added to the comments and time fields.

Most people won't do much here, perhaps setting
a flag that marks the image as finished.
*/
}

Setup¶

You will want to do the I/O initialization before you get into libpng, so if it doesn't work, you don't have anything to undo. If you are not using the standard I/O functions, you will need to replace them with custom writing functions. See the discussion under Customizing libpng.

FILE *fp = fopen(file_name, "wb");

if (!fp)
return (ERROR);

Next, png_struct and png_info need to be allocated and initialized. As these can be both relatively large, you may not want to store these on the stack, unless you have stack space to spare. Of course, you will want to check if they return NULL. If you are also reading, you won't want to name your read structure and your write structure both "png_ptr"; you can call them anything you like, such as "read_ptr" and "write_ptr". Look at pngtest.c, for example.

png_structp png_ptr = png_create_write_struct
(PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
user_error_fn, user_warning_fn);

if (!png_ptr)
return (ERROR);

png_infop info_ptr = png_create_info_struct(png_ptr);
if (!info_ptr)
{
png_destroy_write_struct(&png_ptr,
(png_infopp)NULL);
return (ERROR);
}

If you want to use your own memory allocation routines, define PNG_USER_MEM_SUPPORTED and use png_create_write_struct_2() instead of png_create_write_struct():

png_structp png_ptr = png_create_write_struct_2
(PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
user_error_fn, user_warning_fn, (png_voidp)
user_mem_ptr, user_malloc_fn, user_free_fn);

After you have these structures, you will need to set up the error handling. When libpng encounters an error, it expects to longjmp() back to your routine. Therefore, you will need to call setjmp() and pass the png_jmpbuf(png_ptr). If you write the file from different routines, you will need to update the png_jmpbuf(png_ptr) every time you enter a new routine that will call a png_*() function. See your documentation of setjmp/longjmp for your compiler for more information on setjmp/longjmp. See the discussion on libpng error handling in the Customizing Libpng section below for more information on the libpng error handling.

if (setjmp(png_jmpbuf(png_ptr)))
{
png_destroy_write_struct(&png_ptr, &info_ptr);
fclose(fp);
return (ERROR);
}
...
return;

If you would rather avoid the complexity of setjmp/longjmp issues, you can compile libpng with PNG_NO_SETJMP, in which case errors will result in a call to PNG_ABORT() which defaults to abort().

You can #define PNG_ABORT() to a function that does something more useful than abort(), as long as your function does not return.

Now you need to set up the output code. The default for libpng is to use the C function fwrite(). If you use this, you will need to pass a valid FILE * in the function png_init_io(). Be sure that the file is opened in binary mode. Again, if you wish to handle writing data in another way, see the discussion on libpng I/O handling in the Customizing Libpng section below.

png_init_io(png_ptr, fp);

If you are embedding your PNG into a datastream such as MNG, and don't want libpng to write the 8-byte signature, or if you have already written the signature in your application, use

png_set_sig_bytes(png_ptr, 8);

to inform libpng that it should not write a signature.

Write callbacks¶

At this point, you can set up a callback function that will be called after each row has been written, which you can use to control a progress meter or the like. It's demonstrated in pngtest.c. You must supply a function

void write_row_callback(png_structp png_ptr, png_uint_32 row,
int pass);
{
/* put your code here */
}

(You can give it another name that you like instead of "write_row_callback")

To inform libpng about your function, use

png_set_write_status_fn(png_ptr, write_row_callback);

When this function is called the row has already been completely processed and it has also been written out. The 'row' and 'pass' refer to the next row to be handled. For the non-interlaced case the row that was just handled is simply one less than the passed in row number, and pass will always be 0. For the interlaced case the same applies unless the row value is 0, in which case the row just handled was the last one from one of the preceding passes. Because interlacing may skip a pass you cannot be sure that the preceding pass is just 'pass-1', if you really need to know what the last pass is record (row,pass) from the callback and use the last recorded value each time.

As with the user transform you can find the output row using the PNG_ROW_FROM_PASS_ROW macro.

You now have the option of modifying how the compression library will run. The following functions are mainly for testing, but may be useful in some cases, like if you need to write PNG files extremely fast and are willing to give up some compression, or if you want to get the maximum possible compression at the expense of slower writing. If you have no special needs in this area, let the library do what it wants by not calling this function at all, as it has been tuned to deliver a good speed/compression ratio. The second parameter to png_set_filter() is the filter method, for which the only valid values are 0 (as of the July 1999 PNG specification, version 1.2) or 64 (if you are writing a PNG datastream that is to be embedded in a MNG datastream). The third parameter is a flag that indicates which filter type(s) are to be tested for each scanline. See the PNG specification for details on the specific filter types.

/* turn on or off filtering, and/or choose
specific filters. You can use either a single
PNG_FILTER_VALUE_NAME or the bitwise OR of one
or more PNG_FILTER_NAME masks.
*/
png_set_filter(png_ptr, 0,
PNG_FILTER_NONE | PNG_FILTER_VALUE_NONE |
PNG_FILTER_SUB | PNG_FILTER_VALUE_SUB |
PNG_FILTER_UP | PNG_FILTER_VALUE_UP |
PNG_FILTER_AVG | PNG_FILTER_VALUE_AVG |
PNG_FILTER_PAETH | PNG_FILTER_VALUE_PAETH|
PNG_ALL_FILTERS);

If an application wants to start and stop using particular filters during compression, it should start out with all of the filters (to ensure that the previous row of pixels will be stored in case it's needed later), and then add and remove them after the start of compression.

If you are writing a PNG datastream that is to be embedded in a MNG datastream, the second parameter can be either 0 or 64.

The png_set_compression_*() functions interface to the zlib compression library, and should mostly be ignored unless you really know what you are doing. The only generally useful call is png_set_compression_level() which changes how much time zlib spends on trying to compress the image data. See the Compression Library (zlib.h and algorithm.txt, distributed with zlib) for details on the compression levels.

#include zlib.h

/* Set the zlib compression level */
png_set_compression_level(png_ptr,
Z_BEST_COMPRESSION);

/* Set other zlib parameters for compressing IDAT */
png_set_compression_mem_level(png_ptr, 8);
png_set_compression_strategy(png_ptr,
Z_DEFAULT_STRATEGY);
png_set_compression_window_bits(png_ptr, 15);
png_set_compression_method(png_ptr, 8);
png_set_compression_buffer_size(png_ptr, 8192)

/* Set zlib parameters for text compression
* If you don't call these, the parameters
* fall back on those defined for IDAT chunks
*/
png_set_text_compression_mem_level(png_ptr, 8);
png_set_text_compression_strategy(png_ptr,
Z_DEFAULT_STRATEGY);
png_set_text_compression_window_bits(png_ptr, 15);
png_set_text_compression_method(png_ptr, 8);

Setting the contents of info for output¶

You now need to fill in the png_info structure with all the data you wish to write before the actual image. Note that the only thing you are allowed to write after the image is the text chunks and the time chunk (as of PNG Specification 1.2, anyway). See png_write_end() and the latest PNG specification for more information on that. If you wish to write them before the image, fill them in now, and flag that data as being valid. If you want to wait until after the data, don't fill them until png_write_end(). For all the fields in png_info and their data types, see png.h. For explanations of what the fields contain, see the PNG specification.

Some of the more important parts of the png_info are:

png_set_IHDR(png_ptr, info_ptr, width, height,
bit_depth, color_type, interlace_type,
compression_type, filter_method)

width - holds the width of the image
in pixels (up to 2^31).

height - holds the height of the image
in pixels (up to 2^31).

bit_depth - holds the bit depth of one of the
image channels.
(valid values are 1, 2, 4, 8, 16
and depend also on the
color_type. See also significant
bits (sBIT) below).

color_type - describes which color/alpha
channels are present.
PNG_COLOR_TYPE_GRAY
(bit depths 1, 2, 4, 8, 16)
PNG_COLOR_TYPE_GRAY_ALPHA
(bit depths 8, 16)
PNG_COLOR_TYPE_PALETTE
(bit depths 1, 2, 4, 8)
PNG_COLOR_TYPE_RGB
(bit_depths 8, 16)
PNG_COLOR_TYPE_RGB_ALPHA
(bit_depths 8, 16)

PNG_COLOR_MASK_PALETTE
PNG_COLOR_MASK_COLOR
PNG_COLOR_MASK_ALPHA

interlace_type - PNG_INTERLACE_NONE or
PNG_INTERLACE_ADAM7

compression_type - (must be
PNG_COMPRESSION_TYPE_DEFAULT)

filter_method - (must be PNG_FILTER_TYPE_DEFAULT
or, if you are writing a PNG to
be embedded in a MNG datastream,
can also be
PNG_INTRAPIXEL_DIFFERENCING)

If you call png_set_IHDR(), the call must appear before any of the other png_set_*() functions, because they might require access to some of the IHDR settings. The remaining png_set_*() functions can be called in any order.

If you wish, you can reset the compression_type, interlace_type, or filter_method later by calling png_set_IHDR() again; if you do this, the width, height, bit_depth, and color_type must be the same in each call.

png_set_PLTE(png_ptr, info_ptr, palette,
num_palette);

palette - the palette for the file
(array of png_color)
num_palette - number of entries in the palette

png_set_gAMA(png_ptr, info_ptr, file_gamma);
png_set_gAMA_fixed(png_ptr, info_ptr, int_file_gamma);

file_gamma - the gamma at which the image was
created (PNG_INFO_gAMA)

int_file_gamma - 100,000 times the gamma at which
the image was created

png_set_cHRM(png_ptr, info_ptr, white_x, white_y, red_x, red_y,
green_x, green_y, blue_x, blue_y)
png_set_cHRM_XYZ(png_ptr, info_ptr, red_X, red_Y, red_Z, green_X,
green_Y, green_Z, blue_X, blue_Y, blue_Z)
png_set_cHRM_fixed(png_ptr, info_ptr, int_white_x, int_white_y,
int_red_x, int_red_y, int_green_x, int_green_y,
int_blue_x, int_blue_y)
png_set_cHRM_XYZ_fixed(png_ptr, info_ptr, int_red_X, int_red_Y,
int_red_Z, int_green_X, int_green_Y, int_green_Z,
int_blue_X, int_blue_Y, int_blue_Z)

{white,red,green,blue}_{x,y}
A color space encoding specified using the chromaticities
of the end points and the white point.

{red,green,blue}_{X,Y,Z}
A color space encoding specified using the encoding end
points - the CIE tristimulus specification of the intended
color of the red, green and blue channels in the PNG RGB
data. The white point is simply the sum of the three end
points.

png_set_sRGB(png_ptr, info_ptr, srgb_intent);

srgb_intent - the rendering intent
(PNG_INFO_sRGB) The presence of
the sRGB chunk means that the pixel
data is in the sRGB color space.
This chunk also implies specific
values of gAMA and cHRM. Rendering
intent is the CSS-1 property that
has been defined by the International
Color Consortium
(http://www.color.org).
It can be one of
PNG_sRGB_INTENT_SATURATION,
PNG_sRGB_INTENT_PERCEPTUAL,
PNG_sRGB_INTENT_ABSOLUTE, or
PNG_sRGB_INTENT_RELATIVE.

png_set_sRGB_gAMA_and_cHRM(png_ptr, info_ptr,
srgb_intent);

srgb_intent - the rendering intent
(PNG_INFO_sRGB) The presence of the
sRGB chunk means that the pixel
data is in the sRGB color space.
This function also causes gAMA and
cHRM chunks with the specific values
that are consistent with sRGB to be
written.

png_set_iCCP(png_ptr, info_ptr, name, compression_type,
profile, proflen);

name - The profile name.

compression_type - The compression type; always
PNG_COMPRESSION_TYPE_BASE for PNG 1.0.
You may give NULL to this argument to
ignore it.

profile - International Color Consortium color
profile data. May contain NULs.

proflen - length of profile data in bytes.

png_set_sBIT(png_ptr, info_ptr, sig_bit);

sig_bit - the number of significant bits for
(PNG_INFO_sBIT) each of the gray, red,
green, and blue channels, whichever are
appropriate for the given color type
(png_color_16)

png_set_tRNS(png_ptr, info_ptr, trans_alpha,
num_trans, trans_color);

trans_alpha - array of alpha (transparency)
entries for palette (PNG_INFO_tRNS)

num_trans - number of transparent entries
(PNG_INFO_tRNS)

trans_color - graylevel or color sample values
(in order red, green, blue) of the
single transparent color for
non-paletted images (PNG_INFO_tRNS)

png_set_hIST(png_ptr, info_ptr, hist);

hist - histogram of palette (array of
png_uint_16) (PNG_INFO_hIST)

png_set_tIME(png_ptr, info_ptr, mod_time);

mod_time - time image was last modified
(PNG_VALID_tIME)

png_set_bKGD(png_ptr, info_ptr, background);

background - background color (of type
png_color_16p) (PNG_VALID_bKGD)

png_set_text(png_ptr, info_ptr, text_ptr, num_text);

text_ptr - array of png_text holding image
comments

text_ptr[i].compression - type of compression used
on "text" PNG_TEXT_COMPRESSION_NONE
PNG_TEXT_COMPRESSION_zTXt
PNG_ITXT_COMPRESSION_NONE
PNG_ITXT_COMPRESSION_zTXt
text_ptr[i].key - keyword for comment. Must contain
1-79 characters.
text_ptr[i].text - text comments for current
keyword. Can be NULL or empty.
text_ptr[i].text_length - length of text string,
after decompression, 0 for iTXt
text_ptr[i].itxt_length - length of itxt string,
after decompression, 0 for tEXt/zTXt
text_ptr[i].lang - language of comment (NULL or
empty for unknown).
text_ptr[i].translated_keyword - keyword in UTF-8 (NULL
or empty for unknown).

Note that the itxt_length, lang, and lang_key
members of the text_ptr structure only exist when the
library is built with iTXt chunk support. Prior to
libpng-1.4.0 the library was built by default without
iTXt support. Also note that when iTXt is supported,
they contain NULL pointers when the "compression"
field contains PNG_TEXT_COMPRESSION_NONE or
PNG_TEXT_COMPRESSION_zTXt.

num_text - number of comments

png_set_sPLT(png_ptr, info_ptr, &palette_ptr,
num_spalettes);

palette_ptr - array of png_sPLT_struct structures
to be added to the list of palettes
in the info structure.
num_spalettes - number of palette structures to be
added.

png_set_oFFs(png_ptr, info_ptr, offset_x, offset_y,
unit_type);

offset_x - positive offset from the left
edge of the screen

offset_y - positive offset from the top
edge of the screen

unit_type - PNG_OFFSET_PIXEL, PNG_OFFSET_MICROMETER

png_set_pHYs(png_ptr, info_ptr, res_x, res_y,
unit_type);

res_x - pixels/unit physical resolution
in x direction

res_y - pixels/unit physical resolution
in y direction

unit_type - PNG_RESOLUTION_UNKNOWN,
PNG_RESOLUTION_METER

png_set_sCAL(png_ptr, info_ptr, unit, width, height)

unit - physical scale units (an integer)

width - width of a pixel in physical scale units

height - height of a pixel in physical scale units
(width and height are doubles)

png_set_sCAL_s(png_ptr, info_ptr, unit, width, height)

unit - physical scale units (an integer)

width - width of a pixel in physical scale units
expressed as a string

height - height of a pixel in physical scale units
(width and height are strings like "2.54")

png_set_unknown_chunks(png_ptr, info_ptr, &unknowns,
num_unknowns)

unknowns - array of png_unknown_chunk
structures holding unknown chunks
unknowns[i].name - name of unknown chunk
unknowns[i].data - data of unknown chunk
unknowns[i].size - size of unknown chunk's data
unknowns[i].location - position to write chunk in file
0: do not write chunk
PNG_HAVE_IHDR: before PLTE
PNG_HAVE_PLTE: before IDAT
PNG_AFTER_IDAT: after IDAT

The "location" member is set automatically according to what part of the output file has already been written. You can change its value after calling png_set_unknown_chunks() as demonstrated in pngtest.c. Within each of the "locations", the chunks are sequenced according to their position in the structure (that is, the value of "i", which is the order in which the chunk was either read from the input file or defined with png_set_unknown_chunks).

A quick word about text and num_text. text is an array of png_text structures. num_text is the number of valid structures in the array. Each png_text structure holds a language code, a keyword, a text value, and a compression type.

The compression types have the same valid numbers as the compression types of the image data. Currently, the only valid number is zero. However, you can store text either compressed or uncompressed, unlike images, which always have to be compressed. So if you don't want the text compressed, set the compression type to PNG_TEXT_COMPRESSION_NONE. Because tEXt and zTXt chunks don't have a language field, if you specify PNG_TEXT_COMPRESSION_NONE or PNG_TEXT_COMPRESSION_zTXt any language code or translated keyword will not be written out.

Until text gets around a few hundred bytes, it is not worth compressing it. After the text has been written out to the file, the compression type is set to PNG_TEXT_COMPRESSION_NONE_WR or PNG_TEXT_COMPRESSION_zTXt_WR, so that it isn't written out again at the end (in case you are calling png_write_end() with the same struct).

The keywords that are given in the PNG Specification are:

Title Short (one line) title or
caption for image

Author Name of image's creator

Description Description of image (possibly long)

Copyright Copyright notice

Creation Time Time of original image creation
(usually RFC 1123 format, see below)

Software Software used to create the image

Disclaimer Legal disclaimer

Warning Warning of nature of content

Source Device used to create the image

Comment Miscellaneous comment; conversion
from other image format

The keyword-text pairs work like this. Keywords should be short simple descriptions of what the comment is about. Some typical keywords are found in the PNG specification, as is some recommendations on keywords. You can repeat keywords in a file. You can even write some text before the image and some after. For example, you may want to put a description of the image before the image, but leave the disclaimer until after, so viewers working over modem connections don't have to wait for the disclaimer to go over the modem before they start seeing the image. Finally, keywords should be full words, not abbreviations. Keywords and text are in the ISO 8859-1 (Latin-1) character set (a superset of regular ASCII) and can not contain NUL characters, and should not contain control or other unprintable characters. To make the comments widely readable, stick with basic ASCII, and avoid machine specific character set extensions like the IBM-PC character set. The keyword must be present, but you can leave off the text string on non-compressed pairs. Compressed pairs must have a text string, as only the text string is compressed anyway, so the compression would be meaningless.

PNG supports modification time via the png_time structure. Two conversion routines are provided, png_convert_from_time_t() for time_t and png_convert_from_struct_tm() for struct tm. The time_t routine uses gmtime(). You don't have to use either of these, but if you wish to fill in the png_time structure directly, you should provide the time in universal time (GMT) if possible instead of your local time. Note that the year number is the full year (e.g. 1998, rather than 98 - PNG is year 2000 compliant!), and that months start with 1.

If you want to store the time of the original image creation, you should use a plain tEXt chunk with the "Creation Time" keyword. This is necessary because the "creation time" of a PNG image is somewhat vague, depending on whether you mean the PNG file, the time the image was created in a non-PNG format, a still photo from which the image was scanned, or possibly the subject matter itself. In order to facilitate machine-readable dates, it is recommended that the "Creation Time" tEXt chunk use RFC 1123 format dates (e.g. "22 May 1997 18:07:10 GMT"), although this isn't a requirement. Unlike the tIME chunk, the "Creation Time" tEXt chunk is not expected to be automatically changed by the software. To facilitate the use of RFC 1123 dates, a function png_convert_to_rfc1123(png_ptr, png_timep) is provided to convert from PNG time to an RFC 1123 format string.

Writing unknown chunks¶

You can use the png_set_unknown_chunks function to queue up chunks for writing. You give it a chunk name, raw data, and a size; that's all there is to it. The chunks will be written by the next following png_write_info_before_PLTE, png_write_info, or png_write_end function. Any chunks previously read into the info structure's unknown-chunk list will also be written out in a sequence that satisfies the PNG specification's ordering rules.

The high-level write interface¶

At this point there are two ways to proceed; through the high-level write interface, or through a sequence of low-level write operations. You can use the high-level interface if your image data is present in the info structure. All defined output transformations are permitted, enabled by the following masks.

PNG_TRANSFORM_IDENTITY No transformation
PNG_TRANSFORM_PACKING Pack 1, 2 and 4-bit samples
PNG_TRANSFORM_PACKSWAP Change order of packed
pixels to LSB first
PNG_TRANSFORM_INVERT_MONO Invert monochrome images
PNG_TRANSFORM_SHIFT Normalize pixels to the
sBIT depth
PNG_TRANSFORM_BGR Flip RGB to BGR, RGBA
to BGRA
PNG_TRANSFORM_SWAP_ALPHA Flip RGBA to ARGB or GA
to AG
PNG_TRANSFORM_INVERT_ALPHA Change alpha from opacity
to transparency
PNG_TRANSFORM_SWAP_ENDIAN Byte-swap 16-bit samples
PNG_TRANSFORM_STRIP_FILLER Strip out filler
bytes (deprecated).
PNG_TRANSFORM_STRIP_FILLER_BEFORE Strip out leading
filler bytes
PNG_TRANSFORM_STRIP_FILLER_AFTER Strip out trailing
filler bytes

If you have valid image data in the info structure (you can use png_set_rows() to put image data in the info structure), simply do this:

png_write_png(png_ptr, info_ptr, png_transforms, NULL)

where png_transforms is an integer containing the bitwise OR of some set of transformation flags. This call is equivalent to png_write_info(), followed the set of transformations indicated by the transform mask, then png_write_image(), and finally png_write_end().

(The final parameter of this call is not yet used. Someday it might point to transformation parameters required by some future output transform.)

You must use png_transforms and not call any png_set_transform() functions when you use png_write_png().

The low-level write interface¶

If you are going the low-level route instead, you are now ready to write all the file information up to the actual image data. You do this with a call to png_write_info().

png_write_info(png_ptr, info_ptr);

Note that there is one transformation you may need to do before png_write_info(). In PNG files, the alpha channel in an image is the level of opacity. If your data is supplied as a level of transparency, you can invert the alpha channel before you write it, so that 0 is fully transparent and 255 (in 8-bit or paletted images) or 65535 (in 16-bit images) is fully opaque, with

png_set_invert_alpha(png_ptr);

This must appear before png_write_info() instead of later with the other transformations because in the case of paletted images the tRNS chunk data has to be inverted before the tRNS chunk is written. If your image is not a paletted image, the tRNS data (which in such cases represents a single color to be rendered as transparent) won't need to be changed, and you can safely do this transformation after your png_write_info() call.

If you need to write a private chunk that you want to appear before the PLTE chunk when PLTE is present, you can write the PNG info in two steps, and insert code to write your own chunk between them:

png_write_info_before_PLTE(png_ptr, info_ptr);
png_set_unknown_chunks(png_ptr, info_ptr, ...);
png_write_info(png_ptr, info_ptr);

After you've written the file information, you can set up the library to handle any special transformations of the image data. The various ways to transform the data will be described in the order that they should occur. This is important, as some of these change the color type and/or bit depth of the data, and some others only work on certain color types and bit depths. Even though each transformation checks to see if it has data that it can do something with, you should make sure to only enable a transformation if it will be valid for the data. For example, don't swap red and blue on grayscale data.

PNG files store RGB pixels packed into 3 or 6 bytes. This code tells the library to strip input data that has 4 or 8 bytes per pixel down to 3 or 6 bytes (or strip 2 or 4-byte grayscale+filler data to 1 or 2 bytes per pixel).

png_set_filler(png_ptr, 0, PNG_FILLER_BEFORE);

where the 0 is unused, and the location is either PNG_FILLER_BEFORE or PNG_FILLER_AFTER, depending upon whether the filler byte in the pixel is stored XRGB or RGBX.

PNG files pack pixels of bit depths 1, 2, and 4 into bytes as small as they can, resulting in, for example, 8 pixels per byte for 1 bit files. If the data is supplied at 1 pixel per byte, use this code, which will correctly pack the pixels into a single byte:

png_set_packing(png_ptr);

PNG files reduce possible bit depths to 1, 2, 4, 8, and 16. If your data is of another bit depth, you can write an sBIT chunk into the file so that decoders can recover the original data if desired.

/* Set the true bit depth of the image data */
if (color_type & PNG_COLOR_MASK_COLOR)
{
sig_bit.red = true_bit_depth;
sig_bit.green = true_bit_depth;
sig_bit.blue = true_bit_depth;
}

else
{
sig_bit.gray = true_bit_depth;
}

if (color_type & PNG_COLOR_MASK_ALPHA)
{
sig_bit.alpha = true_bit_depth;
}

png_set_sBIT(png_ptr, info_ptr, &sig_bit);

If the data is stored in the row buffer in a bit depth other than one supported by PNG (e.g. 3 bit data in the range 0-7 for a 4-bit PNG), this will scale the values to appear to be the correct bit depth as is required by PNG.

png_set_shift(png_ptr, &sig_bit);

PNG files store 16-bit pixels in network byte order (big-endian, ie. most significant bits first). This code would be used if they are supplied the other way (little-endian, i.e. least significant bits first, the way PCs store them):

if (bit_depth > 8)
png_set_swap(png_ptr);

If you are using packed-pixel images (1, 2, or 4 bits/pixel), and you need to change the order the pixels are packed into bytes, you can use:

if (bit_depth < 8)
png_set_packswap(png_ptr);

PNG files store 3 color pixels in red, green, blue order. This code would be used if they are supplied as blue, green, red:

png_set_bgr(png_ptr);

PNG files describe monochrome as black being zero and white being one. This code would be used if the pixels are supplied with this reversed (black being one and white being zero):

png_set_invert_mono(png_ptr);

Finally, you can write your own transformation function if none of the existing ones meets your needs. This is done by setting a callback with

png_set_write_user_transform_fn(png_ptr,
write_transform_fn);

You must supply the function

void write_transform_fn(png_structp png_ptr, png_row_infop
row_info, png_bytep data)

See pngtest.c for a working example. Your function will be called before any of the other transformations are processed. If supported libpng also supplies an information routine that may be called from your callback:

png_get_current_row_number(png_ptr);
png_get_current_pass_number(png_ptr);

This returns the current row passed to the transform. With interlaced images the value returned is the row in the input sub-image image. Use PNG_ROW_FROM_PASS_ROW(row, pass) and PNG_COL_FROM_PASS_COL(col, pass) to find the output pixel (x,y) given an interlaced sub-image pixel (row,col,pass).

The discussion of interlace handling above contains more information on how to use these values.

You can also set up a pointer to a user structure for use by your callback function.

png_set_user_transform_info(png_ptr, user_ptr, 0, 0);

The user_channels and user_depth parameters of this function are ignored when writing; you can set them to zero as shown.

You can retrieve the pointer via the function png_get_user_transform_ptr(). For example:

voidp write_user_transform_ptr =
png_get_user_transform_ptr(png_ptr);

It is possible to have libpng flush any pending output, either manually, or automatically after a certain number of lines have been written. To flush the output stream a single time call:

png_write_flush(png_ptr);

and to have libpng flush the output stream periodically after a certain number of scanlines have been written, call:

png_set_flush(png_ptr, nrows);

Note that the distance between rows is from the last time png_write_flush() was called, or the first row of the image if it has never been called. So if you write 50 lines, and then png_set_flush 25, it will flush the output on the next scanline, and every 25 lines thereafter, unless png_write_flush() is called before 25 more lines have been written. If nrows is too small (less than about 10 lines for a 640 pixel wide RGB image) the image compression may decrease noticeably (although this may be acceptable for real-time applications). Infrequent flushing will only degrade the compression performance by a few percent over images that do not use flushing.

Writing the image data¶

That's it for the transformations. Now you can write the image data. The simplest way to do this is in one function call. If you have the whole image in memory, you can just call png_write_image() and libpng will write the image. You will need to pass in an array of pointers to each row. This function automatically handles interlacing, so you don't need to call png_set_interlace_handling() or call this function multiple times, or any of that other stuff necessary with png_write_rows().

png_write_image(png_ptr, row_pointers);

where row_pointers is:

png_byte *row_pointers[height];

You can point to void or char or whatever you use for pixels.

If you don't want to write the whole image at once, you can use png_write_rows() instead. If the file is not interlaced, this is simple:

png_write_rows(png_ptr, row_pointers,
number_of_rows);

row_pointers is the same as in the png_write_image() call.

If you are just writing one row at a time, you can do this with a single row_pointer instead of an array of row_pointers:

png_bytep row_pointer = row;

png_write_row(png_ptr, row_pointer);

When the file is interlaced, things can get a good deal more complicated. The only currently (as of the PNG Specification version 1.2, dated July 1999) defined interlacing scheme for PNG files is the "Adam7" interlace scheme, that breaks down an image into seven smaller images of varying size. libpng will build these images for you, or you can do them yourself. If you want to build them yourself, see the PNG specification for details of which pixels to write when.

If you don't want libpng to handle the interlacing details, just use png_set_interlace_handling() and call png_write_rows() the correct number of times to write all the sub-images (png_set_interlace_handling() returns the number of sub-images.)

If you want libpng to build the sub-images, call this before you start writing any rows:

number_of_passes = png_set_interlace_handling(png_ptr);

This will return the number of passes needed. Currently, this is seven, but may change if another interlace type is added.

Then write the complete image number_of_passes times.

png_write_rows(png_ptr, row_pointers, number_of_rows);

Think carefully before you write an interlaced image. Typically code that reads such images reads all the image data into memory, uncompressed, before doing any processing. Only code that can display an image on the fly can take advantage of the interlacing and even then the image has to be exactly the correct size for the output device, because scaling an image requires adjacent pixels and these are not available until all the passes have been read.

If you do write an interlaced image you will hardly ever need to handle the interlacing yourself. Call png_set_interlace_handling() and use the approach described above.

The only time it is conceivable that you will really need to write an interlaced image pass-by-pass is when you have read one pass by pass and made some pixel-by-pixel transformation to it, as described in the read code above. In this case use the PNG_PASS_ROWS and PNG_PASS_COLS macros to determine the size of each sub-image in turn and simply write the rows you obtained from the read code.

Finishing a sequential write¶

After you are finished writing the image, you should finish writing the file. If you are interested in writing comments or time, you should pass an appropriately filled png_info pointer. If you are not interested, you can pass NULL.

png_write_end(png_ptr, info_ptr);

When you are done, you can free all memory used by libpng like this:

png_destroy_write_struct(&png_ptr, &info_ptr);

It is also possible to individually free the info_ptr members that point to libpng-allocated storage with the following function:

png_free_data(png_ptr, info_ptr, mask, seq)

mask - identifies data to be freed, a mask
containing the bitwise OR of one or
more of
PNG_FREE_PLTE, PNG_FREE_TRNS,
PNG_FREE_HIST, PNG_FREE_ICCP,
PNG_FREE_PCAL, PNG_FREE_ROWS,
PNG_FREE_SCAL, PNG_FREE_SPLT,
PNG_FREE_TEXT, PNG_FREE_UNKN,
or simply PNG_FREE_ALL

seq - sequence number of item to be freed
(-1 for all items)

This function may be safely called when the relevant storage has already been freed, or has not yet been allocated, or was allocated by the user and not by libpng, and will in those cases do nothing. The "seq" parameter is ignored if only one item of the selected data type, such as PLTE, is allowed. If "seq" is not -1, and multiple items are allowed for the data type identified in the mask, such as text or sPLT, only the n'th item in the structure is freed, where n is "seq".

If you allocated data such as a palette that you passed in to libpng with png_set_*, you must not free it until just before the call to png_destroy_write_struct().

The default behavior is only to free data that was allocated internally by libpng. This can be changed, so that libpng will not free the data, or so that it will free data that was allocated by the user with png_malloc() or png_calloc() and passed in via a png_set_*() function, with

png_data_freer(png_ptr, info_ptr, freer, mask)

freer - one of
PNG_DESTROY_WILL_FREE_DATA
PNG_SET_WILL_FREE_DATA
PNG_USER_WILL_FREE_DATA

mask - which data elements are affected
same choices as in png_free_data()

For example, to transfer responsibility for some data from a read structure to a write structure, you could use

png_data_freer(read_ptr, read_info_ptr,
PNG_USER_WILL_FREE_DATA,
PNG_FREE_PLTE|PNG_FREE_tRNS|PNG_FREE_hIST)

png_data_freer(write_ptr, write_info_ptr,
PNG_DESTROY_WILL_FREE_DATA,
PNG_FREE_PLTE|PNG_FREE_tRNS|PNG_FREE_hIST)

thereby briefly reassigning responsibility for freeing to the user but immediately afterwards reassigning it once more to the write_destroy function. Having done this, it would then be safe to destroy the read structure and continue to use the PLTE, tRNS, and hIST data in the write structure.

This function only affects data that has already been allocated. You can call this function before calling after the png_set_*() functions to control whether the user or png_destroy_*() is supposed to free the data. When the user assumes responsibility for libpng-allocated data, the application must use png_free() to free it, and when the user transfers responsibility to libpng for data that the user has allocated, the user must have used png_malloc() or png_calloc() to allocate it.

If you allocated text_ptr.text, text_ptr.lang, and text_ptr.translated_keyword separately, do not transfer responsibility for freeing text_ptr to libpng, because when libpng fills a png_text structure it combines these members with the key member, and png_free_data() will free only text_ptr.key. Similarly, if you transfer responsibility for free'ing text_ptr from libpng to your application, your application must not separately free those members. For a more compact example of writing a PNG image, see the file example.c.

Custom chunks¶

If you need to read or write custom chunks, you may need to get deeper into the libpng code. The library now has mechanisms for storing and writing chunks of unknown type; you can even declare callbacks for custom chunks. However, this may not be good enough if the library code itself needs to know about interactions between your chunk and existing `intrinsic' chunks.

If you need to write a new intrinsic chunk, first read the PNG specification. Acquire a first level of understanding of how it works. Pay particular attention to the sections that describe chunk names, and look at how other chunks were designed, so you can do things similarly. Second, check out the sections of libpng that read and write chunks. Try to find a chunk that is similar to yours and use it as a template. More details can be found in the comments inside the code. It is best to handle private or unknown chunks in a generic method, via callback functions, instead of by modifying libpng functions. This is illustrated in pngtest.c, which uses a callback function to handle a private "vpAg" chunk and the new "sTER" chunk, which are both unknown to libpng.

If you wish to write your own transformation for the data, look through the part of the code that does the transformations, and check out some of the simpler ones to get an idea of how they work. Try to find a similar transformation to the one you want to add and copy off of it. More details can be found in the comments inside the code itself.

Configuring for 16-bit platforms¶

You will want to look into zconf.h to tell zlib (and thus libpng) that it cannot allocate more then 64K at a time. Even if you can, the memory won't be accessible. So limit zlib and libpng to 64K by defining MAXSEG_64K.

Configuring for DOS¶

For DOS users who only have access to the lower 640K, you will have to limit zlib's memory usage via a png_set_compression_mem_level() call. See zlib.h or zconf.h in the zlib library for more information.

Configuring for Medium Model¶

Libpng's support for medium model has been tested on most of the popular compilers. Make sure MAXSEG_64K gets defined, USE_FAR_KEYWORD gets defined, and FAR gets defined to far in pngconf.h, and you should be all set. Everything in the library (except for zlib's structure) is expecting far data. You must use the typedefs with the p or pp on the end for pointers (or at least look at them and be careful). Make note that the rows of data are defined as png_bytepp, which is an "unsigned char far * far *".

Configuring for gui/windowing platforms:¶

You will need to write new error and warning functions that use the GUI interface, as described previously, and set them to be the error and warning functions at the time that png_create_*_struct() is called, in order to have them available during the structure initialization. They can be changed later via png_set_error_fn(). On some compilers, you may also have to change the memory allocators (png_malloc, etc.).

Configuring for compiler xxx:¶

All includes for libpng are in pngconf.h. If you need to add, change or delete an include, this is the place to do it. The includes that are not needed outside libpng are placed in pngpriv.h, which is only used by the routines inside libpng itself. The files in libpng proper only include pngpriv.h and png.h, which %14%in turn includes pngconf.h. in turn includes pngconf.h and, as of libpng-1.5.0, pnglibconf.h. As of libpng-1.5.0, pngpriv.h also includes three other private header files, pngstruct.h, pnginfo.h, and pngdebug.h, which contain material that previously appeared in the public headers.

Configuring zlib:¶

There are special functions to configure the compression. Perhaps the most useful one changes the compression level, which currently uses input compression values in the range 0 - 9. The library normally uses the default compression level (Z_DEFAULT_COMPRESSION = 6). Tests have shown that for a large majority of images, compression values in the range 3-6 compress nearly as well as higher levels, and do so much faster. For online applications it may be desirable to have maximum speed (Z_BEST_SPEED = 1). With versions of zlib after v0.99, you can also specify no compression (Z_NO_COMPRESSION = 0), but this would create files larger than just storing the raw bitmap. You can specify the compression level by calling:

#include zlib.h
png_set_compression_level(png_ptr, level);

Another useful one is to reduce the memory level used by the library. The memory level defaults to 8, but it can be lowered if you are short on memory (running DOS, for example, where you only have 640K). Note that the memory level does have an effect on compression; among other things, lower levels will result in sections of incompressible data being emitted in smaller stored blocks, with a correspondingly larger relative overhead of up to 15% in the worst case.

#include zlib.h
png_set_compression_mem_level(png_ptr, level);

The other functions are for configuring zlib. They are not recommended for normal use and may result in writing an invalid PNG file. See zlib.h for more information on what these mean.

#include zlib.h
png_set_compression_strategy(png_ptr,
strategy);

png_set_compression_window_bits(png_ptr,
window_bits);

png_set_compression_method(png_ptr, method);

png_set_compression_buffer_size(png_ptr, size);

As of libpng version 1.5.4, additional APIs became available to set these separately for non-IDAT compressed chunks such as zTXt, iTXt, and iCCP:

#include zlib.h
#if PNG_LIBPNG_VER >= 10504
png_set_text_compression_level(png_ptr, level);

png_set_text_compression_mem_level(png_ptr, level);

png_set_text_compression_strategy(png_ptr,
strategy);

png_set_text_compression_window_bits(png_ptr,
window_bits);

png_set_text_compression_method(png_ptr, method);
#endif

Controlling row filtering¶

If you want to control whether libpng uses filtering or not, which filters are used, and how it goes about picking row filters, you can call one of these functions. The selection and configuration of row filters can have a significant impact on the size and encoding speed and a somewhat lesser impact on the decoding speed of an image. Filtering is enabled by default for RGB and grayscale images (with and without alpha), but not for paletted images nor for any images with bit depths less than 8 bits/pixel.

The 'method' parameter sets the main filtering method, which is currently only '0' in the PNG 1.2 specification. The 'filters' parameter sets which filter(s), if any, should be used for each scanline. Possible values are PNG_ALL_FILTERS and PNG_NO_FILTERS to turn filtering on and off, respectively.

Individual filter types are PNG_FILTER_NONE, PNG_FILTER_SUB, PNG_FILTER_UP, PNG_FILTER_AVG, PNG_FILTER_PAETH, which can be bitwise ORed together with '|' to specify one or more filters to use. These filters are described in more detail in the PNG specification. If you intend to change the filter type during the course of writing the image, you should start with flags set for all of the filters you intend to use so that libpng can initialize its internal structures appropriately for all of the filter types. (Note that this means the first row must always be adaptively filtered, because libpng currently does not allocate the filter buffers until png_write_row() is called for the first time.)

filters = PNG_FILTER_NONE | PNG_FILTER_SUB
PNG_FILTER_UP | PNG_FILTER_AVG |
PNG_FILTER_PAETH | PNG_ALL_FILTERS;

png_set_filter(png_ptr, PNG_FILTER_TYPE_BASE,
filters);
The second parameter can also be
PNG_INTRAPIXEL_DIFFERENCING if you are
writing a PNG to be embedded in a MNG
datastream. This parameter must be the
same as the value of filter_method used
in png_set_IHDR().

It is also possible to influence how libpng chooses from among the available filters. This is done in one or both of two ways - by telling it how important it is to keep the same filter for successive rows, and by telling it the relative computational costs of the filters.

double weights[3] = {1.5, 1.3, 1.1},
costs[PNG_FILTER_VALUE_LAST] =
{1.0, 1.3, 1.3, 1.5, 1.7};

png_set_filter_heuristics(png_ptr,
PNG_FILTER_HEURISTIC_WEIGHTED, 3,
weights, costs);

The weights are multiplying factors that indicate to libpng that the row filter should be the same for successive rows unless another row filter is that many times better than the previous filter. In the above example, if the previous 3 filters were SUB, SUB, NONE, the SUB filter could have a "sum of absolute differences" 1.5 x 1.3 times higher than other filters and still be chosen, while the NONE filter could have a sum 1.1 times higher than other filters and still be chosen. Unspecified weights are taken to be 1.0, and the specified weights should probably be declining like those above in order to emphasize recent filters over older filters.

The filter costs specify for each filter type a relative decoding cost to be considered when selecting row filters. This means that filters with higher costs are less likely to be chosen over filters with lower costs, unless their "sum of absolute differences" is that much smaller. The costs do not necessarily reflect the exact computational speeds of the various filters, since this would unduly influence the final image size.

Note that the numbers above were invented purely for this example and are given only to help explain the function usage. Little testing has been done to find optimum values for either the costs or the weights.

Removing unwanted object code¶

There are a bunch of #define's in pngconf.h that control what parts of libpng are compiled. All the defines end in _SUPPORTED. If you are never going to use a capability, you can change the #define to #undef before recompiling libpng and save yourself code and data space, or you can turn off individual capabilities with defines that begin with PNG_NO_.

In libpng-1.5.0 and later, the #define's are in pnglibconf.h instead.

You can also turn all of the transforms and ancillary chunk capabilities off en masse with compiler directives that define PNG_NO_READ[or WRITE]_TRANSFORMS, or PNG_NO_READ[or WRITE]_ANCILLARY_CHUNKS, or all four, along with directives to turn on any of the capabilities that you do want. The PNG_NO_READ[or WRITE]_TRANSFORMS directives disable the extra transformations but still leave the library fully capable of reading and writing PNG files with all known public chunks. Use of the PNG_NO_READ[or WRITE]_ANCILLARY_CHUNKS directive produces a library that is incapable of reading or writing ancillary chunks. If you are not using the progressive reading capability, you can turn that off with PNG_NO_PROGRESSIVE_READ (don't confuse this with the INTERLACING capability, which you'll still have).

All the reading and writing specific code are in separate files, so the linker should only grab the files it needs. However, if you want to make sure, or if you are building a stand alone library, all the reading files start with "pngr" and all the writing files start with "pngw". The files that don't match either (like png.c, pngtrans.c, etc.) are used for both reading and writing, and always need to be included. The progressive reader is in pngpread.c

If you are creating or distributing a dynamically linked library (a .so or DLL file), you should not remove or disable any parts of the library, as this will cause applications linked with different versions of the library to fail if they call functions not available in your library. The size of the library itself should not be an issue, because only those sections that are actually used will be loaded into memory.

Requesting debug printout¶

The macro definition PNG_DEBUG can be used to request debugging printout. Set it to an integer value in the range 0 to 3. Higher numbers result in increasing amounts of debugging information. The information is printed to the "stderr" file, unless another file name is specified in the PNG_DEBUG_FILE macro definition.

When PNG_DEBUG > 0, the following functions (macros) become available:

png_debug(level, message)
png_debug1(level, message, p1)
png_debug2(level, message, p1, p2)

in which "level" is compared to PNG_DEBUG to decide whether to print the message, "message" is the formatted string to be printed, and p1 and p2 are parameters that are to be embedded in the string according to printf-style formatting directives. For example,

png_debug1(2, "foo=%d0, foo);

is expanded to

if (PNG_DEBUG > 2)
fprintf(PNG_DEBUG_FILE, "foo=%d0, foo);

When PNG_DEBUG is defined but is zero, the macros aren't defined, but you can still use PNG_DEBUG to control your own debugging:

#ifdef PNG_DEBUG
fprintf(stderr, ...
#endif

When PNG_DEBUG = 1, the macros are defined, but only png_debug statements having level = 0 will be printed. There aren't any such statements in this version of libpng, but if you insert some they will be printed.