How to read and write PNG file using libpng. Covers trivial method calls like png_set_filler.

libpng_test.c

/*
 * A simple libpng example program
 * http://zarb.org/~gc/html/libpng.html
 *
 * Modified by Yoshimasa Niwa to make it much simpler
 * and support all defined color_type.
 *
 * To build, use the next instruction on OS X.
 * $ brew install libpng
 * $ clang -lz -lpng16 libpng_test.c
 *
 * Copyright 2002-2010 Guillaume Cottenceau.
 *
 * This software may be freely redistributed under the terms
 * of the X11 license.
 *
 */

#include <stdlib.h>
#include <stdio.h>
#include <png.h>

int width, height;
png_byte color_type;
png_byte bit_depth;
png_bytep *row_pointers = NULL;

void read_png_file(char *filename) {
  FILE *fp = fopen(filename, "rb");

  png_structp png = png_create_read_struct(PNG_LIBPNG_VER_STRING, NULL, NULL, NULL);
  if(!png) abort();

  png_infop info = png_create_info_struct(png);
  if(!info) abort();

  if(setjmp(png_jmpbuf(png))) abort();

  png_init_io(png, fp);

  png_read_info(png, info);

  width      = png_get_image_width(png, info);
  height     = png_get_image_height(png, info);
  color_type = png_get_color_type(png, info);
  bit_depth  = png_get_bit_depth(png, info);

  // Read any color_type into 8bit depth, RGBA format.
  // See http://www.libpng.org/pub/png/libpng-manual.txt

  if(bit_depth == 16)
    png_set_strip_16(png);

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

  // PNG_COLOR_TYPE_GRAY_ALPHA is always 8 or 16bit depth.
  if(color_type == PNG_COLOR_TYPE_GRAY && bit_depth < 8)
    png_set_expand_gray_1_2_4_to_8(png);

  if(png_get_valid(png, info, PNG_INFO_tRNS))
    png_set_tRNS_to_alpha(png);

  // These color_type don't have an alpha channel then fill it with 0xff.
  if(color_type == PNG_COLOR_TYPE_RGB ||
     color_type == PNG_COLOR_TYPE_GRAY ||
     color_type == PNG_COLOR_TYPE_PALETTE)
    png_set_filler(png, 0xFF, PNG_FILLER_AFTER);

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

  png_read_update_info(png, info);

  if (row_pointers) abort();

  row_pointers = (png_bytep*)malloc(sizeof(png_bytep) * height);
  for(int y = 0; y < height; y++) {
    row_pointers[y] = (png_byte*)malloc(png_get_rowbytes(png,info));
  }

  png_read_image(png, row_pointers);

  fclose(fp);

  png_destroy_read_struct(&png, &info, NULL);
}

void write_png_file(char *filename) {
  int y;

  FILE *fp = fopen(filename, "wb");
  if(!fp) abort();

  png_structp png = png_create_write_struct(PNG_LIBPNG_VER_STRING, NULL, NULL, NULL);
  if (!png) abort();

  png_infop info = png_create_info_struct(png);
  if (!info) abort();

  if (setjmp(png_jmpbuf(png))) abort();

  png_init_io(png, fp);

  // Output is 8bit depth, RGBA format.
  png_set_IHDR(
    png,
    info,
    width, height,
    8,
    PNG_COLOR_TYPE_RGBA,
    PNG_INTERLACE_NONE,
    PNG_COMPRESSION_TYPE_DEFAULT,
    PNG_FILTER_TYPE_DEFAULT
  );
  png_write_info(png, info);

  // To remove the alpha channel for PNG_COLOR_TYPE_RGB format,
  // Use png_set_filler().
  //png_set_filler(png, 0, PNG_FILLER_AFTER);

  if (!row_pointers) abort();

  png_write_image(png, row_pointers);
  png_write_end(png, NULL);

  for(int y = 0; y < height; y++) {
    free(row_pointers[y]);
  }
  free(row_pointers);

  fclose(fp);

  png_destroy_write_struct(&png, &info);
}

void process_png_file() {
  for(int y = 0; y < height; y++) {
    png_bytep row = row_pointers[y];
    for(int x = 0; x < width; x++) {
      png_bytep px = &(row[x * 4]);
      // Do something awesome for each pixel here...
      //printf("%4d, %4d = RGBA(%3d, %3d, %3d, %3d)\n", x, y, px[0], px[1], px[2], px[3]);
    }
  }
}

int main(int argc, char *argv[]) {
  if(argc != 3) abort();

  read_png_file(argv[1]);
  process_png_file();
  write_png_file(argv[2]);

  return 0;
}

Thanks :3

It might be a good idea to call png_destroy_read_struct to avoid leaking memory.

TO avoid memory leaks just add:
in function:

write_png_file()

at the end add:

if (png && info)
        png_destroy_write_struct(&png, &info);

in function

 read_png_file()

at the end add:

png_destroy_read_struct(&png, &info, NULL);
    png=NULL;
    info=NULL;

Thanks for your sample.
I want to ask something about row data. How can I get only row data in png image. No header no info only image data. I want to get a pointer to (char *) type?
Has any idea about it ?

You have image data in row_pointers and you can access it directelly in process_png_file() function

Great code. I am using with xlib and it is working well.

The most useful part of this was process_png_file(); more specifically the lines png_bytep row = row_pointers[y]; and png_bytep px = &(row[x * 4]);. Thanks!

hello,I want to use libpng to parse png file. I have downloaded the libpng, and build it successfully. Could you tell me how to put your c file into the libpng project?

Great when i use clang.
But then i need to compile it using g++-8 and the error is: fatal error: png.h: No such file or directory

I'm on mac and I did:
$ brew install libpng

Please help to solve it.

Hello.
I have used a picture.png to write in a file, and then saved that file successfully as another image (picture1.png). But I dont want to save to a file, i want to send it directly to a usb thermal printer which is connected to my raspberry pi (/dev/usb/lp0). When i do that it doesn't print as an image but some weird Chinese letters.
If anyone can help i would be very grateful.

Very nice example usage. Thank you.

This is awesome. Thank you so much.

png_get_rowbytes was the same, cache that would be better.

I realized my old gist has many comments... and so I decided to update it to cover next things.

• Destroy read and write structs.
• Update comments to use -lpng16, which is current Homebrew provided libpng version.
• Check row_pointers for read and write.

*> Cool, but the png.h is impossible to compile, which makes your code useless.

@Soraiko png.h is impossible to compile, because it is just a header. If you install cmake, you can download and build libpng library from sources.

• download libpng from sourceforge,
• download cmake and install,
• go to libpng sources
• make a build directory
• run cmake in that directory, point at parent directory
• select your host environment (visual studio, mingw, linux gcc) and target environment

that's it, I've done it a million times. On linux it is even easier.

• wget sourceforgeurl.targ.z
• tar zvxf package.tar.gz
• cd package
• mkdir build
• cd build
• cmake ..
• make -j 4
• done

Somehow for me, the script won't execute this line png_read_info(png, info); and png_write_info(png, info);
Can anyone explain what's happening?

Thanks for this snippet! It helped me a lot :)

The original code checks for a valid PNG signature before trying to read the file. You claim to have simplified the code. Omitting sanity checks is a bad way to simplify.

You should add

	//check header magic
	{
		unsigned char header[8];
		fread(header, 1, 8, fp);

		if (png_sig_cmp(header, 0, 8))
		{
			std::cerr << "File " << filename << " is not recognized as a PNG file" << std::endl;
			fclose(fp);
			return;
		}
	}

right before png_structp png = png_create_read_struct(PNG_LIBPNG_VER_STRING, NULL, NULL, NULL); in read_png_file

and

png_set_sig_bytes(png, 8); // we already read the 8 signature bytes right before png_read_info(png, info);

(FYI: I have #include <iostream> that's why i can use std::cerr)

In the setjmp() conditional, it's a good practice to also destroy the read struct and close the FILE*.

if (setjmp(png_jmpbuf(png))) {
    png_destroy_read_struct(&png, &info, NULL);
    fclose(fp);
    return NULL;
}