Error saving PNG after CAST

[jfconekta]

Bug report

Describe the bug
Saving a PNG with the function pngsave creates always an 8bit image, even when we change the format to VIPS_FORMAT_USHORT

To Reproduce
Steps to reproduce the behavior:

1. Read an image
2. Change the format of the image using Cast
3. Save the new image
4. The created image using the new version always create a 8bit image

#include <vips/vips8>

using namespace vips;

int main (int argc, char **argv)
{ 
  if (VIPS_INIT (argv[0])) 
    vips_error_exit (NULL);
  
  VImage out;
  VImage in = VImage::new_from_file (argv[1],
    VImage::option ()->set ("access", VIPS_ACCESS_SEQUENTIAL));
  
  double avg = in.avg ();
  
  printf ("avg = %g\n", avg);
  printf ("width = %d\n", in.width ());
  
  in = VImage::new_from_file (argv[1],
    VImage::option ()->set ("access", VIPS_ACCESS_SEQUENTIAL));
  
  out=in;
  out = out.cast(VIPS_FORMAT_USHORT);
  out.pngsave(argv[2], VImage::option()->set("compression",0));
  
  vips_shutdown ();
  
  return (0);
}

Expected behavior
New image should be 16bit as in previous versions

Actual behavior
A clear and concise description of what actually happen.

Screenshots

Environment

• OS: Ubuntu 22.04
• Vips: vips-8.12.1 / vips-8.14.3

Additional context
With previous versions vips-8.9.1 in ubuntu 20.04 the new image was 16-bits as expected.

[jcupitt]

Hi @jfconekta,

We improved the rules for when to write 16-bit PNG in 8.10 (I think it was). You now need to set the image interpretation to RGB16 or GREY16 to write a 16-bit PNG.

In your code, try:

  out = out.cast(VIPS_FORMAT_USHORT).copy(VImage::option ()->set ("interpretation", "grey16"));

Of course the image will look very dark, I expect you know.

[jfconekta]

Thanks a lot, is perfectly clear now.