Tiff bitdepth ignored

[dnsbty]

Possible bug

Is this a possible bug in a feature of sharp, unrelated to installation?

• Running npm install sharp completes without error.
• Running node -e "require('sharp')" completes without error.

If you cannot confirm both of these, please open an installation issue instead.

Are you using the latest version of sharp?

• I am using the latest version of sharp as reported by npm view sharp dist-tags.latest.

If you cannot confirm this, please upgrade to the latest version and try again before opening an issue.

If you are using another package which depends on a version of sharp that is not the latest, please open an issue against that package instead.

What is the output of running npx envinfo --binaries --system --npmPackages=sharp --npmGlobalPackages=sharp?

System:
OS: macOS 13.5.2
CPU: (10) arm64 Apple M1 Max
Memory: 390.08 MB / 64.00 GB
Shell: 5.9 - /bin/zsh
Binaries:
Node: 18.16.1 - ~/.asdf/installs/nodejs/18.16.1/bin/node
npm: 9.7.2 - ~/.asdf/plugins/nodejs/shims/npm
npmPackages:
sharp: ^0.32.6 => 0.32.6

What are the steps to reproduce?

await sharp("input.gif")
  .tiff({ bitdepth: 1 })
  .toFile("output.tif")

What is the expected behaviour?

When I run exiftool output.tif I would expect to see the following:

Bits Per Sample                 : 1

Instead I see this:

Bits Per Sample                 : 8 8 8

Please provide a minimal, standalone code sample, without other dependencies, that demonstrates this problem

Reproduction

Please provide sample image(s) that help explain this problem

This is the input gif I'm using:

And this is the tiff image that is outputted (compressed since Github doesn't allow tiff uploads):

output.tif.zip

[dnsbty]

In attempting to debug this myself, I tried dropping down to C++ to see if I could find anything. I created this script:

#include <vips/vips8>
using namespace vips;

int
main (int argc, char **argv)
{ 
  if (VIPS_INIT (argv[0])) 
    vips_error_exit (NULL);
  
  if (argc != 3)
    vips_error_exit ("usage: %s input-file output-file", argv[0]);

  VImage in = VImage::new_from_file (argv[1],
    VImage::option ()->set ("access", VIPS_ACCESS_SEQUENTIAL));
  
  in.tiffsave (argv[2], VImage::option ()->set ("bitdepth", 1));
  
  vips_shutdown ();
  
  return (0);
}

When I compile this:

g++ -g -Wall test.cc `pkg-config vips-cpp --cflags --libs` -o test

And then run it:

./test input.gif output.tif

I get this warning:

(process:60900): VIPS-WARNING **: 23:48:04.401: can only set bitdepth for 1-band uchar and 3-band float lab -- disabling bitdepth

I'm guessing this warning is being suppressed by sharp, and is probably the root of the issue. I'm having a hard time figuring out what this actually means though. Any pointers would be greatly appreciated.

[lovell]

The addition of .removeAlpha().toColourspace('b-w') to a pipeline will reduce the input to a single, greyscale band suitable for use with a bitdepth of 1.

https://sharp.pixelplumbing.com/api-channel#removealpha
https://sharp.pixelplumbing.com/api-colour#tocolourspace

sharp supports the NODE_DEBUG feature of Node.js to view warnings, i.e. set the NODE_DEBUG=sharp environment variable. Perhaps this should be an error condition rather than a warning?

I think the miniswhite setting will support an optional alpha channel too. The underlying logic in libvips can be seen at https://github.com/libvips/libvips/blob/0819eccda6abe097c5da16edf8d028c7c0e6b595/libvips/foreign/vips2tiff.c#L1446-L1476

See also #3486 #3655

[dnsbty]

Thank you very much for writing that up. I'm not sure why those other issues didn't turn up when I searched. The NODE_DEBUG tip was especially helpful.

With that I was able to build up a working pipeline, but I think I found an incompatibility. When using .withMetadata({ density: 200 }) as recommended in the docs when working with inches, the bitdepth is ignored once again with the following warnings showing up when using the NODE_DEBUG environment variable:

SHARP 87527: profile incompatible with image
SHARP 87527: can only set bitdepth for 1-band uchar and 3-band float lab -- disabling bitdepth

This is the code I used. I've also updated the reproduction repository

await sharp(inputFile)
  .removeAlpha()
  .toColourspace('b-w')
  .withMetadata({ density: 200 })
  .tiff({ bitdepth: 1, compression: "lzw", predictor: "none" })
  .toFile(outputFile)

[dnsbty]

I should probably note that this has an easy workaround by providing the xres and yres options (converted from inches to millimeters) to the tiff function, but I thought it might be worth mentioning.

[lovell]

If you still need help here, I do intend to revisit this issue once v0.33.0 is made available. There are a number of metadata-related changes in libvips v8.15.0 that we're about to take advantage of in sharp so it might all "just work" ™️ .

[lovell]

@dnsbty Were you able to make any progress with this?

[dnsbty]

Because of the xres + yres workaround that I found, I didn't really need the fix anymore, so I haven't looked into it anymore since then.