Regression in v0.31.3: linear filter is not applied correctly to 16-bit images

[SaraKaminska]

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.

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.

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

System:
OS: Windows 10 10.0.19044
CPU: (8) x64 Intel(R) Core(TM) i7-8565U CPU @ 1.80GHz
Memory: 4.98 GB / 15.75 GB
Binaries:
Node: 19.6.0 - C:\Program Files\nodejs\node.EXE
Yarn: 1.22.19 - ~\AppData\Roaming\npm\yarn.CMD
npm: 9.4.0 - C:\Program Files\nodejs\npm.CMD
npmPackages:
sharp: ^0.32.0 => 0.32.0

What are the steps to reproduce?

1. Read a 16-bit PNG image with Sharp.
2. Apply the linear filter with any custom value.
3. Apply toColorspace('grey16') to ensure the space is maintained.
4. Save the image to file.

What is the expected behavior?

The expected behavior is that the linear filter is applied to the image with the expected result. Instead the output image is completely black. This issue occurred in the version 0.31.3. My guess is that the reson for this issue is setting the option to "uchar" when applying linear.
See src/operations.cc row 148:
return RemoveAlpha(image).linear(a, b, VImage::option()->set("uchar", TRUE)).bandjoin(alpha);

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

const sharp = require("sharp");

const adjusted = sharp('16-bit.png')
  .linear(1.1, 10)
  .toColorspace('grey16')
  .toFile('gamma-16-bit.png')

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

Original

Output

[lovell]

Thanks for reporting this, you're quite right about the uchar in there that's causing it.

This was a regression caused by #3468 in v0.31.3 so you might be able to workaround this by pinning to v0.31.2.

[SaraKaminska]

Thank you for the fast answer. Yes, will continue working on an older version until this bug is fixed.

[lovell]

Commit b9c3851 fixes this and adds a test that would previously have failed to help prevent regression. This will be in v0.32.1.

[lovell]

v0.32.1 now available.