PNG to AVIF. Prebuilt & custom-built sharp outputs larger file size w/ bigger `effort`

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: macOS 13.0
    CPU: (10) arm64 Apple M1 Pro
    Memory: 689.75 MB / 16.00 GB
    Shell: 5.8.1 - /bin/zsh
  Binaries:
    Node: 18.11.0 - /opt/homebrew/bin/node
    Yarn: 1.22.19 - /opt/homebrew/bin/yarn
    npm: 8.19.2 - /opt/homebrew/bin/npm
  npmPackages:
    sharp: 0.31.1 => 0.31.1 

or another system bearing the same problem:

  System:
    OS: Linux 5.15 Amazon Linux 2
    CPU: (8) arm64 unknown
    Memory: 6.85 GB / 7.76 GB
    Container: Yes
    Shell: 4.2.46 - /bin/bash
  Binaries:
    Node: 16.17.1 - /var/lang/bin/node
    npm: 8.15.0 - /var/lang/bin/npm

What are the steps to reproduce?

1. Install sharp with pre-built binaries
2. Use image from example.png with the provided code snippet.
3. Observe the console to see: the higher effort results in bigger file size

On my device(s), it prints as:

Effort 0 size: 342380
Effort 4 size: 356314
Effort 9 size: 361679

What is the expected behaviour?

Increase in effort doesn’t make the size bigger (at least keeps it the same as with the lower effort)

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

const checkEffortAndSize = async (effort) => {
  const cmd = sharp(`./example.png`)
  const buffer = await cmd
    .avif({ quality: 64, effort })
    .toBuffer()

  console.info(`Effort ${effort} size: ${buffer.byteLength}`)
}

await Promise.all([checkEffortAndSize(0), checkEffortAndSize(9)])

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

example.zip

[lovell]

Thanks for the update, and no worries, as you say this is a question for the libaom maintainers.

The small file size difference probably relates to chroma subsampling (sharp defaults to 4:4:4, i.e. none, whereas libheif defaults to 4:2:0).

[o-alexandrov]

Just for the record, libaom maintainers think libaom is working as expected.

• they think cpu-used (their internal parameter that is connected to sharp’s effort parameter) should affect the processing quality, which in turn results in a different output