As the inverse of #8, it would be nice to get the raw pixels of an image for the purpose of detecting a color pallet.

For example, perhaps a new method, getImageData(), which is how it would be solved with <canvas>:

import { readFileSync } from 'fs'
import palette = from 'get-rgba-palette'
import { Image } from '@napi-rs/image'

const decoder = new Transformer(readFileSync('./img.jpg'))
const pixels = decoder.getImageData()
const colors = palette(pixels, 5)

https://github.com/mattdesl/get-rgba-palette

I'm using @napi-rs/image in my project to compress images, and I ran into a bug when trying to compress PNG images with pngquant. When I try to compress very small PNG images, the program crashes. I also tried using imagemin-pngquant and didn't encounter this issue, so I suspect it might be related to the implementation of @napi-rs/image.

You can reproduce this problem by following these steps:

$ git clone https://github.com/Asuka109/napi-rs-image-sample.git
$ cd napi-rs-image-sample
$ pnpm i
$ node transform.mjs
node:internal/process/esm_loader:97
    internalBinding('errors').triggerUncaughtException(
                              ^

[Error: Create image failed BUFFER_TOO_SMALL] { code: 'GenericFailure' }

Is this a known issue, or is there something I can do to fix it?

Thank you for your time.

• [x] Replace AVIF encoder with https://github.com/njaard/libavif-rs

node bench/bench.mjs

@napi-rs/image 202 ops/s
sharp 169 ops/s
In webp suite, fastest is @napi-rs/image
@napi-rs/image 26 ops/s
sharp 24 ops/s
In avif suite, fastest is @napi-rs/image

UV_THREADPOOL_SIZE=10 node bench/bench.mjs 

@napi-rs/image 431 ops/s
sharp 238 ops/s
In webp suite, fastest is @napi-rs/image
@napi-rs/image 36 ops/s
sharp 32 ops/s
In avif suite, fastest is @napi-rs/image

There are a lot of requests for blurhash support in Next.js, and I'm wondering if we could add a method to Image to set raw pixels.

https://github.com/woltapp/blurhash/issues/56#issuecomment-877538877

Perhaps something like setPixels():

import { decode } from 'blurhash'
import Image from '@napi-rs/image'

const pixels = decode('LEHV6nWB2yk8pyo0adR*.7kCMdnj', 32, 32) // Uint8ClampedArray
const image = new Image(32, 32)
image.setPixels(pixels)
const src = image.src  // data:image/png;base64,iVBORw0KGgoAA...

• Docs: https://docs.rs/oxipng/5.0.1/oxipng/enum.Headers.html

• Test: https://github.com/shssoichiro/oxipng/blob/7ffbe1fb7865a8e97512e63cef06bebc245059a1/tests/flags.rs#L224-L230

• safe would remove all auxiliary data blocks that do not potentially affect the rendering of the image. For example, Text Chunks will be removed.

• all would maintain the current behavior and remove all auxiliary data blocks.

https://github.com/Brooooooklyn/Image/blob/e61083611a6e16e07b38eaefee8d2537c93efe2c/packages/binding/src/png.rs#L131-L139

imageoptim uses safe by default.

https://github.com/ImageOptim/ImageOptim/blob/d8d6bc528f0b19253585d824851e320db3261d03/imageoptim/Backend/Workers/OxiPngWorker.m#L31-L33