๐๏ธ This library is work in progress! ๐๏ธ
A writeable in-memory Image JSI Host Object.
JSI-Image is a modern library that provides Image primitives for the native iOS and Android Platforms, neatly packaged together in one single fast JavaScript API.
There are 3 ways to create a JSI-Image instance:
- Load from a file
- Load from a Web-URL
- Returned by another library, such as VisionCamera's
takePhoto(...)function.
Traditionally, Images in React Native could not be handled efficiently. To demonstrate this, let's take a look at how a Camera library might take a photo:
- [js] User taps capture button,
takePhoto(...)is called. - [native] Camera takes a photo. The library now has
UIImageinstance (photo) in-memory. - [native] Library creates a new file on disk. (slow! ๐)
- [native] Library writes the
UIImageinstance to the file. (slow! ๐) - [native] Library returns the path to the file to the caller (JS)
- [js] App now navigates to the "captured media" screen to display the media.
- [js] App passes the file path to a
<FastImage>component. - [native]
<FastImage>component has to load the image from file. (slow! ๐) - [native]
<FastImage>component then displays theUIImagefrom the file.
With JSI-Image, all the unnecessary slow file operations can be skipped, since the Image can be passed around in-memory.
- [js] User taps capture button,
takePhoto(...)is called. - [native] Camera takes a photo. The library now has
UIImageinstance (photo) in-memory. - [native] Library returns the
UIImageinstance to the caller (JS) (fast! ๐ฅ) - [js] App now navigates to the "captured media" screen to display the media.
- [js] App passes the in-memory
Imageinstance to a<FastImage>component. - [native]
<FastImage>component then displays the already in-memoryUIImageinstance. (fast! ๐ฅ)
[log] Successfully took photo in 312ms!
[log] Successfully took photo in 95ms!
JSI-Image improved capture speed (takePhoto(...)) by more than 3x!
These improvements are even greater at more complicated image processing, such as rotating an image, applying image filters, resizing images, etc.
yarn add react-native-jsi-image
cd ios && pod installimport { loadImageFromUrl } from "react-native-jsi-image"
const image = await loadImageFromUrl('https://...')
console.log(`Successfully loaded ${image.width} x ${image.height} image!`)import { loadImageFromFile } from "react-native-jsi-image"
const image = await loadImageFromFile('file:///Users/Marc/image.png')
console.log(`Successfully loaded ${image.width} x ${image.height} image!`)const image = ...
const size = image.width * image.height
const realSize = size * image.scale
const orientation = image.orientation
for (const pixel of image.data) {
console.log(`Pixel: ${pixel}`)
}const image = ...
console.log(image.isFlipped) // false
const flipped = image.flip()
console.log(flipped.isFlipped) // true
if (image.orientation === "up") {
// rotates image in-memory
image.orientation = "right"
}let image = ...
image = rotateImageCorrectly(image)
await image.save('file:///tmp/temp-image.png') // or .jpgTo use JSI-Image in your native library, your functions must be JSI functions.
In your JSI Module:
#include <JsiImage/ImageHostObject.h>
// ...
jsi::Value myFunction(jsi::Runtime& runtime,
jsi::Value& thisArg,
jsi::Value* arguments,
size_t count) {
auto imageHostObject = arguments[0].asObject(runtime).asHostObject<ImageHostObject>(runtime);
auto uiImage = imageHostObject->image;
// use uiImage here
}In your TypeScript declaration:
import { Image } from 'react-native-jsi-image'
export function myFunction(image: Image): voidIn your JSI Module:
#include <JsiImage/ImageHostObject.h>
// ...
jsi::Value myFunction(jsi::Runtime& runtime,
jsi::Value& thisArg,
jsi::Value* arguments,
size_t count) {
UIImage* image = // ...
auto instance = std::make_shared<ImageHostObject>(image, promiseVendor);
return jsi::Object::createFromHostObject(runtime, instance);
}In your TypeScript declaration:
import { Image } from 'react-native-jsi-image'
export function myFunction(): ImageSee the contributing guide to learn how to contribute to the repository and the development workflow.
MIT