react-native-vision-camera/src/CameraProps.ts
Marc Rousavy b6a67d5ced
feature: Frame Processors (iOS) (#2)
* Clean up Frame Processor

* Create FrameProcessorHolder

* Create FrameProcessorDelegate in ObjC++

* Move frame processor to FrameProcessorDelegate

* Decorate runtime, check for null

* Update FrameProcessorDelegate.mm

* Cleanup FrameProcessorBindings.mm

* Fix RuntimeDecorator.h import

* Update FrameProcessorDelegate.mm

* "React" -> "React Helper" to avoid confusion

* Rename folders again

* Fix podspec flattening a lot of headers, causing REA nameclash

* Fix header imports to avoid REA naming collision

* Lazily initialize jsi::Runtime on DispatchQueue

* Install frame processor bindings from Swift

* First try to call jsi::Function (frame processor) 👀

* Call viewForReactTag on RCT main thread

* Fix bridge accessing

* Add more logs

* Update CameraViewManager.swift

* Add more TODOs

* Re-indent .cpp files

* Fix RCTTurboModule import podspec

* Remove unnecessary include check for swift umbrella header

* Merge branch 'main' into frame-processors

* Docs: use static width for images (283)

* Create validate-cpp.yml

* Update a lot of packages to latest

* Set SWIFT_VERSION to 5.2 in podspec

* Create clean.sh

* Delete unused C++ files

* podspec: Remove CLANG_CXX_LANGUAGE_STANDARD and OTHER_CFLAGS

* Update pod lockfiles

* Regenerate lockfiles

* Remove IOSLogger

* Use NSLog

* Create FrameProcessorManager (inherits from REA RuntimeManager)

* Create reanimated::RuntimeManager shared_ptr

* Re-integrate pods

* Add react-native-reanimated >=2 peerDependency

* Add metro-config

* blacklist -> exclusionList

* Try to call worklet

* Fix jsi::Value* initializer

* Call ShareableValue::adapt (makeShareable) with React/JS Runtime

* Add null-checks

* Lift runtime manager creation out of delegate, into bindings

* Remove debug statement

* Make RuntimeManager unique_ptr

* Set _FRAME_PROCESSOR

* Extract convertJSIFunctionToFrameProcessorCallback

* Print frame

* Merge branch 'main' into frame-processors

* Reformat Swift code

* Install reanimated from npm again

* Re-integrate Pods

* Dependabot: Also scan example/ and docs/

* Update validate-cpp.yml

* Create FrameProcessorUtils

* Create Frame.h

* Abstract HostObject creation away

* Fix types

* Fix frame processor call

* Add todo

* Update lockfiles

* Add C++ contributing instructions

* Update CONTRIBUTING.md

* Add android/src/main/cpp to cpplint

* Update cpplint.sh

* Fix a few cpplint errors

* Fix globals

* Fix a few more cpplint errors

* Update App.tsx

* Update AndroidLogger.cpp

* Format

* Fix cpplint script (check-cpp)

* Try to simplify frame processor

* y

* Update FrameProcessorUtils.mm

* Update FrameProcessorBindings.mm

* Update CameraView.swift

* Update CameraViewManager.m

* Restructure everything

* fix

* Fix `@objc` export (make public)

* Refactor installFrameProcessorBindings into FrameProcessorRuntimeManager

* Add swift RCTBridge.runOnJS helper

* Fix run(onJS)

* Add pragma once

* Add `&self` to lambda

* Update FrameProcessorRuntimeManager.mm

* reorder imports

* Fix imports

* forward declare

* Rename extension

* Destroy buffer after execution

* Add FrameProcessorPluginRegistry base

* Merge branch 'main' into frame-processors

* Add frameProcessor to types

* Update Camera.tsx

* Fix rebase merge

* Remove movieOutput

* Use `useFrameProcessor`

* Fix bad merge

* Add additional ESLint rules

* Update lockfiles

* Update CameraViewManager.m

* Add support for V8 runtime

* Add frame processor plugins API

* Print plugin invoke

* Fix React Utils in podspec

* Fix runOnJS swift name

* Remove invalid redecl of `captureSession`

* Use REA 2.1.0 which includes all my big PRs 🎉

* Update validate-cpp.yml

* Update Podfile.lock

* Remove Flipper

* Fix dereferencing

* Capture `self` by value. Fucking hell, what a dumb mistake.

* Override a few HostObject functions

* Expose isReady, width, height, bytesPerRow and planesCount

* use hook again

* Expose property names

* FrameProcessor -> Frame

* Update CameraView+RecordVideo.swift

* Add Swift support for Frame Processors Plugins

* Add macros for plugin installation

* Add ObjC frame processor plugin

* Correctly install frame processor plugins

* Don't require custom name for macro

* Check if plugin already exists

* Implement QR Code Frame Processor Plugin in Swift

* Adjust ObjC style frame processor macro

* optimize

* Add `frameProcessorFrameDropRate`

* Fix types

* Only log once

* Log if it executes slowly

* Implement `frameProcessorFps`

* Implement manual encoded video recordings

* Use recommended video settings

* Add fileType types

* Ignore if input is not ready for media data

* Add completion handler

* Add audio buffer sampling

* Init only for video frame

* use AVAssetWriterInputPixelBufferAdaptor

* Remove AVAssetWriterInputPixelBufferAdaptor

* Rotate VideoWriter

* Always assume portrait orientation

* Update RecordingSession.swift

* Use a separate Queue for Audio

* Format Swift

* Update CameraView+RecordVideo.swift

* Use `videoQueue` instead of `cameraQueue`

* Move example plugins to example app

* Fix hardcoded name in plugin macro

* QRFrame... -> QRCodeFrame...

* Update FrameProcessorPlugin.h

* Add example frame processors to JS base

* Update QRCodeFrameProcessorPluginSwift.m

* Add docs to create FP Plugins

* Update FRAME_PROCESSORS_CREATE.mdx

* Update FRAME_PROCESSORS_CREATE.mdx

* Use `AVAssetWriterInputPixelBufferAdaptor` for efficient pixel buffer recycling

* Add customizable `pixelFormat`

* Use native format if available

* Update project.pbxproj

* Set video width and height as source-pixel-buffer attributes

* Catch

* Update App.tsx

* Don't explicitly set video dimensions, let CVPixelBufferPool handle it

* Add a few logs

* Cleanup

* Update CameraView+RecordVideo.swift

* Eagerly initialize asset writer to fix stutter at first frame

* Use `cameraQueue` DispatchQueue to not block CaptureDataOutputDelegate

* Fix duration calculation

* cleanup

* Cleanup

* Swiftformat

* Return available video codecs

* Only show frame drop notification for video output

* Remove photo and video codec functionality

It was too much complexity and probably never used anyways.

* Revert all android related changes for now

* Cleanup

* Remove unused header

* Update AVAssetWriter.Status+descriptor.swift

* Only call Frame Processor for Video Frames

* Fix `if`

* Add support for Frame Processor plugin parameters/arguments

* Fix arg support

* Move to JSIUtils.mm

* Update JSIUtils.h

* Update FRAME_PROCESSORS_CREATE.mdx

* Update FRAME_PROCESSORS_CREATE.mdx

* Upgrade packages for docs/

* fix docs

* Rename

* highlight lines

* docs

* community plugins

* Update FRAME_PROCESSOR_CREATE_FINAL.mdx

* Update FRAME_PROCESSOR_PLUGIN_LIST.mdx

* Update FRAME_PROCESSOR_PLUGIN_LIST.mdx

* Update dependencies (1/2)

* Update dependencies (2/2)

* Update Gemfile.lock

* add FP docs

* Update README.md

* Make `lastFrameProcessor` private

* add `frameProcessor` docs

* fix docs

* adjust docs

* Update DEVICES.mdx

* fix

* s

* Add logs demo

* add metro restart note

* Update FRAME_PROCESSOR_CREATE_PLUGIN_IOS.mdx

* Mirror video device

* Update AVCaptureVideoDataOutput+mirror.swift

* Create .swift-version

* Enable whole module optimization

* Fix recording mirrored video

* Swift format

* Clean dictionary on `markInvalid`

* Fix cleanup

* Add docs for disabling frame processors

* Update project.pbxproj

* Revert "Update project.pbxproj"

This reverts commit e67861e51b88b4888a6940e2d20388f3044211d0.

* Log frame drop reason

* Format

* add more samples

* Add clang-format

* also check .mm

* Revert "also check .mm"

This reverts commit 8b9d5e2c29866b05909530d104f6633d6c49eadd.

* Revert "Add clang-format"

This reverts commit 7643ac808e0fc34567ea1f814e73d84955381636.

* Use `kCVPixelFormatType_420YpCbCr8BiPlanarVideoRange` as default

* Read matching video attributes from videoSettings

* Add TODO

* Swiftformat

* Conditionally disable frame processors

* Assert if trying to use frame processors when disabled

* Add frame-processors demo gif

* Allow disabling frame processors via `VISION_CAMERA_DISABLE_FRAME_PROCESSORS`

* Update FrameProcessorRuntimeManager.mm

* Update FRAME_PROCESSORS.mdx

* Update project.pbxproj

* Update FRAME_PROCESSORS_CREATE_OVERVIEW.mdx
2021-05-06 14:11:55 +02:00

162 lines
5.5 KiB
TypeScript

import type { ViewProps } from 'react-native';
import type { CameraDevice, CameraDeviceFormat, ColorSpace } from './CameraDevice';
import type { CameraRuntimeError } from './CameraError';
import type { CameraPreset } from './CameraPreset';
import type { Frame } from './Frame';
export interface CameraProps extends ViewProps {
/**
* The Camera Device to use.
*
* See the [Camera Devices](https://cuvent.github.io/react-native-vision-camera/docs/guides/devices) section in the documentation for more information about Camera Devices.
*
* @example
* ```tsx
* const devices = useCameraDevices('wide-angle-camera')
* const device = devices.back
*
* return (
* <Camera
* device={device}
* isActive={true}
* style={StyleSheet.absoluteFill}
* />
* )
* ```
*/
device: CameraDevice;
/**
* Whether the Camera should actively stream video frames, or not. See the [documentation about the `isActive` prop](https://cuvent.github.io/react-native-vision-camera/docs/guides/devices#the-isactive-prop) for more information.
*
* This can be compared to a Video component, where `isActive` specifies whether the video is paused or not.
*
* > Note: If you fully unmount the `<Camera>` component instead of using `isActive={false}`, the Camera will take a bit longer to start again. In return, it will use less resources since the Camera will be completely destroyed when unmounted.
*/
isActive: boolean;
//#region Common Props (torch, zoom)
/**
* Set the current torch mode.
*
* Note: The torch is only available on `"back"` cameras, and isn't supported by every phone.
*
* @default "off"
*/
torch?: 'off' | 'on';
/**
* Specifies the zoom factor of the current camera, in percent. (`0.0` - `1.0`)
*
* **Note:** Linearly increasing this value always appears logarithmic to the user.
*
* @default 0.0
*/
zoom?: number;
/**
* Enables or disables the native pinch to zoom gesture.
*
* If you want to implement a custom zoom gesture, see [the Zooming with Reanimated documentation](https://cuvent.github.io/react-native-vision-camera/docs/guides/animated).
*
* @default false
*/
enableZoomGesture?: boolean;
//#endregion
//#region Format/Preset selection
/**
* Automatically selects a camera format which best matches the given preset. Must be `undefined` when `format` is set!
*/
preset?: CameraPreset;
/**
* Selects a given format. Must be `undefined` when `preset` is set!
*/
format?: CameraDeviceFormat;
/**
* Specify the frames per second this camera should use. Make sure the given `format` includes a frame rate range with the given `fps`.
*
* Requires `format` to be set.
*/
fps?: number;
/**
* Enables or disables HDR on this camera device. Make sure the given `format` supports HDR mode.
*
* Requires `format` to be set.
*/
hdr?: boolean;
/**
* Enables or disables low-light boost on this camera device. Make sure the given `format` supports low-light boost.
*
* Requires `format` to be set.
*/
lowLightBoost?: boolean;
/**
* Specifies the color space to use for this camera device. Make sure the given `format` contains the given `colorSpace`.
*
* Requires `format` to be set.
*/
colorSpace?: ColorSpace;
//#endregion
/**
* Also captures data from depth-perception sensors. (e.g. disparity maps)
*
* @default false
*/
enableDepthData?: boolean;
/**
* A boolean specifying whether the photo render pipeline is prepared for portrait effects matte delivery.
*
* When enabling this, you must also set `enableDepthData` to `true`.
*
* @platform iOS 12.0+
* @default false
*/
enablePortraitEffectsMatteDelivery?: boolean;
/**
* Indicates whether the photo render pipeline should be configured to deliver high resolution still images
*
* @default false
*/
enableHighResolutionCapture?: boolean;
//#region Events
/**
* Called when any kind of runtime error occured.
*/
onError?: (error: CameraRuntimeError) => void;
/**
* Called when the camera was successfully initialized.
*/
onInitialized?: () => void;
/**
* A worklet which will be called for every frame the Camera "sees". Throttle the Frame Processor's frame rate with {@linkcode frameProcessorFps}.
*
* > See [the Frame Processors documentation](https://cuvent.github.io/react-native-vision-camera/docs/guides/frame-processors) for more information
*
* @example
* ```tsx
* const frameProcessor = useFrameProcessor((frame) => {
* 'worklet'
* const qrCodes = scanQRCodes(frame)
* console.log(`Detected QR Codes: ${qrCodes}`)
* }, [])
*
* return <Camera {...cameraProps} frameProcessor={frameProcessor} />
* ```
*/
frameProcessor?: (frame: Frame) => void;
/**
* Specifies the maximum frame rate the frame processor can use, independent of the Camera's frame rate (`fps` property).
*
* * A value of `1` (default) indicates that the frame processor gets executed once per second, perfect for code scanning.
* * A value of `10` indicates that the frame processor gets executed 10 times per second, perfect for more realtime use-cases.
* * A value of `25` indicates that the frame processor gets executed 30 times per second, perfect for high-speed realtime use-cases.
*
* If you're using higher values, always check your Xcode/Android Studio Logs to make sure your frame processors are executing fast enough
* without blocking the video recording queue.
*
* @default 1
*/
frameProcessorFps?: number;
//#endregion
}