實(shí)現(xiàn)原理：
通過(guò)GPUImage濾鏡處理、錄制、保存的思路如圖。
視頻部分：經(jīng)過(guò)filter的視頻幀分兩步，一步用于在屏幕預(yù)覽GPUImageView上顯示，另一步用于寫入GPUImageMovieWriter。
音頻部分：從GPUImageVideoCamera分離的音頻直接寫入GPUImageMovieWriter。另外，如果需要對(duì)音頻進(jìn)行混響、變聲等處理，可以從這個(gè)節(jié)點(diǎn)分支處理寫入。
實(shí)現(xiàn)流程：
攝像頭
videoCamera = [[GPUImageVideoCamera alloc] initWithSessionPreset:AVCaptureSessionPreset640x480 cameraPosition:AVCaptureDevicePositionBack];
videoCamera.outputImageOrientation = UIInterfaceOrientationPortrait;
videoCamera.runBenchmark = YES;
濾鏡和預(yù)覽響應(yīng)鏈
filter = [[GPUImageSaturationFilter alloc] init];
previewView = [[GPUImageView alloc] initWithFrame:self.view.frame];
previewView.fillMode = kGPUImageFillModePreserveAspectRatio;
self.view = previewView;
[videoCamera addTarget:filter];
[filter addTarget:previewView];
[videoCamera startCameraCapture];
錄入本地和保存
//開(kāi)始
unlink([pathToMovie UTF8String]);
movieWriter = [[GPUImageMovieWriter alloc] initWithMovieURL:movieURL size:CGSizeMake(480, 640)];
movieWriter.encodingLiveVideo = YES;
[filter addTarget:movieWriter];
videoCamera.audioEncodingTarget = movieWriter;
[movieWriter startRecording];
//結(jié)束
[filter removeTarget:movieWriter];
videoCamera.audioEncodingTarget = nil;
[movieWriter finishRecording];
[self saveVideoToLibrary:pathToMovie];//存入相冊(cè)方法
核心類：GPUImageMovieWriter
GPUImageMovieWriter實(shí)現(xiàn)了GPUImageInput，一般作為響應(yīng)鏈終點(diǎn)。
init方法
最終調(diào)用
- (id)initWithMovieURL:(NSURL *)newMovieURL size:(CGSize)newSize fileType:(NSString *)newFileType outputSettings:(NSMutableDictionary *)outputSettings;
方法里重寫了void runSynchronouslyOnContextQueue(GPUImageContext *context, void (^block)(void))方法。
錄制方法
//私有方法
1 - (void)initializeMovieWithOutputSettings:(NSDictionary *)outputSettings;
//公共方法
2 - (void)startRecording;
3 - (void)startRecordingInOrientation:(CGAffineTransform)orientationTransform;
4 - (void)finishRecording;
5 - (void)finishRecordingWithCompletionHandler:(void (^)(void))handler;
6 - (void)cancelRecording;
7 - (void)processAudioBuffer:(CMSampleBufferRef)audioBuffer;
8 - (void)enableSynchronizationCallbacks;
在私有方法中初始化了AVAssetWriter，傳入一個(gè)本地存儲(chǔ)地址和文件格式。
assetWriter = [[AVAssetWriter alloc] initWithURL:movieURL fileType:fileType error:&error];
2-6的方法無(wú)甚可說(shuō)，基本就是調(diào)用AVAssetWriter的三個(gè)方法：
- startWriting
- cancelWriting
- finishWriting
還有finishRecording方法中對(duì)音頻和視頻AVAssetWriterInput調(diào)用的方法markAsFinished
第七個(gè)方法
- (void)processAudioBuffer:(CMSampleBufferRef)audioBuffer;比較重要，在整個(gè)GPUImage中，只在兩種響應(yīng)鏈起點(diǎn)GPUImageMovie和GPUImageVideoCamera中作為audioEncodingTarget屬性調(diào)用過(guò)，這一部分也可以進(jìn)行擴(kuò)展，進(jìn)行混音等操作。
幀渲染方法
//都是私有方法
- (void)createDataFBO;
- (void)destroyDataFBO;
- (void)setFilterFBO;
- (void)renderAtInternalSizeUsingFramebuffer:(GPUImageFramebuffer *)inputFramebufferToUse;
GPUImageInput協(xié)議方法
幾個(gè)存取方法
追加
** AVAssetWriterInputPixelBufferAdaptor**：
AVAssetWriterInputPixelBufferAdaptor負(fù)責(zé)將圖片轉(zhuǎn)成的緩存數(shù)據(jù)CVPixelBufferRef追加到AVAssetWriterInput中，（把CVPixelBufferRef視頻貞圖像拼接到視頻中）。
You use an AVAssetWriterInputPixelBufferAdaptor to append video samples packaged as CVPixelBuffer objects to a single AVAssetWriterInput object.//注意overview的部分
