【iOS开发】使用GPUImage框架对接拓幻美颜SDK

关于拓幻的SDK基本使用: 拓幻美颜对接进腾讯云直播过程以及问题

新项目中有一个美颜设置界面, 可提前设置视频时的美颜效果,主要是预览+美颜, 所以找了下, 使用GPUImage图像处理框架来对接拓幻SDK。

一个简单的效果:


想要图像预览, 那就要有一个图像视频流。

在GPUImage框架中, GPUImageVideoCamera作为GPUImageOutput的子类,会把图像作为纹理, 传给OpenGL ES处理,然后把纹理往下传递, 所以我们需要创建一个GPUImageVideoCamera的对象:

@property (nonatomic, strong) GPUImageVideoCamera *videoCamera;
- (GPUImageVideoCamera *)videoCamera {
    if (!_videoCamera) {
        _videoCamera = [[GPUImageVideoCamera alloc] initWithSessionPreset:AVCaptureSessionPreset1280x720 cameraPosition:AVCaptureDevicePositionFront];
        _videoCamera.frameRate = 30; // 调整帧率
        _videoCamera.outputImageOrientation = UIInterfaceOrientationPortrait;
        _videoCamera.horizontallyMirrorFrontFacingCamera = true;
        _videoCamera.horizontallyMirrorRearFacingCamera = false;
        
        [_videoCamera addAudioInputsAndOutputs];
        _videoCamera.delegate = self;
        
        [_videoCamera addTarget:self.previewView];
        
    }
    return _videoCamera;
}

视频流有了, 还需要一个视图来进行渲染,GPUImageView 一个用作GPUImage 输出的端点的UIView子类, 所以我们再来创建一个GPUImageView的对象:

@property (nonatomic, strong) GPUImageView *previewView;
- (GPUImageView *)previewView {
    if (!_previewView) {
        _previewView = [[GPUImageView alloc] initWithFrame:self.view.frame];
        _previewView.fillMode = kGPUImageFillModePreserveAspectRatioAndFill;
    }
    return _previewView;
}

准备就绪后, 初始化TiSDK,将美颜UI 添加到self. previewView上加载,并设置代理, 开始相机拍摄:

- (void)viewDidLoad {
    [super viewDidLoad];
    // Do any additional setup after loading the view.
    
    // 美颜设置
    [TiSDK init:kTuoHuanKey CallBack:^(InitStatus callBack) {
        
    }];
    [[TiUIManager shareManager] loadToView:self.previewView forDelegate:self];
    
    [self setUI];
    
    [self.videoCamera startCameraCapture];
}

GPUImageVideoCamera有个代理GPUImageVideoCameraDelegate,其实就是一个摄像头采集协议,实现协议方法-(void)willOutputSampleBuffer:(CMSampleBufferRef)sampleBuffer

这个方法用于对帧动画的手动处理, 它会在捕获每一帧画面的时候被调用,此时就可以对画面进行一些处理, 那对上我们的需求, 就可以使用TiSDK对每一帧进行纹理渲染:

-(void)willOutputSampleBuffer:(CMSampleBufferRef)sampleBuffer {
    
    // CoreMedia.framework: CMSampleBufferGetImageBuffer
    CVPixelBufferRef pixelBuffer = (CVPixelBufferRef)CMSampleBufferGetImageBuffer(sampleBuffer);
    
    BOOL isMirror = (self.videoCamera.cameraPosition == AVCaptureDevicePositionFront);
    
    CVPixelBufferLockBaseAddress(pixelBuffer, 0);
    unsigned char *baseAddress = (unsigned char *)CVPixelBufferGetBaseAddressOfPlane(pixelBuffer, 0);
    
    UIDeviceOrientation orientation = [UIDevice currentDevice].orientation;
    TiRotationEnum rotation;
    switch (orientation) {
        case UIDeviceOrientationPortrait:
            rotation = CLOCKWISE_90;
            break;
        case UIDeviceOrientationLandscapeLeft:
            rotation = isMirror ? CLOCKWISE_180 : CLOCKWISE_0;
            break;
        case UIDeviceOrientationLandscapeRight:
            rotation = isMirror ? CLOCKWISE_0 : CLOCKWISE_180;
            break;
        case UIDeviceOrientationPortraitUpsideDown:
            rotation = CLOCKWISE_270;
            break;
        default:
            rotation = CLOCKWISE_90;
            break;
    }
    
    // 视频帧格式
    TiImageFormatEnum format;
    switch (CVPixelBufferGetPixelFormatType(pixelBuffer)) {
        case kCVPixelFormatType_32BGRA:
            format = BGRA;
            break;
        case kCVPixelFormatType_420YpCbCr8BiPlanarVideoRange:
            format = NV12;
            break;
        case kCVPixelFormatType_420YpCbCr8BiPlanarFullRange:
            format = NV12;
            break;
        default:
            NSLog(@"错误的视频帧格式!");
            format = BGRA;
            break;
    }
    
    int imageWidth, imageHeight;
    if (format == BGRA) {
        imageWidth = (int)CVPixelBufferGetBytesPerRow(pixelBuffer) / 4;
        imageHeight = (int)CVPixelBufferGetHeight(pixelBuffer);
    } else {
        imageWidth = (int)CVPixelBufferGetWidthOfPlane(pixelBuffer , 0);
        imageHeight = (int)CVPixelBufferGetHeightOfPlane(pixelBuffer , 0);
    }
    
    //todo --- tillusory start ---
    [[TiSDKManager shareManager] renderPixels:baseAddress Format:format Width:imageWidth Height:imageHeight Rotation:rotation Mirror:isMirror];
    //todo --- tillusory end ---
    
//    self.outputImage = [CIImage imageWithCVPixelBuffer:pixelBuffer];
    self.outputImagePixelBuffer = pixelBuffer;
    CVPixelBufferUnlockBaseAddress(pixelBuffer, 0);
    
}
@property(nonatomic, assign) CVPixelBufferRef outputImagePixelBuffer;

在iOS里,我们经常能看到 CVPixelBufferRef 这个类型,在Camera 采集返回的数据里得到一个CMSampleBufferRef,而每个CMSampleBufferRef里则包含一个 CVPixelBufferRef,在视频硬解码的返回数据里也是一个 CVPixelBufferRef。
顾名思义,CVPixelBufferRef 是一种像素图片类型,由于CV开头,所以它是属于 CoreVideo 模块的。
CVPixelBufferRef里包含很多图片相关属性,比较重要的有 width,height,PixelFormatType等。

到这里,预览+拓幻美颜的功能基本就实现了, 当设置了不同的美颜, 就会实时进行纹理渲染。

由于拓幻SDK有缓存, 即每次设置之后都会做缓存,下次使用时会自动读取缓存, 所以我们设置完成后,在视频时就可以直接使用设置好的美颜参数。


GPUImageVideoCamera的几个常用方法:
切换摄像头:[self.videoCamera rotateCamera];
开始视频采集:[self.videoCamera startCameraCapture];
停止视频采集:[self.videoCamera stopCameraCapture];

以上, 结束。

如果对你有帮助, 就给一个赞吧~

你可能感兴趣的:(【iOS开发】使用GPUImage框架对接拓幻美颜SDK)