gpuimage 视频相关使用
基础的图片等使用方法先看一下官方的demo或者其他的教程。这里有的：本地视频添加滤镜（无声，跟官方demo一样，这个要注意释放，我这里没释放了），拍照摄像，本地视频带声预览(ง•̀ω•́)ง✧（感谢同事帮找的https://blog.csdn.net/u011270282/article/details/50354208，不过不支持同步录制）。
另外，预览状态切换滤镜和头像挂件这两个也实现过一下，我比较懒，可能有空会再加一下也可能就这样太监了。
demo地址：本文domo
如果需要边录边播的话，可能你会需要纠正一下视频播放速度：
{
// Do this outside of the video processing queue to not slow that down while waiting
CMTime currentSampleTime = CMSampleBufferGetOutputPresentationTimeStamp(sampleBufferRef);
CMTime differenceFromLastFrame = CMTimeSubtract(currentSampleTime, previousFrameTime);
NSLog(@"%lld %d",currentSampleTime.value,currentSampleTime.timescale);
if( differenceFromLastFrame.value >0) {
CFAbsoluteTime currentActualTime = CFAbsoluteTimeGetCurrent();
CGFloat frameTimeDifference = CMTimeGetSeconds(differenceFromLastFrame);
CGFloat actualTimeDifference = currentActualTime - previousActualFrameTime;
if(frameTimeDifference > actualTimeDifference ){
CGFloat difTime = (frameTimeDifference - actualTimeDifference) - delayoOffsetTime;
if(difTime >0){
doubletime =1000000.0* difTime;
usleep(time);
}
delayoOffsetTime = CFAbsoluteTimeGetCurrent() - currentActualTime - difTime;
if(delayoOffsetTime <0) {
delayoOffsetTime =0;
}
NSLog(@"date:%f %f dif:%f difTime:%f",frameTimeDifference,actualTimeDifference,delayoOffsetTime,difTime);
}
previousFrameTime = currentSampleTime;
previousActualFrameTime = CFAbsoluteTimeGetCurrent();
}
}
替换此处
播放速度纠正
64宫格的图的获取方法：你ps上怎么处理其他图片，把gpu里lookup那张图片拿出来，一样的流程处理一遍，就好了。只处理颜色，不要处理模糊什么的。
64宫格
因为可能不会跟新了，附一下切换滤镜的关键方法
//https://blog.csdn.net/u013488791/article/details/69361818
- (void)filterVideo:(NLGPUImageCustomFilter*)terminalFilter{
if (!gpuMovieFile) {
playerItem = [[AVPlayerItem alloc]initWithURL:videoUrl];
player = [AVPlayer playerWithPlayerItem:playerItem];
gpuMovieFile= [[GPUImageMovie alloc] initWithPlayerItem:playerItem];
gpuMovieFile.delegate = self;
gpuMovieFile.runBenchmark = YES;
gpuMovieFile.playAtActualSpeed = YES;
}
AVAsset *asset = [AVAsset assetWithURL:videoUrl];
NSArray *tracks = [asset tracksWithMediaType:AVMediaTypeVideo];
if([trackscount] >0) {
AVAssetTrack*videoTrack = [tracksobjectAtIndex:0];
CGAffineTransform t = videoTrack.preferredTransform;//这里的矩阵有旋转角度，转换一下即可
CGFloatwidth = videoTrack.naturalSize.width;
CGFloatheight = videoTrack.naturalSize.height;
// 对旋转90和旋转270的视频进行缩放，平移
if(t.a==0&&fabs(t.b) ==1.0&&fabs(t.c) ==1.0&& t.d==0){
t.b= t.b*width/height;
t.c= t.c*width/height;
t.tx=0;
t.ty=0;
gpuPlayView.transform = t;
}
}
if (CMTimeCompare(pausedTime, player.currentItem.asset.duration)&&isPlayingVideo) {
pausedTime = player.currentTime;
}else{
pausedTime = CMTimeMake(0, 600.0);
}
// 滤镜
if(![terminalFilter isKindOfClass:[NLGPUImageCustomFilterclass]]) {
//没点滤镜（直接点击播放按钮）
if ([currentFilter isKindOfClass:[NLGPUImageCustomFilter class]]) {
//已选中过滤镜，直接使用
terminalFilter = [self getFilterByFilter:currentFilter];
}else{
//当前未选中过滤镜
terminalFilter = [[NLGPUImageNormalFilteralloc]init];
}
}else{
terminalFilter = [selfgetFilterByFilter:terminalFilter];
}
[gpuMovieFile cancelProcessing];
[gpuMovieFile removeAllTargets];
[gpuMovieFile addTarget:terminalFilter];
[terminalFilteraddTarget:gpuPlayView]; // gpuPlayView is my GPUImageView
[gpuMovieFile startProcessing];
// Seeking to the point where video was paused
if (CMTimeCompare(pausedTime, player.currentItem.asset.duration) == 0) {
[player play];
}else{
[player seekToTime:pausedTime toleranceBefore:kCMTimeZero toleranceAfter:kCMTimeZero];
[player play];
}
}
多层View录制方法，至于人脸跟踪，这里就不展开讲，苹果有自带的人脸跟踪（AVFoudation的，不要用coreImage，我看有些教程直接用coreimage当摄像头显示层来做人脸跟随，真是服气），另外商用的话，还是需要三方的人脸跟踪，苹果的有点慢。
需要学习的可以参考
