java - 在另一个屏幕元素上设置 ARCore Sceneview

Question

我试图通过简单地使用一个 fragment 的提要并将另一个屏幕元素设置为相同的内容来在屏幕上显示一个 arFragment 两次，但我无法弄清楚要采用哪个元素。

我知道，我通过调用获取当前的Camera图像

ArFragment arFragment = (ArFragment) getSupportFragmentManager()
                           .findFragmentById(R.id.arFragment);
Image image = arFragment.getArSceneView().getArFrame().acquireCameraImage();

但我不知道如何获取另一个屏幕对象并将 View 设置为提要，arFragment 给了我。例如:

TextureView secondView = (TextureView) findViewById(R.id.texture);
secondView.setSurfaceTexture((SurfaceTexture) image);

产生不可转换的类型错误。

我无法使用另一个 arFragment，因为那样会分配另一个相机(显然会产生黑屏和“相机已在使用”错误)。我还没找到

arFrame.assignCamera();

方法，这并不重要，因为Fragment使用的相机只是一个对象，而不是真实的东西。但我无法弄清楚硬件与 fragment 的联系在哪里。如果我没记错的话，我无法在那里读写。

我可以将 feed 转换为位图，或者将其放置到 imageView 上，但我有点害怕每秒执行 60 次。必须有一个简单的解决方案，对吧？...

显示 View 两次并不难 -.-

Answer 1

好的，明白了。转换为 bmp 有点神奇，但我猜确实没有直接的方法。

所以我初始化了一个bytebuffer，分析了android.media.image的YUV分量，将它们转换为Jpeg，然后将其更改为Bitmap，将其旋转90°以匹配原始图片。

// get the arFragment
arFragment = (ArFragment) getSupportFragmentManager().findFragmentById(R.id.arFragment);
ArSceneView arSceneView = arFragment.getArSceneView();

// set up a Listener to trigger on every frame
arSceneView.getScene().addOnUpdateListener(frameTime -> 
{
  try 
  {
   frame = arSceneView.getArFrame();
   androidMediaImage = frame.acquireCameraImage();
   int imageWidth = androidMediaImage.getWidth();
   int imageHeight = androidMediaImage.getHeight();

   // select the target Container to display the image in
   ImageView secondView = (ImageView) findViewById(R.id.imageView3);
   byte[] nv21;

   // an Android.Media.Image is a YUV-Image which is made out of 3 planes
   ByteBuffer yBuffer = androidMediaImage.getPlanes()[0].getBuffer();
   ByteBuffer uBuffer = androidMediaImage.getPlanes()[1].getBuffer();
   ByteBuffer vBuffer = androidMediaImage.getPlanes()[2].getBuffer();

   // set up a Bytearray with the size of all the planes
   int ySize = yBuffer.remaining();
   int uSize = uBuffer.remaining();
   int vSize = vBuffer.remaining();

   nv21 = new byte[ySize + uSize + vSize];

   // Fill in the array. This code is directly taken from https://www.programcreek.com 
   //where it was pointed out that U and V have to be swapped
   yBuffer.get(nv21, 0 , ySize);
   vBuffer.get(nv21, ySize, vSize);
   vBuffer.get(nv21, ySize + vSize, uSize);

   // combine the three layers to one nv21 image
   YuvImage yuvImage = new YuvImage(nv21, ImageFormat.NV21, imageWidth, imageHeight, null);
   // Open a Bytestream to feed the compressor
   ByteArrayOutputStream out = new ByteArrayOutputStream();
   // compress the yuv image to Jpeg. This is important, because the BitmapFactory can't read a 
   // yuv-coded image directly (belief me I tried -.-)
   yuvImage.compressToJpeg(new Rect(0, 0, imageWidth, imageHeight), 50, out);
   // now write down the bytes of the image into an array
   byte[] imageBytes = out.toByteArray();
   // and build the bitmap using the Factory
   Bitmap bitmapImage = BitmapFactory.decodeByteArray(imageBytes, 0, imageBytes.length);

   // use a Matrix for the rotation
   Matrix rotationMatrix = new Matrix();
   // the thing is basically a bunch of numbers which then can be used to compute the new location of each pixel
   rotationMatrix.postRotate(90);
   // the rotatedImage will be our target image
   Bitmap rotatedImage = Bitmap.createBitmap(bitmapImage, 0,0, bitmapImage.getWidth(), bitmapImage.getHeight(), rotationMatrix, true);

   // it's so easy!!!!
   secondView.setImageBitmap(rotatedImage);
  } catch (NotYetAviableException e) 
   {
     e.printStackTrace();
   }
});

如果我完全错了，你显然可以纠正我，并且有一个更简单的解决方案。但至少它有效，所以我很高兴<3