Develop Reference

Android Camera2 API - Video always saved in landscape

I am working on a video recording app in which I want to record videos in portrait. Everything seems fine except for the video which is saved in landscape mode. I tried the implementation using this project: https://github.com/HofmaDresu/AndroidCamera2Sample as an example, but still, the video is being saved in landscape mode.
void PrepareMediaRecorder()
{
if (mediaRecorder == null)
{
mediaRecorder = new MediaRecorder();
}
else
{
mediaRecorder.Reset();
}
var map = (StreamConfigurationMap)characteristics.Get(CameraCharacteristics.ScalerStreamConfigurationMap);
if (map == null)
{
return;
}
videoFileName = GetVideoFilePath();
mediaRecorder.SetAudioSource(AudioSource.Mic);
mediaRecorder.SetVideoSource(VideoSource.Surface);
mediaRecorder.SetOutputFormat(OutputFormat.Mpeg4);
mediaRecorder.SetOutputFile(videoFileName);
mediaRecorder.SetVideoEncodingBitRate(10000000);
mediaRecorder.SetVideoFrameRate(30);
var videoSize = ChooseVideoSize(map.GetOutputSizes(Java.Lang.Class.FromType(typeof(MediaRecorder))));
mediaRecorder.SetVideoEncoder(VideoEncoder.H264);
mediaRecorder.SetAudioEncoder(AudioEncoder.Aac);
mediaRecorder.SetVideoSize(videoSize.Width, videoSize.Height);
int rotation = (int)Activity.WindowManager.DefaultDisplay.Rotation;
mediaRecorder.SetOrientationHint(GetOrientation(rotation));
mediaRecorder.Prepare();
}

Assuming a high-quality video player shows you the video in portrait (if not, your GetOrientation method probably has an error in it), but other players you still care about are stuck on landscape:
You'll have to rotate the frames yourself. Unfortunately, this is messy, since there's no automatic control for this on the media encoder APIs that I know of.
Options are receiving frames via an ImageReader from the camera, and then doing the rotation in Java or via JNI, in native code, and then sending the frame to the encoder either via an ImageWriter to a MediaRecorder or MediaCodec Surface, or writing frames via MediaCodec's ByteBuffer interface.
Or you could send the frames to the GPU via a SurfaceTexture, rotate in a fragment shader, and then write out to a Surface tied to a MediaRecorder/MediaCodec again.
Both of these require a lot of boilerplate code and understanding of lower-level details, unfortunately.

Android Camera2 session creation logic

I am new to Camera2 framework and trying to understand the logic of creation of capture sessions.
I need a simple thing - preview and record video. I also want to set the correct orientation hint at the time I start recording the video. But I came to a chicken/egg problem.
Here is my logic:
In order to start recording, I am doing this:
val recordRequest = session.device.createCaptureRequest(CameraDevice.TEMPLATE_RECORD).apply {
// Add the preview and recording surface targets
addTarget(viewFinder.holder.surface)
addTarget(recorder.surface)
}.build()
session.setRepeatingRequest(recordRequest, null, cameraHandler)
recorder.setOrientationHint(it) // NOT allowed after getSurface()!
recorder.prepare() // NOT allowed after getSurface()!
recorder.start()
However, I already called recorder.surface (or getSurface()) when I added targets above. One can think that I can prepare and then add targets, however, the documentation for addTarget() says, that the surface The Surface added must be one of the surfaces included in the most recent call to CameraDevice#createCaptureSession
That leads to an interesting problem. Whenever I open the app, I need to create the capture session to start previewing camera image. However, at the point of creation the createCaptureSession() needs to include all the surfaces that will come in future capture requests. Which means that I also need to include the recording surface, even if I simply open camera without recording yet. How do I get this Surface for recording? Well, the documentation says I can get it from MediaRecorder or I can get it from MediaCodec. I want to get it from MediaRecorder since I want to use CamcorderProfiles. However, as I showed in the above code, once I get the surface from the recorder at the point of session creation - I cannot do any changes there at the point of starting recording, like setting orientation hint.
The official Camera2Video sample app does a trick - it uses createPersistentInputSurface however in their example the camera is fixed, which allows them to allocate enough memory for it and use that surface throughout the app lifecycle.
How can this be solved? Am I misunderstanding the concepts here? How can I create the recorder at a later point, when I start recording, but still have the surface for it created earlier, when I open the camera for preview?

Using a persistent input surface is the right approach. Create a new MediaRecorder once you know the orientation for recording, and set its Surface using the persistent input surface.
That's exactly what the Camera2Video sample does, as well:
// React to user touching the capture button
capture_button.setOnTouchListener { view, event ->
when (event.action) {
MotionEvent.ACTION_DOWN -> lifecycleScope.launch(Dispatchers.IO) {
// Prevents screen rotation during the video recording
requireActivity().requestedOrientation =
ActivityInfo.SCREEN_ORIENTATION_LOCKED
// Start recording repeating requests, which will stop the ongoing preview
// repeating requests without having to explicitly call `session.stopRepeating`
session.setRepeatingRequest(recordRequest, null, cameraHandler)
// Finalizes recorder setup and starts recording
recorder.apply {
// Sets output orientation based on current sensor value at start time
relativeOrientation.value?.let { setOrientationHint(it) }
prepare()
start()
}
and recorder is created with an earlier-created persistent surface:
/** Saves the video recording */
private val recorder: MediaRecorder by lazy { createRecorder(recorderSurface)
}
When you say the camera is fixed, do you mean the app orientation being fixed, or that the sample doesn't support switching front/back cameras? None of that should particularly matter for persistent surfaces; you can create a new one when you switch cameras or change orientations, if you need to.

Recording darkens preview and ratios

This is coded in NativeScript, so I'll try my best to adapt the scenario to Java. I have created an in-app video view with support to record the video.
This is done as follows:
First I create a SurfaceView that will hold the preview of the camera:
this.mSurfaceView = new android.view.SurfaceView(this._context);
this.mHolder = this.mSurfaceView.getHolder();
this.mHolder.setType(android.view.SurfaceHolder.SURFACE_TYPE_PUSH_BUFFERS);
Then I create an instance of the Camera, and sets the video surface:
var mCamera = android.hardware.Camera;
var camera = mCamera.open(1);
this.camera = camera;
this.camera.setDisplayOrientation(90);
var parameters = camera.getParameters();
parameters.setRecordingHint(true);
if( parameters.isVideoStabilizationSupported() ){
parameters.setVideoStabilization(true);
}
camera.setParameters(parameters);
this.camera.setPreviewDisplay(_this.mHolder);
this.camera.startPreview();
this.camera.startFaceDetection();
Now, all is good. I have the camera preview in the view that I want it to be. The color is good and I think the image aspect ratio is good too.
However, when I initiate the recording, as I do with the following code:
this.mediarecorder = new android.media.MediaRecorder();
// Step 1: Unlock and set camera to MediaRecorder
this.camera.unlock();
this.mediarecorder.setCamera(this.camera);
// Step 2: Set sources
this.mediarecorder.setAudioSource(android.media.MediaRecorder.AudioSource.CAMCORDER);
this.mediarecorder.setVideoSource(android.media.MediaRecorder.VideoSource.CAMERA);
//this.mediarecorder.setOutputFormat(android.media.MediaRecorder.OutputFormat.MPEG_4);
// Step 3: Set a CamcorderProfile (requires API Level 8 or higher)
this.mediarecorder.setProfile(android.media.CamcorderProfile.get(android.media.CamcorderProfile.QUALITY_HIGH));
// platform.screen.mainScreen.widthDIPs
// platform.screen.mainScreen.heightDIPs
// Step 4: Set output file
var fileName = "videoCapture_" + new Date() + ".mp4";
var path = android.os.Environment.getExternalStoragePublicDirectory(android.os.Environment.DIRECTORY_DCIM).getAbsolutePath() + "/Camera/" + fileName;
this.file = new java.io.File(path);
this.mediarecorder.setOutputFile(this.file.toString());
this.mediarecorder.setOrientationHint(270);
try {
this.mediarecorder.prepare();
this.mediarecorder.start();
} catch( ex ) {
console.log(ex);
}
Then, the image suddenly becomes darker, and my face (its what's in focus when I'm trying it out) gets wider. So the aspect ratio changes, and so does the lighting somehow.
I have tried setting setPictureSize on the camera parameters, and setVideoSize on the MediaRecorder with no luck. And for the lighting change, I have simply no clue as to whats going on. Now I've been googling myself half way to heaven, and still found nothing, so I hope someone here has got any tip on what to pursue next?

Video recording generally tries to run at a steady frame rate, such as 30fps. Camera preview often slows itself down to 10-15fps to maintain brightness, so if you're in a darker location, video recording will be darker (since it can't expose for longer than 1/30s instead of 1/10s that camera preview can).
Did you call setVideoSize before or after calling setProfile? The setProfile call changes many parameters, including preview size; most video recording sizes are 16:9, and the default camera preview resolution is likely a 4:3 size. So when you start the recording, the aspect ratio switches.
Most video recording apps use 16:9 preview sizes even before starting recording so that they're consistent. You can also record 4:3 video, but that's generally not what people want to see.

Android MediaRecorder saving empty file and weird camera behaviours

I am having a headache over the Camera API 1 for android. After reading all of the Internet content, I made some sample app that works OK. It creates a service, which then is used to operate with the camera in the background, so there is no preview or activity enabled. To achieve this I use a dummy SurfaceHolder, like this:
protected class MySurfaceHolder implements SurfaceHolder {
private final Surface surface;
private final SurfaceTexture surfaceTexture;
public MySurfaceHolder () {
int[] textures = new int[1];
GLES20.glGenTextures(1, textures, 0);
if (textures.length > 0) {
this.surfaceTexture = new SurfaceTexture(textures[0]);
this.surface = new Surface(this.surfaceTexture);
} else {
this.surface = null;
this.surfaceTexture = null;
}
}
[...]
}
and then I use it like this
// simplified version of my code
try {
initializeCamera(); // open camera and set Camera.Parameters
camera.setPreviewDisplay(new MySurfaceHolder());
camera.startPreview();
camera.unlock();
initializeMediaRecorder(); // create MediaRecorder, set video/audio parameters
mediaRecorder.prepare();
mediaRecorder.start();
// wait until recording finish and exit
} finally {
stopRecording();
}
the Camera and MediaRecorder initialization methods are just like the documentation states they should be (and they work).
Everything works and operates as it should. Almost everything - sometimes, under unknown circumstances the MediaRecorder creates empty files, like 32kB containing only headers and info about the video - no frames. The longer I record like this, the bigger is the file (few kB every few seconds). After 1 minute, the file weights about 80kB. Funny thing is I know that the camera is working and capturing frames (I debugged it a little showing preview frames), but the frames are not written into the output file.
Also when it happens I am not able to record in FHD (1920x1080) - I get the "start failed" message - at this time camera is not capturing frames. The same thing could happen when I use wrong (not supported) video size. I suppose in this case the message is thrown at the mediaRecorder.start(); line, and stopRecording(); is invoked but I am not sure.
After some time or after unknown action the problem is suddenly gone (I don't know when, I don't know how). It happens for sure on Android 5.1, but may happen on other versions as well.
Could this bug be related to my custom surface code?
What could cause the MediaRecorder to not write frames into a file?
Why I am not able to record in FHD, but in the same time I am able to record in HD (1280x720)?
Is there any alternative for MediaRecorder, so I can avoid these bugs?
May it happen when another app is trying to get Camera object, thus distrupting current recording? If so, how to regain access to the Camera object (I apparently am not able to do this now on some devices).
EDIT:
I think I might have a clue. I am calling
camera.setOneShotPreviewCallback(new Camera.PreviewCallback() {
// ... get current frame
}
camera.startPreview();
to get preview frame of current recording. It appears that the bug occurs when I am using this method to get preview frame (at random times). It seems flawed, because not all devices react to this thing properly (sometimes there is no preview frame...). Is there any other, better method of handling current preview frame without the real surface?

Slow focusing on Nexus 6 with camera2 API

I've bumped into the issue with slow focusing on Nexus 6.
I develop camera application and now I'm using camera2 API.
For application needs we create preview request with 2 surfaces
- SurfaceView (viewfinder)
- YUV ImageReader surface (to use data in hstogram calculation)
And there is a critical point! If just add only viewfinder surface, focusing occurs as normal. But with 2 those surfaces focusing occurs very slow with visual steps of lens moving!
Code is quite standard, written according google documentations:
mImageReaderPreviewYUV = ImageReader.newInstance(previewWidth, previewHeight, ImageFormat.YUV_420_888, 2);
previewRequestBuilder = camDevice.createCaptureRequest(CameraDevice.TEMPLATE_PREVIEW);
previewRequestBuilder.addTarget(getCameraSurface()); //Add surface of SurfaceView
previewRequestBuilder.addTarget(mImageReaderPreviewYUV); //Add ImageReader
mCaptureSession.setRepeatingRequest(previewRequestBuilder.build(), captureCallback null);

Does the system logcat show any warnings about buffers not being available?
Is the preview frame rate slow, or is smooth (~30fps) but focusing just works oddly?
If the former, you may not be returning Image objects to the ImageReader (by closing them once done with them) at 30 fps, so the camera device is starved for buffers to fill, and cannot maintain 30fps preview.
To test this, implement the minimal ImageReaderListener.onImageAvailable(ImageReader reader) method that just returns the image immediately:
public class TestImageListener extends ImageReaderListener {
public void onImageAvailable(ImageReader reader) {
Image img = reader.acquireNextImage();
img.close();
}
}
...
mImageReaderPreviewYUV.setOnImageAvailableListener(new TestImageListener());
If this lets you get fluid preview, then your image processing is too slow.
As a solution, you should increase the number of buffers in your ImageReader, and the nuse the reader.acquireLatestImage() to drop older buffers and only process the newest Image each time you calculate your histogram.

I had the same issues on the N6 and I think it works smoother now - add the ImageReader surface before the camera surface:
previewRequestBuilder = camDevice.createCaptureRequest(CameraDevice.TEMPLATE_PREVIEW);
previewRequestBuilder.addTarget(mImageReaderPreviewYUV); //Add ImageReader
previewRequestBuilder.addTarget(getCameraSurface()); //Add surface of SurfaceView
I also tested my camera app with a N4/5.0.1 and both ways work perfectly there.