0

I want to join-encode several videos with different frame rates in the same output video with constant frame rate using FFmpegFrameRecorder, as fast as possible. I am thinking to duplicate frames in case I want to increase frame rate, and skip frames when I need to decrease frame rate. I am using JavaCV.

Is there any better way of duplicating frames except for calling record(image) twice? (the output is always h264 stored in mp4 file format) Something like a flag in h264 stream indicating that current frame is a duplicate of last one, or something else to skip the redundant encoding.

Improve this question

1 Answer 1

Reset to default
0

  1. FrameGrabber is a class that feeds you with Audio-Frames,Video-Frames,FrameRate, Sample Rate etc. You just need to create an instance of this class with the path of the video file as a parameter.

  2. then, FrameGrabber.grab() will give you frames of the audio at the same frame rate as the input video. If FrameGrabber.grab() returns null, then this would mean that the video has been iterated to its last frame.

  3. Now you can set up a new grabberObject, with the path of your second video as parameter to its constructor and repeat step 2

  4. You can repeat step 3 untill all your videos have been read.

  5. Now to record the video; You can initialize the recorder with parameters returned from the FrameGrabber. Its Height,width,frameRate,samplerate all can be set exactly to match the grabber object. Also you need to set a path where the recorder will save the video. After each FrameGrabber.grab(), you need to call recorder.record(FrameGrabber.grab()).

  6. Now the the point where you feed the new video to the FrameGrabber(), following will be the reason of quality-lags:

a) The frameGrabber takes time to set up. The transition/shift to the new video path that will be assigned to the grabber, will not be instantaneous. This can take time in the order of seconds.

b) This transition/shift to the new video path, will also have to change parameters of recorder-Framerate,SampleRate,Height & Width.

To handle both these issues you will need to handle the first frame added after a transition/shift to the new video path, as a special case. Here you would reset the properties of recoderFramerate,SampleRate,Height & Width to match the new frameGrabber properties.

This implementation should work fine.

Moreover duplicating frames in case you want to increase frame rate, and skip frames when I need to decrease frame rate. Both these will distort the video completely. You can never match up to the original quality of the video. More over the lag in video frames and sound would clearly evident.

Improve this answer
Sign up to request clarification or add additional context in comments.

3 Comments

Thanks for your time to answer this. I am not using FrameGrabber, but rather MediaCodec to decode the video/audio. I am also resampling audio to the desired output sample rate, which may be different than the one of current video. All that I need is to change the frame rate of video in case video 2 fps is different than video1 fps, and I am trying to do this as optimal as possible. The obvious case is to call twice record on the current frame. Are you aware of any better way?
MediaCodec will deal with this very efficiently, but still it is not clear what benefit one can achieve by duplicating frames.
By duplicating frames, I can increase framerate if fps in original video is lower than output fps, so that animations blended to the video have smooth motion. I was using JavaCV, because MediaCodec encoder requires to support many color formats if encoding from Buffer and not Surface, and I needed to support android 4.1 devices.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.