Advanced search

On other websites (9629)

• iOS - How can I stream Encoded Video Frames(from AVFoundation and VideoToolBox) from device to server via RTP

  2 October 2015, by ASP Peek

  I am trying to stream live video from my iPhone device to server using RTP.
  Using AVFoundation’s AVCaptureVideoDataOutput, I was able to get CMSampleBuffer for video. I then feed these frames as they arrive into VideoToolBox’s VTCompressionSessionEncodeFrame() and is able to get Encoded CMSampleBuffer.

  Now to send these encoded Frames via RTP, I came across FFMPEG and found its built library for iOS device. (https://github.com/kewlbear/FFmpeg-iOS-build-script)

  However I am not able to find any iOS example or sample code or any documentation that explains the process of sending the encoded frames via RTP for iOS apps.

  Is there any existing example or documentation that can explain me how can I send the encoded CMSampleBuffers to server via RTP using FFMPEG.

  Thanks in Advance :)

• How can I stream video in swift using FFmpeg in real time?

  13 November 2019, by rIn

  I want to make a real time video streaming app but I’m new to Swift and live streaming..

  Video input captured by AVCaptureSession is from iPhone and I want to encode the input from MPEG-4 to MPEG-2.
  I will use FFmpeg library to encode the video input in swift.

  Here’s my questions.

  1. I don’t know how to deliver video data to FFmpeg functions. Actually I’m not certain that I have to use FFmpeg library to encode video data from MPEG-4 to MPEG-2 transport stream. Is there any API that can encode video data in Swift by Apple?

  2. How can I deal with video data from AVCaptureSession? Are there frames or h.264 video in CMSampleBuffer? I want to know what type of data is in CMSampleBuffer.

  I’m struggling to solve this problem. Please let me know anything if you have some experiences about this kind of project. Any ideas or advice for a better approach are welcome.

• FFmpeg encoding slow for 4K HDR content

  1 March 2023, by Geno Diaz

  When processing 4K input with the following configuration it is taking upwards of 2 minutes to process a 35s, 60fps, 4K HDR clip recorded from an iPhone. Is this the expected performance or is there an inefficiency within the configuration that is causing this?

  


  In comparison, running this configuration on a 35s, 30fps, 4K non-HDR clip, only takes about 20 seconds.

  


  ffmpeg 
-i "input path" 
-y 
-filter:v scale=w=2160:h=3840 
-threads 4 
-r 59.94 
-c:v libx264 
-preset veryfast 
-vsync 1 
-tune film 
-maxrate 6000k 
-bufsize 5400k 
-g 60 
-x264opts no-scenecut 
-c:a aac 
-af aresample=async=1:min_hard_comp=0.100000:first_pts=0 
-ac 2 
-b:a 128k 
-ar 44100 
-vf zscale=transfer=linear:npl=100,
  format=gbrpf32le,
  zscale=primaries=bt709,
  tonemap=tonemap=hable:desat=0,
  zscale=transfer=bt709:matrix=bt709:range=tv,
  format=yuv420p 
-sws_flags full_chroma_int+full_chroma_inp 
-pix_fmt yuv420p 
"outputfile".mp4