Advanced search

On other websites (1741)

• FFmpeg AVI to MP4 transcode not playing fully in Quicktime

  11 August 2014, by Joel Kennedy

  I’m trying to transcode an AVI video to an MP4 stream on the fly using FFmpeg which can be watched on a webpage with a HTML5 video tag. I already know that MP4 isn’t designed for streaming, however it is possible if segmented so I’m still trying to use MP4 as it’s the method that I’ve found is best supported by multiple browsers.

  My stream plays fine in many browsers except for Safari on OS X, which is using Quicktime to play the HTML5 video.

  

  You can see in the image above that Quicktime only plays the video for a second, and then stops. I’ve tested the same video source in Google Chrome on OS X, and it continues playing fine even after a second has elapsed.

  I believe this is something to do with the set Duration in the tkhd and mdhd atoms? User vbence described that "FFMpeg’s muxer will set the Duration in both tkhd and mdhd atoms to 0xffffffff for each track. This causes problems in some players (for example Quicktime will not play such files). You should find a tool and change it to zero (0x00000000)" in an answer here.

  Is there a way of getting FFmpeg to set a longer duration on the video to get Quicktime to continue playing it instead of stopping? Because I’m transcoding on the fly, I wouldn’t want to use a separate tool apart from FFmpeg as in my situation I can’t transcode and save the video file, I have to transcode and pipe it to the client on request.

  The command that I’m using with FFmpeg to transcode the video is the following:

  ffmpeg -re -i
  http://www.example.com/video/stream?path=videopathhere -g 52
  -reset_timestamps 1 -vsync 1 -flags global_header -vcodec libx264 -strict -2 -f mp4 -b:v 512k -s 1280x720 -movflags frag_keyframe+empty_moov+faststart pipe:

  Any help or pointers in the right direction would be greatly appreciated.

• Change the frame order/sequence of a video (avconv/ffmpeg)

  2 November 2014, by Simon Streicher

  I want to use Python and a similar method explained by http://zulko.github.io/blog/2013/09/27/read-and-write-video-frames-in-python-using-ffmpeg:

  import subprocess as sp
  
  command = [ 'ffmpeg',
  
          '-y', # (optional) overwrite output file if it exists
  
          '-f', 'rawvideo',
  
          '-vcodec','rawvideo',
  
          '-s', '420x360', # size of one frame
  
          '-pix_fmt', 'rgb24',
  
          '-r', '24', # frames per second
  
          '-i', '-', # The imput comes from a pipe
  
          '-an', # Tells FFMPEG not to expect any audio
  
          '-vcodec', 'mpeg'",
  
          'my_output_videofile.mp4' ]
  
  
  
  pipe = sp.Popen( command, stdin=sp.PIPE, stderr=sp.PIPE)
  
  pipe.proc.stdin.write(image_array.tostring())

  to write an image array as frames to ffmpeg. In my application I will write the frames to the buffer as they are completed (not all at once as above).

  want to apply time-distortion to the output video from a "frame-map":

  Frame order In : [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
  
  Frame order Out: [1, 1, 2, 3, 3.5, 4, 4.5, 6, 7, 8, 9.5 10]

  Is there any way I can pipe in the frame sequence?

  I know I can just program my script to sent the frames in the correct order, but I was kind of hoping to discard the frames as I am done with them (I am talking about long HD videos) and let avconv/ffmpeg handle the ordering and inter-frame averaging (for example frame 3.5).

  My other option is to read the input frames at the output frame speed (by playing catchup) and keep the last 300 frames or so. Beforehand I can make sure my pipe wouldn’t need a 301-or-more-frames-back frame. I can even make it more robust by keeping track of any 301-or-more-frames-back frames and make an exception for storing them.

  So could ffmpeg/avconv handle all this drama on its own, or do I have to code this up?

• How to create a fixed image video with ffmpeg that avoids pixelization when it's not maximized?

  11 August 2022, by Nelson Teixeira

  I have a waiting video in the project I develop. This video is a fixed waiting image with music. I was asked to change the image in the video due to deprecated logo.

  


  So initially I tried just to replace the image in the video using this command:

  


  ffmpeg -i old_waiting_video.mp4 -i new_image.png -filter_complex "[1][0]scale2ref[i][v];[v][i]overlay" -c:a copy new_waiting_video.mp4


  


  This lead to the image became pixelated when the video isn't maximized. When maximized the image is OK. But it looks terrible in the small player.

  


  Here is an image of the problem:

  


  

  


  The above image is the image without zoom out. The left bottom image is a screen capture when I resize it in an image tool and the right bottom is a screen capture of how it looks like in the player.
The image has more than this text, but I thought that wouldn't be appropriate to display my company's logo here. The logo itself is even more pixelized and looks terrible.

  


  So to try to solve this problem I extracted the audio from the file and try to add the image and the audio together to see if it removed the pixelization. So I used this command:

  


  ffmpeg -i audio.m4a -i new_image.png new_waiting_video.mp4


  


  When it failed, searching a bit I came across this other command:

  


  ffmpeg -loop 1 -i new_image.png -i audio.m4a -c:v libx264 -tune stillimage -c:a copy -pix_fmt yuv420p -shortest new_waiting_video.mp4


  


  Again the pixelization ocurred. I know it's possible because this problem doesn't occur in the old logo.

  


  What would be the correct command that allowed that the zoomed out image in the player doesn't pixelate ?

  


  Edit 1

  


  I would also be interested in a command that substitutes the image on the old video as long it doesn't pixelate in the non-maximized player.

  


  Edit 2

  


  

  • Video resolution: 1920x1080

  


  • New image resolution: 1920x1080

  


  • Non-maximized video player size: 640x360

  


  • Video tag: <video src="my-video-url.mp4" style="width: 640px;"></video>