Recherche avancée

Sur d’autres sites (6791)

• HLS Content Throwing 403 Errors on iPhone Indeterminately [closed]

  10 août 2023, par Shishir Jakati

  I am facing an issue with video content that I have converted to HLS playlist content (using ffmpeg) added to an S3 bucket that is shared through a Cloudfront Distribution. My scenario is the following :

  


  I have a bucket called bucket-a, with a "folder" video-1 which contains the following files :

  


  

  • output.m3u8

  


  • output0.ts

  


  • ...

  


  • output15.ts

  


  • audio/

    

    • audio.aac

    


    


  


  • image.jpg

  


  


  All items in bucket-a are blocked from public access through S3. Content is only vended through a Cloudfront distribution which has origin bucket-a. I am able to access https://<distro>.cloudfront.net/path/output.m3u8</distro> on a desktop browser without fail, and no errors thrown. But the file output.m3u8 and all .ts files are not available on iPhone mobile browsers. The part that is peculiar is that this is not true for all playlist content in bucket-a. For example, I have a "folder" video-2 within bucket-a that has the same file structure as video-1 that is completely accessible through all mobile browsers.

  


  Here is an example master playlist error : https://dbs3s11vyxuw0.cloudfront.net/bottle-promo/script_four/output.m3u8

  


  Even more head-scratching is that I am able to access all the playlists that are within this playlist.

  


  What I've tried :

  


  

  • Initially, I believed the issue to be due to the way the video was transcoding so I standardized the video transcoding.

  


  • Then I believed the issue to be due to CloudFront permissions, though those seem to be fine.

  


  • I've validated my stream here : https://ott.dolby.com/OnDelKits_dev/StreamValidator/Start_Here.html

  


  


  Not sure which way to turn.

  


• FFmpeg - Channel relabelling of multi audio tracked video file [closed]

  8 avril 2024, par Mahi s

  I'm trying to relabel the audio channels of the multitrack audio, video files. For some reason I could not relabel the audio channels or audio streams using FFmpeg command. Can someone please help me with a ffmpeg command that can retain all the audio properties like audio count and retain the tracks unaltered(some ffmpeg commands that I have tried just picks one out of multiple streams and provides the output).

  


  Example scenario : I have a video file which has multi stream audio's (6 audio streams, 6 channel) file with audio format pcm, which was wrongly labelled as C C C C C C (when checked in media info - channel layout). But I need to change the labeling of audio tracks as L R C LFE Ls Rs (The desired audio codec is 5.1 (surround)).

  


  The scenarios that I have faced when trying to relabel this video file is :

  


  

  1. FFMpeg relabels correctly as L R C LFE Ls Rs, but it just takes one audio stream and places them in C channel (when checked in Audacity I can see the audio is present only in C channel), and leaves out other channels.

  


  2. Or it just merges all the audio streams into one (i.e.,the audio count becomes 1- media info output on the resulting file) resulting in a 5.1(LRC) audio codec- matrix.

  


  


  The command that got me close was

  


  ffmpeg -i input.mov -map 0:v -map 0:a:0 -c:a:0 copy -metadata:s:a:0 channel_layout="L" -map 0:a:1 -c:a:1 copy -metadata:s:a:1 channel_layout="R" -map 0:a:2 -c:a:2 copy -metadata:s:a:2 channel_layout="C" -map 0:a:3 -c:a:3 copy -metadata:s:a:3 channel_layout="LFE" -map 0:a:4 -c:a:4 copy -metadata:s:a:4 channel_layout="Ls" -map 0:a:5 -c:a:5 copy -metadata:s:a:5 channel_layout="Rs" -c:v copy output.mov"


  


  But output.mov file does not have any channel layout labeled to it (when checked in media info).

  


  Kindly help me with the proper command, so that I can convert the following file into a properly labeled (L R C LFE Ls Rs) 6 audio track, 5.1 (surround) file.

  


  Expecting a FFMpeg command to relabel the audio channel layout (in media info) from "C C C C C C" (array) to "L R C LFE Ls Rs" (array) and audio codec to be "5.1 (Surround)", audio count to be "6" retaining all video properties from parent file.

  


• ffmpeg : Continuous streaming plus recording at random time

  21 juin 2014, par Ralf Anderson

  It’s possible to stream and record the same input at the same time with ffmpeg. What I want to do though, is only start (and stop) recording at some random point later.

  Is it possible to somehow tell the active ffmpeg process (which is already streaming) to also start recording ?
  My first alternative was to simply start another ffmpeg process just to record, but the source input will be blocked by the first ffmpeg process streaming it and the stream will be blocked by the viewer displaying the stream on screen, so I can’t access anything to record. Can I change something so the stream/source input is not reserved for just one receiver ?

  In my concrete scenario, the source input is a webcam, which I want to continuously display plus be able to record at user-driven time. I stream the webcam using ffmpeg via udp to the localhost and can view it with a suitable player. I just can’t record it while also viewing and vice versa.

  I also tried the "Tee Muxer", which can generate multiple outputs with the same ffmpeg instance. I streamed it to the localhost on different ports, one to view and one to record.

  ffmpeg.exe -f dshow -i video="ManyCam Virtual Webcam" -map 0 -vcodec h264 -r 25 -f tee "[f=mpegts]udp://127.0.0.1:33333/|[f=mpegts]udp://127.0.0.1:33334/"

  This works fine, except for the delay of about 500ms, which I only get when I stream, not when I access the Webcam directly with ffplay e.g.. I was expecting no delay at all, since it’s all local, but I suppose this comes from the encoding ? I’ve tried streaming the raw video (setting -vcodec rawvideo), but I can neither view it with ffplay nor record it. Are there some special parameters needed for rawvideo ? After all, the input for the stream is rawvideo too, according to ffmpeg.

  I have not yet looked at ffserver, but I assume it would be no different to generating multiple outputs with tee muxer in my use case ?

  If I somehow could be able to display/record the stream generated by

  ffmpeg.exe -f dshow -i video="ManyCam Virtual Webcam" -map 0 -vcodec rawvideo -r 25 -f tee "[f=mpegts]udp://127.0.0.1:33333/|[f=mpegts]udp://127.0.0.1:33334/"

  then that would already suffice. Or changing the encoding in such a way that no compression is done at all should work just as well (if that’s causing the delay).