Recherche avancée

Sur d’autres sites (5889)

• FFmpeg truncates untrunc'ed video

  18 mars 2020, par petrpulc

  Background : Colleague's Panasonic consumer camera had issues with high-capacity SD card and yielded two different types of truncated files (mp4 and mdt) with both header and index corrupted. Both were "fixable" with following steps :

  



  

  1. Locate with hex editor the ascii sequence "mdat" in both a truncated and a corresponding known good file. In mp4 the "mdat" is quite deep in the file as it contains some (but bad) headers, the mdt type almost starts with it.

  


  2. Replace everything up to the "mdat" sequence with known good header. BUT this also means that metadata on length are copied over.

  


  3. Run https://github.com/ponchio/untrunc on the file with corrected header (it refused to run on original files).

  


  4. Profit ! VLC plays the file fine, ffprobe shows correct length. BUT ffmpeg encoding stops at the end timecode of the file I taken the known good header from in step 2.

  


  5. A somewhat ugly solution is to recode the file in VLC (video copy seems to be breaking file up again).

  


  



  I am wondering if I overlooked something ; big time. I guess I do, but simple timecode operations on input file do not seem to have any effect at all.

  



  Is there a way to persuade ffmpeg to encode the whole file that ffprobe and VLC see ?

  



  Files may be provided for analysis on personal basis only, sorry.

  


• Different results between windows and Linux in opencv c++

  24 mars 2020, par Samer

  I have tried this opencv and ffmpeg C++ code to demux a video and edit it and mux it back again but it doesn’t seem to work properly on windows (using Visual Studio)

  the problem occurs by skipping some frames especially at the end of the video and increasing the video output size not on all windows PCs. We have tried it on a 35 MB video and when we did the editing the output video size was 2.8GB on some windows PCs and the last frames skipped (the problem) and some other PCs the output video size was 500MB and there were no skipped frames (all fine).

  so the question is, why would the same code with the same build and windows versions behave differently and produce problems on some Pcs ?

  (we also did it on linux ubuntu and macos and it was working fine on different PCs)

  https://github.com/WajdiMuh/parallelalgo for the files

  https://gjuedujo-my.sharepoint.com/personal/m_albizreh1_gju_edu_jo/_layouts/15/onedrive.aspx?id=%2Fpersonal%2Fm%5Falbizreh1%5Fgju%5Fedu%5Fjo%2FDocuments%2FParallel%20Project%2Fbinr&originalPath=aHR0cHM6Ly9nanVlZHVqby1teS5zaGFyZXBvaW50LmNvbS86ZjovZy9wZXJzb25hbC9tX2FsYml6cmVoMV9nanVfZWR1X2pvL0VzaHFWNDJCVThkQ25KOFV0dm82NHJNQmdoOWdCeEZOblkwQWtRMkQ0MU5UV3c_cnRpbWU9TTl4Zi1YblAxMGc for the release version (vidd.MP4 is the video)

  Thanks in Advance

• How to mix multiple audios from the end ?

  24 avril 2020, par Vijay Singh Kholiya

  In my project I have a feature of recording the meeting which is in a form of group discussion consisting of multiple attendees. All attendees will be having their own personal system and they will be accessing that web app through their personal account. Their is one meeting controller(only one person) who starts the recording and for every attendee whether he is present in the same room or is remote, their microphone will be opened and recording will be stored in S3 bucket.

  



  Since I need a single audio file which consists of the whole meeting recording, so I am using SOX for this. But the problem is, what if any attendee attends meeting in between the discussion so he's audio recording length will be smaller than everyone else.

  



  Since I am mixing the audio files to get one file, but for this late commer attendee the mixing will be wrong. So how can I mix the audios from the end instead of starting ?