Customer came to us with a 21 hour, 27GB large mp4-file captured by a Logitech C920 camera (using Logitech's own camera software). This video captured the customer's point of entry to their parking. A car parked there for an hour, in the middle of this camera recording, and they need the car's registration number to pursue the issue.

However, the video file in question seems to be corrupt. My best guess here is that the software has an internal limit of mp4 length and overwrote chunks of data, corrupting a lot of headers/metadata. VLC, Premiere, Windows Media player, (any software), everyone reports "Unknown data". The 27GB file is definetly not empty or full of zeroes, so somewhere in it must there be a P-frame that contains the car of interest.

Does anyone know what we can do to try to extract all P-frames into separate png-files (or something like that), so we can analyse them?

If i remember correctly, R-Studio is able to identify frames of which this type of videos are composed as PNG files.
If you scan with R-Studio you may find thousands of PNG files all very similar.

_________________
My firmware database:
https://mega.nz/folder/O01DkBRI

If that's correct then it's definitely a good incitament for us to finally purchase a license to R-studio (it's been on the maybe-list very long).

Will look into it. Thanks.

I never heard of this. Maybe there were PNG files to be detected and saved in that instance.

Why not try repair the MP4? Or try examine how much damage there is using HxD or 010 editor (comes with MP4 template)?

To get some idea with HxD, https://youtu.be/LjKRegGXIHQ, skip my bitflip ramblings, skip to 09:40

_________________
Joep - http://www.disktuna.com - video & photo repair & recovery service

https://trac.ffmpeg.org/wiki/Create%20a ... he%20video
https://superuser.com/questions/604858/ ... p-b-frames

Should be worth to record a new file with this camera to see what it writes exactly or anything at all. mp4 ist just a container and doesn't mean much.