I have a relatively new Samsung HMX-S10 camcorder that I have just started using. I captured a dozen or so video clips of my daughter since she was born, but today when I got done recording one, the camera popped up a message that the SD card wasn't formatted. The card was indeed formatted when I put it into the camera, and had not been removed before this message popped up.
Knowing this was a bad sign, I immediately went to my computer and ran a few different image/video undelete tools. Each one was able to find a bunch of mp4 and mov files. The mp4 files are all around 66kB and the mov files vary. On my mac, Finder shows that the mov files are 1280 x 720 and some number of seconds long (between 10 and 40, roughly). The mov files will play in quicktime, but are just a black screen. I can't get the mp4 files to play in VLC or QT.
Also recovered is an XML file that describes the codecs. It shows H.264 as the video codec and AAC as the audio codec. I'm thinking that either the mov and mp4 files are the audio and video streams or that one is the container and the other should be wrapped in it. Is there any way to tell this? I have a macbookpro and a Fedora desktop available to work on these files. If anyone can help me out, it would be greatly appreciated.
+ Reply to Thread
Results 1 to 13 of 13
If it is helpful, when I try to play the mp4 files with VLC, this is the first error that I get:
main debug: looking for demux module: 52 candidates macosx debug: input has changed, refreshing interface mp4 debug: found Box: ftyp size 32 mp4 debug: found Box: free size 65496 mp4 debug: skip box: "free" mp4 debug: dumping root Box "root" mp4 debug: | + ftyp size 32 mp4 debug: | + free size 65496 mp4 debug: unrecognized major file specification (avc1). mp4 error: MP4 plugin discarded (no moov box) macosx debug: notification received in VLC with name VLCOSXGUIInit and object VLCEyeTVSupport avformat debug: detected format: mov,mp4,m4a,3gp,3g2,mj2 avformat error: av_open_input_stream failed
The mov and mp4 are normally the containers, while the video and audio data is identified as you mentioned above.
See this post about a free utility:
Also, some have had success with a program called photorec, but I'm not sure if it runs on Mac or Linux.
This has become an increasingly common topic lately. If you could upload one of the files the group could take a crack at it. I was just working on one with a incomplete header.
Thanks for the tips. I'll take a look at the utility. Photorec does run on Mac, Windows, and Linux and is the program I tried first. I made an image with ddrescue before doing anything else, so I am now looking to see if I can just pull out the files manually from the binary file. Will post results.
What I have determined from looking at a working video file from the same camera is that the *.mp4 files and the *.mov files that I am getting are two parts of the same video. It would appear that I could concatenate them somehow to fix the video file. However, some quick tries using cat (eg. cat file1.mp4 file2.mov > testvid.mp4) failed to produce a working video. Since this is just looking at the binary file, I decided that maybe rather than trying to piece back together the two files, I might be able to cut one binary file out of my binary disk image. I've found the pattern that marks the beginning of the file, so now I just need to figure out how to find the end of the file. Any thoughts?
Post the two files here or at a file hosting site (and put the links in here) For example, mediafire.com -
and somebody will take a look. Always good to get a second opinion.
OK, here are the files I am currently working with. Let me know if any clarification is needed. These are the first two in the recovery directory so I am assuming they go together (if not, it's going to be challenging to figure out which go with which).
On a related note, anyone know about endian-ness of data coming off of a camera? I've managed to cut some videos out of the binary disk image, but they are just random blocks of color. I'm wondering if it could be an issue of the video data needing to be re-ordered.
Not sure if this is an improvement or not, but this (http://cbeal.stanford.edu/research/testclip.mp4) is a clip cut contiguously from the binary disk image, with the approach I mentioned before. It doesn't want to play on VLC or QuickTime, but it does appear to have the exact same structure as the working video (http://cbeal.stanford.edu/research/workingclip.mp4). With mPlayer for Linux, it plays, but shows as random color blocks. Can anyone discern an underlying difference between the two? The working hypothesis is, of course, that the data itself is not corrupted. This could be incorrect, in which case I am hosed....
cbeal, I wasn't able to get anything out of the files you posted above.
They're so small, not likely to be worthwhile, perhaps a second or less.
The video I posted last (testclip.mp4) is of the same form and has a lot more data. I think the recovery tool only recovered a small part of the data for the video. How were you checking to see if there is any content? I am only looking with a hex editor, but perhaps running this through some video tools might give more information?
The two file I looked at were the 64 and 102 KB files you posted above. I didn't say they had no content. I looked using a hex editor also. I see something that looks like a header, some binary zeroes, and some data.
However, I couldn't get anything useful to play.
I had success with a tool somebody posted earlier in an older thread. Give it a try on yours.
Last edited by davexnet; 20th Apr 2011 at 10:12.