-
Corrupted ProRes and QuickTime Atoms
We have recurring problems of corrupted files MOV/ProRes.
It looks like the image below :
We have tens of examples.
Opening a file with FCP7 alters it, you drop it in the bin and it is altered. The modification date and checksum was updated. It’s a serious issue.
The connection between opening, corruption and artifacts is proved true.The corruption by FCP7 consists in inserting an atom type moov which is added to the one of origin, I read Benoît Joonsen’s blog, in particular the post about NLE softwares adding a moov without deleting the previous one.
But many things show the atom is bad :
location : it is changing but far enough in the file to be abnormal
parsing (with QT Atom Parse) confirms this, the moov is not parsed and is right in the middle of the mdat atom
feature : except one, all files weighed over 4 Gio. , supernumerary moov always displays an atom type stco and not co64 as the original moov at the end of the file.
Artifacts TC do not always match with offsets (rule of three)
injection is inevitably destructive because there is no file rewriting. Moreover, I have not seen any padding atom, type free or wide, at the original location of the moov injected into the modified file
FFMpeg notes (verbosity level error) on some of the affected files errors type “invalid data size” (many) and one error type “invalid frame”Furthermore, opening the file another time causes a new change at the same place or elsewhere. I see thus feature films (> 100 Gio) with up to 60 of these additional moov.
Amazingly, playback on Windows or Mac PC, whether with VLC or QT, corrected a lot of flaws. I guess it is the robustness of the QT engine. But on a Linux machine, VLC or MPlayer with some files presents a lot of of artifacts on several frames.
Transcoding / re-export may remove moov, it can also be blank the artifacts or make them visible. Repeated play with QT with rewind, frame by frame playback and the artifact may appear different.
I have submitted the case to Apple with an affected file. They could not read it either with QT or VLC MPEGStreamClip. We have no problem reading or transcoding.
Since, moreover, that the media seems to have understood nothing, it’s black out despite my mail or phone reminders.
Two of our softwares, with their own ProRes codec, does not accept these ProRes.
PremierePro, and other CC’s softwares, have no problem.Given the magnitude of the problem, it is surprising that we have no more feedbacks on the web. I have heard both companies having found the similar problems. But an analysis of several files, including from outside, before any intervention, shows that files with moov supernumerary is majority.
I conclude that we may be a widespread professional software that corrupts files and QuickTime systematically is “hiding the dust under the carpet”.
Do you know this issue ? Is this trivial for you ? How to cure that ?Done tests :
- FCP7, FCPX seemingly not affected
- local/NAS (NFS & AFP)
Context :
- OSX.6, OSX.8, OSX.10
- MP3,1, 4,1 & MP5,1
- FCP7, 7.0.0 & 7.0.3
- QT7Pro
- Ethernet MTU : 9 kB
.
-
5 Replies
-
This is easily resolved. Stop using 7 year old software that was discontinued 5 years ago.
(That’s my snarky response for the day…)
Shane
Little Frog Post
Read my blog, Little Frog in High Def -
Thanks for the answer.
But :
– OSX.6 is affected, obsolete too, yes I know, but FCP7 is supported until OSX.7
– a professional video software modify a file just by opening
– this modification is a corruption
– this corruption leads sometimes to visible artifacts
– this artifacts are hidden, most of time, by QuickTime during playback. Another prolem.
– FCPX have many inconvenients (i.e. MXF playback, no VTR ingest or output)After that, we have found alternative (FCPX, PP CC and BM MediaExpress). But given several files from outside are affected, FCP7 is still widespread.
This message is both a question (have you meet this issue ?) that a feedback.
Sorry for my english writing.
Sorry for my english writing.
-
I have never seen this issue. And I edited many broadcast TV shows with it over the course of 6 years. First I heard of any of this.
Shane
Little Frog Post
Read my blog, Little Frog in High Def -
We wonder since how long we have this problem. We process too many programs for broadcast (TV etc.), without issue.
FCP7, like QT, hides most of artifacts. No editor in our company has seen fault. It’s detected before or after process or editing.
Most recent example : trailer in ProRes (HQ) : one artifact, one bad moov. Transcoding (PRHQ always) with PP CC : just one, good, moov, no artifact.
We don’t know when a bad moov will induce artifact, rarely. It depends on this location on the mdat flow, if it overwrites a essential part of frame (icpf atom).
Thanks.
-
Just to say how to check ProRes/MOV.
OSX.6*, in Terminal :
grep -bao moov fileOutput :
Offset:moov
...If you get more one moov, the last or the first, the nearest of beginning or end of file is the right.
And, this file is potentially affected by artifacts.
* : Strangely, grep command doesn’t work on big files on a OSX.10.
Reply to this Discussion! Login or Sign Up