Dreamboard

There is a new plugin, ReconstructApSc, which makes it possible to reconstruct missing or corrupt .ap and .sc files of recorded movies. These files are used for precise seeking and winding in the movie, and also for finding the cut positions when using the plugin MovieCut to cut movies. This plugin scans through .ts files, and stores the structural information into the .ap and .sc files.

Examples of situations when this plugin may be useful include:

• Movies recorded with Enigma2 from before the spring of 2009 doesn't have .sc files, so rewind/fast forward won't work properly.
• The previous version of MovieCut did not know about .sc, and therefore produced no such files for the cut movies.
• If you have cut your movies on a PC, or downloaded movies from the web, you may not get any .ap or .sc files.
• If a specific movie is impossible to seek correctly in, or just gets black or a frozen picture when trying to fast forward or rewind, then one may suspect that the coresponding .ap and .sc have got corrupt for some reason.
• After a disk crash it may be possible to rescue the .ts files (because their specific structure) but perhaps not the others.
• Maybe you have removed the .ap and .sc files because they took up disk space and you didn't know what they were used for.

Once installed, the plugin shows up in the menu in the movie list, as "Reconstruct AP/SC". When started, it will give you three alternatives:
The plugin is available from the Plugins menu (the experimental feed). I hope that it can be of use.

Best Regards
Anders Holst

Quoted

But where exactly can I download it?

If you use the "experimental image" (as opposed to the more stable "official release"), then the Plugins menu will automatically give you plugins from the experimental feed. With the Plugins menu I meant:
Main menu -> Plugins -> Download -> Extensions -> reconstructapsc

Alternatively you can fetch it directly from http://dreamboxupdate.com/opendreambox/1…5/experimental/
Scroll down to "enigma2-plugin-extensions-reconstructapsc_experimental", select the version of latest date, and install it manually on the box.

Best Regards
Anders Holst

I have tried to look into this, but have not been able to repeat it, mainly because I have no access to a movie of the kind that causes problems. (HelgeBS didn't have any suitable movies for the moment, unfortunately). If anyone can offer me a small piece of a problematic movie, some 20 seconds or so would be sufficient, I could look into it again.

There were some speculation on another forum that it had to do with a missing or corrupt .meta file, but that seems not to be the case - that should have slightly different symptoms. My guess is that the movie has a format that is not recognized by the current version of ReconstructApSc, and that the .ap file gets too small, maybe empty. My guess, without being able to confirm it, is that removing .ap only, but leaving .sc, will make the movie viewable again.

Anyway, the behavior is not as designed, so if I can have a look at the format, I could at least make ReconstructApSc recognize it and just skip .ap if it can't produce it correctly.

Best Regards
Anders Holst

DartMaul40 wrote:

Quoted

I don't have a .meta file either! That is the problem?

No. A corrupt .meta file could be a problem. A missing .meta should not.

Anders

Quoted

If you want, I will cut a few seconds from one of those movies and send it to you.

I tried to send you an email, but maybe it failed.

Yes please, but if you first email me the full .ap file I can check it to find an interesting section to cut out, and email you an appropriate command to cut it out from the movie.

Best Regards
Anders Holst

I have now analyzed two example movies that I got (thanks DarthMaul40 and HelgeBS). There were in total three different problems with them. Below are the technical details for those who care about such things.

For those who don't, here is the synopsis: Download a new ReconstructApSc with the plugin manager. It should at least make the movies playable. Patches have been sent to DMM for the problems with winding and length calculation.

Details:

One of the movies did not play at all with the .ap file constructed by ReconstructApSc, although it could be played without that file. The movie was SD but very loosely packed so it was large as a typical HD. The problem turned out to be that normally the sequence start markers and picture start markers are in the same ts-frame, whereas in this case they were in consecutive ts-frames. ReconstructApSc has up to now stored the position of ts-frames with picture starts, but apparently it has to store it to sequence starts instead to work properly. I have updated ReconstructApSc to reflect this, and it is available for download the normal way.

But there was another problem too with this movie: After having corrected the .ap-file, it was possible to play it, but not rewind it nor wind faster that x8 forward. For this the .sc-file is used. However, the problem is not with the contents of that file, which is correct, but with the way it is used by enigma2: It jumps to consecutive I-frame picture starts when fast-forwarding or rewinding, but just as above it should jump to sequence starts instead. I have prepared a fix for this and submitted to DMM. It may as usual take some time until it takes effect though.

Now to the second movie. This was a HD movie (or at least it was in the H264 format) but very highly compressed so it was not larger than a typical SD. It played alright, but the problem was that with the .ap-file from ReconstructApSc the displayed movie length was far too short, and the position gauge jumped back and forth during playback. It turned out that this was because the I-frames was very far apart, up to almost 10 seconds simetimes. Currently enigma2 interprets gaps between I-frames over 5 seconds as a "discontinuity" in the movie, and when an .ap-file exists, it is used to find those gaps and compensate for them in the length calculation and position gauge. Compare a situation when this feature works as intended: A movie with a two hour discontinuity in the middle, which without the .ap-file would seem two hours longer than it is, but with the .ap-file this gap is detected and compensated for. However, if up to 10 seconds can occur normally in the movie, the threshold for when it is considered a discontinuity should be increased to (at least) 10 seconds. I have submitted such a patch too to DMM, and again it may take some time before it is applied.

Something I have not done is to make ReconstructApSc construct the .meta file as well when it is missing. This might possibly be done some day in the future.

Best Regards
Anders Holst