Use MediaInfo to read metadata 'tags' in video files and import into datbase/library

[Gamester17]

Could Boxee integrate MediaInfo (http://mediainfo.sourceforge.net) metadata reader code and use it when scanning video files into the database?

MediaInfo supplies technical and can read tag/atom information about a video or audio file containers.

MediaInfo is coded in C++, is cross-platform (Win32, Linux, and Mac), open source code (GPL/LGPL dual-license), and can be compiled into a .dll (Dynamicly Linked Library) on Windows, .so (Shared Object) on Linux, .dynlib (Dynamic Library) on Mac.

What information MediaInfo will get from looking at a container:
General: Title, Author, Director, Album, Track Number, Date, Duration, and much more.
Video: Codec, Resolution, Pixel Aspect Ratio, FPS, Bit-rate, and much more.
Audio: Codec, Sample Rate, Channels, Language, Bit-rate, ReplayGain, and much more.
Text Language of Subtitles (SRT, SSA, ASS, SAMI, and many more).
Chapters: Number of Chapters, List of Chapters (names).

Examples of just some of the supported formats:
Video: AVI, MKV, OGM, DV (RAW), MPEG-1, MPEG-2, MPEG-TS, MPEG-4, WMV, MOV, IFO/VOB (DVD-Video), CDXA (VCD/SVCD), Blu-ray and HD DVD files, and more.
Audio: MP3 (Id3v2), OGG, WAV, PCM, RA, AC3, DTS, AAC, M4A, AU, AIFF, WMA, APE/MAC, FLAC, and more.


about integration of metadata in MediaInfo, my goal is to have all the
metadata available in a file, so if data is present, what is request ("Right
now it allows for the following atoms (new atoms can be defined as well) Show
Name, Episode ID, Episode Number, Season Number, Cast Members, Producers,
Directors, Writers, Original Air Date, Original Network, Long Description (Plot),
Artwork, Video Kind (TV Show, Movie, Music Video) Rating, Genre.") is
easy to have.

Checkout the MediaInfo website for more information:
http://mediainfo.sourceforge.net

Source code (and SVN):
http://sourceforge.net/projects/mediainfo

PS! MediaInfo could also be used to read tags in audio files but that is not what I am requesting here.

You can follow the dicussion about this in the XBMC forum here:
http://xbmc.org/forum/showthread.php?t=12721 :)

[iBog]

I also recommend the ability to read in meta data from existing files.

I spend a lot of time and effort to tag my files and don't always like the data that Boxee finds (especially for the descriptions which are too long to be displayed).

If a file is not tagged, it would be useful for Boxee to write meta data to the file.

[Gamester17]

I also recommend the ability to read in meta data from existing files.

If a file is not tagged, it would be useful for Boxee to write meta data to the file.That is exactly what my feature request is ;)
Was that not clear enough :confused: too technical?

Anyway, yes we are on the same page :D

[marcel]

yes too technical...please make it simple for guys like me:)

[Gamester17]

More people requesting this, see:
http://forum.boxee.tv/showthread.php?t=690
Boxee has the same problem that Plex and everything else based on XBMC has. It completely ignores all the metadata in my video files,...

...In short, read metadata from my video files, and sort accordingly.MediaInfo's author, Jerome Martinez (a.k.a. Zenitram), have now even said that he would be willing to work with any volunteering XBMC/Boxee/Plex C++ developer(s) in order get MediaInfo library working in XBMC/Boxee/Plex, (his e-mail address is "zen at mediaarea dot net" if a C++ programmer would like to contact him about this).

;)

[iBog]

Gamester... I got the intent of your original post. Just wanted to say I supported the idea and also clarified it a bit.

I have another post (http://forums.boxee.tv/showpost.php?p=4843&postcount=13) that describes the workflow I'd like to see for supporting metadata.

I'd like to see Boxee try to read the metadata first before attempting to look it up and if it looks it up I'd like the option of writing the missing or updated data back to the file so if I have to rebuild the library DB it will automatically get the latest data.

I wish I were a hard core programmer to help out.

[marklight]

This is my #1 feature request too...

[Gamester17]

I have another post (http://forums.boxee.tv/showpost.php?p=4843&postcount=13) that describes the workflow I'd like to see for supporting metadata.My own idea of a IMHO better process chain is described and discussed in the XBMC Community Forum here:
http://xbmc.org/forum/showthread.php?t=37206
and also spread over to this other discussion thread:
http://xbmc.org/forum/showthread.php?t=14186
and here too:
http://xbmc.org/forum/showthread.php?t=30161
as well touching the subject here:
http://xbmc.org/forum/showthread.php?t=33710

;)

[iBog]

I read your first link and you provided a very detailed and thorough explanation of the process and completely agree with it.

The only additions I'd suggest are:

1. Use mediainfo to read data from the file tags. This should be done before imdb scrapping.
2. Use as much information from the file tags as possible to improve the accuracy of imdb scrapping. I'm glad you acknowledge that this may not be perfect.
3. I think there could be a confidence or accuracy rating when using imdb scrapping. This could also be combined with a preference for when the user is willing to be prompted (for more information, select from a list or skip).
4. From the library, manually launch the imdb scrapper for selected entries.
5. There could also be the ability to write metadata to the file.

What do you think?

[Gamester17]

5. There could also be the ability to write metadata to the file.This is the only part I do not agree with. As a media center software it is not Boxee's job to write metadata to the files, in fact Boxee's media center software should not touch any of the files or folder structure. Boxee media center software should only write to its SQL databases and at most be able to export those SQL databases into XML formatted NFO files like XBMC can, see:
http://xbmc.org/wiki/?title=Import_-_Export_Library

Editing the tags of media files is the job of a other software that is not a media center. Maybe in the future Team Boxee could produce such tag editor software as well as a sort of 'media companion', but in my opinion that should be a separate software application from Boxee's Media Center software application. I have created a separate forum discussion thread about that here: http://forum.boxee.tv/showthread.php?t=991

:)

[iBog]

Boxee is already doing so much work to try and identify the media and look up metadata. In some cases, the user may even have been prompted for missing data or to confirm what the media is.

I am not totally convinced with Boxee's accuracy so I wouldn't want it writing the tags. However, once confirmed to be correct I think writing the data there would be a great service to the user. Once tagged, the media could be reimported into boxee (if you have to rebuild the library), imported into another boxee (if it resides on a NAS) or used by any other media player.

I expect writing the tag to have a significant impact on the performance of Boxee. This is the only reason I may not agree with it. Exporting to a separate text file could be useful (if other applications support import) but not as elegant. I currently use a few different metadata editors to tag all my media (the reason I want Boxee to be able to read it). These editors don't always lookup the data so it's still time consuming to tag files the way I want. Since Boxee is already going through significant effort to do this I'd like to see the effort not wasted by tagging the file when done (and when I want it to happen).

I don't disagree with a standalone companion but then it may no longer a complete or integrated solution.

[erikveland]

Again, another voice added to the choir. Even writing metadata if confirmed correct by the user first!

[agentlame]

i'm with gamester17 on this... i don't want boxee messing with my files.

[giyad]

This is the only part I do not agree with. As a media center software it is not Boxee's job to write metadata to the files, in fact Boxee's media center software should not touch any of the files or folder structure. Boxee media center software should only write to its SQL databases and at most be able to export those SQL databases into XML formatted NFO files like XBMC can, see:
http://xbmc.org/wiki/?title=Import_-_Export_Library

Editing the tags of media files is the job of a other software that is not a media center. Maybe in the future Team Boxee could produce such tag editor software as well as a sort of 'media companion', but in my opinion that should be a separate software application from Boxee's Media Center software application. I have created a separate forum discussion thread about that here: http://forum.boxee.tv/showthread.php?t=991

:)

I'm a little torn... I agree, I don't think that Boxee should touch the metadata of the file, but I recently connected my video files and Boxee recognized all the files and downloaded their info/cover art. If that stuff gets written to the actual files, that would save me from ever having to do it.

I think iBog's suggestion is a perfect solution. The user should be prompted to confirm a tag, but Boxee should only write metadata to files that 1) are not already tagged (this would save from having to confirm every file), or 2) the user wants to write over the original tag (maybe the tag was incorrect).

[DGMurdockIII]

this is perfect

[Carniphage]

Currently the movie library shows only identified (or misidentified) content.

Placing movie files in the right directory may result in the movie appearing in the library. But if it does not get identified, the file does not appear. There are no error messages. No notification. There is no way to play the content. And no way to trigger a manual identification.

The only way to solve the unidentified movie issue is to go to the file-browsing screen. The file browsing screen does not clearly mark unidentified files either.

A better way would be to have unidentified movies appear in the library - but flagged as such. The title should be the filename.

With this proposed method, the user would open the movie page, browse all the content - and with unidentified movies, the user could trigger an identification. Or just play the file.

Identified files would be marked (with full cover art).
Unidentified files would have a question mark.
Files in the process of being identified would have a syncing gear or something.

C.

[solefald]

pretty much agree with everyting, except the question mark for unidentified movies. Movie thumbnail should be sufficient here.

Also please give us ability to assign our own movie posters (for all those foreign or generic movies not listed in imdb). None of my snowboarding movies get recognized because they are not in imdb.

[alexreusch]

Don't forget, that many people who want to manage their online DVD collections normally rip their DVD's into the native TS folder structure (as it is on the original DVD), without any compromises on quality etc. This is a best practise here.

So there should no info to be written directly in the media files. What is normally used is a DVD collection management software, which stores data in a local database or writes additional meta data in the movie folder (for example: separate file in XML format). Also, there is very often the possibility to synchronize the local database to the web. So in other words, the data is also available in a online database as well.

So what we need is, that BOXEE supports other online databases beside IMDB. Because IMDB is not optimized for DVD collections and does not support localization, different versions, languages, localized cover arts, bonus features, multiple disc editions etc.

For example: Online access to the "My Movies" database, "DVD Profiler", "Collectorz" etc.

The above examples are written for DVD collection management and therefore optimized for this purpose.