[reaven]

if the copy is a 1:1 I dont see any problem at all,

BTW whats the word Ultra you haven't receive any answer about this on the trac?

[msbob]

Well, CRC is calculated from stuff like file datetime and content of some of the files. So if these are kept intact you can do it without the original disc.
Now, some of these files can be encrypted with CSS and therefore give you a wrong checksum, if CSS has been removed from your files, but afaik mymovies gets by this by having multiple discid's and it's own algorithm for this.
So mymovies recognizes your backup's too - and so could XBMC, if mymovies webservice is used. I know this because I tested with both DVD and Blu-ray backup's I had done

Thanks for the info. Ran Process Monitor against it and it definately reads a pattern of offsets in VIDEO_TS.IFO among other things.

[davilla]

Is this ID the same ID as the DVD serial number ?

see http://lists.mplayerhq.hu/pipermail/dvdn...00963.html

[ultrabrutal]

BTW whats the word Ultra you haven't receive any answer about this on the trac?

He never replied, so I'm afraid it's a dead end for now.

Maybe someone else can reverse engineer the function?

[ultrabrutal]

This is most incorrect. The DVDID's on the site are user contributed. Where would we find DVDIDs to steal from if we are the largest DVDID database? That's just silly. Proper credit is also given for meta data sources and several Microsoft employees regularly use and contribute to DVDxml.com.

Yes, the DVDID's might be user contributed, but the meta data is obtained from Microsoft's service for MCE. This data comes from AMG (All Movie Guide) which Microsoft has a license for. Now you don't have to use MCE or even Windows to view this data without having any software or license yourself. Examples:
http://dvdxml.com/download/view.php?xml_id=47610
http://dvdxml.com/download/view.php?xml_id=47636
http://dvdxml.com/download/view.php?xml_id=47968

Notice that "..., All Movie Guide" at the end of the overviews?

Do DVDXML.COM have a license for AMG? That's the question to be answered.

I'm not Ultrabrutal, but I can answer this for you. The CRC64 data requires the ORIGINAL DVD. If you have a copied, burned, ISO'd or ripped DVD, it changes the CRC64. You MUST have the original disc.

The files used for the calculation must be unchanged yes. But like I write in another post, there are ways around that when using mymovies instead of DVDXML.com

[ultrabrutal]

Is this ID the same ID as the DVD serial number ?

see http://lists.mplayerhq.hu/pipermail/dvdn...00963.html

No I don't think so without looking at what mplayer refers to as DVD serial number. DVDID is also refered to as CRC64.

Do you post mean you are interested in this automatic and correct identification of movies?

[ultrabrutal]

Thanks for the info. Ran Process Monitor against it and it definately reads a pattern of offsets in VIDEO_TS.IFO among other things.

Sound's like you are on the right track to reverse engineer this yourself

[reaven]

He never replied, so I'm afraid it's a dead end for now.

Maybe someone else can reverse engineer the function?

you could maybe pm one of the developer so you can submit a patch for that ticket since you have the solution.

[ultrabrutal]

you could maybe pm one of the developer so you can submit a patch for that ticket since you have the solution.

I cannot give out this code sorry. Someone else needs to reverse engineer the code for XBMC use. Not even Wine has this function implemented to loan from

[ultrabrutal]

DvdId only works for dvd's and not blu-ray/hd-dvd's. Of course it would be nice to have proper matching of DvdId against mymovies/dvdxml to get metadata, but wouldn't we rather be able to match all different kinds of media?

I suggest we develop a new Open ID which we get users to submit to all the different meta data databases. If we are even cleaver enough, this ID can cover containers also (MKV for example).

Databases
How do we get interest for users to submit our open id to databases and how do we get database owners to add support for our id? Well the benefit for automatic correct matching of movies speaks for itself. No more need to search titles and end up with automatic picking a wrong title and having to manually correct this in the library.

TheMovieDB only has movie data and not info about releases.
MyMovies has info about releases (DVD, Blu-ray, HD DVD) and initial movie support (Posters and some of the data).
DVDXML has movie data.

I believe that MyMovies provides the best data and will give the best result in a media center. It has a huge dedicated following and a very big and complete database. The service is blazing fast compared to TMDB for example.

Uniqueness
Now what can uniquely identify a media? This is the really tricky part.
We know from CDDB that "chapter" length's are not enough to get a unique hit. This means we cannot use the chapters in a movie - solely.
A movie can be in wide or full screen, but still have the same chapters. This is not a problem if you scrape movie data (info not specific to the release, and using a poster), but if you scrape media data (eg. want the data from the DVD cover and the DVD cover it self depicted in the gui), it's not good enough.
If we use the resolution this will hit rips in containers since black bars are cropped off and resolution might also be lowered. DVD's have the same resolution for full-/widescreen but a flag to indicate 4:3/16:9.

Any ideas guys?

[msbob]

Concerning MM and DVDId, I looked and it is calling IDvdInfo2::GetDiscID through managed code. Blu-ray/HD-DVD identification is standard and easily to replicate.

I can probably reverse GetDiscID based on what process monitor is telling me it accesses. But, this is a big but, I don't DEV for XBMC. I work on Open Media Library. Since I'm windows only the only benifit of reversing GetDiscID for me is if it can be used against a VIDEO_TS folder (and process monitor is telling me this doesn't look promising). If you can tell me otherwise I'll spend the time to reverse.

As far as OpenId is concerned, an alternate approach to having TVDB and TMDB support OpenId would be to have a secondary lookup repository. My logic for this is;
1. You are not tied to a specific source for meta
2. You will have many formats and releases for a single movie tied to one meta resource (mkv, dvd, etc.)

As far as generating the OpenId, I almost suggest extending the CRC64 logic to whatever format the media is in. As in do a checksum "fingerprint" of the main video file.

The above is based on two assumptions. We are either talking about direct rips of DVD/Blu-ray discs to the FS or we are talking about scene releases. If we are talking about Joe Blow's provate rip of Full Metal Jacket from Blu to avi, there really wouldn't be anything to match against.

[ultrabrutal]

Concerning MM and DVDId, I looked and it is calling IDvdInfo2::GetDiscID through managed code. Blu-ray/HD-DVD identification is standard and easily to replicate.

I can probably reverse GetDiscID based on what process monitor is telling me it accesses. But, this is a big but, I don't DEV for XBMC. I work on Open Media Library. Since I'm windows only the only benifit of reversing GetDiscID for me is if it can be used against a VIDEO_TS folder (and process monitor is telling me this doesn't look promising). If you can tell me otherwise I'll spend the time to reverse.

Well if you reverse GetDiscID the algorithm can be shared with Wine and XBMC developers, right? CRC64 uses stuff like file datetime stamps and hashes of file content. Don't know if you can get all that from process explorer or if the DLL must be disassembled.
But still DiscID only works for DVD's, so it might not be a good idea to even spend time on this. Open ID is the solution if we can define a standard which all parties can agree on (huge undertaking I know).

As far as OpenId is concerned, an alternate approach to having TVDB and TMDB support OpenId would be to have a secondary lookup repository. My logic for this is;
1. You are not tied to a specific source for meta
2. You will have many formats and releases for a single movie tied to one meta resource (mkv, dvd, etc.)

As far as generating the OpenId, I almost suggest extending the CRC64 logic to whatever format the media is in. As in do a checksum "fingerprint" of the main video file.

The above is based on two assumptions. We are either talking about direct rips of DVD/Blu-ray discs to the FS or we are talking about scene releases. If we are talking about Joe Blow's provate rip of Full Metal Jacket from Blu to avi, there really wouldn't be anything to match against.

Of course we are never talking about scene releases. Open ID should work on DVD, Blu-ray and HD-DVD, but if cleaver enough then if you rip your own media to for example a MKV and keep the properties needed for Open ID calculation in the container (chapters and video info or whatever), you will be able to get a 100% match on your MKV file. Open ID could of course also be tagged into the MKV which could be used instead of calculating a new one (speed or if properties are missing). Ripper software should support Open ID so it could be embedded automatically.

[msbob]

Well if you reverse GetDiscID the algorithm can be shared with Wine and XBMC developers, right?

Right, if I end up going that route (and am actually successful), what I do is open source. I wouldn't be providing you guys a patch, but as I said it is open source.

But still DiscID only works for DVD's, so it might not be a good idea to even spend time on this. Open ID is the solution if we can define a standard which all parties can agree on (huge undertaking I know).

As I mentioned above, the identifier for HD-DVD and Blu-ray used by MM is easy to replicate. A DiscID is valuable because it is an identifier that has already been assigned meta.

Open ID should work on DVD, Blu-ray and HD-DVD, but if cleaver enough then if you rip your own media to for example a MKV and keep the properties needed for Open ID calculation in the container (chapters and video info or whatever), you will be able to get a 100% match on your MKV file. Open ID could of course also be tagged into the MKV which could be used instead of calculating a new one (speed or if properties are missing). Ripper software should support Open ID so it could be embedded automatically.

Don't get me wrong here, I think the idea is great. It just doesn't seem like a realistic solution. You are relying on either rippers to maintain consistent info in the container or rippers to generate the id from the disc and add it to the container.

[reaven]

As I mentioned above, the identifier for HD-DVD and Blu-ray used by MM is easy to replicate. A DiscID is valuable because it is an identifier that has already been assigned meta.

even when almost everyone media library its digital now days this is still a good idea now or then you can pop up a disc a rental or whatever movie in disc format and still get almost all the benefits of a digital one, this is the whole purpose of this "patch". why would you want openid, discid or dvdid to stick to the movie if you rip it you have the scraper for that anyway.

this is not a high priority "patch" but is a must have to still have one of the best media center !

[ultrabrutal]

why would you want openid, discid or dvdid to stick to the movie if you rip it you have the scraper for that anyway.

Why? Well here are two points:

1. Scrapers don't match a rip on anything but the Title or IMDB ID. This means you get movie meta data and not media meta data.
2. You get a 100% match instead of a likely match. Scraper will search for title and it might get 50 back from the server. From this list it must pick the most likely, and it can go wrong here. Also if the title is not 100% correct then you will get a faulty result or perhaps none at all.