I would like to have as a feature of DVD profiler (or some other program included with the download package) a way to play a DVD and do the following:

1. Mark the start and end of the opening and/or closing credits.

2. After marking the start and end of the opening and/or closing credits start copying the text of the credits to a temporary text file while playing the credits, proceeding one screenful at a time and using Optical Character Recognition (OCR) if necessary. While doing this, the original movie credit text would be displayed on one side of the screen and the copied text on the other. The text on the copied side of the screen would be editable, so that any errors in the copy process could be corrected right away. When done with a screenful of info, you could press a key (such as "Page Down") or click a mouse button to advance to the next screen.

A spelling and grammar checker for this text editor would help to quickly spot errors.

3. When done with this, prompt to save the text file where you want to save it.

4. Import, or provide a means for importing, the text from this file into the appropriate sections of DVD profiler.

This would make it MUCH EASIER AND QUICKER to get full cast and crew for contributions, and would SUBSTANTIALLY REDUCE THE NUMBER OF CONTRIBUTIONS WITHOUT PROPER CAST AND CREW INFO.

Did you check the DVDPca thread? I already have code in progress, but nobody seemed interested when I mentioned that I was working on it.

So, now I guess I'll get back to working on it.

Unfortunately, whatever I do will probably be free. 

What I've done so far:

- embedded the Microsoft OCR package
- coded a technique for converting the image to black and white and enhancing the contrast
- run the OCR and put into a text window

What remians:

- make a convenient way to mark/capture successive credit frames, or automatically detect the scrolling speed, plus top / bottom of the screen. I haven't tried that yet.
- detect the orientation of the credits. If I can't, the user will have to tell me what the "format" is
- experiment with editing techniques. It will probably be easiest to copy into MS Word window, so you can use all of its editing features, then provide a way to scrape back into the profile.

Example: after the imperfect OCR, maybe the text looks like

Jobn Joe .udio nginee...
Jane Doa .irecto..

The user would edit the file to look like:

[F] John [L] Doe [R] Audio Engineer
[F] Jane [L] Doe [R] Director

Where [F] = first name, [L] = last name, [R] =role

Something along those lines. it remains to be seen how much I can do, and whether it will be any easier than simply typing (or speaking) the information in.

Any ideas along these lines, please post here and/or in the DVDPca thread.

(BTW I have also thought about grabbing the Overview from the back cover scan.)

@Parsec,

Well I haven't exposed the OCR part yet. But the incomplete code is in there. And the issue you bring up is one of the reasons that it isn't finished. I have been waiting for an idea of how to make it easy to correct the OCR mistakes.

Now as far as the other method, I thought that there was already a couple of good Cast/Crew editors already in place, e.g. the one by DJ Doena. So I haven't played in that turf. There is enough work to go around that I try to avoid re-inventing the wheel.

So, take a look at that one first, and report on how your idea is the same or different. Then maybe DJ would be willing to make some changes. If not, we could explore putting something like your idea into DVDPca, but it would not be anywhere near as sophisticated as the other plugins.

(I like your idea, but I'm not sure how / whether it is different from what is already available - not to mention that DVD Profiler itself has a pretty easy way to add cast/crew not unlike what you suggest.)

Also, no offense to DJ Doena, but the freestyle cast/crew editor is just that. It doesn't pull new entries from a db or have any name completion and when the plugin is active you can't do anything else within DVDProfiler until you close it's window.

OK, so let's keep thinking. Based on what I have learned here, there are two cases that we could explore:

(1) There is no data that can be grabbed from elsewhere, cause the DVD is new / rare / old or whatever.
(2) User prefers NOT to get data from elsewhere, for whatever reason.

In these cases, perhaps it would be useful to have a front-end to the headshot DB, allowing creation of cast/crew data for import into FCCE or DVDPca or directly into DVDP, per Parsec's suggestion.

I would still like to play with OCR to see where it leads, per the original suggestion.

Any other thoughts?

Hmm we could really open up a can of worms here. Personally I like databases. ok it's a weird like to have, and normal people look at you strange when you say that.... I like the headshot db because just through sheer luck it has an enormous list of cast/crew within it. It also has the accepted BY's as well - no need to go checking for that.

I think if we had DVDProfiler built extra db's like headshot and plugins that could access these extra public supported databases for various purposes, that would be great. I'd even like something to use the headshot db without having to import the entire thing just to get the headshots for the cast/crew in my own profiles, I really only need the ones for the listings I have.

Ok I'm straying a bit here, but I kinda like the vision of having huge extra db's of cast, studios, distribution (media) companies, even retailers, featurette and audio tracks info - all created by and maintained by DVDProfiler owners, like headshot is, and then have plugins that can access these in order to assist creating correct profile entries... The extra db's could even hold extra info not used in dvdprofiler - for various things (imagine looking up a media company to find their website address so you can go check their listing of a certain profile)

Wish I could program like you guys do.

Quoting mediadogg:

Quote:
...
I don't even understand the issue that surfeur51 brings up, why it is caused by one method of entering names, and avoided by the other ... sigh ...

I'll try to explain with one example :

Movie 1: credits have François Berléand (this is correct), you enter François Berléand
Movie 2: credits have François  BERLEAND (this is correct), rules ask to enter François Berleand (this is incorrect), and you create a name variant with a spelling mistake.
Movie 3: credits have FRANCOIS BERLEAND (this is correct), rules ask to enter Francois Berleand (this is incorrect), and you create a second name variant, with two spelling mistakes.

So, using Invelos rules, you create two incorrect variants that exist only in Invelos database (and in fact not in credits). The three variants are in CLT and do not link.

If your plugin uses Invelos rules, you'll have this problem.

DJ Doena's plugin uses, for the three movies, IMDb name, which is, very simply, the name of the actor correctly spelled, François Berléand. No variant, perfect linking, but forbidden by rules.

Some contributors have the skill and inclination to use program(s) mentioned here, but not the source material. Others of us have the source material, but not the time, skill or inclination to do the good works, so ...

I was thinking about whether it might be possible to do credit video clipping on the side by one user that would then share the clip with another.

It's obvious that DVDProfiler could have nothing to do with the exchange process, but could benefit from a more complete database.

I also think this would be rather lower on the list of priorities for MPAA / RIAA legal persecution, since ultimately the project is about fostering the buying, collecting and enjoying of movies, versus pirating the content.

I'm sure this could be done, but don't know exactly how