question

ashleyxxx

1: I have two movie clips of the same file 1 movie clip is 30 seconds long and the second movie clip is 5 minutes long.

2: The 30 second movie clip is located on my server

3: The 5 minute movie clip is located at a different site which I do know.
But lets say that I do not know where it is.
(I think I need a web crawler here to find the site with the full video)

4: I would like to know if it is possible to parse the 30 second movie clip which is a flv file, and find the full clip on another site if it exist.

I would like to know if that is possible to do in php.
If yes ,Then how would I go about it ?

etully

This is a very difficult task and is mostly impossible.

There's no "trick" or simple script that we're going to be able to share with you that's going to make this work.

To make this work, you'd need tons of computers and tons of bandwidth (in the order of millions of dollars) to crawl the Internet, search for video, and fingerprint the video. By fingerprinting the video, I mean that you have to take a frame, look at all the pixels in that frame, find some general patterns, and reduce this 1/30th of a second down to something like: 00 AA 34 F2 23 9D E7 33 B2 B9 05 C4 88 38 7A

Then you store that in a database.

Then, when someone provides you with the 30 second clip, you do the same process of fingerprinting a handful of frames and look for that fingerprint in your database.

30 seconds of video is (30 seconds) X (30 frames per second) = 900 frames of video. Searching your database to see if these 900 fingerprints exist in your database is way too time consuming so you're going to have to cheat and take every 10th frame or something like that.

This is going to require a computer science degree - not a PHP script that you can download off the Internet. Oh, and the millions of dollars of startup money, don't forget about that part too.

ashleyxxx

etully;10974304 wrote:
This is a very difficult task and is mostly impossible.

There's no "trick" or simple script that we're going to be able to share with you that's going to make this work.

To make this work, you'd need tons of computers and tons of bandwidth (in the order of millions of dollars) to crawl the Internet, search for video, and fingerprint the video. By fingerprinting the video, I mean that you have to take a frame, look at all the pixels in that frame, find some general patterns, and reduce this 1/30th of a second down to something like: 00 AA 34 F2 23 9D E7 33 B2 B9 05 C4 88 38 7A

Then you store that in a database.

Then, when someone provides you with the 30 second clip, you do the same process of fingerprinting a handful of frames and look for that fingerprint in your database.

30 seconds of video is (30 seconds) X (30 frames per second) = 900 frames of video. Searching your database to see if these 900 fingerprints exist in your database is way too time consuming so you're going to have to cheat and take every 10th frame or something like that.

This is going to require a computer science degree - not a PHP script that you can download off the Internet. Oh, and the millions of dollars of startup money, don't forget about that part too.

Thanks I figured that much

ashleyxxx

Lets say i just want to do that for one particular video and index the video on a site that i am aware of just for proof of concept can that be done in php ?
Is that the only way is there not header information inside the video?

etully

Anyone who understands what "proof of concept" means is going to see that your presentation is not proving the important part. It would be like building a rear view mirror to prove that an automobile is possible.

So let's say that there is header info and you scan it on your local search clip and you scan it on the remote target clip and you compare the two and they match. What are you going to say, "See? It can be done. So all we have to do now is index all the video on the Internet..."

If the remote site is friendly... that is, they are willing to cooperate with you... then you might be able to pay them enough money to scan the videos at their end and keep track of what they have and then it's a simple matter of saying, "Hey YouTube, do you have any clips from Shrek?" And they respond yes, and you're done. However, if your goal is to monitor the Internet to find out if anyone is using your video clip anywhere, then you're either going to have to (A) download billions of hours of video so that you can index it all, or (😎 get millions of web sites to cooperate with you via the Shrek example.

As for header information, every format is going to be different and not every clip is going to have reliable, correct information. For example, if I rip a Shrek DVD and post it somewhere, I'm not necessarily going to take great care to make sure that the video headers correctly identify the clip as Shrek so you'll have a low chance of spotting your clip.

You're really focused on whether or not this can be done in PHP. The answer is that PHP can work as a sort of shell language to control whatever video player codec reading binary header gathering gizmos might be necessary to get some info out of the clip.

Please investigate ffmpeg. It can read TONS of data about video clips. More data than you need. And yes, ffmeg can be controlled by PHP but also by any other language (Python, ASP, Java, etc etc etc.) The shell language that controls ffmpeg (or whatever tool you use to extract data from video files) is only a secondary concern.

ashleyxxx

etully;10974309 wrote:
Anyone who understands what "proof of concept" means is going to see that your presentation is not proving the important part. It would be like building a rear view mirror to prove that an automobile is possible.

So let's say that there is header info and you scan it on your local search clip and you scan it on the remote target clip and you compare the two and they match. What are you going to say, "See? It can be done. So all we have to do now is index all the video on the Internet..."

If the remote site is friendly... that is, they are willing to cooperate with you... then you might be able to pay them enough money to scan the videos at their end and keep track of what they have and then it's a simple matter of saying, "Hey YouTube, do you have any clips from Shrek?" And they respond yes, and you're done. However, if your goal is to monitor the Internet to find out if anyone is using your video clip anywhere, then you're either going to have to (A) download billions of hours of video so that you can index it all, or (😎 get millions of web sites to cooperate with you via the Shrek example.

As for header information, every format is going to be different and not every clip is going to have reliable, correct information. For example, if I rip a Shrek DVD and post it somewhere, I'm not necessarily going to take great care to make sure that the video headers correctly identify the clip as Shrek so you'll have a low chance of spotting your clip.

You're really focused on whether or not this can be done in PHP. The answer is that PHP can work as a sort of shell language to control whatever video player codec reading binary header gathering gizmos might be necessary to get some info out of the clip.

Please investigate ffmpeg. It can read TONS of data about video clips. More data than you need. And yes, ffmeg can be controlled by PHP but also by any other language (Python, ASP, Java, etc etc etc.) The shell language that controls ffmpeg (or whatever tool you use to extract data from video files) is only a secondary concern.

Thank you so much for your help. Im trying to get something done for my school project. Thats something I thought would be fun to do and learn.