Export to Tape... then scene detection? at DVinfo.net

Chris Rackauckas · April 5th, 2009, 10:37 PM

Hey, so I have a lot of MJPEG files from a digital camera that I take videos with while out and about, and those files are huge for what they do. I tested an MPEG-2 compression on them and they look alright with it, sound only alright too, but livable. Just some shadows gone and it's not as crisp. So I decided I wanted to store them in this format...

And I also know that dvtape is really cheap. I want to know if I could just put them all in sony vegas and export to tape, and then on the capture would it still have scene detection?

Tripp Woelfel · April 6th, 2009, 05:54 AM

If you're going to store on DV tape, you'll have to convert to the DV format. MPEG won't work. If you want to archive the footage, you might consider burning the original files to data DVDs. You won't lose any quality due to a transcode.

As I understand it, scene detection uses breaks in the date/time information in a video file to detect scenes. Transcoding the files will likely lose the ability to scene detect unless that data is preserved. Unfortunately, I don't know how to do that.

John Miller · April 6th, 2009, 06:15 AM

Recently I have been developing a widget for someone to do something quite similar. In his case, he has a DV file of clips also without any time/date info and wants to store them on tape and retrieve them as individual clips later on.

He uses some optical scene detection software to identify the original clips (his DV file is actually a converted MPEG file). The detection software spits out a text file that contains information about where the breaks occur.

I have written a small program that uses our DV processor behind the scenes to send the DV file to tape while simultaneously changing/adding the date/time information such that each clip has its own unique date/time. Subsequent capture from tape yields the original clips instead of one long file.

Something similar could be done in your case - getting at the original clip information would be different but not impossible. e.g., scanning each MJPEG clip in a folder and looking at the header to get the frame count etc.

Chris Rackauckas · April 6th, 2009, 10:35 AM

Well, this is for archiving footage off of a Canon Powershot A720. I have 70 gigs of video from this year and it's all in an MJPEG avi format that is WAY to large. MPEG-2 seems to get 1/6 - 1/7 the size of the video on average, and even more sometimes. So, I was thinking of a way to archive it since I don't have enough dvds for 70 gigs.

Would there be much difference in the compression by shooting it over to dv?

John Miller · April 6th, 2009, 11:05 AM

DV is always compressed at a fixed rate of approx. 27Mbps whereas MJPEG can be compressed at a variety of levels (like JPEG images). So it depends on your existing compression. 10 minutes of DV is about 2GB.

(DV and MJPEG are very similar compression schemes.)

Chris Rackauckas · April 6th, 2009, 11:27 AM

Thank you. It seems this will be my storage way instead of just having one copy of all 70 gigs of video. It's just home video/ some stuff for other people but if it's not a bad change in compression then I'm fine. It's not ultra professional by any standards anyways... it's from a Powershot camera :P.

Here's the workflow for this:

Capture using Powershot.
Copy to Hard Drive.
Put in a Sony Vegas project. Put a bunch of files with about 1 second of space between them. Fill to about 58-59 minutes.
Export to Tape.

When needed: Copy back to harddrive using optical scene detection (easy since there is a second of space in between). This will be done with scenalyzer.

I'll try this out and post how it works. Thanks for John for leading me on the track to find optical scene detection and the confirmation that MJPEG and DV are similar.

April 5th, 2009, 10:37 PM	#1
Chris Rackauckas Regular Crew Join Date: Feb 2009 Location: USA Posts: 113	Export to Tape... then scene detection? Hey, so I have a lot of MJPEG files from a digital camera that I take videos with while out and about, and those files are huge for what they do. I tested an MPEG-2 compression on them and they look alright with it, sound only alright too, but livable. Just some shadows gone and it's not as crisp. So I decided I wanted to store them in this format... And I also know that dvtape is really cheap. I want to know if I could just put them all in sony vegas and export to tape, and then on the capture would it still have scene detection?

April 6th, 2009, 05:54 AM	#2
Tripp Woelfel Trustee Join Date: Jan 2008 Location: North Conway, NH Posts: 1,745	If you're going to store on DV tape, you'll have to convert to the DV format. MPEG won't work. If you want to archive the footage, you might consider burning the original files to data DVDs. You won't lose any quality due to a transcode. As I understand it, scene detection uses breaks in the date/time information in a video file to detect scenes. Transcoding the files will likely lose the ability to scene detect unless that data is preserved. Unfortunately, I don't know how to do that.

April 6th, 2009, 06:15 AM	#3
John Miller Major Player Join Date: Oct 2005 Location: Hillsborough, NC, USA Posts: 968	Recently I have been developing a widget for someone to do something quite similar. In his case, he has a DV file of clips also without any time/date info and wants to store them on tape and retrieve them as individual clips later on. He uses some optical scene detection software to identify the original clips (his DV file is actually a converted MPEG file). The detection software spits out a text file that contains information about where the breaks occur. I have written a small program that uses our DV processor behind the scenes to send the DV file to tape while simultaneously changing/adding the date/time information such that each clip has its own unique date/time. Subsequent capture from tape yields the original clips instead of one long file. Something similar could be done in your case - getting at the original clip information would be different but not impossible. e.g., scanning each MJPEG clip in a folder and looking at the header to get the frame count etc.

April 6th, 2009, 10:35 AM	#4
Chris Rackauckas Regular Crew Join Date: Feb 2009 Location: USA Posts: 113	Well, this is for archiving footage off of a Canon Powershot A720. I have 70 gigs of video from this year and it's all in an MJPEG avi format that is WAY to large. MPEG-2 seems to get 1/6 - 1/7 the size of the video on average, and even more sometimes. So, I was thinking of a way to archive it since I don't have enough dvds for 70 gigs. Would there be much difference in the compression by shooting it over to dv?

April 6th, 2009, 11:05 AM	#5
John Miller Major Player Join Date: Oct 2005 Location: Hillsborough, NC, USA Posts: 968	DV is always compressed at a fixed rate of approx. 27Mbps whereas MJPEG can be compressed at a variety of levels (like JPEG images). So it depends on your existing compression. 10 minutes of DV is about 2GB. (DV and MJPEG are very similar compression schemes.)

April 6th, 2009, 11:27 AM	#6
Chris Rackauckas Regular Crew Join Date: Feb 2009 Location: USA Posts: 113	Thank you. It seems this will be my storage way instead of just having one copy of all 70 gigs of video. It's just home video/ some stuff for other people but if it's not a bad change in compression then I'm fine. It's not ultra professional by any standards anyways... it's from a Powershot camera :P. Here's the workflow for this: Capture using Powershot. Copy to Hard Drive. Put in a Sony Vegas project. Put a bunch of files with about 1 second of space between them. Fill to about 58-59 minutes. Export to Tape. When needed: Copy back to harddrive using optical scene detection (easy since there is a second of space in between). This will be done with scenalyzer. I'll try this out and post how it works. Thanks for John for leading me on the track to find optical scene detection and the confirmation that MJPEG and DV are similar.