Scene Detection after Capture? at DVinfo.net

Nicholas de Kock · May 14th, 2009, 01:47 PM

Is it possible to capture a full HDV tape (13GB) in one file and have a NLE do scene detection in post? In other words having software automatically trim video into splits in the software and leaving the caputured file as one big file.

Harm Millaard · May 15th, 2009, 04:05 AM

I don't think that is possible, at least not with CS4. The only possible appraoch, that defeats your purpose of keeping one large file, would be to use HDVSplit on the long file and have it split the file based on scene detection into many smaller ones. But that is not what you want, so AFAIK the answer is no.

Nicholas de Kock · May 17th, 2009, 05:17 PM

It's a shame there isn't a way to do this, would have been useful if a NLE would capture one large file and track the timecode for each scene and split in the timeline.

Adam Gold · May 17th, 2009, 06:12 PM

I believe Premiere Elements, not Pro, does this. Also, if I recall correctly, there was an old package called Scenalyzer (not the newer Scenalyzer Live) that did this after capture, not during, as the Live version does. Perhaps a Google search will reveal if the old version is still available.

Edit: Just found the site for Scenalyzer and it appears this is for DV only (although I could have sworn I did this for HDV at one time); however, the HDVSplit site seems to imply splitting can be done retroactively as well as during capture, as Harm mentioned, and this shouldn't be too much more work than doing it inside the NLE.

Harm Millaard · May 18th, 2009, 03:17 AM

Adam,

I know it is not much trouble to split the long file into many smaller ones, based on scene detection with HDVSplit, but I assume Nicholas does not want that, because it makes loading the project much slower with many small files instead of one large file.

OTOH, I think that even if loading the project is much slower, it makes the editing much easier. Personally, I find that date and timestamp added to the filename is a handy feature as well and you definitely avoid out of sync issues the Adobe capture had, at least in CS3. I only use HDVSplit for capture, not CS4.

Jon Shohet · May 18th, 2009, 06:29 AM

The major flaw with capturing with HDV split IMO is that you lose time-code in the captured clips. I still prefer using it than capturing with PP but Adobe definitely should add scene detect to hdv!
Another huge advantage in splitting a tape into small clips is that you can easily reduce project size by deleting unneeded clips. this is especially useful when archiving large projects.

Stuart Graham · May 18th, 2009, 11:15 AM

Avid Media Composer will autodetect and split clips during capture

Nicholas de Kock · May 18th, 2009, 11:41 AM

Jon I rarely delete clips before I finish a project, lately it's been such a hassle to track and share multiple clips with my editor that sometimes I wish that I had one big file to keep things simple, however scene detect is a vital part of editing so I'll make my peace with hundreds of files.

Harm Millaard · May 18th, 2009, 12:06 PM

Stuart, what is the relevancy of that statement for the OP?

Jon Shohet · May 18th, 2009, 05:46 PM

There is a free After Effects script written by Lloyd Alvarez :

After Effects Scripts Magnum - The Edit Detector

It takes one layer of video and splits it into multiple layers based on scene detection. If you then import the AE composition into Premiere, you get a sequence with all the scenes cut up on the timeline, without cutting up the source footage.
If you're on CS4, you can create a subclip from each cut. For some reason, Audio and Video are unlinked when imported from AE, so you'll have to relink AV for each clip manually before you can subclip. That's a big bother, but not impossible. Especially if you make keyboard shortcuts for "Link" and "Make Subclip".

The bigger problem is that this script does optical scene detect, and not timecode scene detect - it has to analyze every frame, and that takes a loooong time. So I think it's not really a practical solution for entire HDV tapes. However, you can try to contact Lloyd, or maybe someone can write a script that does timecode scene detection. It's worth a shot.

Stuart Graham · May 19th, 2009, 02:02 AM

Sorry Nicholas I didn't read your question properly, my answer wasn't very useful

Jack Bellford · May 21st, 2009, 05:19 AM

Both Avid Liquid and Pinnacle Studio will do what you're asking but there is a twist.

These two programs do not actually split your hdv tape. Instead they capture the tape as a whole and write a separate file parallel to the captured tape with all the scene split info in it. This enables you to do things lie frame accurate batch capturing, split by time... etc.

The catch is that the scene split info file is only good for the respective program so if you do a scene split in liquid for example... you will only be able to use liquid for your editing.

One other thing... both these programs capture the audio and video as two separate files (M2V and WAV)

Mike Browning · June 11th, 2009, 02:32 PM

I believe that the scene-detection you can select during capture works because of a time-stamp on the tape itself. In other words, if I'm not mistaken, it checks for lapses in time in order to determine the break, not the time-code itself. If you've ever auto-captured a tape from a camera with no time/date set on the camera, it will not auto-detect the next clip.

All that to say, I don't think you can do this after the fact without that time/date data, unless you NLE captures that as meta information on the master clip. I could just be adding to the confusion, though.

Also, assuming you found a way to split the master clip, you still might have to face render time to render out the sub clips, unless your collaborating editors use the same NLE and project file. Just a thought.

May 14th, 2009, 01:47 PM	#1
Nicholas de Kock Major Player Join Date: May 2007 Location: KLD, South Africa Posts: 983	Scene Detection after Capture? Is it possible to capture a full HDV tape (13GB) in one file and have a NLE do scene detection in post? In other words having software automatically trim video into splits in the software and leaving the caputured file as one big file.

May 18th, 2009, 11:15 AM	#7
Stuart Graham Major Player Join Date: Aug 2008 Location: Selby UK Posts: 274	Avid Media Composer will autodetect and split clips during capture __________________ Stuart Graham www.magentapictures.com

May 19th, 2009, 02:02 AM	#11
Stuart Graham Major Player Join Date: Aug 2008 Location: Selby UK Posts: 274	Sorry Nicholas I didn't read your question properly, my answer wasn't very useful __________________ Stuart Graham www.magentapictures.com

May 15th, 2009, 04:05 AM	#2
Harm Millaard Trustee Join Date: Aug 2006 Location: Rotterdam, Netherlands Posts: 1,832	I don't think that is possible, at least not with CS4. The only possible appraoch, that defeats your purpose of keeping one large file, would be to use HDVSplit on the long file and have it split the file based on scene detection into many smaller ones. But that is not what you want, so AFAIK the answer is no.

May 17th, 2009, 05:17 PM	#3
Nicholas de Kock Major Player Join Date: May 2007 Location: KLD, South Africa Posts: 983	It's a shame there isn't a way to do this, would have been useful if a NLE would capture one large file and track the timecode for each scene and split in the timeline.

May 17th, 2009, 06:12 PM	#4
Adam Gold Inner Circle Join Date: Jan 2007 Location: Woodinville, WA USA Posts: 3,467	I believe Premiere Elements, not Pro, does this. Also, if I recall correctly, there was an old package called Scenalyzer (not the newer Scenalyzer Live) that did this after capture, not during, as the Live version does. Perhaps a Google search will reveal if the old version is still available. Edit: Just found the site for Scenalyzer and it appears this is for DV only (although I could have sworn I did this for HDV at one time); however, the HDVSplit site seems to imply splitting can be done retroactively as well as during capture, as Harm mentioned, and this shouldn't be too much more work than doing it inside the NLE.

May 18th, 2009, 03:17 AM	#5
Harm Millaard Trustee Join Date: Aug 2006 Location: Rotterdam, Netherlands Posts: 1,832	Adam, I know it is not much trouble to split the long file into many smaller ones, based on scene detection with HDVSplit, but I assume Nicholas does not want that, because it makes loading the project much slower with many small files instead of one large file. OTOH, I think that even if loading the project is much slower, it makes the editing much easier. Personally, I find that date and timestamp added to the filename is a handy feature as well and you definitely avoid out of sync issues the Adobe capture had, at least in CS3. I only use HDVSplit for capture, not CS4.

May 18th, 2009, 06:29 AM	#6
Jon Shohet Major Player Join Date: Dec 2007 Location: London, UK Posts: 277	The major flaw with capturing with HDV split IMO is that you lose time-code in the captured clips. I still prefer using it than capturing with PP but Adobe definitely should add scene detect to hdv! Another huge advantage in splitting a tape into small clips is that you can easily reduce project size by deleting unneeded clips. this is especially useful when archiving large projects.

May 18th, 2009, 11:41 AM	#8
Nicholas de Kock Major Player Join Date: May 2007 Location: KLD, South Africa Posts: 983	Jon I rarely delete clips before I finish a project, lately it's been such a hassle to track and share multiple clips with my editor that sometimes I wish that I had one big file to keep things simple, however scene detect is a vital part of editing so I'll make my peace with hundreds of files.

May 18th, 2009, 12:06 PM	#9
Harm Millaard Trustee Join Date: Aug 2006 Location: Rotterdam, Netherlands Posts: 1,832	Stuart, what is the relevancy of that statement for the OP?

May 18th, 2009, 05:46 PM	#10
Jon Shohet Major Player Join Date: Dec 2007 Location: London, UK Posts: 277	There is a free After Effects script written by Lloyd Alvarez : After Effects Scripts Magnum - The Edit Detector It takes one layer of video and splits it into multiple layers based on scene detection. If you then import the AE composition into Premiere, you get a sequence with all the scenes cut up on the timeline, without cutting up the source footage. If you're on CS4, you can create a subclip from each cut. For some reason, Audio and Video are unlinked when imported from AE, so you'll have to relink AV for each clip manually before you can subclip. That's a big bother, but not impossible. Especially if you make keyboard shortcuts for "Link" and "Make Subclip". The bigger problem is that this script does optical scene detect, and not timecode scene detect - it has to analyze every frame, and that takes a loooong time. So I think it's not really a practical solution for entire HDV tapes. However, you can try to contact Lloyd, or maybe someone can write a script that does timecode scene detection. It's worth a shot.

May 21st, 2009, 05:19 AM	#12
Jack Bellford Regular Crew Join Date: Mar 2009 Location: Winnipeg Canada Posts: 68	Both Avid Liquid and Pinnacle Studio will do what you're asking but there is a twist. These two programs do not actually split your hdv tape. Instead they capture the tape as a whole and write a separate file parallel to the captured tape with all the scene split info in it. This enables you to do things lie frame accurate batch capturing, split by time... etc. The catch is that the scene split info file is only good for the respective program so if you do a scene split in liquid for example... you will only be able to use liquid for your editing. One other thing... both these programs capture the audio and video as two separate files (M2V and WAV)

June 11th, 2009, 02:32 PM	#13
Mike Browning Regular Crew Join Date: Apr 2008 Location: Golden Isles, GA Posts: 25	I believe that the scene-detection you can select during capture works because of a time-stamp on the tape itself. In other words, if I'm not mistaken, it checks for lapses in time in order to determine the break, not the time-code itself. If you've ever auto-captured a tape from a camera with no time/date set on the camera, it will not auto-detect the next clip. All that to say, I don't think you can do this after the fact without that time/date data, unless you NLE captures that as meta information on the master clip. I could just be adding to the confusion, though. Also, assuming you found a way to split the master clip, you still might have to face render time to render out the sub clips, unless your collaborating editors use the same NLE and project file. Just a thought.