HDV scene capture problem in Vegas 7.0x at DVinfo.net

Christian de Godzinsky · December 12th, 2006, 03:17 PM

Hi,

Anyone else that has problems with the following?

I am capturing HDV in Vegas 7.0c from a HDR-HC1E (PAL) using FireWire. The audio is missing for a couple of frames at the end of every scene-detected clip. There are no dropped frames no whatsoever, and the video capture is perfect. This is very annoying and renders the clips unusable, unless you manually crop each of them at the very end.

I had big hopes that Sony would have fixed this in 7.0c - but no!!! I reported this problem at the very first day i got the first 7.0 loaded and tested the HDV scene detect feature. No comments from them since. Every 7.0 x version suffers from this problem, even the just recently released C-bugfix.

Everything else works fine, DV-import is no problem and works great.

Please anyone with this combination of equipent & sw, can you confirm this malfunction.

I think that Sony has not yet to release a fix for the HDV scene detection to work properly... Would like to get this confirmed.

Please see the attached screendump where the missing audio can clearly be seen in the expanded audio track...

Best regards,

Christian

James Harring · December 12th, 2006, 05:14 PM

I do see the next frame at the end in preview, I suspect this may be part of Vegas cutting on the I frame.

I'm using an A1u (your cam's first cousin) and have not noticed the audio issues you cite... but I tend to always have some extra "lead in/out" video, so tail end issues usually never notice, since I don't use that video anyway. Tape is cheap.

If you want to really do a comparison HDV split does a fine job, I'd say better than Vegas actually IMO. It has less overhead and it's possible your PC may be part of the issue you cite. Be a good comparison tool.

Matti Remonen · December 13th, 2006, 09:06 AM

Hi,
I just checked my stock of m2t-files which I've captured using HDVSplit-app. Some of them show the missing audio at the end of the take from 0.5 to 7 frames (in my material). So it seems to be more like a feature than a bug. Of the camera or of how HDV material gets coded.

I wouldn't worry about it too much because almost never you should have anything that significant at so close of end-of-take.

One thing is to capture material w/o scene detection and check whether the audio really is dropping between takes.

Edit: I shoot with A1E

John Rofrano · December 13th, 2006, 02:01 PM

I've seen it too. This is most likely because the capture is completing the 15-frame GOP even though the audio stopped exactly when the scene changed. Not really a bug, just one more reason why not to edit M2T files and use an intermediary like CineForm.

~jr

Matti Remonen · December 14th, 2006, 01:42 PM

Hi John.

I just tried ConnectHD first time and compared the results with material captured by HDVSplit. For some reason the material captured and converted with HDLink are always shorter (5-7 frames and at least in one scene more than 10 frames) than m2t-material.

Also the sound was lagging approx. one frame when compared to HDVSplit m2t-files. Or it seems to when eyeballing the peak-files.

I posted this same issue to Sony's forum and I can report back if something turns out. Even if it is that I've been mistaken and/or do not know how to do things right.

Christian de Godzinsky · December 14th, 2006, 02:42 PM

Hi,

Nice to see that I am not alone with this problem ;)

However, one WOULD assume that the same scene detection SHOULD and COULD be implemented in such a way that it works exactly the same as in DV capturing. Yes, I fully understand the nature of GOP's and how difficult it is to cut and split m2t files. Think how difficult it is for the cam itself to make nice insert and assembly edits on tape. If that can be solved (no dropping audio - nor video) one would again assume that this could be achieved on a computer that is just working with files - not a helical scan tape recording system.

Frankly speaking, the HDV recording on tape IS continuous during playbakc -and does not have these dropouts.

It should not be a big issue to stop - even a GOP series - at any point. That is actually very easy. It is more difficult to start the next scene and stitch its start to the ending of the former. This works however perfectly in Vegas.

I am fully confident that this IS a bug and will be fixed - but probably at very low priority. Even Sony confirmed to me that this is a recognized bug and it will be fixed in some future release. But that was just after 7.0 was released. B and C versions did not fix the problem - yet :(

You are also right, normally you do not use the tail anyhow. So why bother? But sometimes you do use it and it would be prudent to that precision in a (semi?) professional software as Vegas is supposed to be...

Thanks for your comments so far.

Christian

Matti Remonen · December 14th, 2006, 09:59 PM

Hi Christian.

I'm not that sure if there is anything wrong with Sony VidCap since I get the same behaviour with HDVSplit. I assume that there does not exist any sound at the very end of the scene.

What worried me in the first place was that ConnectHD, which converts the MPEG -stream to more lossless format for editing, is actually completely dropping the last frames.

I agree totally with you that this HDV-thing is still on its very early stage.

BR,
MRe

Matti Remonen · December 15th, 2006, 12:37 PM

OK, I have now made some thorough(?) investigation of the matter. I've tested following apps and file formats:

HDVSplit -> m2t
Vegas VidCap -> m2t
Cineform HDLink -> m2t
Cineform HDLink -> AVI (CFHD)

Material is PAL 25 fps.

I captured the same tape with the same equipment (HVR-M15E) except with HDVSplit where HVR-A1E was used.

It really is that the capture app is affecting to how the sound is managed. I do have screenshots to prove this but for some reason I cannot find a button to upload any material here.

Cineform HDLink capture to m2t did not split scenes so I got uniterrupted reference material of the sound on tape. Both Sony and HDVSplit will mute the sound before video ends. Sony being worse on this. Cineform CFHD-material did not have any mute on sound but the scene detection (or decoding algorithm) cuts the file well before any muting happens in Sony or HDVSplit. So with Cineform you actually lose material.

HDVSplit scene detection let material to "bleed" over the scene change at least in one place.

And one notable thing also: Cineform CFHD -material has the audio lagging approx two frames compared to m2t material.

So Christian, you were right from the beginning.

Christian de Godzinsky · December 18th, 2006, 01:47 PM

Hi Matti,

Thank's for your thorough(!) testing. It clearly shows that the handling and splitting of mt2 files is not an easy task - for any software.

I rather have the audio muted at the end of the clips, instead of out of sync material or completely cropped clips (at the end).

Since Sony has confirmed it is a bug - I live in the hope that some day and in some revision this is fixed.

Well, this IS the type of problem you can call an anomality - and thus can live with. If the HD scene detection would have worked as advertised it would have saved me lots of work in a project where I just needed to reorganize a zillion HD clips without any further editing. Now I had to cut the end of every &#¤"/&%#¤ clip manually. But - I survived this anomality ;)

Thank's again all for you feedback!

Christian

December 12th, 2006, 03:17 PM	#1
Christian de Godzinsky Regular Crew Join Date: Dec 2005 Location: Helsinki/Finland Posts: 131	HDV scene capture problem in Vegas 7.0x Hi, Anyone else that has problems with the following? I am capturing HDV in Vegas 7.0c from a HDR-HC1E (PAL) using FireWire. The audio is missing for a couple of frames at the end of every scene-detected clip. There are no dropped frames no whatsoever, and the video capture is perfect. This is very annoying and renders the clips unusable, unless you manually crop each of them at the very end. I had big hopes that Sony would have fixed this in 7.0c - but no!!! I reported this problem at the very first day i got the first 7.0 loaded and tested the HDV scene detect feature. No comments from them since. Every 7.0 x version suffers from this problem, even the just recently released C-bugfix. Everything else works fine, DV-import is no problem and works great. Please anyone with this combination of equipent & sw, can you confirm this malfunction. I think that Sony has not yet to release a fix for the HDV scene detection to work properly... Would like to get this confirmed. Please see the attached screendump where the missing audio can clearly be seen in the expanded audio track... Best regards, Christian Attached Thumbnails __________________ The future does not exist - we must invent it !

December 12th, 2006, 05:14 PM	#2
James Harring Major Player Join Date: Aug 2006 Location: Tampa, FL Posts: 210	audio upcut on capture I do see the next frame at the end in preview, I suspect this may be part of Vegas cutting on the I frame. I'm using an A1u (your cam's first cousin) and have not noticed the audio issues you cite... but I tend to always have some extra "lead in/out" video, so tail end issues usually never notice, since I don't use that video anyway. Tape is cheap. If you want to really do a comparison HDV split does a fine job, I'd say better than Vegas actually IMO. It has less overhead and it's possible your PC may be part of the issue you cite. Be a good comparison tool.

December 13th, 2006, 09:06 AM	#3
Matti Remonen Regular Crew Join Date: Oct 2006 Location: Helsinki, Finland Posts: 54	Hi, I just checked my stock of m2t-files which I've captured using HDVSplit-app. Some of them show the missing audio at the end of the take from 0.5 to 7 frames (in my material). So it seems to be more like a feature than a bug. Of the camera or of how HDV material gets coded. I wouldn't worry about it too much because almost never you should have anything that significant at so close of end-of-take. One thing is to capture material w/o scene detection and check whether the audio really is dropping between takes. Edit: I shoot with A1E Last edited by Matti Remonen; December 13th, 2006 at 12:37 PM.

December 13th, 2006, 02:01 PM	#4
John Rofrano Sponsor: VASST Join Date: May 2005 Location: New York Posts: 516	I've seen it too. This is most likely because the capture is completing the 15-frame GOP even though the audio stopped exactly when the scene changed. Not really a bug, just one more reason why not to edit M2T files and use an intermediary like CineForm. ~jr __________________ Developer: VASST Ultimate S, Scattershot 3D, Mayhem, FASST Apps, and other VASST Software plug-ins Web Site: www.johnrofrano.com

December 14th, 2006, 02:42 PM	#6
Christian de Godzinsky Regular Crew Join Date: Dec 2005 Location: Helsinki/Finland Posts: 131	Hi, Nice to see that I am not alone with this problem ;) However, one WOULD assume that the same scene detection SHOULD and COULD be implemented in such a way that it works exactly the same as in DV capturing. Yes, I fully understand the nature of GOP's and how difficult it is to cut and split m2t files. Think how difficult it is for the cam itself to make nice insert and assembly edits on tape. If that can be solved (no dropping audio - nor video) one would again assume that this could be achieved on a computer that is just working with files - not a helical scan tape recording system. Frankly speaking, the HDV recording on tape IS continuous during playbakc -and does not have these dropouts. It should not be a big issue to stop - even a GOP series - at any point. That is actually very easy. It is more difficult to start the next scene and stitch its start to the ending of the former. This works however perfectly in Vegas. I am fully confident that this IS a bug and will be fixed - but probably at very low priority. Even Sony confirmed to me that this is a recognized bug and it will be fixed in some future release. But that was just after 7.0 was released. B and C versions did not fix the problem - yet :( You are also right, normally you do not use the tail anyhow. So why bother? But sometimes you do use it and it would be prudent to that precision in a (semi?) professional software as Vegas is supposed to be... Thanks for your comments so far. Christian __________________ The future does not exist - we must invent it !

December 14th, 2006, 01:42 PM	#5
Matti Remonen Regular Crew Join Date: Oct 2006 Location: Helsinki, Finland Posts: 54	Hi John. I just tried ConnectHD first time and compared the results with material captured by HDVSplit. For some reason the material captured and converted with HDLink are always shorter (5-7 frames and at least in one scene more than 10 frames) than m2t-material. Also the sound was lagging approx. one frame when compared to HDVSplit m2t-files. Or it seems to when eyeballing the peak-files. I posted this same issue to Sony's forum and I can report back if something turns out. Even if it is that I've been mistaken and/or do not know how to do things right.

December 14th, 2006, 09:59 PM	#7
Matti Remonen Regular Crew Join Date: Oct 2006 Location: Helsinki, Finland Posts: 54	Hi Christian. I'm not that sure if there is anything wrong with Sony VidCap since I get the same behaviour with HDVSplit. I assume that there does not exist any sound at the very end of the scene. What worried me in the first place was that ConnectHD, which converts the MPEG -stream to more lossless format for editing, is actually completely dropping the last frames. I agree totally with you that this HDV-thing is still on its very early stage. BR, MRe

December 15th, 2006, 12:37 PM	#8
Matti Remonen Regular Crew Join Date: Oct 2006 Location: Helsinki, Finland Posts: 54	OK, I have now made some thorough(?) investigation of the matter. I've tested following apps and file formats: HDVSplit -> m2t Vegas VidCap -> m2t Cineform HDLink -> m2t Cineform HDLink -> AVI (CFHD) Material is PAL 25 fps. I captured the same tape with the same equipment (HVR-M15E) except with HDVSplit where HVR-A1E was used. It really is that the capture app is affecting to how the sound is managed. I do have screenshots to prove this but for some reason I cannot find a button to upload any material here. Cineform HDLink capture to m2t did not split scenes so I got uniterrupted reference material of the sound on tape. Both Sony and HDVSplit will mute the sound before video ends. Sony being worse on this. Cineform CFHD-material did not have any mute on sound but the scene detection (or decoding algorithm) cuts the file well before any muting happens in Sony or HDVSplit. So with Cineform you actually lose material. HDVSplit scene detection let material to "bleed" over the scene change at least in one place. And one notable thing also: Cineform CFHD -material has the audio lagging approx two frames compared to m2t material. So Christian, you were right from the beginning.

December 18th, 2006, 01:47 PM	#9
Christian de Godzinsky Regular Crew Join Date: Dec 2005 Location: Helsinki/Finland Posts: 131	Hi Matti, Thank's for your thorough(!) testing. It clearly shows that the handling and splitting of mt2 files is not an easy task - for any software. I rather have the audio muted at the end of the clips, instead of out of sync material or completely cropped clips (at the end). Since Sony has confirmed it is a bug - I live in the hope that some day and in some revision this is fixed. Well, this IS the type of problem you can call an anomality - and thus can live with. If the HD scene detection would have worked as advertised it would have saved me lots of work in a project where I just needed to reorganize a zillion HD clips without any further editing. Now I had to cut the end of every &#¤"/&%#¤ clip manually. But - I survived this anomality ;) Thank's again all for you feedback! Christian __________________ The future does not exist - we must invent it !