Captured Digital8 tapes, many scenes have errors?

[mike86325]

Hi all,

I recently captured 32 Digital8 tapes using this setup:

-Old i7 4770k desktop
-Windows 7 ultimate 64bit, clean install
-PCIe Firewire card with TI chipset
-Sony DCR-TRV140 camcorder
-ScenalyzerLive 4 (Type 1 DV with automatic scene detection on - min time between scenes 0h00m01s)

Three questions, as I really want to make sure I end up with the absolute best capture I can get:

1) I saw lordsmurf mention, and hopefully I'm not misinterpreting him, that Digital8 capture can/will have seconds missing from the beginning of each scene.

Quote:

It depends on how close recording were to cuts. Digital transfer loses up to 2-3 seconds of footage per cut. If you had lead time, no issue. But if you hit record right when grandma was singing Happy Birthday, then part of the song will be missing. Seconds can matter.

Is there a way to check if this was the case with all of my captured tapes? Would this show up in some way using DVanalyzer 1.4.2? It would be pretty devastating to realize this and have to recapture 32 tapes due to missing seconds from each scene.

2) Also, I maybe had 3 dropped frames total after all 32 tapes, but I ended up with a lot of scenes with errors. Some were totally clean and others had very minimal errors, but every tape had at least a scene or more with some form of errors. Given the amount of tapes and scenes, what is the best way to workflow to capture everything I can from the data on the tape? My uneducated first guess was something like this:

-Analyze folder of all scenes on a given tape using DVAnalyzer
-Clean heads, ffwd and rewind tape in question
-For scenes with errors, manually find on camcorder and capture
-If errors still present, capture again
-If errors still present, attempt to find exact frames with errors and attempt to splice the "best" frame available from the multiple captures into that scene.

This seems incredibly tedious and will take an obscene amount of time. Hopefully there is a better method that someone can recommend. I know that the DVRescue project's program does this automatically - rewinds and recaptures bad frames and merges best frames together - but unfortunately despite claiming to be Windows and Linux compatible, I couldn't get their software to run on either. It appears to be functioning properly only on Mac at the moment.

3) As I mentioned, I essentially had zero dropped frames on most tapes. If no frames were dropped, I should have all of the tape's contents captured, correct? Is there a way that certain scenes or sections of the tapes could have been skipped in some way? I just want to make sure I'm not missing anything off of these tapes, but I also don't want to have to sit down and watch the tape alongside these clips on my computer in order to make sure it didn't miss a thing. Is there a program I can use to verify the full contents of the tape was captured? Would looking at the frame counts on DVAnalyzer solve this (making sure the next scene starts exactly 1 frame after the previous scene)?

Thank you all in advance for your help and advice! I have put off this project for so long and am hoping I can wrap it up soon with your help.

[latreche34]

Unfortunately there is no other way to reduce tape dropouts other than cleaning the heads and trying again, If the dropouts are consistent that means data is lost forever.

[Hushpower]

Quote:

that Digital8 capture can/will have seconds missing from the beginning of each scene.

I have never heard of this, nor encountered it with any of my capture programs including WinDV and Scenalyzer. Check a few scenes. It will be obvious if you've lost a few seconds; talking starting half way through the conversation, illogical panning movement. If it's not obvious, you are not "missing" anything, so forget it. You could also pick one tape and check the scenes. I'd be very very surprised if they are not all there.

What's reporting the errors? You'll be watching each tape anyway to make sure it's a good capture, and any stutters or drops will be obvious. If the audio is in sync and nothing is obvious, move on.

What program was Lordsmurf using?

[latreche34]

Scene detection works pretty well I myself never had a missing frame, Usually camcorders allow enough amount of lead in before starting to record a footage at the beginning of a tape, Detecting the first frame is not a problem at least for Sclive as long as the tape has no damage or has not been shortened due to a repair.

[timtape]

Quote:

Originally Posted by mike86325

Hi all,


2) Also, I maybe had 3 dropped frames total after all 32 tapes, but I ended up with a lot of scenes with errors. Some were totally clean and others had very minimal errors, but every tape had at least a scene or more with some form of errors. Given the amount of tapes and scenes, what is the best way to workflow to capture everything I can from the data on the tape? My uneducated first guess was something like this:

-Analyze folder of all scenes on a given tape using DVAnalyzer
-Clean heads, ffwd and rewind tape in question
-For scenes with errors, manually find on camcorder and capture
-If errors still present, capture again
-If errors still present, attempt to find exact frames with errors and attempt to splice the "best" frame available from the multiple captures into that scene.

This seems incredibly tedious and will take an obscene amount of time. Hopefully there is a better method that someone can recommend. I know that the DVRescue project's program does this automatically - rewinds and recaptures bad frames and merges best frames together - but unfortunately despite claiming to be Windows and Linux compatible, I couldn't get their software to run on either. It appears to be functioning properly only on Mac at the moment.

3) As I mentioned, I essentially had zero dropped frames on most tapes. If no frames were dropped, I should have all of the tape's contents captured, correct? Is there a way that certain scenes or sections of the tapes could have been skipped in some way? I just want to make sure I'm not missing anything off of these tapes, but I also don't want to have to sit down and watch the tape alongside these clips on my computer in order to make sure it didn't miss a thing. Is there a program I can use to verify the full contents of the tape was captured? Would looking at the frame counts on DVAnalyzer solve this (making sure the next scene starts exactly 1 frame after the previous scene)? ...

Knowing WHY parts of the data werent captured is key. If the problem is a dropout in the tape at various points, due say to some minute dust on the camera tape as it was actually filmed, constantly going over that section of tape will not reveal anything more as that section of data was never recorded to the tape in the first place.

Just because no frames were registered as dropped doesnt mean we have captured everything on the tape. Parts of frames may not have been fully captured.

Tapes recorded at the slowest LP speed are regularly the ones most difficult to capture without error. The reason is the recorded tracks are very narrow and it's easy for a video head misalignment to occur. Everything has to be "just right" in the tape playback.

Perhaps the biggest error people make these days is assuming that data (picture or sound) not fully extracted at capture stage can be "restored" in post. Part of this seems to come from watching too many CSI type TV shows where restoration "magic" regularly happens.

[mike86325]

Quote:

Originally Posted by Hushpower

I have never heard of this, nor encountered it with any of my capture programs including WinDV and Scenalyzer. Check a few scenes. It will be obvious if you've lost a few seconds; talking starting half way through the conversation, illogical panning movement. If it's not obvious, you are not "missing" anything, so forget it. You could also pick one tape and check the scenes. I'd be very very surprised if they are not all there.

What's reporting the errors? You'll be watching each tape anyway to make sure it's a good capture, and any stutters or drops will be obvious. If the audio is in sync and nothing is obvious, move on.

What program was Lordsmurf using?

Ok, that may be a good idea to check captured scenes side by side with what is showing directly on the camcorder. Scenalyzer stated there was errors with each capture, so I used a program called DV Analyzer (https://mediaarea.net/DVAnalyzer) that is able to state exactly which frames are affected and what type of error is present.

Lordsmurf didn't mention a program specifically, but was stating this as a general issue with digital capture vs. S-Video. Here is the forum thread https://www.digitalfaq.com/forum/vid...ital8-usb.html

Quote:

Originally Posted by latreche34

Scene detection works pretty well I myself never had a missing frame, Usually camcorders allow enough amount of lead in before starting to record a footage at the beginning of a tape, Detecting the first frame is not a problem at least for Sclive as long as the tape has no damage or has not been shortened due to a repair.

Is there an objective way to determine if a frame has been missed? I was wondering this in my 3rd question, whether a program could be used to determine that all frames were captured.

I actually saw you mentioned this in response to Lordsmurf's comment in a previous thread when he claimed that digital capture will result in missing seconds from each scene. He followed up saying

Quote:

Nope, it can miss. DV transfer auto detects the video on the tape, because it tries to avoid snow (blank tape). There is a delay in the detection, hence missing seconds of footage.

. If this is true, it's a huge problem for all of us capturing DV and means that I probably need to start from scratch with S-Video unless I can prove that no frames/scenes/seconds have been skipped.

Quote:

Originally Posted by timtape

Knowing WHY parts of the data werent captured is key. If the problem is a dropout in the tape at various points, due say to some minute dust on the camera tape as it was actually filmed, constantly going over that section of tape will not reveal anything more as that section of data was never recorded to the tape in the first place.

Just because no frames were registered as dropped doesnt mean we have captured everything on the tape. Parts of frames may not have been fully captured.

Tapes recorded at the slowest LP speed are regularly the ones most difficult to capture without error. The reason is the recorded tracks are very narrow and it's easy for a video head misalignment to occur. Everything has to be "just right" in the tape playback.

Perhaps the biggest error people make these days is assuming that data (picture or sound) not fully extracted at capture stage can be "restored" in post. Part of this seems to come from watching too many CSI type TV shows where restoration "magic" regularly happens.

This is very true, I know that there isn't any magic bullet to fix everything in post, which is why I'm trying to be as meticulous as possible with my capture process. I'm just not sure if my plan laid out in question 2 of my original post is the most efficient and robust for getting as much of the data as possible.

[Hushpower]

Quote:

If this is true, it's a huge problem for all of us capturing DV and means that I probably need to start from scratch with S-Video unless I can prove that no frames/scenes/seconds have been skipped.

We need more factual info from Lordsmurf.

I very much doubt that Scenalyzer is sitting there doing nothing and suddenly says "oh dear, here is some real video instead of snow, get your act together you lot and start capturing", seconds later.

Quote:

Scenalyzer stated there was errors with each capture, so I used a program called DV Analyzer (https://mediaarea.net/DVAnalyzer) that is able to state exactly which frames are affected and what type of error is present.

So what did DVAnalyzer say?

[Hushpower]

I've just done a quick check of a DV file I have here with DVAnalyzer. It has given multiple warnings about non-sequential files when in fact, when I checked the actual file in VDub (accurate frame number display) it is just a scene change and not an error at all.

In another tape I checked (D8 camcoder with Video8 tape) the file has multiple Scenalyzer yellow marks indicating problems, but on inspection the yellow marks are at dead spots in the tape, so once again, no issue.

To be absolutely positive, you could capture a tape with S-Video and then check each scene, side by side with your DV files (two instances of VDub open), to see if you got the "whole scenes". I won't be.

[latreche34]

If you don't use scene detection Sclive will capture the entire tape including blank spaces but will throw errors in those areas. Using the detection method comes down to saving time, If you have a lot of tapes, capturing the entire tape as one file meant that you have to sit and edit the entire tape looking for those blank segments and cut them off and you do this for each tape.

As to the beginning of the tape, As I mentioned before Sclive will detect the first frame in the footage and will register it, and even if there is a delay it won't be more than 2 frames but I highly doubt it.

[Hushpower]

Quote:

Originally Posted by Latreche

If you have a lot of tapes, capturing the entire tape as one file meant that you have to sit and edit the entire tape looking for those blank segments and cut them off and you do this for each tape.

No, any decent editor will auto-detect scenes. Even Scenalyzer detected 100 scenes in a 60 minute tape of mine in 10 seconds. Have a cup of coffee while it commits and writes 100 DV scene files.

[latreche34]

Not really, Once the frames are legalized in the stream they are no longer corrupted so detecting frames just because they are black is not a trigger, Sclive may have the option to detect scenes even if it is not instructed to do so, But if you feed it one hour DV files from another computer I don't see how that works, At least it didn't work for me and I don't trust that approach anyway. I enjoy my coffee with scene detection ON.

[Hushpower]

I didn't understand any of that, but it appears you don't understand what a DV file is. A 60 minute one-tape capture file still contains all the individual timecodes of each scene. When you run Scene-detect, the program simply splits the video at all the scene changes as determined by timecode changes. This happens in seconds.

[mike86325]

Quote:

Originally Posted by Hushpower

We need more factual info from Lordsmurf.

I very much doubt that Scenalyzer is sitting there doing nothing and suddenly says "oh dear, here is some real video instead of snow, get your act together you lot and start capturing", seconds later.


So what did DVAnalyzer say?

I agree, it doesn't make sense. In DVAnalyzer, I had it scan several folders (each representing one tape's worth of scenes). The errors I'm getting are all type A or labeled as "arbitrary bit inconsistency", usually in low percentages within certain scenes. It's not very helpful, but at least let's me know what scenes I might need to attempt to recapture in some way.

Quote:

Originally Posted by Hushpower

I've just done a quick check of a DV file I have here with DVAnalyzer. It has given multiple warnings about non-sequential files when in fact, when I checked the actual file in VDub (accurate frame number display) it is just a scene change and not an error at all.

In another tape I checked (D8 camcoder with Video8 tape) the file has multiple Scenalyzer yellow marks indicating problems, but on inspection the yellow marks are at dead spots in the tape, so once again, no issue.

To be absolutely positive, you could capture a tape with S-Video and then check each scene, side by side with your DV files (two instances of VDub open), to see if you got the "whole scenes". I won't be.

That's interesting, I have not seen that non-sequential file error pop up for me, but I also had scene detection on so that may be why it's not showing those "errors". When I import my folder of scenes from a tape into DVAnalyzer, I can see under the "File Summary" tab that the DV timecode of sequential scenes does not have any time gap. I'm not sure if this is accurate, but if this DV timecode reading is from the tape itself, it appears that it is not missing frames.

For instance:
1st Scene
DV Timecode 00:00:07;05-00:00:13;18

2nd Scene
DV Timecode 00:00:13;19-00:00:19;16

So it appears that there is zero time loss between scenes captured from the tape. I'm not sure if this information is 100% accurate, but if so it could be a quicker way of checking things instead of comparing against the S-Video capture. That is a good idea though. Can virtualDub import a folder worth of DV files and allow me to arrange them in the order of the tape? Or will I need to find a video editor to do that?

Quote:

Originally Posted by latreche34

If you don't use scene detection Sclive will capture the entire tape including blank spaces but will throw errors in those areas. Using the detection method comes down to saving time, If you have a lot of tapes, capturing the entire tape as one file meant that you have to sit and edit the entire tape looking for those blank segments and cut them off and you do this for each tape.

As to the beginning of the tape, As I mentioned before Sclive will detect the first frame in the footage and will register it, and even if there is a delay it won't be more than 2 frames but I highly doubt it.

I initially went back and forth a lot on whether to capture the tape with or without scene detection on. I eventually chose scene detection because I figured if/when I have errors during capture, it would be much easier to determine the exact scene and the exact length of that scene that I need to recapture. As long as nothing is getting lost using scene detection then I think it was the right option.

[Hushpower]

Quote:

Originally Posted by Mike86325

Can virtualDub import a folder worth of DV files and allow me to arrange them in the order of the tape?

VDub does have an "Append video segment" feature on the file menu, but you'll have to try it to see if you can add multiple files at once, otherwise it might be tedious.