Audio sync problems, dropped frames, with TBC?

[jdp90000]

The Project
Capture home movies on 25 Video8 tapes, 25 Hi8 tapes, and 30 VHS tapes.

The Problems
1. Audio/video not in sync.
2. Dropped frames, usually 7 to 12 per 2-hour capture. Evenly spread out across capture time (e.g., one every 10 minutes or so).
3. Inserted frames, usually 7 to 12 per capture. All happen immediately when capture starts, before pressing PLAY on camcorder. (There is no parity or apparent correlation between inserts and drops. Sometimes there is more of one than another.)

System

Camcorder -> TBC -> Capture Device -> Computer

Camcorder
CCD-TPV68 (owned since new; excellent condition)

TBC
BigVooDoo BVTBC (this one). Purchased from lordsmurf.

Capture Device
ATI TV Wonder 600 USB. Also purchased from lordsmurf.
Software installed per lordsmurf's instructions.

Computer
Dell Optiplex 7040
Windows 7, SP1. Clean install.
CPU: i7-6700 3.4GHz
RAM: 16GB
Storage:
500GB SSD (SATA) (operating system)
1TB SSD (PCIe M.2) (capture destination, otherwise empty)
All unnecessary software disabled, bloatware uninstalled.
No internet. No WiFi. Ethernet unplugged during capture.
USB 2.0 port at rear of computer used for ATI.
Computer/keyboard/mouse not touched during capture.

Cables
Video: Monster s-video from camcorder to TBC. Acoustic research s-video from TBC to ATI.
Audio: Monster RCA. Mono from camcorder, split to “stereo” before ATI.

Software
Virtualdub 1.9.11 and Huffyuv installed per lordsmurf's instructions.

What I have tried?

Forum Advice
Spent the last month reading dozens of prior forum threads about sync issues, including (probably) all that involved an ATI USB device. Tried all recommended settings, etc. (detailed below).

Read lordsmurf’s article on troubleshooting dropped frames. Here is how I worked through each of the “21 fixes.”
1. Multi-tasking. Inapplicable, and all background software is deleted/disabled.
2. Program Settings. See below regarding VirtualDub settings I have tried.
3. Anti-Virus and other Background Software. None.
4. Heat / Overheating. Inapplicable. PC remains cool. CPU never exceeds 5-20% during capture. Only a small portion of a single core is used.
5. Reboot! Shutdown! Computer is rebooted before each capture and shut down when not in use (not my primary machine).
6. LAN and Internet. Unplugged. No WiFi.
7. Use the Best Drivers. ATI software and VirtualDub installed per lordsmurf's instructions. I am open to what other “best” drivers I may be lacking.
8. Software vs. Hardware Encoding. CPU use does not seem to be an issue (see #4).
9. Slow or Old Computers. If anything, this PC is probably overkill for the application. I thought overkill would help avoid these issues, but look where that got me.
10. CPU Usage. See #8 and #4.
11. VHS and other Analogue Videotape Source. I have been testing with Hi8 tapes, which are the best of the bunch. They were stored properly. The video plays back great live, with or without a TBC. I have tested the sync/drops issue with and without the TBC. Whether or not I use the TBC does not seem to affect the result. In fact, I am not even sure how to tell that it is working.
12. Hard Drive Fragmentation. Inapplicable (SSDs).
13. Separate Hard Drives Suggested! Check.
14. Check Your Hard Drive Settings. The guide suggests checking hard drive settings, but that seems to apply to older hardware. One unresolved hard drive question that I have is whether the partitioning schemes I am using might be an issue (Disk1 with OS is MBR, while Disk2 for capture is GPT).
15. Capture Software and Codecs. The advice is to use VirtualDub, which I am trying to do here. I have seen recommendations to try AmaRecTV, which I am open to, but not without exhausting VirtualDub capture options.
16. Preview Window. I am using Overlay and not Preview (though I have also tried Preview and No Display--it seems to make no difference).
17. Sound Cards. The advice seems to apply to older systems. This machine has onboard sound. Not sure whether that even matters as the ATI USB is capturing sound (not the sound card).
18. Desktop Graphics Settings. This machine can support two 4k monitors. I have one 4k monitor connected to it. To rule that out as causing a problem, I tried several captures using instead an old-school 1920x1080 TV as a monitor. I also tried with my normal monitor turned down to non-native 1920x1080 in both 16-bit and 32-bit color. These experiments have had no apparent effect.
19. Memory / RAM. LS says anything over 2GB does not matter. This machine has 16GB.
20. Hard Drive Cache. I checked this setting for both drives.
21. BIOS Settings. Not sure what the specific guidance is regarding the BIOS, but I have been through it thoroughly.
22. And If None of That Works . . . . LS Suggests that sometimes cards are just bad. But I assume the ATI sourced from him is fine.

What are my VirtualDub Settings?
As best as I can gather, the “recommended settings” in VirtualDub for my hardware are as follows (but as described further below, my mileage has varied):

Device
- ATI TV Wonder selected as device

Video
- Overlay
- Video source: s-video
- Capture pin: screenshot attached below
- Compression: screenshot attached below

Audio
- Enable Audio Playback: Unchecked
- No compression, PCM format
- “Capture device” selected as device

Capture
- Timing: screenshot attached below

What other settings have I tried?
I have experimented with capturing the same Hi8 tape with a variety of timing and other settings to see if I could eliminate dropped frames or get audio in sync. Please reference the screenshots (attached below), which I describe here.

Timing - Tried This 1.png
This improved audio sync somewhat. Recall that on the recommended settings the sync gets worse as the capture progresses. Here, sync is off by about 480ms, but is basically consistent throughout the capture. 13 inserts and 7 dropped frames.

Timing - Tried This 2.png
Here I reverted to the “recommended” settings in the top section, but tried some new DirectShow options. The result was the worst capture to date. The information panel during capture showed bizarre readings (e.g., 180fps average video rate). The audio sync was off by several seconds by the end of the 2-hour capture.

Timing - Tried This 3.png
I read LS’ recommendation in one thread to check “disable timestamps for preview.” Combined with the “sync audio to video” setting, this produced my best capture yet. Audio is almost perfectly synced to video (about 7 frames ahead of video). The audio/video skew is consistent throughout the two-hour capture, so applying a 233ms skew correction produces a great end result. 10 frames dropped and 8 inserts.

Timing - Tried This 4.png
For this round, I had just found lordsmurf’s recommendation to disable DirectX in the main preferences. I attempted a capture with DirectX disabled and with the “recommended” (i.e., “do not resync”) settings. The result was quite poor. Sync is off by only a few frames at the start of the two-hour capture, but at the end of the capture sync is off by about two seconds. 12 dropped frames and 11 inserts.

Note: I do not have screenshots or notes, but I have also tried enabling/disabling the following options made any difference (it did not):
- “Correct video timing for fewer frame drops/inserts”
- “Automatically disable resync when integrated audio/video capture is detected”

My Questions

My primary concern is correcting the audio sync issue. Yes, I care about dropped frames, but only to the extent that may be causing the sync issue. A dropped or duplicate frame (or 10), on its own, is not a problem for this project (this is not the Zapruder film). My questions are, thus, as follows:

1. Are there further settings or settings combinations that I should try to correct the sync/frame drop issue?
2. Are there any hardware issues that I am overlooking?
3. As I described, my best captures have come from selecting “sync audio to video” in the timing options. In that case, sync is still off (audio still precedes video by 7 frames), but it is consistent throughout the capture and that is easily corrected. Also, there are still 7-12 dropped and inserted frames (in two hours), but for my purposes the videos are perfectly fine. I would be satisfied with that result and workflow, but I would like to confirm whether “resampling the audio to a faster or slower rate” may cause any problems later (e.g., when editing/encoding). The resampling is apparently so minor that I cannot hear any distorted audio (no chipmunks or Barry White)--it sounds great. I would just like to confirm that this process is not overly destructive and will not create future headaches.

Thank you!!

[latreche34]

Try AmarecTV just to rule out a software setting issue.

[jdp90000]

Quote:

Originally Posted by latreche34

Try AmarecTV just to rule out a software setting issue.

Thank you. I will try that tomorrow.

I remembered some other things that I have tried with no improvement:
- Virtualdub2, 32bit and 64 bit (I know, not recommended)
- Lagarith for capture
- Capturing to operating system drive (SATA SSD - 500GB) (I know, not recommended)

[lordsmurf]

AmarecTV won't solve anything.

Your VCR, TBC, and capture card are all fine.

Look at the USB ports, either the hardware or drivers. I've run into that on these old Dell systems. What often happens is USB 3 ports operate at USB 2 speeds, or USB 2 ports dip and drop well below sustainable norms. It's most specific to Win7 on these, but I have seen it on Win10 at least once. Sometimes the latest/updated USB drivers are the problem, and you have to roll back earlier. Sometimes latest works best, and prior are the problem. Dell systems can work great for capture systems, but sometimes their prioprietary drivers and hardware implementations get fiddly.

[mrmuy97]

Quote:

Originally Posted by jdp90000

14. Check Your Hard Drive Settings. The guide suggests checking hard drive settings, but that seems to apply to older hardware. One unresolved hard drive question that I have is whether the partitioning schemes I am using might be an issue (Disk1 with OS is MBR, while Disk2 for capture is GPT).

I can't think of any possible way that the partition table type could be directly to blame for this issue. While I also have never personally seen any data read/write latency issue caused by a GPT disk in Win7, I will say that in the case of a dedicated WinXP and/or Win7 capture desktop build, I suggest using only MBR partition tables. This is not a question of one being better than the other in general, but simply a matter of using the one with the lowest risk of issues for the task at hand. Again, I can't imagine it is the culprit for you, just a detail to note all the same.

Quote:

Originally Posted by jdp90000

21. BIOS Settings. Not sure what the specific guidance is regarding the BIOS, but I have been through it thoroughly.

If you'd like to post a <100MB video, or pictures of the screens with user-adjustable options, then I'm happy to give you my suggested values for them. An audio or video call is also fine.

Quote:

Originally Posted by jdp90000

I remembered some other things that I have tried with no improvement:
- Virtualdub2, 32bit and 64 bit (I know, not recommended)
- Lagarith for capture
- Capturing to operating system drive (SATA SSD - 500GB) (I know, not recommended)

Well, not recommended to use as a permanent setup. But you're talking about for troubleshooting, in which case you're doing the right thing by testing the variables and analyzing the results.

[lordsmurf]

Quote:

Originally Posted by mrmuy97

I can't think of any possible way that the partition table type could be directly to blame for this issue.

I can, seen it. But it's way far down the troubleshooting solution list.

Quote:

While I also have never personally seen any data read/write latency issue caused by a GPT disk in Win7

Same here, seen it. But it depends on the drivers (usually Intel or Microsoft), and how it plays with the hardware. Dell is a PITA, screwy, fiddly, but it works eventually. I've build many capture systems, both desktop and laptop, on certain Dell systems.

BTW, I do not recommend using Dell's proprietary cases, power supplies, or fans (for desktops). Extract it to a "real" case, get a good CPU cooler (Cooler Master 212 or better), quality PSU (Corsair RM850x or better), and get adapters (as needed) from Harbin Repair.

[jdp90000]

Quote:

Try AmarecTV just to rule out a software setting issue.

Quote:

AmarecTV won't solve anything.

Last night I installed AmarecTV and captured the same 2-hour Hi8 tape used for all previous tests (probably bad for the tape but want to constrain variables). Interesting (and perhaps unexpected?) result: no dropped frames and perfect audio/video sync. I mean, really perfect sync. Perfect at the beginning of the capture. Perfect in the middle. Perfect at the end.

One odd observation. I had a few false starts where I started the capture in AmarecTV and it immediately reported 3-5 dropped frames. This was before I started the camcorder playback. I just canceled and tried again, maybe two or three times, and finally I got a start without dropped frames. At that point, I pressed play on the camcorder and it captured the whole thing without any drops.

Also odd, during capture, AmarecTV reports "30fps" at the bottom of the window. Occasionally it jumps down to 29.97fps, then back up. The capture settings were properly set to 29.97fps. I am assuming this is just a rounding error in the figure displayed on the status bar, but wanted to mention it.

Question: putting aside the bad English translation and somewhat clunky user interface, is there any disadvantage to capturing in AmarecTV? I was able to use the same Huffyuv codec that LS provided. I would not mind using AmarecTV to capture, but want to know whether the result is inferior to what VirtualDub would produce if working properly.

Quote:

I can, seen it. But it's way far down the troubleshooting solution list.

Tonight I will convert Disk2 to MBR.

Quote:

Look at the USB ports, either the hardware or drivers.

I will work on the USB driver issue.

Question: To rule out the Dell USB hardware/drivers as a potential fault, might I just install a USB 2.0 PCIe card? For example, this card is USB 2.0 only (no 3.0). According to reviews, it works natively with Windows 7. Several reviewers use it for legacy hardware that does not play well with 3.0 ports running in 2.0 mode. If that would rule out the Dell USB issue, I'll just do it.

[Feedbucket]

Amarec glosses over platform flaws and maintains sync by very liberally inserting duplicate frames. It Just Works if your goal is to get something done quick and dirty, but it's problematic if you're looking for per-frame accuracy to the source material.

[CaptureGuru]

Quote:

Originally Posted by Feedbucket

Amarec glosses over platform flaws and maintains sync by very liberally inserting duplicate frames. It Just Works if your goal is to get something done quick and dirty, but it's problematic if you're looking for per-frame accuracy to the source material.

User jdp90000 can easily verify if there are inserted frames in his amarec capture; a frame inserted is not "written" by any capture software, but there is an indication (few bytes) in the stream to repeat the previous.

A method to find them:

1. check the amarec log file
2. open the capture in virtualdub and use the menu commands Go ~> Next drop frame
3. run a specific avisynth script to find identical frames, there are many available
4. compare side-by-side the virtualdub capture and the amarec capture in avisynth and check the frames sequence (if the virtualdub capture has dropped frames, this should be taken into account)

Very very easy to assess!

[latreche34]

Vdub hasn't been updated for ages, it's bloated with features and settings that should no longer be there, which makes it buggy with some computer hardware, I mean who still records analog TV channels from a tuner card? I personally could never get vdub to work properly on any of my machines, I abandoned it long time ago, not just because it doesn't work, but because I use a proprietary lightweight capture app for my specific hardware to capture lossless AVI. I do use vdub2 for trimming and processing.

[lordsmurf]

And yet, VirtualDub is actually more updated than AmaRecTV, which is suddenly getting undeserved recommendations. It has more bugs, and fails with more cards, as compared to VirtualDub. Prior to this, it had been forgotten, Japanese software from a now-gone era of online gaming.

The only reason it doesn't have TV tuner ability is because it was made for analog streamcasting video games -- which is also why it's not suggested for analog videotape capture. Sort of like how OBS is for digital streamcasting, and treats all capture cards like webcams, not fit for videotapes.

In fact, your SDI card isn't newer, and I doubt the capture app is either.

This is a legacy task. These are 2010s updates to 2000s hardware/software, for capturing 1980s-90s videtapes. "Mine's newer than yours" is a playground pissing contest.

It's all bloated and buggy in some way. I've never met a capture app that wasn't.

I tried to work with a software dev to de-bloat VirtualDub as a capture-only app, but he ghosted during the pandemic. I'm not sure what happened to him.

[CaptureGuru]

User lordsmurf, why do not wait until the user jdp90000 can provide some details about the captures? This will explain a lot

User jdp90000, I can help if needed, because I am very interested in the subject

[latreche34]

I never said newer is better, I was talking about updates, Media express is old but it gets updated regularity. Had vdub been updated and made to work with newer platforms I would be using it today, Yes it's a legacy task but some features are no longer needed, hence the need for updates.

[lordsmurf]

Quote:

Originally Posted by latreche34

Media express is old but it gets updated regularity.

Interesting. What are those for?

Quote:

Had vdub been updated and made to work with newer platforms
but some features are no longer needed, hence the need for updates.

Agree.

[jdp90000]

Quote:

Originally Posted by jdp90000

Question: To rule out the Dell USB hardware/drivers as a potential fault, might I just install a USB 2.0 PCIe card? For example, this card is USB 2.0 only (no 3.0). According to reviews, it works natively with Windows 7. Several reviewers use it for legacy hardware that does not play well with 3.0 ports running in 2.0 mode. If that would rule out the Dell USB issue, I'll just do it.

I ordered this USB card and will install it tomorrow. Would be good to know whether people think that is worthwhile to try. I am assuming that the drivers for the PCIe USB card are different than the suspect drivers that operate the onboard Dell USB ports. Is that a fair assumption?

Quote:

I suggest using only MBR partition tables.

I have re-partitioned the capture drive to be MBR. Have not done another capture to test as that seems like an unlikely cause of the problem. I will run another test capture after installing the USB card.

Quote:

compare side-by-side the virtualdub capture and the amarec capture in avisynth and check the frames sequence (if the virtualdub capture has dropped frames, this should be taken into account)

I am also comparing the Virtualdub and Amarac captures using the Avisynth stackhorizontal function. Preliminary results may be disappointing to both Virtualdub and Amarac proponents.

Before using stackhorizontal, I first trimmed the beginning of each capture to start at the same frame. I scrubbed through manually and made the following observations.

(1) Only a few seconds into the footage, Amarac had a duplicate frame. By that I mean it was an exact copy of the same frame, appearing twice in a row. That caused the two captures to be out-of-sync from each other during the stackhorizontal playback.

(2) A few minutes later, the Virtualdub capture experienced a frame insert. That brought the captures back into sync. But unlike the Amarac insert, it was not a perfect duplicate of the prior frame. It was actually a combination of two adjacent frames spliced together vertically separated by a vertical zig-zag border (zig-zag like scan lines). I can post a screenshot tonight.

(3) At the end of the footage (2h:04m), the Virtualdub capture had 40 more frames than the Amarac capture. I have deduced that this could have been caused by Amarac drops and/or Virtualdub inserts. But my capture notes state that during the Virtualdub capture there were only 11 reported inserts (not 40), so the reported Virtualdub inserts do not account for the discrepancy. Also recall that Amarac reported 0 drops during the capture. That means either the Amarac drop reporting is inaccurate or the Virtualdub insert reporting is inaccurate, or both.

I need to perform a more thorough comparison to identify where the remaining drops and inserts are. A member has recommended a duplicate detection tool that may save me some time over comparing the footage manually. Will report back. Any further thoughts are welcome.

[CaptureGuru]

Quote:

Also recall that Amarac reported 0 drops during the capture.

This is probably correct, and does not relate to inserted frames. Drop frames and inserted frames are different. A drop frame is a frame present in the source that has been dropped because it arrived late in time, with the capture software having no other option than discharge it. An inserted frame is because the current frame does not arrive in time, but the capture software has the chance to introduce a replica of the previous frame to keep a/v synch, or because some jitter is present in the a/v packets and is compensated by the capture software, also to keep a/v synch.

Once a video has been captured, it is very easy to detect inserted frames, because they are somehow "marked" in the capture stream, and can be found using this property, or with a simple avisynth script checking the difference/matching between frames (use this as double check)

The dropped frames cannot be detected after capture, only a visual inspection of the whole frame sequence can sometimes reveal them because you see a sort of "jump" in the sequence, but it is not always so obvious. Multiple captures of the same tape may help in identifing dropped frames, but only if there is not a problem in the tape repeating at each capture and introducing each time a dropped frame(s).

Keep us up to date with your progress, so users here can have some data

[jdp90000]

Thank you. The frame numbers in Virtualdub are marked [K], [D] at the bottom of the window. I read elsewhere that [D] marks a dropped frame. If that is true, does that mean there were additional frames in the source before or after the [D]-marked frame? It does not make sense that the frame marked [D] itself would be the dropped frame (as we can look at it).

You are saying that dropped frames can be hard to detect. Do you mean non-marked dropped frames?

Unlike the [D] marked frames, I have not been able to find any way that inserted frames are marked. Is there some marking that I do not see? I will try the Avisynth tool, but so far I have only found duplicates by manual comparison.

[CaptureGuru]

A dropped frame is not present in the capture, and cannot be "marked" in any way; is reported by virtualdub or amarec as "dropped", giving the time when this happened.

An inserted frame is not a real frame in the capture, but an instruction to the player to repeat the previous frame. This is the "marker" and can be easily detected by a hex editor or by a software, like vdub using its menus. Be careful that what virtualdub calls "dropped frames" in its menus are in fact "inserted frames" (maybe this is the reason for your doubt)

[jdp90000]

Disappointing news. The PCIe USB 2.0 card that I ordered was delivered today and I ran a test capture with the capture card connected through it. No improvement with dropped frames and audio sync is still way off. I confirmed that the USB card is using different drivers (Microsoft) than the Dell onboard USB ports (Intel).

Now what?

[jdp90000]

Putting aside the debate about the advantages of Virtualdub vs. Amarec, I am looking to further troubleshoot why I am getting frame drops and inserts in both programs now that I have ruled out the on-board USB ports as causing the problem by installing a dedicated USB 2.0 card.

I am not committed to using this particular PC if another system would perform better. I would just need to know what system is known to play well with the ATI 600. I just want to begin capturing and will buy the system I need. What specific system is working well for others with this capture card? Suggestions welcome on the following:
- Specific operating system (XP, 7) (SP1, SP2, etc.)
- Processor
- Using a pre-built system that's working for you? Please share the specific model and what changes you have made. Below I received advice about problems with Dell's on-board USB ports. Wish I had known that before outfitting a Dell.

It seems less likely that the Hi8 camcorder I am using for playback is causing dropped frames or sync issues. But again, if that is possible, I will get another one to rule it out.

That leaves only the capture card and TBC, for which I paid handsomely and am told are fine.