JVC settings and VC500 vs. ATI 600?

[Nunkin]

Workflow:

I have sourced myself a VC500 and an ATI 600 TVWonder USB (VisionTek), running from a JVC SR-VS30U with S-Video, into my Windows 7 64bit pro pc, using VirtualDub.

I'm working on sourcing an external full-frame TBC, but that is taking a while.


What I've tried:

I have tried using each capture card (A) on a home video recording of a broadcast (NTSC), and (B) on a commercial video tape (NTSC). Both are non-cartoons.

On my VCR, I have left everything at default settings apart VCR Picture Control Setting which I put into "EDIT", and "Digital TBC/NR" and "Video Stabilizer" which I ran in 3 settings scenarios:

1. Digital TBC ON, Video Stab OFF
2. Digital TBC OFF, Video Stab ON
3. Both OFF

(They can't both be on at the same time)

Full list of relevant options for reference
VIDEO CALIBRATION - [ON/OFF]
VCR PICTURE CONTROL SETTING - [AUTO/EDIT/SOFT/SHARP]
AUTO SP->EP TIMER - [ON/OFF]
BLUE BACK - [ON/OFF]
DIGITAL TBC/NR - [ON/OFF]
VIDEO STABILIZER - [ON/OFF]
DIGITAL R3 - [ON/OFF]


Results:
After de-interlacing only. Bearing in mind my untrained eye, I thought max detail quality was 5/10, same as TV.


***** Home Video: *****

Best result: ATI600 Scen 3 - I was very surprised by this result given it has both the TBC and the Stabilizer off.
Gave minimal horizontal and vertical jitters
Vertically the picture stayed straight
Bottom 3% of image seemed messed up (back porch?)
Quality of detail was ok ~5/10
Video lagged audio substantially, which seemed to be a general problem I had with this card

ATI 600 Scen 1
Side-to-side (see image 1 below) wave to the right then left at the top 10% of the screen image that was very jittery
Some minor up down wobbles/jitters on the entire image - I was surprised given this had the VCR TBC on.
Bottom 5% of image had a wierd black/white image that didn't tie too well with the rest of the video (back porch?)
Quality of detail was ok ~5/10.
Video lagged behind audio.

ATI 600 Scen 2
Had a smaller side-to-side wave in the top 6-7% of the screen image (see image 2 below)
No up-down wobbles or jitters
Image bent to the left from bottom to top such that the top of the image was about 3-5% to the left of the bottom.
Quality of detail was ok ~5/10.
Top 3% of image looked like black/white ants (front porch?). No bottom 3-5% mess though.
Video lagged behind audio.

VC500 Scen 1
Massive bend to the right then left (see image 1 below) in the top 15-20% of the image.
A bit of vertical jitter
Bottom 3% of image looked weird blur (back porch?)
Detail quality was 4/10 to 5/10.

VC500 Scen 2
Really bad up-down jitter - watchable if desperate but will give you a headache after 10mins.
Image bent to the left from bottom to top such that the top of the image was about 3-5% to the left of the bottom.
Bottom 3% of image was black.
Detail quality 3/10 or 4/10 (if you ignore the jitter) - particularly grainy, and detail wasn't v clear.

VC500 Scen 3
Image seemed to bend gradually to the left from bottom to top such that the top strip was about 5% further left than the bottom strip of the image. Otherwise would have given it 1st ahead of ATI 600 as there was no video/audio mismatch. Again surprised as this had no TBC or stabilizer on.
3% at bottom messed up (back porch?).
Detail quality 5/10


***** Commercial Video: *****

Best result: VC500 Scen 1
No horizontal or vertical waves / jitters / bending
Bottom 1-2% of image didn't make much sense - back porch?
Detail quality 5/10 - same as TV
However, see VC500 question 1 below about brightness vs the ATI600

ATI 600 Scen 1 - would also have been best result, if not for what I think might be Macrovision
No horizontal or vertical waves / jitters / bending
Video lagged audio
Detail quality 5/10 - same as TV
See ATI 600 Q2 below about what I think was macrovision
White balance seemed right

ATI 600 Scen 2
Entire image had some side-to-side waviness (see image 2 below)
Minor side to side jittering
No issues resembling what I think could be back porch / front porch captures
Video lagged audio
Detail quality 4/10

ATI 600 Scen 3
Entire image had some side-to-side waviness (see image 2 below)
Slight bend to the left from middle to top such that the top of the image was about 0.5-1% to the left of the bottom
Bottom 1-2% of image didn't make much sense - back porch?
Video lagged audio
Detail quality 4/10



VC500 Scen 2
Quite bad vertical jitter - not something you would be able to watch for more than 5-10 mins without a seizure
Slight bend to the left from middle to top such that the top of the image was about 1-2% to the left of the bottom
White balance very off (much more so than the other VC500 captures).
Detail quality 4/10
No issues resembling what I think could be back porch / front porch captures

VC500 Scen 3
very slight horizontal jitter
Slight bend to the left from middle to top such that the top of the image was about 1-2% to the left of the bottom
Bottom 1-2% of image didn't make much sense - back porch?
Detail quality 5/10
Same white balance question


Other notes on my tests
Sound often seemed crackly / to have noise in the VirtualDub preview screen, however when I looked at the output file, the sound was clean, so not something to worry about.


Questions

ATI 600 Issues

1. Has anyone seen / experienced video lagging behind the audio significantly on the ATI 600?
2. On the commercial video, every 5 seconds the image would get totally messed up. Stopping and pressing play would clear it up, but this would happen again in about 5 seconds' time. I believe this is Macrovision that the external TBC will sort out? This did not happen with the VC500 though - so is VC500 Macrovision friendly? (See image 3 below)

VC500 Issues

1. The VC500 seemed to give a slightly brighter image than the ATI600 - such that some of the info was lost (See image 4 below of a bookcase with glass front window). Interestingly, the TV image was actually a match with the VC500 - does that mean that's the way it's supposed to be?!
2. With the VC500, I played with the brightness (4950 vs 5000 default) and the contrast (4750 vs 5000 default) to get something similar to the colour results that the ATI600 produced - should I have played with the contrast that much?
3. With the VC500, the proc amp settings didn't seem to stick. I don't mean the fact that I have to reset them every time I open VirtualDub - I mean that when I stopped and played the video, it looked like the settings were not there. Only if I paused / stopped / rewound and then played again, did the settings seem to take effect. Does anyone know why this could be?

General
Any general thoughts on the results of my tests would also be greatly appreciated.

1. Does anyone know why with the home video recording worked better without the TBC or Stabilizer turned on? This was essentially true for both the VC500 and the ATI 600.
2. I consistently saw this top/bottom 3-5% bar of weirdness across the captures - is this back porch / front porch signal being captured? Why doesn't this come up on a TV? Why would the commercial video have so much less of it?
3. Will any of the issues I saw be cleared up by the external TBC?
4. How do I identify 'crushed blacks'? Is the Snell & Wilcox test video the only way to do that properly?

Pictures:
I haven't posted full images in all cases, but tried to give you an idea of how much of the image you can see in each case:

Image 1 - Large side-to-side wave at top of image - full width, top 20-30% of image


Image 2 - smaller side-to-side wave at top of image, or throughout image - full width, top 20-30% of image


Image 3 - White balance comparison ATI 600 vs VC500 - ~20% height, 40% width


Image 4 - Is this Macrovision?

[traal]

I'm commenting only on the VC500 issues because I have one.

Quote:

Originally Posted by Nunkin

1. The VC500 seemed to give a slightly brighter image than the ATI600 - such that some of the info was lost (See image 4 below of a bookcase with glass front window). Interestingly, the TV image was actually a match with the VC500 - does that mean that's the way it's supposed to be?!

No, you should adjust brightness and contrast before capture, then both should look similar.

Quote:

Originally Posted by Nunkin

2. With the VC500, I played with the brightness (4950 vs 5000 default) and the contrast (4750 vs 5000 default) to get something similar to the colour results that the ATI600 produced - should I have played with the contrast that much?

On my VC500, I adjust them much more than that. Don't worry about what the numbers are, worry about getting a good histogram.

Quote:

Originally Posted by Nunkin

3. With the VC500, the proc amp settings didn't seem to stick. I don't mean the fact that I have to reset them every time I open VirtualDub - I mean that when I stopped and played the video, it looked like the settings were not there. Only if I paused / stopped / rewound and then played again, did the settings seem to take effect. Does anyone know why this could be?

It could be the VC500's built-in AGC. I found that adding in a good external TBC helps to stabilize luma.

[jwillis84]

Image 1 and image 2 are examples of "Top Curl"

Some people call it "Flag waving" because it resembles a flag on a pole waving in the wind at the top of the image.

This article explains some of this:

Analog video time base correction and processing for nonstandard TV signals

Basically the article separates "timing errors" into those common for Broadcast captured video [and] then those common for VCR captured video.

They are a little different.

Broadcast is vulnerable to more line to line noise which jitters back and forth, this can also occur for VCRs but for different reasons.

Broadcast picks up genuine RF noise which confuses the "sync" detection circuits for horizontal and vertical sync pulses.

VCRs do not pickup noise due to RF but they have mechanical errors which can induce their own form of problems for the "sync" detection circuits for horizontal and vertical sync pulses.

Dumbing it all down, it comes down to "gating" or "filtering" the places where the sync pulses "should exist" but for whatever reason may not exist, or can't be separated from the noise.

TBC correction gets defined by different things and fixed by different methods.

But in the broad view, they "free run" a clock that is approximately synchronized to what the fields and frames "should be" running and adjust these clocks slowly.. if ever with a feedback loop.

This clock is the "timebase" and its used to "insert" the corrected sync pulses so they are clear and sharp.

The video capture device then has no trouble seeing these sync pulse and lock on.

Flag waving or "Top Curl" happens because in a VCR there are two heads, one for each "field" they sweep the video tape at an angle from bottom to top. One sweep contains all of the horizontal lines for that field.

The two fields that make up a frame are called "Odd" and "Even"

What they have in common is they both start their sweep from the bottom edge of the video tape.

Next to the start of those sweep is a linear "tracking" signal that is recorded linearly along the edge of the tape.

Over time as the tape gets played, aged, or stressed the tracking pulses begin to bleed or effect the ability of the sync detection circuits to get a clean lock on the start of horizontal sync pulses for the first few lines. That causes the "drift" that waves like a flag. Later lines further up in the tape away from the Top Edge are better preserved and consistent with themselves.

The fix is for the time base corrector to "delay" or "gate" resetting it horizontal sync pulses based on the other pulses from previous lines in the previous field and frames horizontal lines.

That's way over simplifying things.. but that's the TBC correction for sync signals specific for VCR video.

The difference in the way a TBC correction for sync signals for Broadcast video is mostly the same, but the adjustments tend to occur at a faster rate.. which screws up when its trying to remove Top Curl from a VCR video signal. The Top Curl remains.

So there are TBC's for VCRs and TBC's for Broadcasters

Another type of TBC correction is stretching and compressing each line so it fits in the time interval a horizontal line should exist.

They do this by sampling for an extra long time into a memory buffer and they "normalizing" it so samples are provided at the appropriate rate to turn the line back into the appropriate length. Basically its a game of don't fill up the output buffer too fast, but also make sure it doesn't run out of samples to provide to output of the TBC. That way when the Video capture device resamples the pseudo "analog signal" it seems a nice clean uniform set of horizontal lines across the whole stack of horizontal lines that make up each field and frame of the picture.

Some VCR's have this last feature. They basically "re-time" their output stage so the horizontal lines are all of uniform length. But generally they do not "re-time" or "re-construct" their vertical and horizontal sync signals and are vulnerable to "Top Curl".

That's why you need [both] a TBC that re-times VCR horizontal lines and a TBC that reconstructs VCR sync signals.. they are basically two very different things.

And you need the TBC that reconstructs VCR sync signals be "designed" for a consumer VCR, and not Broadcast or Broadcast VCRs, because the Broadcast quality equipment will "over correct" looking for RF interference.. rather than expecting a slower "drift" due to problems with the tracking signal and edge of the video tape.

Up until now I haven't mentioned the Vertical "sync" pulses.. which are usually far easier to catch, and harder to mess up with noise or mechanical errors. Basically each sweep of the drum by a video head generates a clear vertical sync pulse preambled by a bunch of "video blanking lines" which precede the vertical sync pulse, and take place "off camera" as it were after the head switching noise at the bottom.

Originally AGC - Automatic Gain Control circuits used that blank area to "Reference" a null zone for resetting the Amps to Black. But Macrovision started "Writing" non-zero pulses in that area to deliberately screw up Automatic Gain Control, such that every field might have a different set of pulses and alternately start flashing or totally crashing the AGC. TV's AGC can tolerate the extra "noise". Video Capture devices generally "can't" or "they are not supposed to" since legally this is a protected method of Copyright enforcement.

As part of their job correcting vertical and horizontal "sync" errors, TBCs correcting sync "clamp" or "drive to null" the vertical sync errors by assuming they are noise "squashing" Macrovision as a side effect.

This has other downsides.

Close Captions and EPG programming information is also encoded in some of those vertical blanking lines, so a TBC will squash those lines too if its not sophisticated enough to only clamp some lines and not others.

[lordsmurf]

Long post.

- Can be good VCR. Have one myself.
- 600 USB fine.
- VC500 is just a variable card, all over the place in terms of values, AGC, etc. So not recommended for that reason. Some are good, some not.

Non-cartoons? Boo!

Default bad.
EDIT almost always bad, disables quality-improving features, like chroma NR.
R3 bad, causes halos.
Superimpose bad.
TBC almost always on. Rarely off.

Deinterlacing? Why?

Bottom 3% is head switching noise. Normal.

Audio lag/sync is lack of frame TBC. When TBC in use, settings error (user error),

Image 1 is tearing. If JVC TBC fails, ES10/15 needed.
Same for Image 2.
"curl" or "flagging" is laymen terms. Technical jargon is "tearing". (analog tape tearing, not to be confused with digital tearing, or other types of tearing)

Image 4 can be anti-copy detection. Sometimes the anti-copy is actually present on tape, sometimes not (false positive). Again, this is due to lack of frame TBC, such as DataVideo TBC-1000.
BTW, in other cases (not here), it can be VirtualDub preview vs. overlay issue.

This is just a "lack of TBC" post. All the errors are rooted there.

Buy a TBC. Problem solved.

[Nunkin]

Quote:

Originally Posted by traal

No, you should adjust brightness and contrast before capture, then both should look similar.

Thanks - I forgot about adjusting to the histogram - will try that out next time I test the VC500.

Quote:

Originally Posted by jwillis84

Image 1 and image 2 are examples of "Top Curl"

Broadcast is vulnerable to more line to line noise which jitters back and forth, this can also occur for VCRs but for different reasons.

• Interesting, so the crazy jitters I was seeing on my broadcast home-recording could have been because of this then?

Quote:

Originally Posted by jwillis84

Image 1 and image 2 are examples of "Top Curl"

The fix is for the time base corrector to "delay" or "gate" resetting it horizontal sync pulses based on the other pulses from previous lines in the previous field and frames horizontal lines.

• Isn't this what the JVC VCR line TBC does? But more importantly, on the home video, why would I get better results without the VCR TBC on?

Quote:

Originally Posted by lordsmurf

Long post.

Took inspiration from jwillis84...! And hoping that if I describe in detail, it will help someone else in future.

Quote:

Originally Posted by lordsmurf

Non-cartoons? Boo!

They are coming. Don't you worry!

Quote:

Originally Posted by lordsmurf

Default bad.
EDIT almost always bad, disables quality-improving features, like chroma NR.
R3 bad, causes halos.
Superimpose bad.
TBC almost always on. Rarely off.

Deinterlacing? Why?

• Which mode do you suggest for VCR Picture Control Setting - Auto?
• Will turn off R3 - what are halos?
• I don't think I have a superimpose setting right? Or is it called something else here?
• To my question to jwillis above though, having the TBC off actually helped in one of the two situations - do you know why..?
• Oh - why not de-interlacing? You have the deinterlacing filter in your version of VirtualDub I think?

Quote:

Originally Posted by lordsmurf

If JVC TBC fails, ES10/15 needed.

• How can I tell if the JVC line TBC is what's at fault?

Quote:

Originally Posted by traal

It could be the VC500's built-in AGC. I found that adding in a good external TBC helps to stabilize luma.

Quote:

Originally Posted by lordsmurf

Buy a TBC. Problem solved.

Hopefully on its way!

Thanks everyone!

[Nunkin]

Project update and questions:

My workflow is now a JVC VS30U -> TBC-1000 -> ATI 600 / VC500 -> Win 7 + VDub 1.9.11 32bit (lordsmurf's) + 3tb internal HDD non-OS.

In case it matters, I have either the TBC or JVC feeding to a TV as well, to monitor what's going on with the signal pre-capture card.

I've started out by testing a 4hr EP set of tapes.

Visual Quality:

• I don't see any difference between the two capture cards, but it could just be my untrained eye.
• I can confirm from some test videos that the VC500's proc amp settings don't seem stable. Proc amp settings that gave a good output histogram could end up with a bad histogram on the next play through - I guess this is the AGC at play?

Audio Quality:

• If I have audio preview on, the sound comes out noisy but it seems to be fine in the actual capture file - more of an issue with ATI 600.
• With both cards, I have some video lag (behind audio), but the lag I get with the ATI seems to be significantly worse. See below the stat outputs from the ATI card and my timing settings (which I took from Sanlyn's VirtualDub Settings Guide. The sample video I took here started with a ~1s lag 2 minutes into the video, and ended with a ~4s lag at the end of the ~3.5hr video.

Other:

• When using the ATI 600, VirtualDub (whether v1.9.11 32bit or 1.10.4 32 or 64bit) is noticeably slow. e.g. when I press F6 (capture) it takes between 10 and 20s for the capture to start, prior to which virtual dub is frozen and the mouse cursor shows it is thinking. Similarly if I change settings, turn on or turn off previewing, the volume meter, basically anything at all, I have to wait for 5-10s for virtual dub to respond again. This happens a bit with the VC 500 but to a much lesser extent (fraction of a second delay).
• I also noticed that if I plug in or disconnect USB devices, this creates a break in the timing graph, and correspondingly the capture is broken at that point too. This confirms what is widely known - don't do anything with your computer when you are capturing! Even if it's a multi-threaded machine with low CPU usage during capture, like mine.

Q1: Has anyone seen poor VirtualDub responsiveness when using the ATI 600? I assume it's either the card or the driver that I'm using given VC500 doesn't do this.

I'm using the driver that msgohan posted (#3136), which I understand is just a slimmed down version of what is on the full CD that lordsmurf posted (#3136) at the start of that thread.

Tried this one lordsmurf posted here (#3070), but for whatever reason it didn't seem to work at all.

I didn't use the full CD version as Kaspersky came up with "V:\Install\XP\AtiCimUn.exe;V:\Install\XP\AtiCimUn .exe;Trojan.Win32.Inject.anfmy" ; Trojan program - has anyone else seen this before?




And just to consolidate some of my previous questions, for which any thoughts would be much appreciated!

Q2: Isn't this what the JVC VCR line TBC does? [Was referring to fixing top-curl] But more importantly, on the home video, why would I get better results with the VCR TBC OFF? In general having the VCR TBC off seemed to help in a couple situations (and have confirmed this to remain the case even with an external frame TBC) - why might this be the case? Elsewhere on this forum it's also noted that sometimes (rarely) TBC off can be better - does anyone know why?

Q3: How can I tell if the JVC line TBC is faulty?


Separately I think I have also figured out some answers to previous questions above - so for future readers' benefits:

Q4: I don't think I have a superimpose setting right? Or is it called something else here?
A: It's under "Mode" and controls whether or not the VCR's symbols are superimposed on the image or not (e.g. when you press play whether or not it stamps "PLAY" on the image)

Q5: Which mode do you suggest for VCR Picture Control Setting - Auto?
A: Yes lordsmurf usually recommends Norm/Auto, but it seems this has to be decided on a tape by tape basis. Others have suggested Edit to be the preference, but because it turns off noise reduction (NR) can be worse.
See this post (#8455), where EDIT seems to be the better mode.

Q6: Oh - why not de-interlacing? You [lordsmurf] have the de-interlacing filter in your version of VirtualDub I think?
A: De-interlacing loses information which you will never get back - hence it's not favoured for archiving. However if your goal is to compress your video into a small size video for consumption, and never worry about the 100GB lossless file ever again, then I believe de-interlacing is what you want. Someone please correct if wrong.

Q7: What are halos?
A: See this post (#4634) for a good example. Though I must admit I couldn't identify halos that sanlyn spoke of in this post (#8455)...

-- merged --

I should have added a bullet to the visual quality section. Admins please merge.

A lot of the issues I was having - e.g. the suspected macrovision mashup, and vertical lines not being straight, have been solved by the addition of the TBC.

[traal]

Quote:

Originally Posted by Nunkin

I can confirm from some test videos that the VC500's proc amp settings don't seem stable. Proc amp settings that gave a good output histogram could end up with a bad histogram on the next play through - I guess this is the AGC at play?

Just to be clear, that is WITH the TBC-1000, right?

[Nunkin]

Quote:

Originally Posted by traal

Just to be clear, that is WITH the TBC-1000, right?

Yes, with the TBC-1000. I just tested it to double check again. Used the same proc amp settings as my first test in my log that I found to give a good histogram (6440 brightness, 4596 contrast). Initially had a lot of overflow on the white side. Rewound the tape about 1 minute back, played, and progressively watched the red overflow on this histogram reduce. By the time I got back to the spot where i first started, there was no red overflow.

[rflagler]

Quote:

Originally Posted by Nunkin

Yes, with the TBC-1000. I just tested it to double check again. Used the same proc amp settings as my first test in my log that I found to give a good histogram (6440 brightness, 4596 contrast). Initially had a lot of overflow on the white side. Rewound the tape about 1 minute back, played, and progressively watched the red overflow on this histogram reduce. By the time I got back to the spot where i first started, there was no red overflow.

Wow, I'm going through a very similar issue with my setup.
JVC 9800
TBC-1000
Diamond VC500

Will start a tape and adjust to get a good histogram and then rewind, replay, and it's different....very annoying. I'm going to try a different capture card next and see if it helps.

[Nunkin]

I changed to the ATI 600 USB stick on my windows 7 system and all the brightness problems disappeared.

The only downside is that the ATI 600 USB seems to take a few (10-25s) to actually start capturing from the moment I press the F6 key on VirtualDub. I haven't figured out if this is VirtualDub being slow or the card doing something funny, but 25s is workable as most tapes have a blank section at the beginning anyway.

Hope that helps!!