Youtube comparisons of VHS captures?

[Gary34]

I was just wondering what you guys thought of this. https://youtu.be/3y20wXQziUI?si=QFVZaxjDV_v1Zox5

To me plugging an Intensity Shuttle up without using the proc amp settings then putting it through Topaz doesn’t seem like a fair representation of conventional capture. They blame clipping on the old outdated processing of the VCR.

The guy that does the video is hopefully pretty knowledgeable considering he restores and sells all of these models on eBay.

Conventional capture is limited range YUV. You gotta stay within the legal limits. The Intensity Shuttle will capture outside of the legal limits but you have to adjust that before going to RGB. I’ve seen other examples besides this where it’s just clipping and the conventional capture method is blamed.

I wouldn’t trust Topaz with not screwing up a comparison. https://community.topazlabs.com/t/ur...findings/54770

[Aya_Rei]

Ah, this is the guy from this thread

Clipping can still happen when the footage was being recorded (IE turning on the camera's light when the room is already brightly lit) But at that point I believe adjusting the proc amp is more so to not make the baked in clipping any worse.

I haven't really seen that much benefit to having the values not be clipped during capture when it comes to camcorder recordings, more so recordings done from a digital device or game console. I can 100% see recovered detail when the black and whites (moreso blacks) are within my capture card's legal limit of 16-235

Still try and have the values set so that the levels are within legal limits of course, no matter the tape. Usually I see recovery of lost detail in the form of added noise.

Hm, guess it is harder to find reputable refurbishers, I'd say my only option going forward is armakolt..

[latreche34]

I believe the DDD comparison is not fair, though it still shows chroma artifacts not available in conventional methods, So yeah, intensity shuttle and downsizing to 640 is wrong from the get go.

[Aya_Rei]

Good points Latreche, yeah it's really difficult to see the samples clearly thanks to YouTube compression.

Even then on the topic of sharpness and detail gain, this begs the question, what were the settings used on each VCR? Was it EDIT or NORM? We may never know..

I think for comparisons like this, and in general if restoration work is planned, it is best to leave the VCR in EDIT mode rather than NORM.

[aramkolt]

Definitely hadn't seen this one, it's stirring a lot of thoughts for me, most of them being poor comparison design.

@5:21 - 7600 is unquestionably sharper than 9600 - that shouldn't be representative of all 9600's and I believe his either has significantly worn heads or some other problem. Check out the stripes on his shirt which are visible on the right (7600) but not the left (9600).

@7:28 -9800 again is much sharper than the 9600 based on the visibility of the stripes on the shirt. I really think that 9600 is defective.

@6:28 - I think he has "dropout compensator" and "TBC" confused. The 9600u simply appears to have a better dropout compensator to me. You usually don't get black horizontal lines in the middle of an image with.anything a TBC can correct, though I suppose I could be mistaken.

@9:26 - TBC on the 9900 is showing extreme inferiority to the 9800. Watch the date code with vertical jitter on the right 9900 sample. Vertical jitter typically means lack of a frame TBC, but the 9800 does not have the issue at all meaning you could probably get away without one with it, whereas you can't with the 9900.

@14:40 - The striped shirt shows that sharpness takes a hit with the particular firewire chip that is used here. If the same DVS3U was not used for both of these comparisons, this test would be invalid, so would have probably been better to verify this. DV usually doesn't look that inferior when it comes to vertical sharpness, so wonder if something is up with his deinterlace method that differs with DV maybe.

@18:00 - DD shows that the girl's shirt was light pink checkerboard all along. He does say later in the video that the DdD footage was captured on a modified 9600, but doesn't sound like it is the same 9600 that I believe is defective due to the lack of sharpness.

The fact that luma and chroma levels are quite different for the "non-firewire/DdD" suggests that his refurb process does not adjust luma/chroma to any sort of standard output level. This is possible to do via the service remote and his listing descriptions suggests that they are "fully rebuilt" - which is just an odd term to use since there's no way the full mechanism has been fully disassembled and reassembled. Way more potential for harm than good if it works well already. I would suspect most machines just have their transports cleaned, DD disabled, pinch roller replaced if needed (rarely needed on the models mentioned), and some minor alignment.

I question the unit-to-unit variation of the same part number with this specific comparison, most likely in the form of the degree of head wear or possibly capacitors drifting out of spec. If he has a defective 9600 or it happens to have worn heads, it's not going to be representative of a fully functional 9600 in general just as an example. What it really needs is at least two of each model side by side to show that one 9600 isn't inferior to another.

The types of samples he's showing aren't particularly useful to determine if any of the machines have better output characteristics/sharpness at a glance. For that, you'd need to use test pattern recordings that show lines of resolution and other things such as frequency response and chroma bandwidth. SW2 pattern would show all of that in one place.

Other observations:
Doesn't say whether he's using composite or S-Video. You'd assume he's using S-Video, but that'd would explain why DdD looks a lot better here than most of the other head to heads online if you are really comparing composite to the DdD.

Sad to think someone bought that specific 9600U for the $1300 he charges on ebay if the sharpness is that poor.

Frames aren't exactly aligned on his comparisons which is frustrating.

Odd he brings in the DdD since in this specific comparison, it makes his other machines look like garbage. He does the DdD mod to those machines for $300 when it's like 15 minutes and $20 worth of materials at most to do it. Seems like further evidence that he's just fleecing people who don't know what they are buying.

I guess people are willing to pay what they are willing to pay, so hard to get upset about that. What I don't like is that they are paying for the "expertise" when he doesn't sound very knowledgeable in the video and can't tell that 9600 has issues that he's using as the standard of comparison.

I maintain that there's no way he'd use that specific 9600 in the video to transfer his own home tapes if they all look like that given all of the other machines he could use instead (he states he uses a 9600 on his own tapes in the video).

That's my rant, feel free to disagree.

[latreche34]

In the DDD samples I believe he is tipping the scale, There is no way that the blond teenage girl's shirt would lose completely the red dots in a conventional capture and appears completely white, I can challenge him to send me the tape and show him the red dots in my capture.

Aramkolt, Worn heads don't cause the video to loose details, they can cause noise because at the RF level the video hasn't been decoded yet, But the softness or loss of details can be caused by aging electronics in the video processing stages. I would take that comparison with a grain of salt unless he can travel back in time and do it on brand new VCRs, or at least have several used VCRs of each model and can prove the results are consistent.

[lordsmurf]

Quote:

Originally Posted by Gary34

The guy that does the video is hopefully pretty knowledgeable considering he restores and sells all of these models on eBay.

Knowing how to use gear and fix gear are two separate concepts. Rarely do these meet.

Never give credit where none is due.

A lot of these vhs-decode folks wouldn't know a quality capture if it poked them in the eye. They mostly like to tinker with the guts of a VCR, and are wholly inexperienced at image quality (what to expect, what's possible, what is legal/true in the content). Many just believe what their "dear leader" tells them, and/or believe highly misleading "samples" (which are more a statement about the person or the VCR/gear used, not the method as performed by a skilled user with refurb'd in-spec decks/gear).

A lot of VCR/capture users are afraid to even take off the lid of a VCR, fearful that a monster will jump out and eat their face. But they know how to use the gear, in order to extract every last drop of quality.

Sometimes you get a blend of use/fix (use/tinker) skills, but never have I seen somebody that has mastered both. I'm not even a master of both. That's a hard skill, as it require both sides of the brain to fire up. Usage requires the artistic side, and fixing needs the analytical side. Nobody uses all of both, but many use some of both, and many are very binary (only artistic, or or analytical).

I know full well where my knowledge and/or ability to fix gear ends. Unlike many, I don't suffer from Dunning-Kruger effect. I know what I know, I know what I don't know. (And I'm always cautious, because I'm know there can be things where I don't know what I don't know.)

Quote:

Originally Posted by latreche34

Aramkolt, Worn heads don't cause the video to loose details, t

No, I'm fairly certain that bad heads can cause smearing. Not softness, but smearing. The difference in micro-smearing and softness can get fuzzy. You can't discern any of that from a Youtube sample.

Quote:

Originally Posted by aramkolt

poor comparison design.
@5:21 - 7600 is unquestionably sharper than 9600 - that shouldn't be
. I really think that 9600 is defective.
@6:28 - I think he has "dropout compensator" and "TBC" confused.
suggests that his refurb process does not adjust luma/chroma to any sort of standard
"fully rebuilt" - which is just an odd term
more potential for harm than goodThe types of samples he's showing aren't particularly useful
Sad to think someone bought that specific 9600U for the $1300 he charges on ebay if the sharpness is that poor.
Frames aren't exactly aligned on his comparisons which is frustrating.
he doesn't sound very knowledgeable in the video and can't tell that 9600 has issues
That's my rant, feel free to disagree.

Agree.

Quote:

@9:26 - TBC on the 9900 is showing extreme inferiority to the 9800. Watch the date code with vertical jitter on the right 9900 sample. Vertical jitter typically means lack of a frame TBC, but the 9800 does not have the issue at all meaning you could probably get away without one with it, whereas you can't with the 9900.

9800 = 4mb SDR
9900 = 2mb DDR

I've talked about 2mb vs. 4mb differences since at least 2001. Honestly, it's about time that somebody noticed the difference. I'm not entirely sure if the 9911 fixed this gap in performance, but the SR all did fine.

Neither perform well without frame TBC.

Frame TBC is about the tape, the signal, the VCR -- not only the VCR.

[latreche34]

4MB vs 2MB as seen in his example affected only the DOC, I don't see it had any effect on the TBC functionality. Larger memory buffer means more lines to be replaced, small buffer fewer lines.

[lordsmurf]

2mb vs. 4mb has always "broken" quicker, buffer underrun, and errors slip in.

The JVC TBC doesn't do more than line-by-line timing correction, not multi-line (as that would be a field TBC, like the AG-1980P). This is why it sometimes add tearing to line length mismatch.

The decks were designed with 4mb in mind, but 2mb DDR was supposed to be "double". It wasn't.

Although I do forget offhand if the 9900 was 2mb or 4mb. That was right around the breaking point. The 7800 is not as good as the 7600, image-wise, for that reason.

I'm having to go off memory with some of this. I've not dug deep on the 2mb/4mb topic for 5+ years now.

[Gary34]

Quote:

I'd say my only option going forward is armakolt..

I’m going to go to whoever is recommended by this forum at the time because that is based on a lot of repairs and the users giving the recommendation are experienced and the people I’m seeing following forum advice wind up with working gear. I personally wouldn’t go to Nathan Brown. That whole thing just seems kinda strange. I agree with this thread that was linked earlier https://www.digitalfaq.com/forum/vid...-ebay-vcr.html. I don’t know for sure but my BS meter is going off too much to pay for that.

I don’t trust the random reviews. I have seen this too he says that he just got his 1980 refurbished. https://youtu.be/AmYQzYZJc8M?si=u4QBx3qwTx8Yjnvy At 8:30 the Panasonics 1980s aren’t supposed to ghost when they are in norm or detail. Edit in a 1980 isn’t preferred for capture. Alitek (the guy he bought the 1980 from) refurbished his 1980. I wouldn’t wanna buy someone else’s failed project.

[aramkolt]

This is from JVC's SR-MV30 technical description of DigiPure Technology:

Quote:

JVC's exclusive digital technologies maximize Super VHS benefits by reproducing vivid colors most realistically, and offering crisp rendition of image contours.
-Digital Wide TBC (Time Base Corrector)
-3-Dimensional Digital Circuit with 2 MB Frame Memory to make the following possible.
-Precision 3-D Color Circuit for clearer color separation.
-Digital 3-D YNR/CNR to improve S/N ratio.
-Digital 3R Picture System to enhance detail.

Given the ordering of the bullet points, it would suggest the TBC itself is not affected by the amount of RAM it has. Seems the RAM has more to do with noise reduction features only if I'm interpreting that correctly.

It is possible that the TBC was a completely separate chip than the rest and doesn't access the separate RAM at all, but it's kind of hard to say for sure as I don't think the service manual really gives a block diagram for how digipure works. The TBCs definitely do act differently in certain situations, so I don't think they are all the same between the various models.

Also, I don't think that the memory is likely to affect dropout compensation. It is very rare for the actual DOC to be digital, even in professional VCRs, see here: https://www.digiommel.fi/images/Vide...log%20VCRs.pdf

Somewhat counterintuitively, the above paper says that dropouts *IMPROVE* with up to a 90% reduction in frequency after playing the tape back ten times. Their theory is that dirt or debris may be removed with each playback. So could depend in which order the machines were tested in terms of whether one showed a dropout or not. Some factory calibration test tapes actually have purposeful dropouts to adjust internal dropout compensators in order to best mask them. Would be nice to have a more specific technique on how to make a test tape like that yourself, but I imagine it'd just be killing or grounding the output of the heads for short blips of time during a regular recording.

On another note, I don't think softening of the entire image uniformly is likely to be due to aging electronics, but it wouldn't be hard to narrow down that possibility of it being worn heads by swapping the decks in a machine that is soft vs one that is sharp of the same model and seeing if the softness follows the deck or not. The 9800 and the rest of the machines tested are all roughly the same age in the scheme of things, so it being much softer than all of the rest, so I still think this suggests a problem of some sort that they don't "usually" develop just with age (since the other machines tested aren't that way).

While not the greatest proof that worn heads can give an overall soft image, this JVC article describes that worn heads do have poorer frequency response. http://pro.jvc.com/prof/attributes/t...&feature_id=02

The line to highlight is here: "In addition, an automatic equalizer is provided to prevent deterioration of the luminance signal frequency response with worn heads, or when using tapes that have different characteristics or have been overplayed."

The pattern that would have shown if there was a frequency response problem is the "Multiburst" pattern below:

Multiburst.jpg

VHS is only expected to perform as well as the luma bandwidth of VHS allows, which is generally noted to be around 3-4MHz with SVHS being higher yet and you'll be able to distinguish the lines further to the right in the image.

[latreche34]

DOC is pure digital memory, at least after mid 90's, I know it is for JVC's later models, so the size of the memory has an effect on how many lines can be replaced, Whether there is more than one memory that I don't know, from a design perspective all digital processing are done in one step therefore one memory is all that is needed, This is why TBC and DNR cannot be done separately.
If you look at this screenshot you will notice on the TBC board on pin 25 it is labeled DOC:

[lordsmurf]

Quote:

Originally Posted by aramkolt

Given the ordering of the bullet points, it would suggest the TBC itself is not affected by the amount of RAM it has. Seems the RAM has more to do with noise reduction features only if I'm interpreting that correctly.

No, I don't believe you are. As I understand it, the RAM is shared. So if the buffer is max'd out with timing wiggle, and DOC, and NR settings, etc -- you're screwed. To me, finding the breaking points of gear is fun, and is why I have certain test tapes (because if anything plays it, I'm impressed).

Quote:

It is possible that the TBC was a completely separate chip

No.

Quote:

Also, I don't think that the memory is likely to affect dropout compensation. It is very rare for the actual DOC to be digital, even in professional VCRs, see here: https://www.digiommel.fi/images/Vide...log%20VCRs.pdf
Somewhat counterintuitively, the above paper says that dropouts *IMPROVE* with up to a 90% reduction in frequency after playing the tape back ten times.

I remember that paper from the 2000s (or at least the example images). But in the 2020s (even the 2010s), that paper is stupid. Each tape playback will continue to incur damage, and at a nasty curve. Tapes can go 100% unplayable in 1-3 plays, when the tape has excessive dropouts. That suggests tape deterioration.

Quote:

On another note, I don't think softening of the entire image uniformly is likely to be due to aging electronics, but it wouldn't be hard to narrow down that possibility of it being worn heads by swapping the decks in a machine that is soft vs one that is sharp of the same model and seeing if the softness follows the deck or not. The 9800 and the rest of the machines tested are all roughly the same age in the scheme of things, so it being much softer than all of the rest, so I still think this suggests a problem of some sort that they don't "usually" develop just with age (since the other machines tested aren't that way).

I agree ... I think.
Don't use "age" as the descriptor here. Deck usage hours is what matters. Age mostly affects items that alter with time, such as lube, motors, alignment, etc. Not head wear.

Quote:

Originally Posted by latreche34

DOC is pure digital memory, at least after mid 90's, I know it is for JVC's later models, so the size of the memory has an effect on how many lines can be replaced,
Whether there is more than one memory that I don't know, from a design perspective all digital processing are done in one step therefore one memory is all that is needed, This is why TBC and DNR cannot be done separately.
If you look at this screenshot you will notice on the TBC board on pin 25 it is labeled DOC:

This.

[aramkolt]

Interesting evidence with the DOC pin, though I'm not certain that DOC having a pin on a digital board necessarily means that the DOC itself is digital. It could be possible that the digital part triggers the analog DOC to kick in when it sees low RF on that pin. I agree that it would make sense that it would be digital though, but I don't think that explains the discoloration on replaced lines (to me it looks like a light purple or grey discoloration) like most/many machines will do.

If the DOC was digital and replayed from stored information in RAM, would you agree that there should be no line-discoloration for repeated lines as there would be no reason for those digital stored values to change with repeating indefinitely? A digital DOC would only need to hold maybe one or two lines in memory to draw from to repeat whatever missing sections from the last "good" line. Holding more lines in memory I don't think actually helps anything since it only repeats the last good line?

I suppose what could sort of explain the above is if only the luma information is stored for DOC purposes and color is just ignored when lines need replacing.

[latreche34]

You provided evidence in your own post that DOC is digital in later VCRs, I just posted a snippet from a service manual of a 1999 model I own to back it up. It's not really a big deal, we know that YT comparison is flawed anyway.