TBC still required for RF capture? (VHS-decode)

[latreche34]

Quote:

Originally Posted by Novgorod

From a technical point of view, how does tape degradation affect the signal? If it's something like spontaneous demagnetization (some grains flip their magnetic domain orientation), it results in an overall reduced magnetic contrast, akin to a reduced recording current. This shouldn't really affect the luma signal (other than overall SNR) because it's FM. Chroma, on the other hand, is QAM and the amplitude carries the saturation information. Wouldn't it be at least conceivable that a reduced SNR could affect this amplitude-modulated signal non-uniformly (i.e. frequency- or amplitude-dependent), which would manifest not just as noise but also as reduced saturation contrast in the image? I don't know how severely magnetic degradation affects tapes in the real world (as opposed to e.g. physical damage or mold) and it certainly depends on storage conditions, but I wouldn't straight-out discard the possiblity of loss of color definition as hallucinations...

There is no doubt that over time the wear from playback and poor storage condition affect the overall condition of the tape, However people think that tape fade or color fade is when you have a cherry red shirt in the video over time becomes light red or pink, This is impossible. To be able to change the chroma tones you would have to decode the RF signal, extract the chroma, up convert it and then change the chroma primary values, It's called procamp, does it ring a bell?

Chroma can suffer from the poor conditions stated above, in term of timing, If the chroma burst is weak the chroma timing is lost, delayed or advanced, Noise can affect chroma if the carrier frequency dropping out or weak. But it is usually a combination of chroma and luma problems not just one or the other.

[aramkolt]

Small fish in a big pond here, but I love head-to-head comparisons of different conversion methods.

The premise that modern software decoding that doesn't have to work in real time to decode a signal straight from the heads should theoretically produce the most accurate representation of the original signal. It is the absolute shortest "video chain" hence does not have to go through old components within the VCR that may not be performing like they used to when they were new 20 years ago.

I am relatively new to following VHS decode, but I have seen recent feature adds that would be useful to the end user - namely support for Apple Silicon which is relatively easy to get a hold of and is very energy efficient which may be much easier to install and use than previous iterations for the average user. Seems to be getting towards a one-app solution without a complicated install. I haven't tried it myself to know though, but I plan to soon. I applaud their efforts on this project and will absolutely shoot a donation their way if that ends up becoming my ultimate capture method.

Lord Smurf - Why not digitize a couple of your more difficult tapes that you don't mind losing (in a worst case scenario in cases something happens in shipping) via your best conversion method and mail them to those at VHS-Decode and each post YouTube videos separately of the conversion results?

There was some site that I've seen that loads and synchronizes two different YouTube videos in the same browser window, so you could watch them side by side even if they were uploaded by different users.

Personally, I don't think LS is against the idea of VHS decode, I just don't think there are a lot of good head-to-head comparisons out there and his method I am sure produces an excellent result, so his thought process is probably "why change when I've spent decades getting it right" and there's a good chance that he's never seen a process yielding a better result than what he's currently doing and he can't endorse it without that. Some of it may also come down to ease of use if the result is very similar. I also think a lot of the relatively few comparisons out there that they are not of the most modern version of VHS decode, hence, improvements since then are probably not well known.

I will say that a lot of the various hardware changes I've made in my chain with my relatively minimal testing were often hard to tell expensive equipment from consumer grade stuff as a lot of the quality comes down to the source material and the capture card which are the two things that we aren't typically varying. Those being equal, it's often hard to tell what the hardware does better than others, so I do think results of a certain workflow have to be "seen to be believed".

[lordsmurf]

Quote:

Originally Posted by aramkolt

The premise that modern software decoding that doesn't have to work in real time

Real-time vs. non-RT doesn't really matter as some think. It's about the algorithm quality, not the time required. Even the word "modern" doesn't mean much -- ie, the old phrase "don't make them like they used to" can be valid at times, or not.

Quote:

Lord Smurf - Why not digitize a couple of your more difficult tapes that you don't mind losing (in a worst case scenario in cases something happens in shipping) via your best conversion method and mail them to those at VHS-Decode

I've had some tapes in mind for years now. But I need to see better results. As mentioned earlier, an actual upgrade, not a sidegrade.

Quote:

Personally, I don't think LS is against the idea of VHS decode, I just don't think there are a lot of good head-to-head comparisons out there and his method I am sure produces an excellent result, so his thought process is probably "why change when I've spent decades getting it right"

No, that's not it. vhs-decode over-promises, and under-delivers.

The "official" wiki is also laden with misleading "facts" and outright nonsense, in an effort to make it seem better than it really is. There's zero reason to do that, other than ego from some project members.

Quote:

Some of it may also come down to ease of use if the result is very similar.

Yep. Most of what you see is both inferior + superior, so it's a wash. But the inferior makes it harder to watch, while the superior is buried. Many people don't seem to understand how VCRs and TBCs process video, erring on the side of not making the output unviewable, even if at the sacrifice of otherwise "better" aspects.

Quote:

I also think a lot of the relatively few comparisons out there that they are not of the most modern version of VHS decode, hence, improvements since then are probably not well known.

A huge portion is due to junk VCRs in use. Noting that even the capture cards used aren't great. The RF is still affected by the base components in use. RF is "off the heads", not "on the tape".

Quote:

I will say that a lot of the various hardware changes I've made in my chain with my relatively minimal testing were often hard to tell expensive equipment from consumer grade stuff as a lot of the quality comes down to the source material and the capture card which are the two things that we aren't typically varying. Those being equal, it's often hard to tell what the hardware does better than others, so I do think results of a certain workflow have to be "seen to be believed".

Several basic needs must be met:
- good VCR (actual good, not random ideas of "good")
- some form of TBC
- a capture card (and capture software) that doesn't butcher the video

The better the gear, the less quality that is lost. It's not really about "quality gains" once the minimum is met. Better gear also gives less usage issues.

Source matters, and in fact determines what minimum workflow gear spec you can use.

[Novgorod]

Quote:

Originally Posted by lordsmurf

Welcome.

Thanks for the detailed reply!
I'm pretty much an outsider (to both sides), but I'm intrigued by the academical value of the RF capture and software decode concept because it enables the same type of tinkering like SDR or how people have been able to extract the best possible picture from retro consoles (I know, it's not apples to apples, just an inspiration). Commercial viability or even practicality in the "real world" are secondary concerns to me; I value the research effort of reverse-engineering this old format with the potential of recovering the "unaltered" source image, just like others have a passion for hacking their old NES.

Quote:

You're somewhat mixing up concepts here: PAL vs. NTSC, write/storage vs. read, the quality and features of the players, etc.

The magnitude component of the chroma is the same for PAL and NTSC, so fading magnetic contrast should lead to fading (or at least dynamic range reduction) in the same way. Phase distortions (affecting hue) should manifest differently between PAL and NTSC for sure. I'm also not saying that high-end playback machines can't compensate for this to some extent (turn up the saturation, maybe even nonlinearly etc.), but that's essentially image post-processing. On a simple machine without that much post-processing I wouldn't find it unbelievable that magnetic degradation could lead to reduced saturation contrast (i.e. "faded colors") compared to the same tape in good condition.

Quote:

No ... at least not as typically presented ("use bestest ever vhs-decode, not the old poopy method").

I didn't even mention VHS-decode in my post .. I was only talking about making an RF capture as a digital image/backup of your tapes while they're still readable.

Quote:

capture 1st pass in JVC, 2nd pass in Panasonic, capture 3rd in whatever, 4th, etc. Combine best parts. That tedious process is often what is required for some projects, and it's not something 99% of people will need/want to do.

That sounds peculiarly similar to "it's a tedious process to tune your digital signal processing chain [in the software of your choice] and see what produces the best image quality for a given digitized RF signal."

Quote:

Something to keep in mind is that tapes may be "one and done", literally self-destructing on capture. So you get one shot. Don't waste it hobby tech that is unproven and known to be variable output. Those must be captured in the best known-quality workflow. I'd no more suggest current RF/vhs-decode than I would a thrift store VCR in this situation.

Fortunately for all of us it's not either-or - most players, in particular the professional ones, allow non-invasive RF tapping, so you can get more out of your precious museum-grade tape if you capture the RF simultaneously with the hardware-decoded S-video.

Quote:

The problem here is that it's somewhat BS. vhs-decode does not bypass as much as you're led to believe. Yes, it takes out some steps, that might potentially (maybe) cause signal degrade. But the main difference is simply native sharpness. Most other samples that show difference in dynamic range, etc, and comparisons using a junk/crap VCR with the RF method. It's not a true comparison.

Forget about the samples, I'm talking purely academically. If you capture exactly the same RF signal from the tape that the VCR "sees" and feeds into the long analog processing chain, you have by definition the most "lossless" version of what's on the tape (short of scanning the tape with a magnetic force microscope). Of course you have to make sure that the signal doesn't degrade on the way from the RF test point to your ADC, but that's relatively easy to control (BNC cable and a decent front-end for the ADC) and also applies for any analog processing chain. So RF capture does bypass all the processing, obviously, and I don't think there is any controversy about that. The only debate is about how well the decoding software can replace the analog processing, for which you have absolutely valid criticisms regarding VHS-decode (both in terms of usability and results), and that's fine. My point is that software is flexible, can be improved or reimplemented (by anyone) and the decode can be repeated as often as you want in order to reach perfection without ever touching the physical original. It's like forensic work on a hard drive image rather than on the fragile "real thing". Hardware tools, on the other hand, while being very mature thanks to enormous R&D budgets back in the day, are frozen in the way they perform and will only degrade over time, just like the tapes (I'm also talking "overall" here; good storage and maintenance prolongs their lives, of course).

Quote:

Not at all. The RF methods still need quality decks, not thrift store junk. So no cost savings there. The method ignores TBCs, to its own detriment, so I guess you call that a "savings". And then a capture card is still needed. Furthermore, it's not plug-and-play, and has a very steep learning curve, with a large time requirement. So that's not "more accessible" whatsoever.

I have to disagree here .. RF capture needs a quality deck up to the RF tap (i.e. transport, heads, tracking), it does not need any futher processing, conversion or convenience features at all. I guess it can be a hit or miss, but towards the end of the format the transport and pickup systems matured enough to become more or less commodity parts, and the only difference between player models dicatating the price were (largely) the features after the RF pickup. Ignoring all analog processing steps is the whole point of RF capture. The decoding software is the TBC (in addition to RF demodulator, filter, chroma decoder, RGB converter, scaler and anything else you can think of). And regarding steep learning curves, the particular software (VHS-decode) definitely has one for the moment, as does traditional high-end capture where you have to know the quirks of each hardware model. The hardware side of RF capture is very accessible though (unless you're scared to open up a case) - connect a cable from the RF test point to the capture card, that's all, oftentimes even without any soldering. And that's talking about inexperienced hobbyists - adding the RF capture workflow should be trivial for "enthusiasts" who already have a high-end traditional setup.

Quote:

No, not at all. Loss is about 35gb/hour, and RF captures are way larger sizes.

Depends on the resolution, sampling rate and actual bandwidth (which affects FLAC compressibility). In my test with Video8 RF files (higher bandwidth than VHS) I get 65GB/h for 20Msps 10bit, 43GB/h for 20Msps 8bit and 34GB/h for 16Msps 8bit, which is a sufficient sampling rate for VHS. RF captures also tend to be larger because they capture the full field, not just the image area. Instead of throwing around "Nyquist" I'll just subtly remind that "vertical" resolution (bitness) is as important and the required bitness is dictated by the SNR (like the sampling rate is mandated by the bandwidth thanks to Nyquist). It's not clear-cut what's the required bitness and sampling rate to "fully" capture the RF signal as it is on the tape (FM has different requiremens than QAM), so it's up to the individual to either skimp on drive space and use just enough resolution to capture the physical dynamic range (given by the peak-to-noise ratio) and bandwidth of the RF, or overdo it and waste space. There is no one-fits-all recipe yet, that will come with experience. It's at least in the same ballpark as lossless video though.

Quote:

The issue is impatience. Most of us have stated "DO NOT THROW AWAY YOUR TAPES!" fro years now, decades. Someday, the RF method may mature to the point of reliable usability.

I don't believe even the biggest VHS-decode fan would advise anyone to throw away their tapes once they are captured.

Quote:

Yes, there is risk with tape loss/failure, but the lifespan of VHS tape is 35-65 years. Some go quicker (rarely, mostly junk grade off-brand tapes), some may last longer than 65 (likely). There should not be a rush, but simply mindfulness of impending timelines. You can probably wait 5-10 more years for vhs-decode to mature, without many (or any) tapes being lost.

Even if you can prevent the tapes from degrading with proper storage, vintage players in good condition will only be harder to come by with every passing year. It seems like you're equating RF capture with VHS-decode - one is a format, the other is one particular software approach to the format. Just as you don't need the blessings of JVC to homebrew your own VHS player, you don't have to wait for VHS-decode to work to your expectations in order to RF-image your tapes, especially when it costs (almost) nothing compared to a traditional setup apart from drive space.

Quote:

That's a problem. Software, especially open-source, has a way of just .... ending. Not with a bang, or even a whimper. Just poof, gone. Most here can readily think of software that did this, including video, especially something scripted like Avisynth plugins.

Again, a software is not the format. If one project dies, others can fork it or make a new one from scratch. Traditional capture didn't disappear just because one particular capture tool stopped updating. The worst that could happen is that you can free up your hard drives if you believe that there will never be an adequate decoding tool in your lifetime. As I said, it's a low-cost speculative investment (of drive space and a bit of capture effort). Considering the time and money hardcore enthusiasts invest in potentially getting a slightly better image, RF-imaging as an additional insurance policy seems like a no-brainer.

Quote:

That's a false equivalency there. Not wanting to use RF doesn't mean we want $5 Chinese junk. Discussions here should compare a quality setup to RF, not junk to RF.

What do you mean by "we"? I never said "we" or "you" because none of this is "mainstream", not RF, not a fancy TBC setup. And of course a "quality setup" must be the reference to tune the digital signal processing chain using the RF.

Quote:

I've long been intrigued by the project, but I'm vehemently anti-myth and anti-BS.

Totally agree. But the project is the format, the software is just one part of it (even if an important one) and doesn't deserve the drama. Even if some "over-optimistic" claims have been made, it takes nothing away from the concept which is academically sound and might turn out to be a valuable tool for no-compromise high-end video preservation with an accessible way to RF-backup tapes already today. The "capture now decode later" approach is very reasonable and also applies to other conservation fields where some artefact is "stabilized" until proper restoration is possible. The speculative nature of it (delayed results) might make it not worth it for some, which is totally fine - I don't mind it to be a niche.

Quote:

Originally Posted by latreche34

There is no doubt that over time the wear from playback and poor storage condition affect the overall condition of the tape, However people think that tape fade or color fade is when you have a cherry red shirt in the video over time becomes light red or pink, This is impossible. To be able to change the chroma tones you would have to decode the RF signal, extract the chroma, up convert it and then change the chroma primary values, It's called procamp, does it ring a bell?

Chroma can suffer from the poor conditions stated above, in term of timing, If the chroma burst is weak the chroma timing is lost, delayed or advanced, Noise can affect chroma if the carrier frequency dropping out or weak. But it is usually a combination of chroma and luma problems not just one or the other.

There are many things that can happen to the chroma, and I didn't go into the phase stuff (affecting hue). The entire signal on the tape is quite complex and noise or loss of magnetic contrast can have complicated interactions with all the signals in different frequency bands. Still, all the signal components (luma, chroma, FM audio in the case of V8/Hi8 and whatever else) are simply heterodyned together, and the saturation component of the chroma (U or V, I always get them wrong) is an AM signal which is significantly more susceptible to amplitude degradation (such as loss of SNR) than FM (luma) or PM (hue) signals, though there are additional parasitic effects like amplitude-to-phase coupling which would also affect hue...

[lordsmurf]

Quote:

Originally Posted by Novgorod

That sounds peculiarly similar to "it's a tedious process to tune your digital signal processing chain [in the software of your choice] and see what produces the best image quality for a given digitized RF signal."

This stuff can get as tedious as you want, with very diminished returns.

Quote:

allow non-invasive RF tapping, so you can get more out of your precious museum-grade tape if you capture the RF simultaneously with the hardware-decoded S-video.

Perhaps. The problem here is that a lot of the "pro" decks work quite poorly with consumer-made VHS tapes. The JVC series decks from about 95-05 have been proven all-around best quality and playback, with a deck that tends to hold up better than most Panasonics. Since the current RF methods (ie, vhs-decode) seem to work poorly with JVCs, it's a problem.

Quote:

Forget about the samples, I'm talking purely academically.

I've long stated that RF capture is a sound theory. The problem is the "in practice" isn't there. It may someday come, or may never come. That's the issue.

Quote:

My point is that software is flexible, can be improved

But noting it's not guaranteed, especially with open-source. Zero financial incentives has consequences. Even for-pay software eventually quits working, but tends to receive more updates and patches, especially for new OS/hardware.

Quote:

Hardware tools, on the other hand, while being very mature thanks to enormous R&D budgets back in the day, are frozen in the way they perform and will only degrade over time,

I know for a fact that's not true. (I'm not able to say more on the matter.) It can be, but certainly does not have to be.

Quote:

I guess it can be a hit or miss,

That's my only point. It's not as perfect as some want to insist. The samples have repeatedly shown that to be true.

The RF capture method reminds me of trying to play emulators on a 486. It's nifty and fun, but hit or miss. And that fact has never changed in the past 25-30 years. Newer hardware "runs" newer games, better, faster. But it still has lots of errors. Nobody (with any sense) will ever claim emulators work better than the original hardware. Some is better, some is worse, and it's mostly just different. Definitely cheaper, sometimes. Unlike video, video games never elicit the same sort of whining about money. It's a hobby, all hobbies have costs.

Quote:

There is no one-fits-all recipe yet, that will come with experience.

Again, that's the point. Certain people are touting it as more than it is, refusing to acknowledge imperfections, issues, errors, failings, etc. Then they further make false claims about workflow hardware, especially TBCs. Then get angry and defensive when it's called out. That's not helpful to the end users, but merely to stroke egos.

Quote:

I don't believe even the biggest VHS-decode fan would advise anyone to throw away their tapes once they are captured.

And yet, it is happening.

Quote:

It seems like you're equating RF capture with VHS-decode - one is a format, the other is one particular software approach to the format.

No, not at all. I've stated several times that RF is sound, vhs-decode has issues, mostly with certain devs.

Quote:

Again, a software is not the format. If one project dies, others can fork

RF is not really a "format", but an extraction method. And longer term, I think it has to fork to improve.

Quote:

What do you mean by "we"?

Visitors to this site, who come here looking for solutions to achieving quality video conversions.

Quote:

Even if some "over-optimistic" claims have been made, it takes nothing away from the concept which is academically sound and might turn out to be a valuable tool for no-compromise high-end video preservation with an accessible way to RF-backup tapes already today.

The problem is that "today" is not accurate. The output results are all over the place. Again, to many RF devs/fans are ready to proclaim "mission accomplished" (a Bush reference, for the younger folks reading).

Quote:

The "capture now decode later" approach is very reasonable

I doubt most would agree. It's not a NASA trip, where it takes decades to finally get where you want. That sort of patience is not extended to video conversions, and I think it's unreasonable to expect others to agree.

Quote:

and also applies to other conservation fields where some artefact is "stabilized" until proper restoration is possible. The speculative nature of it (delayed results) might make it not worth it for some, which is totally fine - I don't mind it to be a niche.

I think the ideal use is as a secondary (or later) attempt to recover footage, but also understanding you may just get scenes. Or nothing. Maybe a whole something, but unlikely. Regardless, lots of time is required.

Anyway...

I think this has been an informative thread for all involved. But I'm starting repeat myself now, and probably have nothing more to add here for a while. Busy anyhow, video projects to do, not just talk about it!

Overall, I think many people allured by the RF (or vhs-decode) have good intentions: the best video quality. A few egos and a few cheapskates shouldn't detract from it -- just be aware that they exist, and have bias that may not align with seeking quality.

[Novgorod]

I think we're getting to the core of the issue, so I'll try to keep it brief ..

Quote:

Originally Posted by lordsmurf

The problem is that "today" is not accurate. The output results are all over the place. Again, to many RF devs/fans are ready to proclaim "missions accomplished" (a Bush reference, for the younger folks reading).

But what's the mission? If it's getting a better image with software RF decoding under all circumstances for all formats, it's a long way to go. If it's a lossless digital preservation of the tape signal as seen through a VCR read head, that's absolutely doable today. It's not dummy-proof, it needs some (but not all) of the same experience as high-end traditional capture, and it's not cheap (decent ADC and storage) if you don't want to "risk" it (I don't really trust the cheap 8bit cards to capture the full dynamic range of the signal). When (or even if ever) the decoding software will surpass the quality of analog processing is a completely separate issue - that's the speculative part of the digital RF format (it's definitely a format because it's agnostic of the method how you extracted the signal from the tape or whether it even came from a tape, see LD, but I'm merely being academical here again).

Quote:

I doubt most would agree. It's not a NASA trip, where it takes decades to finally get where you want. That sort of patience is not extended to video conversions, and I think it's unreasonable to expect others to agree.
[...]
I think the ideal use is as a secondary (or later) attempt to recover footage, but also understanding you may just get scenes. Or nothing. Maybe a whole something, but unlikely. Regardless, lots of time is required.

I laid out the effort required for digital tape imaging - it's "simple" enough to implement for any high-end traditional capture enthusiast as an additional archiving channel, not as a replacement for traditional capture (of course!), so indeed it's not a NASA trip. Noone needs to agree, it's obviously a personal preference to go the extra mile. But just as claims like "software decoding is solved" are very misleading, so are recommendations like "give them another 10 years with the software, only then start thinking about RF capture". Take data recovery from an important flash drive as an analogy: Would you rather get a raw flash image while you can, even if you don't have the tools yet to reverse-engineer the sector mapping in order to reconstruct the data, or put the broken drive in a drawer until maybe one day the recovery software can process it perfectly and only then attempt the physical transfer? That's how I see the role of the digital RF format - lots of time required indeed and somewhat based on hope. Nonetheless, the best time is now, while tapes can still be read and working players can be found outside museums.

Side note:

Quote:

The RF capture method reminds me of trying to play emulators on a 486. It's nifty and fun, but hit or miss. And that fact has never changed in the past 25-30 years. Newer hardware "runs" newer games, better, faster. But it still has lots of errors. Nobody (with any sense) will ever claim emulators work better than the original hardware. Some is better, some is worse, and it's mostly just different. Definitely cheaper, sometimes. Unlike video, video games never elicit the same sort of whining about money. It's a hobby, all hobbies have costs.

Funny you mentioned it - the recent emergence of FPGA emulators is pretty insane because they can simulate the original hardware exactly and in real time. I'm not comparing anything here (the sheer fact of digital vs. analog), it's just a neat example of how far hardware abstraction/emulation in software can come if there's enough general interest and R&D money for a certain niche...

[latreche34]

I think the true core problem is RF pickup in itself is neither finalized nor can it be trusted, How do I know I got exactly what's been read by the video heads? What if there is a miss representation of analog signal being digitized, How do I verify the data? What if there is bits of dust came across the heads and caused short data loss? Your analogy with a USB drive is not correct, because I can verify the integrity of data in a digital storage medium but I can't verify an analog signal sampled into digital bits. So unless there is a reliable and robust decoding scheme to go along the RF pickup to verify that what I got is all there, there is no way of trusting an RF backup with no verification, Therefore "capture RF today and encode in the future" is flat out wrong.

In a conventional capture, even if you don't monitor the capturing process you can still check the captured files and if a recapture is needed you would know right away, not 10 years later.

[lordsmurf]

Quote:

Originally Posted by Novgorod

so I'll try to keep it brief ..

I'm afraid to see the long version. (I can be the same way, brief intention fails by longer end.)

Quote:

It's not dummy-proof, it needs some (but not all) of the same experience as high-end traditional capture, and it's not cheap (decent ADC and storage)

That's it. Not better, not cheaper, not easier -- just different.

Some different good, some different bad. The "bad" is in terms of time and results. If you have unlimited time, then you may enjoy the alternative manual tinker approach, and can squeeze out that last % that truly exists on the tape beyond what standard quality workflows do. I think it's very warranted in some instances, not at all in most others.

For many, simply getting them into a quality workflow (ie, resisting the urge to buy crap from Amazon and Goodwill) takes some convincing (for their own good). Trying to expand their timeline for a % gain is impossible.

Quote:

capture enthusiast as an additional archiving channel, not as a replacement for traditional capture (of course!),

I fully agree.

Quote:

"give them another 10 years with the software, only then start thinking about RF capture".

In many ways, it reminds me of RISC (eventually ARM), Tesla/EVs (EV still in infancy; power grids will not sustain it easily for many yeasr), etc. And those are just the successful ones. Early results, then some folks makes claims about "the future". But the actual future is different. Enjoy early successes, the early adopters in a tiny niche, many of them too vocal. But then you have massive headwinds for actualization in any meaningful way, and it just takes time to get there. Patience required.

Quote:

Take data recovery from an important flash drive as an analogy: Would you rather get a raw flash image while you can, even if you don't have the tools yet to reverse-engineer the sector mapping in order to reconstruct the data, or put the broken drive in a drawer until maybe one day the recovery software can process it perfectly and only then attempt the physical transfer?

None of the above. I'd use/pay Ontrack. (again!)

Quote:

That's how I see the role of the digital RF format - lots of time required indeed and somewhat based on hope. Nonetheless, the best time is now, while tapes can still be read and working players can be found outside museums.

I think most users and would-be users would wholly agree with this, as do I.

It's just a vocal few that make claims that are not realistic. But these days, that's almost to be expected, with anything. The rest of us just need to tell them to simmer down a bit, get facts straight, don't embellish, don't BS. And that's really been the mantra of this thread.

... but also noting "before it's too late!" is not a reason to rush, it's just something to be mindful of. We have time still, probably multiple decades. I'm fairly certain whatever RF is done now will be done again (and better) later.

That reminds me of scanning photos in the 90s, or even digitizing videos to DVD in 00s (and later again as lossless). Now is fine, but later you realize that "now" was actually pretty crappy. But we do it, and redo it, because most people want quality. But eventually "quality" is diminishing returns, and more time/money investment loses charm. I think wider RF adopter for VHS will meet resistance because of it.

I know I'm not re-scanning my photos for a 3rd time, regardless of how much better dynamic range or res/interpolation gets on flatbed scanning. Maybe a few special photos for framing, but no more. I'm definitely not creating a DSLR rig to shoot the prints, which some do as well (and many are vocal for and against). I think DSLR scanning is a decent analog to RF VHS capture, as it's both better and worse at the same time, takes lots of time and tinkering, and has hidden costs ("just use a camera" instead of a buying a scanner; but oops, buying a quality camera has costs).

Final note...
Too much emphasis is on VHS, which is fairly trivial to get quality results with quality gear. What I want to hear about is all the other formats, like Betamax, where available hardware is truly gone or garbage. That's what RF devs should focus on first, if "before it's too late" is really a concern. Because for some rare formats, we're well into the "too late" phase.

There's even some tape-based digital formats where players are rarer. And the ancient software to transfers/"capture" is difficult to find, install, or use.

Quote:

Funny you mentioned it - the recent emergence of FPGA emulators is pretty insane because they can simulate the original hardware exactly and in real time. I'm not comparing anything here (the sheer fact of digital vs. analog), it's just a neat example of how far hardware abstraction/emulation in software can come if there's enough general interest and R&D money for a certain niche...

So ... dedicated hardware required. Sound familiar?

And then gaming emulation is about 25-30 years old now. Hmmm, that sounds familiar too!

I've not kept up with emulation for several years now. I was considering a Pi for retro gaming around 2018, but wanted to hold out for a Pi 4, due to too many issues with games I had interest in playing on it. That never happened, mostly due to the chip shortage. I don't really have spare time to play games anyway.

Quote:

Originally Posted by latreche34

I think the true core problem is RF pickup in itself is neither finalized nor can it be trusted, How do I know I got exactly what's been read by the video heads? What if there is a miss representation of analog signal being digitized, How do I verify the data? What if there is bits of dust came across the heads and caused short data loss? Your analogy with a USB drive is not correct, because I can verify the integrity of data in a digital storage medium but I can't verify an analog signal sampled into digital bits. So unless there is a reliable and robust decoding scheme to go along the RF pickup to verify that what I got is all there, there is no way of trusting an RF backup with no verification, Therefore "capture RF today and encode in the future" is flat out wrong.

In a conventional capture, even if you don't monitor the capturing process you can still check the captured files and if a recapture is needed you would know right away, not 10 years later.

All true.

Quote:

So unless there is a reliable and robust decoding scheme to go along the RF pickup to verify that what I got is all there,

Yep, back to more hardware needed. You can "easily" have a 2nd pickup downpath that verifies the 1st. Of course, it has to be done in clean conditions, otherwise dust is likely. This has analogy to HDD recovery, open platters = dust magnet, tapes are no better.

[Novgorod]

Quote:

Originally Posted by latreche34

I think the true core problem is RF pickup in itself is neither finalized nor can it be trusted, How do I know I got exactly what's been read by the video heads? What if there is a miss representation of analog signal being digitized, How do I verify the data? What if there is bits of dust came across the heads and caused short data loss? Your analogy with a USB drive is not correct, because I can verify the integrity of data in a digital storage medium but I can't verify an analog signal sampled into digital bits. So unless there is a reliable and robust decoding scheme to go along the RF pickup to verify that what I got is all there, there is no way of trusting an RF backup with no verification, Therefore "capture RF today and encode in the future" is flat out wrong.

In a conventional capture, even if you don't monitor the capturing process you can still check the captured files and if a recapture is needed you would know right away, not 10 years later.

That's absolutely a valid concern, that's why I called it a speculation to do it (somewhat!) "blind". In order to perfectly verify your RF captures, you need to perfectly decode them. If that would be available today (emphasis on "perfectly"), we wouldn't have this discussion. Still, there are basic rules and approaches for digitizing signals in general to ensure you're not altering the RF signal on the way to the ADC. I already pointed out how to assess the resolution requirements. Pushing a few-MHz signal through a shielded BNC cable to an impedance-matched input on the ADC is not that much of a challenge nowadays and it's been done in consumer electronics even with cheap RCA cables for decades. There are ways to check the RF quality without actually decoding the image, just by looking at spectrograms and noise figures. Interference is easy to spot as are drop-outs, and of course you can monitor the playback with a hardware decoder. Apart from the lack of a perfect verification method (yet), it's not harder to get right than digitizing traditional analog S-video on a PC. A good ADC (good SNR + linearity) with a good frontend is crucial, but keep in mind that this applies just as well to traditional capture and the TV capture PC-cards don't always have the best ADCs in them and don't offer much control over the S-video-to-RGB/YUV decoding and quantization process.

The analogy with the flash recovery is specifically that you can't verify the imaged data until some decoding software is developed which is able to reconstruct it. But you can do the imaging to the best of your ability with sound methods. So would you still do the imaging without having the software (yet) to extract and verify the data? It's speculative and therefore everyone should make up their own cost-benefit analysis, but it's definitely not wrong.

Quote:

Originally Posted by lordsmurf

... but also noting "before it's too late!" is not a reason to rush, it's just something to be mindful of. We have time still, probably multiple decades. I'm fairly certain whatever RF is done now will be done again (and better) later.

Yes, mindful, that's what I'm trying to convey. But while being mindful, also consider one's own individual situation in judging how much time is actually left, because not everyone has ideal storage conditions or a museum full of old parts for player maintenance (speaking of the "real world" again). When/if the perfect RF decoder software comes out, your RF captures from today will work with it just fine (even better) because it's necessarily the same format. There's no need to re-capture if a better software comes out, that's why I was stressing so much that RF capture and decoding are very separate things. The only scenario where re-capture would yield a better quality is if somehow the capture hardware improves significantly, but in terms of ADCs there are already very diminishing returns in increasing the resolution (considering the limiting signal quality on the tape). A better magnetic pickup would have the potential to improve the results (for both RF and traditional capture) but I doubt a modern transport/head system will be ever developed because it's far outside the scope of hobbyists; might as well send the tapes to a magnetic force microscopy lab and lose all grasp on reality .. So I think my favoring the "two birds one stone" approach (traditional + RF in one go) is not unreasonable, especially if you want to avoid re-capture in another 10 years...

Quote:

That reminds me of scanning photos in the 90s, or even digitizing videos to DVD in 00s (and later again as lossless). Now is fine, but later you realize that "now" was actually pretty crappy. But we do it, and redo it, because most people want quality. But eventually "quality" is diminishing returns, and more time/money investment loses charm. I think wider RF adopter for VHS will meet resistance because of it.

I know I'm not re-scanning my photos for a 3rd time, regardless of how much better dynamic range or res/interpolation gets on flatbed scanning. Maybe a few special photos for framing, but no more. I'm definitely not creating a DSLR rig to shoot the prints, which some do as well (and many are vocal for and against). I think DSLR scanning is a decent analog to RF VHS capture, as it's both better and worse at the same time, takes lots of time and tinkering, and has hidden costs ("just use a camera" instead of a buying a scanner; but oops, buying a quality camera has costs).

Yeah, I remember a time when I was young and perfectly fine with audio tapes to MP3 .. But joking aside, I wouldn't compare scanning photos with RF capture but rather with traditional capture (analog in, digital picture out). Without getting too philosophical, the "analog signal" of a physical photo is tremendously more complex than the magnetic signal on the tape. You'd need to capture the visible light spectrum of every film grain to get a "lossless" digital image in terms of the analog information on the photo. Of course that's impractical and you have to resort to some reasonable compromise of approximating the photo with an image sensor. The RF signal on a tape, on the other hand, is already a reasonable approximation (for its time) of a physical image with all the required limitations in bandwidth, and can therefore be captured with a modern digitizer "exactly" (or rather with negligible loss) as it was recorded. Again, that's academic (the dreaded magnetic force microscope) but nonetheless a fundamental difference between physical photos and analog signals generated by electric circuits. If there won't ever be a new generation of "modern and improved" read heads, then the RF signal from a current player is the closest we can ever get to the original recording.

Quote:

What I want to hear about is all the other formats, like Betamax, where available hardware is truly gone or garbage. That's what RF devs should focus on first, if "before it's too late" is really a concern. Because for some rare formats, we're well into the "too late" phase.

Personally I'm in it (almost) exclusively for V8 and Hi8 home video tapes which contain the most valuable personal memories but on the other hand aren't of much interest for "corporate" level investment in archiving R&D. The post-2000s Sony camcorders do a reasonably good playback job with integrated TBC and all but you have to live with the baked-in image post-processing and they cut a bit of the image on the left edge and at the bottom (around the head switch), which can both be recovered with RF decoding. So there has to be some incentive (personal investment or otherwise) to get involved with a specific tape format, that's why open-source progress is slow in a niche. But that's even more reason to RF-capture the more obscure formats "before it's too late", especially if decent-quality traditional decoding has been lost to time - with RF there's at least a possibility...

Quote:

So ... dedicated hardware required. Sound familiar?

FPGAs are more software than hardware. The code can run on a PC with the same results, just not in realtime. That's why the only commercial RF capture solution I'm aware of implements the signal processing on an FPGA. For hobbyists the speed penalty of doing RF decoding without hardware acceleration is definitely worth it; but for a game emulator sub-realtime is kind of a deal breaker ..

[lordsmurf]

Quote:

Originally Posted by Novgorod

that's academic
..

I'm just glad this conversation has become sensible, honest, friendly. That's all I ever wanted.

I think most users will be this way -- though some may take longer to reach that conclusion, due to being a bit blinded by overoptimism. Like a kid getting a new puppy -- then learning he has to feed it, walk it, scoop poop, etc. It takes work, more than you realize, and will have both elation and disappointments. It just is what it is, no more, no less.

And on that zen note, g'day.

[Shakedown St.]

Quote:

Originally Posted by Novgorod

All analog processing is permanently baked into the capture, so wouldn't it be prudent to bypass as much of it as possible by digitizing the signal directly from the tape and not after the analog processing?

That was the opposite of my understanding. What worries me is people relying on software instead of hardware to get the job done correctly. I don't see any circumstance where I wouldn't want the best VCR for the job, rather than creating more work for me post process. I want the cleanest signal out of my VCR as possible. This includes using DNR, built in line/field TBC, proc controls, dynamic drums etc.

You don't need to spend money on expensive equipment because insert VHS decode. It does "work" and will continue to improve, but I disagree with the workflow. It's simply my preference and not knocking it for those who use it and appreciate what it does.

Quote:

Originally Posted by Infrid

They will break, not now, maybe not in 10–20 years, but one day they will become so fragile to handle like nitrogen films. Let's not glamorise the analogue format, the VHS is made of plastic, electronics of reproducing them can fail. From e leaky capacitor to a breaking down pinch roller that will ruin the tape.

In 10-20 years many VHS tapes will be too far gone to be archived. We just need the current SVHS decks and TBCs to last another decade or two. We are at the final stretch of this hobby. As Smurf alluded to before, we don't have time to wait for software to catch up. The best results RIGHT NOW are hardware.

I don't want to sound discouraging. It looks like a very interesting project with a lot of dedicated people. I would just disagree with those who say it replaces TBCs and SVHS decks. It's much easier for me to plug in my TBC, press play on my VCR, and let my hardware do all the work. No coding skills required. That's what you work in a transfer house. Consistent results.

I enjoyed reading through the debate and have a lot to learn. I will keep an open mind.

[Novgorod]

Quote:

Originally Posted by Shakedown St.

That was the opposite of my understanding.

That's a misconception and the only reason to bother with RF in the first place. Both methods (traditional and RF capture) start with an identical signal, which is the RF picked up by the read head. For software decoding, it's digitized right there with high enough resolution to have negligible loss (which is not trivial but relatively approachable, just as digitizing the processed S-video or component signal in a traditional workflow). At this stage nothing is baked into the signal yet other than the magnetic pickup characteristics of the VCR, which is the same for both methods. All necessary further processing (demodulation, PAL/NTSC decoding, TBC etc.), whether done in hardware or in software, will introduce loss and bake its result into the final image.

The difference with software decoding is that you can go back to the digitized RF "original" whenever you want and as often as you want, even long after the tapes and hardware players have been lost to time, and improve your decoded picture with the latest available (digital) signal processing technology. In contrast, with a traditional setup you are stuck with the way the hardware decoded and processed the picture that time and the settings you used. You can't "re-tune" anything in the signal processing chain retroactively (e.g. change the TBC, insert another filter etc.) apart from post-processing the final image, which you can also do after software decoding, of course. An analogy (to some extent at least) would be the "raw" image format that digital cameras use vs. the processed/decoded RGB image.

Quote:

What worries me is people relying on software instead of hardware to get the job done correctly. I don't see any circumstance where I wouldn't want the best VCR for the job, rather than creating more work for me post process. I want the cleanest signal out of my VCR as possible. This includes using DNR, built in line/field TBC, proc controls, dynamic drums etc.

Agreed, a good VCR is mandatory for both methods. However, for RF capture it "only" needs to be good up to the RF test point and doesn't require any particular image processing features. But as I said many times, RF capture should not replace traditional capture (because the reliability of the available decoding software is not there yet) but supplement it to preserve the RF original indefinitely and potentially get a better picture out of it in the future due to the flexibility of software decoding. Theoretically you could even feed the digitized RF signal back into your VCR and do your analog processing without a physical tape. My point is, whoever takes tape preservation seriously already has a (more or less) decent traditional setup, so adding RF capture to do both simultaneously might be worth the (manageable) additional effort. Of course everyone has to make an informed judgment whether the decoding software will ever "catch up" (or make their own), but you don't have to wait for the software to mature in order to start RF capturing.

[latreche34]

For this I have to agree with Novgorod, VCR's that have good solid picture are the ones from the mid to late 90's that use Soc (system on the chip) like the JVC, Panasonic and Sony ones, They digitize the RF signal using an ADC, process it (DNR,DOC..), time it and store it as lines or fields in a memory buffer, and convert it back to analog using a DAC, So the near lossless digitizing concept is not a bad idea, So that wasn't the issue we are discussing here.

[Novgorod]

Quote:

Originally Posted by latreche34

For this I have to agree with Novgorod, VCR's that have good solid picture are the ones from the mid to late 90's that use Soc (system on the chip) like the JVC, Panasonic and Sony ones, They digitize the RF signal using an ADC, process it (DNR,DOC..), time it and store it as lines or fields in a memory buffer, and convert it back to analog using a DAC, So the near lossless digitizing concept is not a bad idea, So that wasn't the issue we are discussing here.

Are you sure they digitize the RF directly? I thought the processing is analog at least until after the demodulation step and probably also after de-emphasis (all inside a chip but still analog) and only the demodulated proper PAL/NTSC Y/C signal is digitized and processed with the internal TBC and all the image filtering stuff, like e.g. the later Sony camcorders did.

[latreche34]

To be honest I'm not sure what's going on in their chips, But I know they are more digital than their counterparts from the 80's. You can't tell from their schematic either, because the RF pickup is going into the main chip.

[Shakedown St.]

Quote:

Originally Posted by Novgorod

At this stage nothing is baked into the signal yet other than the magnetic pickup characteristics of the VCR, which is the same for both methods. All necessary further processing (demodulation, PAL/NTSC decoding, TBC etc.), whether done in hardware or in software, will introduce loss and bake its result into the final image.

Simple solution then

Get a VCR with Dynamorphous heads that has such good tracking, line TBC is not required.

Buy a DVHS deck with built in SD MPEG encoders that accepts analog inputs, and buy some blank DVHS tape. Hit the record button and watch your analog signal be recorded onto digital tape without using a frame TBC. Now output to HDMI or firewire for capturing.

All joking aside I wonder if this works. My BV10 TBC is transparent and happy with it.

[latreche34]

Quote:

Originally Posted by Shakedown St.

Buy a DVHS deck with built in SD MPEG encoders that accepts analog inputs, and buy some blank DVHS tape. Hit the record button and watch your analog signal be recorded onto digital tape without using a frame TBC. Now output to HDMI or firewire for capturing.

I did that already, sort of. Nah the old school way is way better, Albeit I did not use the same VCR for both samples.

[Novgorod]

Quote:

Originally Posted by Shakedown St.

Simple solution then

Get a VCR with Dynamorphous heads that has such good tracking, line TBC is not required.

Buy a DVHS deck with built in SD MPEG encoders that accepts analog inputs, and buy some blank DVHS tape. Hit the record button and watch your analog signal be recorded onto digital tape without using a frame TBC. Now output to HDMI or firewire for capturing.

All joking aside I wonder if this works. My BV10 TBC is transparent and happy with it.

You mean digitize the raw RF signal with a DVHS recorder? I like the outside-the-box thinking but it will result in a big mess (or, more realistically, in an empty recording) because the DVHS recorder expects a PAL/NTSC compliant input and would probably just reject anything else. Even if you could somehow override everything and trick the recorder to use some internal clock source unrelated to your input signal, the MPEG encoding would kill it. RF capture has to be lossless (apart from quantization), kind of like PCM audio. Lossy compression which is optimized for raw RF doesn't exist and probably never will because it's too complex and too niche. So maybe some SDI capture hardware might be bastardized for RF capture, but I think at the moment nothing is economically competitive with a PC based ADC and a bunch of hard drives or data tape if you have a huge library...

[lordsmurf]

Quote:

Originally Posted by Novgorod

the MPEG encoding would kill it.

Nope. Not if it uses broadcast profiles, like 4:2:2 @ 50+ mbps. That's always been a holy grail of capturing, nothing does it, aside from appliances that have added downsides.

[Novgorod]

Quote:

Originally Posted by lordsmurf

Nope. Not if it uses broadcast profiles, like 4:2:2 @ 50+ mbps. That's always been a holy grail of capturing, nothing does it, aside from appliances that have added downsides.

I agree that it's plenty sufficient for SD RGB (or YUV) video, but I meant trying to put a raw RF waveform through any type of (lossy) image or video compressor. Technically any waveform signal can be interpreted as a grayscale "image" but the high-frequency RF signal will behave very poorly with lossy compression that is optimized for "human vision". That said, it may be worth exploring lossy RF compression with a modern audio codec like opus or aac with a custom quantization table optimized for the spectral information distribution in the RF signal. This should significantly reduce the file size compared to FLAC with a negligible quality loss. But I think that's not of the highest priority for the RF capture/decode project at the moment ...