Current status of ld-decode / vhs-decode (true "backup" of RF signals)

[lollo]

To Titan_91: if you edit your posts (#987, to which I replied before your edit) entirely changing the content, nobody can follow

[Titan_91]

To Titan_91: if you edit your posts (#987, to which I replied before your edit) entirely changing the content, nobody can follow

Yeah sorry about that, I was clarifying what I meant at the same time you had replied. I did mean using the hardware you had mentioned.

[lollo]

I did mean using the hardware you had mentioned.

The hardware I mentioned cannot do what you described, that's why I love vhs decode or better its evolution for future usage.

But what is really important is the final result, without that arsenal of weapons.

What about the excellent idea of dellsam34 to perform on our side the capture of your same tape and compare? I can pay for the shipping if needed.

[Titan_91]

I did mean using the hardware you had mentioned.

The hardware I mentioned cannot do what you described, that's why I love vhs decode or better its evolution for future usage.

But what is really important is the final result, without that arsenal of weapons.

What about the excellent idea of dellsam34 to perform on our side the capture of your same tape and compare? I can pay for the shipping if needed.

That is true. Those adjustments do make a difference in the final result though, but yeah depending on what tools and adjustments you're referring to. We do have to use a base value for the filtering and processing, and we have been trying to match the VHS spec as close as possible. Let's see if we can both find a copy of a retail tape and compare. Anything in particular you want to use? We have also been testing Disney movies. We have a capture of the opening scene of Spider-Man, I have one of Tom & Jerry The Magic Ring, there's several we could try. We don't want to pick the title, to keep it fair.

[lollo]

Live action is better for start, I suppose. Just send me a good tape you already captured with vhs_decode (I will pay the shipping). I will capture it with my existing workflow and send the tape to dellsam34 at my expenses for him to capture with his JVC-7600 and Brighteye 75.

[Titan_91]

I also want to clarify my posts aren't intended to be misleading. It's often hard to tell where difference in quality comes from, whether it's the tape or the VCR's processing, or processing done by the capture hardware or software. I'm being genuine here, and if it seems my screenshots are unfair then thanks for pointing that out. No foul play intended. Just interested in getting the best results.

[lollo]

I'm being genuine here

Yes, is clear.

What we really want to do is to compare the best existing workflows with the actual outcomes of vhs_decode. Knowing that vhs decode concept is sound, so once you have the raw RF capture there is room for future improvements if and when the post-processing will evolve

[Titan_91]

I don't know if we have anyone that can ship to Italy, I can ask.

[Titan_91]

Is dellsam in the US?

[lollo]

Yes

[Titan_91]

Just did a search for Spider-Man (2002) on eBay. That one is pretty plentiful and available NOS still in shrink wrap. Instead of running the same tape through a VCR to get an RF capture, then shipping it to him via USPS sorting and magnetic scanning machines, wouldn't it make more sense to start with brand new cassettes? To be fair to you guys? Also a lot easier logistically. Everyone can just order brand new tapes to start fresh.

[Bogilein]

I like the idea of forum members working together to make an actual comparison with vhs-decode.

However, even if everyone is using the same film (Spiderman), it is still different source material and PAL/NTSC version should be taken into account.

To make a really meaningful comparison, everyone would have to use the same cassette as Lollo suggested.

Our colleague Brad has already used the film as a reference in this post:

https://forum.videohelp.com/threads/379450-VCR-comparison-SP-mode-

Lollo & I could at least create the PAL version with the different hardware we use.

[Titan_91]

Then that's not going to be feasible, I'm afraid. Mainly being the folks in Italy. Even in the US, it would be difficult to try and pass the same tape around the country anyway, using multiple routes and multiple carriers. For cost and logistics reasons, and physical handling implications from the shipping and playback process. The closest we can get is new old stock copies of the same movie, shipped from eBay sellers once to each person who wants to give it a try. There may even be a seller with multiple copies of Spider-Man from an overstock box from the same duplication house and same run printed at the same time, all stored together in the same environmental conditions. But I haven't found him yet if that's the case.

I did find this, however. A few people in the Discord server have already looked at Ronald McDonald Scared Silly and I believe the quality has been consistent. Here's a box of unsold overstock from McDonald's stores, all together. The nice thing about these programs is they have a full motion skit at the beginning followed by a cartoon. 10 available, 1 sold.

https://www.ebay.com/itm/186050219801

[Orikson]

I would say that I have a decent capture setup with various VCRs for Video8/Hi8 and VHS (EV-S1000E, EV-S9000E and SLV-T2000, all Sony and PAL as I'm in germany) and capture options (Canopus Edius NX PCIe Card for direct S-Video capture and Panasonic ES10ish like DVD-Player with Blackmagic Intensity Pro for HDMI capture). But I just ordered some of the stuff that is required for the RF capture with vhs decode, so I need some time to set up everything. Also, I'm not a perfect expert in the video topic. But I would like to share the result about a comparison, as soon as I get everything to work

[peppi001]

Vhs-decode is a great project, but I've got many PAL VHS LP tapes that want to "digitize". With SP tapes the quality is very good, but with LP tapes the quality is worse than the conventional capture method. Now I use the cvbs-decode for LP tapes, and I wait for quality enhancement of LP tapes in vhs-decode.

[Titan_91]

But I just ordered some of the stuff that is required for the RF capture with vhs decode, so I need some time to set up everything.

If you need any help with the setup we are available in the Discord server.

Vhs-decode is a great project, but I've got many PAL VHS LP tapes that want to "digitize". With SP tapes the quality is very good, but with LP tapes the quality is worse than the conventional capture method. Now I use the cvbs-decode for LP tapes, and I wait for quality enhancement of LP tapes in vhs-decode.

LP and EP are definitely tricky. I have gotten some better results using EP from an NTSC tape with animation. Because of how much noise is on an LP tape, some may end up looking better with conventional capture due to slight softening of the noise. The RF signal on these tapes is quite a bit weaker than SP. I did post some screenshots and details to Discord on tweaks I did a few months ago to improve my EP capture from one tape by quite a bit. I wish the mobile web app would let me search or I would post the stuff here. But I'll try and share those steps in this thread from the desktop client.

[lollo]

Lollo & I could at least create the PAL version with the different hardware we use.

Yes. Let choose a new PAL tape to buy (Spiderman?) for which I can pay the expenses, capture it with Canopus NX (for instance), send it to me, I will capture with a consumer workflow and then I will send the tape to Titan_91 and once captured he will send the tape to dellsam34.

[Titan_91]

I'm going to opt out of that. I did ask anyone else in the Discord server if they want to participate. Pretty unlikely, but we'll see.

[dellsam34]

Anyone wants to send me a tape just PM me for the address.

[Titan_91]

Well, again, nobody on the software side is going to. Maybe one of my posts was missed?

https://forum.videohelp.com/threads/394168-Current-status-of-ld-decode-vhs-decode-(tru...34#post2703672

The process is going to, 95% at least, have the biggest impact on the result. Why not try one of the above we suggested? All fresh new tapes, all from the same McDonald's overstock lot, same duplication house, same seller. Each tape is shipped once, handled once, played once, and not passed around the country or the world like a bong. It won't be lost or mangled by shippers and won't be exposed to 10 differents kinds of environmental and electromagnetic conditions. That stuff has more of an impact on the imprinted magnetic RF signal than one may realize, as we have come to learn by looking directly at the RF. See the listing photo, it's legit. The guy has 10 available.

And there's also the matter of sharing names, addresses, and payment information for everyone. Buying separate tapes from eBay eliminates all of that.

If still no, we can agree to disagree. But we'll still keep things updated in this thread as the software improves

[lollo]

Well, again, nobody on the software side is going to. Maybe one of my posts was missed?

https://forum.videohelp.com/threads/394168-Current-status-of-ld-decode-vhs-decode-(tru...34#post2703672

That's a NTSC tape. I have to check if my actual workflow can properly capture a NTSC tape. On the other hand, if I am not wrong, vhs decode while capturing RF does not care about NTSC or PAL. The same maybe applies to Bogilein. That's why we proposed a PAL tape.

[Titan_91]

Correct, the signal is just RF on the capture side. The main thing there is just ensuring the VCR can properly play and track the envelope on the tape itself. If it can, then region doesn't matter to the software side. VHS-decode can handle both. In fact there is a transform process that is more developed for PAL that gives slightly better results for PAL sources.

[lollo]

OK, then as Bogilein suggested, I will search for a Spiderman (or similar) new VHS PAL tape. Onc eI found and capture it, I will send it to you or Bogilein.

Let's start this cross-comparison to provide facts to the community

edit: Found this for just 3 euro https://www.ebay.it/itm/273823575211 I hope its video master is good enough to appreciate some details. Is it ok for you?

[VideoMem]

Hi,

In the comparisions we could still not in the apples to apples domain.
All the conventional methods uses black box hardware that nobody knows the internals, and some of them even feature that as a trade secret.
We do not know how much of that final result it is just inpainting and DNR.
It could look great, but it is not real.

The final question should be.
If the conventional method 'looks' better, can you tune vhs-decode to look alike?
If not, what is the limitation and why?

This kind of comparisions (without context) are risky, biased, and with audiophile kind of dynamic.
Of course a millon dollar setup will sound good.
But what makes it sound good depends of a lot of factors.

[VideoMem]

The way vhs-decode does TBC cannot be done cheaply by any means using technology made in the era where tape video was mainstream.

That includes VCR's internal TBCs.

Bear in mind there are a lot of compromises that are introduced when designing systems for realtime processing and mass production.

They're great achievements in that regard.

I think vhs-decode has a lot of advantages and room for improvements.

In realtime, you cannot know what comes next on a signal.
Then, all the adaptive filters rely on heuristics.

Most of the decoding process emulates a lot of that.
But it also can decode with arbitrary delay lines, meaning it can adjust the current parameters based on what it sawn previously.

The capture of the RF is the act of preservation by itself.
It is not capturing video, it is an image dump of the media surface.

Vhs decode can improve and make the look of these dumps awesome.
It has been doing that.
A conventional capture is not a dump of the media, thus it cannot be decoded again without running the real tape again.

This project, is mature enough to tell us if our RF captures are bad.

That's a lot.
By the way, Titan_91's decode results are awesome.

[Titan_91]

OK, then as Bogilein suggested, I will search for a Spiderman (or similar) new VHS PAL tape. Onc eI found and capture it, I will send it to you or Bogilein.

Let's start this cross-comparison to provide facts to the community

edit: Found this for just 3 euro https://www.ebay.it/itm/273823575211 I hope its video master is good enough to appreciate some details. Is it ok for you?

Again, I'm not going to participate in a pass-the-tape effort. I've explained why. Even if I was, I don't have any PAL equipment. But, feel free to circulate that Spider-Man tape. I'm curious to see the results. Be sure to get the animated into sequence as well. The threads in the spider web are a good indicator of sharpness, ringing, etc.

Here's a screenshot of what we have on the VHS-decode side. And to be honest these are pretty old screenshots with no adjustments to parameters, just the stock settings. Even better results can probably be achieved. This image is scaled down as well. I would get better stills but I'm not exactly sure which RF capture was used for this sample. This example was decoded as a combined composite video waveform using the 3D comb filter from LD-decode. It may look better if using separate luma/chroma channels, akin to s-video.

The way vhs-decode does TBC cannot be done cheaply by any means using technology made in the era where tape video was mainstream.

That includes VCR's internal TBCs.

Bear in mind there are a lot of compromises that are introduced when designing systems for realtime processing and mass production.

They're great achievements in that regard.

I think vhs-decode has a lot of advantages and room for improvements.

In realtime, you cannot know what comes next on a signal.
Then, all the adaptive filters rely on heuristics.

Most of the decoding process emulates a lot of that.
But it also can decode with arbitrary delay lines, meaning it can adjust the current parameters based on what it sawn previously.

The capture of the RF is the act of preservation by itself.
It is not capturing video, it is an image dump of the media surface.

Vhs decode can improve and make the look of these dumps awesome.
It has been doing that.
A conventional capture is not a dump of the media, thus it cannot be decoded again without running the real tape again.

This project, is mature enough to tell us if our RF captures are bad.

That's a lot.
By the way, Titan_91's decode results are awesome.

Agree 100%. We can't even get past the PAL/NTSC thing. Another issue is we have people from around the world. It's not like we can't make comparisons, just not ideal ones because there are just too many variables.

[Bogilein]

I would use the Spiderman movie as a source.
Brad has already picked out scenes that you can take for comparison. The tape is also without copy protection as far as I can see. First of all, I already got the original Spiderman movie on VHS and DVD after reading Brad's posts.

Maybe everyone should get a copy of the movie first and provide the captures of the selected scenes. The film should be cheap to buy on VHS for everyone. Then there can always be an exchange of the sources by mail if necessary.

Maybe then other users with different hardware (Alwyn, Dellsam, Sharc &........?) will upload their examples for a comparison.

[lollo]

VideoMem, I see your point but only partially agree.

I understand the fact that vhs_decode in not yet established as the ultimate preservation/capture tool because its post-processing not yet at his best (while the RF capture is, more or less).

Comparing with existing now, may suffer of TBC correction not fully adequate on vhs_decode side, we now that and are aware of the limitations.

On the other hand, the final goal is to capture videos and watch them, whatever "principle of preservation" is supporting it. Otherwise is like never buying a TV because each year a better model hits the market.

So, IMO that comparison would have made a sense, to see how we stand today and have a fair comparison versus the best "classic" workflows, and because some of the most knowleadgable videohelp could have been involved, but if you guys do not want so, then is fine to decide to leave this attempt.

Keep us up to date with your progress on vhs_decode here!

[VideoMem]

Thanks for the reply.

I wish we can do all the tests, that will help us to understand better where the errors could be, but there is a lot of logistics and costs involved when dealing with physical stuff.
I'm sure the test comparing same machine conventional and RF is more likely to produce meaningful results.
The scenario to avoid is cross machine comparisons.

Machines can differ in their counstruction, alignments, output voltages at head amplifier test points.
Some requires external amplifiers.

But same machine, with amplifier and filter RF capture vs conventional could be used to profile post processing.

I think the only limits are how many are involved in the project, the way interest in vhs decode and tape archival has intensified in the present.
Many of these tapes have emotional memories for all of us, if that's the case go with the most practical solution first. Then think about if RF is worth the effort.

There is a lot of technical information lost about dynamic deemphasis and noise reduction on a lot of formats.
The TBC sometimes does misbehave because the video processor/proc amp implementation.

The current code have some parts of it done, but it need an optional regenerator part that can be activated to replace sync and burst with their broadcast quality ones, when they're missing or degraded.
That way the thing would keep an image even on white noise during a big dropout and solve the missing VBI from the Type C format.

I hope people will continue contributing, the project grows and there is a lot of decoding issues growing too.
They're not easy to solve, requires time of study and work, and most of us have other responsibilities too.

[dellsam34]

In my opinion the only meaningful test will be using a high end machine in a conventional manner using its TBC/DNR via its Y-C output and then use it at the same time for vhsdecode with the same tape, That's the only way to differentiate a quality VCR processing of the RF signal vs the processing of the RF decode. Using a low end VCR via composite will never come close to what its own RF is capable of, There is just no comparison there. I would argue that even the RF output of a low budget VCR will be inferior to that of a high end VCR, but don't take me on this, it's just my theory based on the manufacturing quality requirement for high end machines.

But hey, I'm in for tests if you guys are still up for it.