Decoding VHS RF signal in software?
1 Attachment(s)
I thought some of you may be interested in this.
As have been mentioned before on these forums, the people behind the domesday86 project have been working on decoding the rf signal off laserdiscs for a while now. I've restarted work on using the software to do the same with VHS (and maybe other tape formats eventually). It's still very much early stage work in progress, though I got some interesting results already. Attachment 9556 https://github.com/happycube/ld-decode/issues/16 |
I don't know what value this has. Explain. :)
Offhand, I have to wonder if access to the "full" signal might allow software-based timebase correction. Other than that (far-fetched?) idea, I'm not sure why it would be done. For LD, isn't it essentially allowed LD ripping? (True ripping, not "ripping" misused to mean a capture.) |
Software-based timebase correction is indeed one advantage of this, as the decoder will has access to the same signal as a VCR or capture device would. The ld-decode software already does correct the horizontal sync to a degree, but as you can see it's a bit inaccurate on the VHS capture decode as of now. You can also do other stuff like dropout compensation (which is partly implemented for the laserdiscs) as well.
Doing it in software means it doesn't have the limitations of a decoder chip that has to do all this with very limited memory space within a very short time, and it can be tweaked to suit each individual capture in post. Also it means that you are not as reliant on needing the top tier SVHS decks with all the processing bells and whistles, as you are bypassing the player processing entirely. As for the reasoning behind the project for capturing laserdiscs, you can see the FAQ on their website. I suppose you could use the term ripping, The Domesday discs, and many other laserdiscs utilized some of the unique features of the format to do more than just straight video playback. There is also the fact that Laserdiscs don't store luma and chroma separately like VHS (and other tape formats) so having excellent comb filtering is very crucial, which further complicate capturing the "traditional" way. |
Actually the VHS one is more important than laser disc, very few laser discs didn't make it to DVD or Blu-ray so not much interest there when you can just get the HD version of the movie. But for home videos this sounds like a good ideas with great potential for better quality. Especially if the process involves one digitization only vs 3 with the traditional workflow assuming the line and full frame TBC's are used.
|
This looks like software defined radio in a FPGA chip, driven completely by a firmware upload to the FPGA on boot up. Though I suppose we're getting to the point where CPUs have enough cores and speed to do it by pinning a task off to one part of one core while the rest of the operating system runs on the rest of the CPU.
The trick then becomes one of a proper signal conditioner and an Analog to Digital converter.. which can also be defined in silicon by software.. or as a stock add-on in many system on chip with bundled "peripherals" like the CPU in a cell phone. A CCD camera chip is after all a very simple ADC with many many converters in row and column arrays. Its not like the old days when you needed an analog circuit tied to a photodiode to generate a single "bit" of data.. now you have acres of bits and a flood of digital data to work with in software. |
Quote:
This process could be important for a variety of analog sources. Thanks for sharing! I find the Domesday Duplicator project to be fascinating. |
Quote:
A decent VRC will be required, but you can avoid the purchase of an expensive TBC, outdated capture card and complicated driver/capture software setup. With an high sampling frequency and bit depth you can be more precise on how the colour is filtered and decoded. Not to mention the software time base correction, to me it's a godsend. The ability to re-process the sampled footage when a new version of the decoder is out, it's simply amazing. A new denoise algorithm is out? just replay the file! A more precise TBC? replay! If I can sample all the signals from my VHS collection, I won't worry too much about rotting and degradation of the media. I captured lots of VHS using my limited equipment, dropped frames, A/V sync problems, unstable image , etc etc. I could solve those problems with an investment of 1-2k, but I don't want to spent that money on the equipment. Projects like this can cut down the cost of digitalising your VHS collection, I am looking forward to this. I really do. |
Yes I am also following this closely. The documentation on the LD decode project is nice and gives a nice tutorial on building your own DAQ system (think it uses a 40 Mega Samples / second ADC chip). I am checking out some old used NI products that are marketed as "digitizers" but which are basically a DAQ with a couple GS/s and see if I can use one of those instead of building the the thing on the LD Decode project (although that looks like a fun/reasonably easy task given the documentation).
I don't know anything about the RF signal that comes off the tape head, though. Have to research that. I know all about the NTSC composite video signal, but I am guessing the signal coming off the head amp is not simply just that signal. I believe this project is particularly useful for laserdisc because they can sometimes contain auxiliary information that is not simply audiovisual (like how teletext is contained in the vertical blanking interval, maybe?). The question I have for VHS is whether or not software can extract the audiovisual information from the raw RF head amp signal as well as the electronics in the AG1980P can. |
This PDF gives an overview of how the signal is stored on tape: https://www.av-iq.com/avcat/images/d...20-%204611.pdf
|
Quote:
Also, the ability to tap into the directs signal off the tape before the electronics interpret it according to being either PAL or NTSC would mean that any system will be able to read any system, as long as cotrol track and tracking are fine, simply grab the RF before it is converted to CVBS and audio signal. And regarding audio, the Hi-Fi stereo needs to be interpreted from the wave then. |
What's the latest on this? Dedicated TBC devices are ridiculously expensive and if this something like this matures it would really democratise the process of good-quality VHS video capture.
|
Most current progress is discussed here: https://forum.videohelp.com/threads/...nals%29/page13
Lots of impressive progress in the last year and a half, but probably still years away from a practical kit. |
Quote:
Yes, this project is neat. But I can say the same about lots of failed projects, lots of vaporware. I want to encourage developers, but at the same time casual users need a bucket of cold water thrown on them. Something like this won't be coming soon -- and maybe never. This particular project still has special hardware requirements, so don't start thinking it's some sort of magic software-only solution. It's not. |
I started to think that way too, There was this guy on youtube that was trying to make a software PCM encoder to record digital audio on a VHS tape as video, so I responded to him and told him that this is already out there and gave him the github link so he said oh ok cool didn't know that. He seem to be knowledgeable, So I told him instead of wasting time on PCM software why don't you come join the VHS decode project team, he came over and gave it a little push but now he is gone (so far).
There are two or three active members currently working on the project but the progress is dragging. Even the Laser decode project is not a ready product yet, you would have to have the skills to build and run such hardware and software. |
Yeah things are progressing, though very slowly at the moment, though hopefully I can spend a bit more time on implementing stuff after christmas this year.
|
We abused another thread by going way off the rail so I though this is the right thread to discuss the matter.
I want to respond to harrypm from his last post on that thread: Quote:
Quote:
Quote:
Quote:
Quote:
Digital Audio Tape recovery using a DDS drive, Working for me so it must be winning. No, the masses are still using expensive DAT home decks and digital cable. VHS PCM recovery and recording, Working for me so it must be winning. No, Archivists still using expensive Sony PCM adapters. I can go on and on but this shows that you have no grasp of reality, just because some of us have some ways of doing things differently doesn't make them a marketable product that everyone can buy and use. |
Quote:
Part of the issue with VHS-Decode I personally think is the name, it's a lot more than just VHS. V2000 was never released in NTSC markets, but it did take a chunk of the PAL market so that's one of the cases where CVBS is useful. It had died by 1984, so well before YC outputs were adopted on PAL VHS, also the PAL markets had N1500/N1700/SVC etc which were abandoned before VHS was widely adopted. The Western Europe PAL markets have videotapes (although, I fully admit, uncommon) that exist well before VHS & Betamax. Technically the Japanese formats were the 2nd/3rd generation home recording formats in this market. That's counting Telcan as the first domestic system (1963) -> The Philips/Grundig N' systems (1972) -> Betamax/VHS/V2000 (1977 onwards). So, yes, very high quality CVBS capture is appreciated maybe more here. |
Quote:
|
Quote:
Quote:
Quote:
Quote:
Quote:
The main thrust (at least to many, most) seems to be cheapness, not quality. However, that cheapness carries negative economics and ROI, so it's really dumb. The more realist devs of the project are not doing this for cheapness. As I mentioned elsewhere, until the projects shakes out the cheap/stupid, it will flounder. I've seen this before. It may need to fork before it gets mature. Or if. Cheapness will create a "good enough" mantra (ie, not good, still many issues), while the quality seekers will grow disgruntled. Fork. Quote:
But we are worried about rarer formats. We'll do whatever is necessary to salvage. Again, it's very much like handling old undeveloped films. It's be costly, very manual. And not something we'd mess with if paths of lesser resistance were available (aka, traditional VHS transfer workflows). Quote:
Quote:
cvbs-decode is a composite variant, like vhs-decode for Y/C. And it's really rough right now, not really even as usable as vhs-decode is. BTW, the "cvbs" is a giveaway that it's yet another very PAL heavy project. That acronym was rarely used in North America, but more heavily in Europe and Asia. While that's great for PAL users, there are enough differences between PAL and NTSC that NTSC is getting less attention. Sometimes discussing NTSC is talking to wall, in various discussions online. |
I was told that laser-decode can capture any composite in nature format, I understand how CVBS decode works, I was just making a point that if ld-decode is ready why bother with CVBS decode. Anyway, I've seen CVBS decodes samples and I can assure you a good FPGA composite capture device like those from Grass Valley or the previous model of Sing-Mai can do a better job, the comb filtering in those devices is so advanced that even implementing any similar technique in software would not yield that much more of quality difference, remember composite is a hot mess.
I would argue that vhs-decode (not CVBS decode) is the best method for CVBS only formats, because it avoids the composite stage all together. Not so much for Y-C formats, the difference in quality from Y-C out vs RF is not worth the hassle right now until it is stable and finalized as a working product. |
Site design, images and content © 2002-2024 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2024 Jelsoft Enterprises Ltd.